UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The persistence question of the species problem Amitani, Yuichi 2010

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2011_spring_amitani_yuichi.pdf [ 1.31MB ]
Metadata
JSON: 24-1.0071552.json
JSON-LD: 24-1.0071552-ld.json
RDF/XML (Pretty): 24-1.0071552-rdf.xml
RDF/JSON: 24-1.0071552-rdf.json
Turtle: 24-1.0071552-turtle.txt
N-Triples: 24-1.0071552-rdf-ntriples.txt
Original Record: 24-1.0071552-source.json
Full Text
24-1.0071552-fulltext.txt
Citation
24-1.0071552.ris

Full Text

The Persistence Question of the Species Problem  by Yuichi Amitani BA, Hiroshima University, 1995 Master of Literature, Kyoto University, 2001  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  Doctor of Philosophy in THE FACULTY OF GRADUATE STUDIES (Philosophy)  The University Of British Columbia (Vancouver) December 2010 c Yuichi Amitani, 2010  Abstract The species problem is the longstanding puzzle in biology regarding the nature of the species category or how to correctly define ‘species.’ Although biologists and philosophers have grappled with it for quite some time, they have thus far failed to reach a general consenus on the matter. My dissertation explores the question of why this problem persists —— why biologists and philosophers remain bothered by the species problem, and have found no general solution to it. I call this the persistence question of the species problem. This dissertation aims to answer this persistence question. My strategy is to divide this question into several sub-questions and to answer them. Those questions include: (a) Why does no definition command universal support? (b) How could biologists conduct their research concerning species without a unanimously accepted solution? For question (a), I put forward the argument from interest-relativity. The premises are: (I) biologists have different interests in species, (R) under some interest, biologists erect a set of species criteria and accept only the definition(s) that best satisfy them, and (N) a taxon satisfying one criterion often fails to satisfy another, at least partly. It follows that there will be very few, if any, unanimously accepted definitions of species, because one definition under one interest will fail to satisfy criteria provoked by others. For the question (b), I pointed out two factors. First, when there is no unanimously accepted definition of species, biologists may fail to communicate effectively about species, because the definition of ‘species’ may vary among biologists. ii  But biologists often agree regarding the reference of ‘species’ and species names (like ‘Homo sapiens’), and we see that this has enabled them to avoid the possible communication breakdown in the history of biology. Second, the way in which biologists represent the notion of species is also relevant. There is a tendency in biologists to represent the species category with its prototype, good species and infer various attributes of the former from those of the latter. When biologists make this attribute substitution, they tend to ignore the complexities of the species problem and do their research without being bothered by them.  iii  Table of Contents Abstract  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  ii  Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  iv  List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  viii  List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  ix  Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  x  1  The Persistence Question of the Species Problem . . . . . . . . . . .  1  1.1  Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  1  1.1.1  The Importance of the Species Problem . . . . . . . . . .  2  1.1.2  Landscape of the Species Problem: Historical Background  3  The Persistence Question . . . . . . . . . . . . . . . . . . . . . .  14  1.2.1  Formulation of the Persistence Question . . . . . . . . . .  14  1.2.2  Sub-Questions . . . . . . . . . . . . . . . . . . . . . . .  15  1.2.3  A Restriction on the Answer to the Species Problem and  1.2  1.3  1.4  1.5  the Persistence Question . . . . . . . . . . . . . . . . . .  18  Extant Answers to the Persistence Question . . . . . . . . . . . .  19  1.3.1  Essentialism vs. Graduality —— the Vagueness Account .  19  1.3.2  Other Accounts . . . . . . . . . . . . . . . . . . . . . . .  22  Problems with Extant Views . . . . . . . . . . . . . . . . . . . .  25  1.4.1  Vagueness Account . . . . . . . . . . . . . . . . . . . . .  25  1.4.2  Other Accounts . . . . . . . . . . . . . . . . . . . . . . .  31  Outline of the Project . . . . . . . . . . . . . . . . . . . . . . . .  40  iv  2  Sharing a Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 2.2  Introduction —— How Can Biologists Do Their Business; How Can They Communicate? . . . . . . . . . . . . . . . . . . . . . .  45  Case Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  47  2.2.1  Strickland and Grey——Natural History in the 19th Century England . . . . . . . . . . . . . . . . . . . . . . . .  47  Darwin’s Strategy for ‘Species’ . . . . . . . . . . . . . .  52  Guy Bush and Sympatric Speciation . . . . . . . . . . . . . . . .  55  2.2.2 2.3  2.3.1  Mayr on the Priority of Defining Species and Studying Speciation . . . . . . . . . . . . . . . . . . . . . . . . . .  2.3.2 2.3.3  60  Bush on the Priority of Defining Species and Studying Speciation . . . . . . . . . . . . . . . . . . . . . . . . . . . .  62  2.3.4  Coyne & Orr and Bush on Rhagoletis pomonella . . . . .  65  2.3.5  Why Does Their Disagreement on the Priority Issue Not Affect the R. pomonella Case? . . . . . . . . . . . . . . .  69  Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  73  2.4.1  3  57  Coyne and Orr on the Priority of Defining Species and Studying Speciation . . . . . . . . . . . . . . . . . . . .  2.4  45  The Incommensurability Problem and the Communication Breakdown . . . . . . . . . . . . . . . . . . . . . . . . .  74  Dual-Process Theory and the Concept of Species . . . . . . . . . . .  79  3.1  Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  79  3.2  Overview of the Dual-Process Theory . . . . . . . . . . . . . . .  82  3.2.1  The Basic Claim of Dual-Process Theory . . . . . . . . .  82  3.2.2  Characterization of the Two Systems . . . . . . . . . . . .  84  3.2.3  Summary . . . . . . . . . . . . . . . . . . . . . . . . . .  88  Elusive Transparency . . . . . . . . . . . . . . . . . . . . . . . .  89  3.3.1  Elusive Transparency . . . . . . . . . . . . . . . . . . . .  90  3.3.2  Luckow and McDade’s Observation . . . . . . . . . . . .  92  Notion of ‘Good Species’ . . . . . . . . . . . . . . . . . . . . . .  95  3.4.1  Survey from Taxacom . . . . . . . . . . . . . . . . . . .  96  3.4.2  Survey from Abstracts in Professional Journals . . . . . . 102  3.3  3.4  v  3.4.3 3.5  ‘Good Species’: The Variety of Usages . . . . . . . . . . 110  Good Species and Dual-Process Theory . . . . . . . . . . . . . . 112 3.5.1  Prototype Effects . . . . . . . . . . . . . . . . . . . . . . 112  3.5.2  Prototype —— Exemplar or Clustered Properties? . . . . 116  3.5.3  Good Species Is a Prototype of Species . . . . . . . . . . 119  3.5.4  How Biologists Reason With the Help of Good Species— —Attribute Substitution . . . . . . . . . . . . . . . . . . 123  3.6  3.5.5  Good Species Reasoning Involves System 1 Processing . . 126  3.5.6  Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 128  Psychological Essentialism . . . . . . . . . . . . . . . . . . . . . 129 3.6.1  Psychological Essentialism and its Widespread Use in Taxonomy . . . . . . . . . . . . . . . . . . . . . . . . . . . 130  3.7  3.6.2  Is Psychological Essentialism Processed in System 2? . . 136  3.6.3  Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 144  Dual Processes in Inferences Concerning Species . . . . . . . . . 145 3.7.1  Good Species, Psychological Essentialism, and Elusive Transparency . . . . . . . . . . . . . . . . . . . . . . . . . . . 145  3.7.2  System 2 Reasoning: Definition-Centered Reasoning on Species . . . . . . . . . . . . . . . . . . . . . . . . . . . 147  4  Answering the Persistence Question . . . . . . . . . . . . . . . . . . 152 4.1  Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152  4.2  Why Do Biologists Believe They Need To Define ‘Species’? . . . 154 4.2.1  Question 6: Why Aren’t We Similarly Bothered By Analogous Issues, Such As the Nature of Life? . . . . . . . . . 155  4.3  Why Does No Definition Command Universal Support? . . . . . . 157 4.3.1  Argument from Interest-Relativity . . . . . . . . . . . . . 158  4.3.2  Argument from Interest-Relativity and the Vagueness Account . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164  4.3.3 4.4  Argument from Interest-Relativity and Species Pluralism . 165  Study of Species and Speciation Without a Unanimously Accepted Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166 4.4.1  Substitution and the Species Problem . . . . . . . . . . . 167 vi  4.4.2 4.5  Reference-Sharing and Prototypical Reasoning . . . . . . 170  Conclusions and Summary . . . . . . . . . . . . . . . . . . . . . 172  Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180 A Further Description of the Dual-Process Theory . . . . . . . . . . . 199 A.1 Elaboration of “System”: Token or Type? . . . . . . . . . . . . . 199 A.2 Relationship Between Systems 1 and 2 . . . . . . . . . . . . . . . 200 A.3 The Unity of Systems and Causal Mechanism Behind It . . . . . . 202 A.3.1 The Unity of System 2 . . . . . . . . . . . . . . . . . . . 202 A.3.2 Neurological Basis . . . . . . . . . . . . . . . . . . . . . 203 A.4 Transfer (Consolidation) . . . . . . . . . . . . . . . . . . . . . . 204 B Sources Surveyed for Usages of ‘Good Species’ . . . . . . . . . . . . 205 B.1 List of Entries Surveyed from the Taxacom Mailing List . . . . . 205 B.2 List of Papers Surveyed on ‘Good Species’ . . . . . . . . . . . . 207  vii  List of Tables Table 1.1  A List of Leading Species Concepts (1) . . . . . . . . . . . . .  4  Table 1.2  A List of Leading Species Concepts (2) . . . . . . . . . . . . .  7  Table 1.3  Sub-Questions of the Persistence Question . . . . . . . . . . .  15  Table 3.1  Characteristics of Two Processes . . . . . . . . . . . . . . . .  85  Table 3.2  Summary of the Usages of ‘Good Species.’ . . . . . . . . . . . 110  Table 3.3  Typicality Ratings of Instances of Categories . . . . . . . . . . 114  Table 3.4  Response Time for Prototypical and Non-Prototypical Members of Categories . . . . . . . . . . . . . . . . . . . . . . . . . . . 114  Table 3.5  Features of System 1 in Biologists’ Reasoning . . . . . . . . . 147  Table 3.6  Features of System 1 and System 2 in Biologists’ Reasoning . . 150  Table 4.1  Essentialism Across Different Types of Concepts . . . . . . . . 154  Table 4.2  Answers to Sub-Questions . . . . . . . . . . . . . . . . . . . . 174  Table B.1  List of Entries from the Taxacom Mailing List . . . . . . . . . 205  viii  List of Figures Figure 1.1  A Unknown Phylogeny Among Terminal Taxa A, B, and C . .  7  Figure 1.2  A Phylogenetic Tree and a Monophyletic Group . . . . . . . .  9  Figure 1.3  Incompatibility of Species Classifications by Different Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  11  Figure 1.4  Anagenesis . . . . . . . . . . . . . . . . . . . . . . . . . . .  35  Figure 1.5  Cladogenesis . . . . . . . . . . . . . . . . . . . . . . . . . .  37  Figure 4.1  Speciation Process and Species Criteria . . . . . . . . . . . . 161  ix  Acknowledgments This is a great opportunity to express my respect to all of the faculty, staff and graduate students in the Department of Philosophy who have all been of great assistance throughout my time at the University of British Columbia. I would like to specifically thank all the participants of the philosophy of biology reading group ——Taylor Davis, Eric Desjardins, Chris French, Andrew Inkpen, John Koolage, Alirio Rosales, and Brendan Smith—— for all of the thought-provoking and helpful discussion. I benefited immensely from discussion and written comments on the topic from Marc Ereshefsky, Dominic Lopes, Shunkichi Matsumoto, Gordon McOaut, Nobuhiro Minaka, Hisashi Nakao, Shun’ichiro Naomi, and Margaret Schabas. I would not have been able to commence my studies at the University of British Columbia without the assistance from Yasuo Deguchi, Kazuyuki Ito, Merrilee Salmon, and Soshichi Uchii. I would also like to thank my editors: Jillian Isenberg and, especially, the late Brian Laetz. Brian read most of earlier drafts and carefully edited them. I am very grateful for all the assistance I received from him. Gemma Celestino, Josh Johnston, Sahasra Pedersen, and Roger Stanev have been a great source of encouragement, discussion and criticism. Wayne Maddison provided me an opportunity to become acquainted with biological practice. Sean Graham corrected my misunderstanding of taxonomical practice. I am very grateful for the funding I received from the Tina and Morris Wagner Foundation Fellowship. Part of this disssertation was presented at the 2009 meeting of the International Society for the History, Philosophy, and Social Studies of Biology in Brisbane, Australia. I thank the audience for helpful comments and discussion. x  It is not possible to do justice to the debt I owe to my parents, Ryokichi and Nobuko, and my sister, Noriko, for a lifetime of unconditional love and support. I would especially like to thank my father, Ryokichi, who passed away while I was preparing for the defense. I dedicate this dissertation to him. Most importantly, I would like to acknowledge the members of my Dissertation Committee ——Chris Stephens, Paul Bartha, and John Beatty—— for their advice, guidance and insight. I owe particular thanks to my supervisor John Beatty, for his indispensable insight and guidance in research, for his critical feedback that helped hone my work above and beyond, and for his emotional support.  xi  Chapter 1  The Persistence Question of the Species Problem 1.1  Introduction  In this introductory chapter, I outline my entire project, which focuses on what I call the persistence question of the species problem. The species problem is the longstanding puzzle regarding the nature of the species category, or how to correctly define species. Although biologists and philosophers have grappled with the species problem for quite some time, they have thus far failed to reach a general consenus on the matter. The persistence question is simply why lack of consensus persists and why biologists and philosophers remain bothered by it. Naturally, I will begin this project with a brief description of the species problem and its importance to biologists and philosophers. Then I give a formulation of the persistence question. I also offer a preliminary analysis of the persistence question by showing that it can be divided into several sub-questions. As it happens, biologists and philosophers have being attempting to answer the persistence question for quite some time, though none have bothered to clearly identify or name it until now. In section 1.3, I survey these answers, subjecting each to critical scutiny. I show none fully accounts for all important aspects of the persistence question. This will clearly indicate the significance of this project. In the last section (section 1.5), I briefly outline what I plan to argue in the subsequent chapters. 1  1.1.1  The Importance of the Species Problem  Importance for Biologists The species problem can be briefly put as a question, namely: what is the nature of the biological species category?1 Here I will describe the importance of this problem and the various concerns driving it. Many biologists have thought that species are the basic components of the living world above the level of organism. Various systematists, who otherwise disagree on many aspects of the species problem and related matters in the philosophy of classification, nonetheless agree that species alone are objective and real, and that higher taxonomic ranks such as genus and family are artificial (for example, Eldredge & Cracraft 1980, Mayr 1969, Hull 1997, Simpson 1961). Many systematists also regard taxa at levels lower than species (such as subspecies and varieties) as artificial (Mayr 1970, Stamos 2003). Accordingly taxonomists have used species as the ultimate “unit of classification,” the building block of their classification system (Stevens 1992, Hull 1998). Moreover, a substantial number of evolutionary biologists have studied how new species form from within the framework of Darwinian theory (Mayr 1963, Otte & Endler 1989, Grant 1981, Howard & Berlocher 1998, Coyne & Orr 2004). Therefore, determining what species are means understanding a central building block of the living world and how evolution works. In addition, it has notable practical import. Since one of the global environmental problems facing us is the challenge of protecting biodiversity, and species are thought of as the basic unit of biodiversity (Claridge et al. 1997), any answer to the species problem should inform controversies regarding how we should understand and protect biodiversity (Maclaurin & Sterelny 2008). For example, if species are only artificially distinguished entities, like constellations, this would clearly compromise the movement for protecting endangered species (Hull 1997). The species rank is also widely taken to be a “currency” to measure biodiversity; 1 The word ‘species’ has taxon/category ambiguity. A taxon (pl. taxa) is a particular taxonomical entity. For example, Mammalia (mammals), Homo, Aves (birds), Homo sapiens and Felis silvestris (wild cat) are taxa. Each taxon typically locates itself at some rank ——such as species, genus, and order—— in the hierarchy of the classification system. The species category is a collection of taxa at the rank of species. The word ‘taxon’ (or ‘taxa’) may have a name of a rank before it as an attributive (e.g., ‘a species taxon’), and this refers to a taxon (or taxa) at that rank. For example, Homo sapiens is a species taxon. The species problem as I discuss here concerns the nature of the species category: what unites all species taxa, and distinguishes them from all other biological ranks.  2  if we have two natural habitats for many species in a country, but can only protect one from development, we may well be more justified in protecting the area containing more species. This is (at least partly) based on the idea that species are not artifacts, but rather real things, and that our grouping of organisms into species reflects the objective natural order (Sterelny & Griffith 1999). Philosophical Dimensions The species problem also has philosophical dimensions. Species have been seen as exemplars of natural kinds. For example, in his famous defense of essentialism regarding natural kinds, Saul Kripke (1980) assumes that biological taxa, such as tigers, qualify as so-called “natural kinds.” But many philosophers of biology believe this assumption does not fit with some basic tenets of Darwinism, and they have therefore rejected it (Hull 1978, Ghiselin 1966, 1974, Ruse 1987, Dupr´e 1981; but the controversy still continues, see, for example, Okasha 2002, Crane 2004, Devitt 2008, Ereshefsky forthcoming). Richard Boyd (1999) has recently mounted a new defense of species as natural kinds, which has drawn much attention (see also, for example, Wilson 1999, Griffiths 1999, Wilson et al. 2007, Brigandt 2009).  1.1.2  Landscape of the Species Problem: Historical Background  Biologists and philosophers have offered many different species concepts over the years (Wilkins 2006b reports there are now over 20 species concepts). However, it is widely agreed that biologists are unlikely to reach agreement on the “correct” account of species any time soon. In this section, I will first review three of the major alternative species concepts, and second, I will describe species pluralism as an attempt to deal with this “messy situation” regarding species concepts.2 2 This project surveys mostly species definitions proposed after the 1940s.  See Wilkins (2009) for the history of the concept of species. I will also limit my discussion to eukryotic species, but this does not mean that there is no controversy about bacterial species. In fact, the issue of bacterial species has been much discussed by philosophers in recent years. See, for example, Wilkins (2006a), Franklin (2007), Ereshefsky (2010).  3  Table 1.1: A List of Leading Species Concepts (1)——See table 1.2 for phylogenetic definitions of species.  Name Biological Species Concept (BSC) Recognition Species Concept (RSC) Ecological Species Concept (EcSC)  Genotypic Cluster Concept (GCSC)  Species  Cohesion (CSC)  Concept  Species  Phenetic Species Concept (PhSC) Morphological (Taxonomic) Species Concept (MSC, or TSC)  Definition (Source) “Groups of actually or potentially interbreeding natural populations, which are reproductively isolated from other such groups.” (Mayr 1942) “Most inclusive population of individual biparental organisms which share a common fertilization system.” (Paterson 1985) “A lineage (or a closely related set of lineages) which occupies an adaptive zone minimally different from that of any other lineage in its range and which evolves separately from all lineages outside its range.” (Van Valen 1976) “We see two species rather than one if there are two identifiable genotypic clusters. These clusters are recognized by a deficit of intermediates.” (Mallet 1995b) “The most inclusive group of organisms having the potential for genetic and/or demographic exchangeability” (Templeton 1989) “The species level is that at which distinct phenetic clusters can be observed.” (Sneath 1976) “The smallest groups that are consistently and persistently distinct, and distinguishable by ordinary means.” (Cronquist 1978)  Background 1: Leading Species Concepts — Proposals and Problems One reason for this bleak prediction regarding the species problem is that proposed species concepts are largely incompatible with each other. Several taxonomies of species concepts have been proposed. As all parties will readily attest, it is not easy to produce a “unified” species concept that all can agree on. Another reason is that each species concept has its own conceptual and/or practical problems. Thus, it is hard to identify one as the very best. Here let us briefly review some leading species concepts. There are, of course, so many species concepts that it is best to just focus on a few. I will review three of the more widely accepted accounts: the morphological (taxonomic) species concept (MSC or TSC), the biological species concept (BSC) and the phylogenetic species concept (history-based) (PSCh). For other species concepts, see Tables 1.1 and 1.2. Wilkins (2006b) provides an overview of various species concepts.  4  Morphological (Taxonomic) Species Concept  Let us begin with the morpho-  logical (or taxonomic) species concept (MSC or TSC). This concept describes the traditional way in which taxonomists identify a species taxon. According to this concept, species is taken to be a group of organisms which resemble each other and differ sufficiently from other such entities (mainly) in morphological characters, i.e., the (external) form and structure of organisms. For example, Arthur Cronquist (1978) defines species in the following way (see also Blackwelder 1967, for a similar definition): Species are the smallest groups that are consistently and persistently distinct, and distinguishable by ordinary means. (p. 3) Cronquist suggests by “ordinary means” that a taxonomist mainly uses morphological characters when she identifies a species under this concept. Notice that the species classification will be entirely subjective if one adopts this definition because it will be up to one’s judgment of “distinctness” and its “consistency” whether a group of organisms is a species. One can imagine a situation in which different taxonomists have different ideas of how much distinctness and consistency is required for a taxon to become a species, and it does obtain frequently. Many authors take this subjectivity of taxonomic practice as evidence that taxonomy is somewhat inferior to other sciences. This is why most of other definitions of species aim to aviod subjectivity in species classification when it is possible. Nevertheless, many taxonomists have traditionally used and still use this conception, partly because of the availability of morphological characters to field taxonomists. Information of external form and structure of organisms is almost always available to taxonomists. In contrast, some definitions of species, such as the biological species concept, use other kinds of information ——the existence of reproductive isolation, for example—— which are inferred from information which is more readily available, including morphological patterns. This difference could be practically important to field taxonomists, because it is not always easy for taxonomists to make such an inference when they do not have appropriate kinds of information.3 3 The supporters of MSC often claim that their concept is more robust than other definitions. I discuss this shortly.  5  Biological Species Concept Ernst Mayr (1942) proposed the biological species concept (BSC; see also Dobzhansky 1937). According to BSC, Species are groups of actually or potentially interbreeding natural populations, which are reproductively isolated from other such groups. (p. 120). Within a species, individuals exchange genes through sexual reproduction, and thereby each species maintains phenotypic unity, while remaining distinct from species that are reproductively isolated from them. That is how the BSC explains the coexistence of different groups of organisms in sympatry4 (this is one of the BSC’s objective. Coyne & Orr 2004, p. 26). Early supporters of the BSC held that it better explained sympatric coexistence than attempts to identify species simply from morphological discontinuity (MSC). A striking example is sibling species. Sibling species are a pair of closely related species, which are (almost) indistinguishable from their morphology, but are still reproductively isolated. The case for the BSC is that those “species” are genetically different entities which will take independent evolutionary paths. The BSC recognizes them as distinctive species, while the morphological criterion alone fails to do it because they might lump them together into a single species. On the other hand, the supporters of the morphological species concept (MSC) claim that MSC has advantages over BSC in other regards. They point out that MSC does not presuppose any evolutionary process and is thereby more universally applicable; it is supposed to be a pattern species concept. In contrast, the biological species concept makes an assumption about evolutionary processes through which one species diverges into two species or maintains its homogeneity. For example, the supporters of BSC assume that genetic homogeneity maintained by reproductive isolation is responsible for phenotypic homogeneity found in many species. If the definition of species refers to a particular process (reproductive isolation), then, by definition, there is no species which can be caused by other processes (such as natural selection). However, it is possible (at least in experiment) for natural selection to cause divergence between two groups even without isolation (Endler 1973). Therefore, for a supporter of the morphological species concept, a pattern-based 4 That  is, they occupy the same habitat in the same geographical area.  6  Table 1.2: A List of Leading Species Concepts (2) Name Phylogenetic Species Concept (Character-based) (PSCc)  Hennigian Species Concept (HSC)1 Phylogenetic Species Concept (History-based) (PSCh)2 Genealogical Species Concept (GSC) Evolutionary Species Concept (EvSC)  1 2  Definition (Source) “The smallest aggregation of populations (sexual) or lineages (asexual) diagnosable by a unique combination of character states in comparable individuals (semaphoronts).” (Nixon & Wheeler 1990) “A segment of a population phylogenetic lineage between nodes.” (Ridley 1989) “The smallest (least inclusive) monophyletic groups–either individual populations or groups of populations.” (de Queiroz & Donoghue 1988) “A basal group of organisms all of whose genes coalesce more recently with each other than with those of any organisms outside the group.” (Baum & Donoghue 1995) “A single lineage of ancestral descendant populations of organisms which maintains its identity from other such lineages and which has its own evolutionary tendencies and historical fate.” (Wiley 1978)  This is also called “Cladistic Species Concept.” Also called “Monophyletic Phylogenetic Species Concept.”  Figure 1.1: A, B, and C are terminal taxa of the phylogenetic analysis. Terminal taxa are often extant taxa. We know the combination of characters each unit possesses, but do not know what historical relationships (phylogeny) among them are like.  species concept, which minimizes the number of assumptions on causal processes, is preferable to process-based concepts (Levin 1979). This is one motivation behind the morphological species concept.5  5 This  is not to say that the supporters of MSC do not have any assumption about evolutionary processes. They do. Their point is that when we choose a definition of species, we should prefer the one which does not assume a particular evolutionary process to the one which does.  7  Phylogenetic Species Concept (History-Based) (PSCh) Then let us turn to phylogenetic species concept (history-based) (PSCh). This version of phylogenetic species concept is one of the two species definitions called “phylogenetic species concept” (I call the other definition phylogenetic species concept (character-based) (PSCc)). The supporters of this definition focus on phylogenetic relationship among populations and typically take a species as the least inclusive monophyletic group of organism. For example, Mishler & Brandon (1987, p. 406) define ‘species’ as follows (see also Mishler & Theriot 1999, de Queiroz & Donoghue 1988, for similar definitions): A species is the least inclusive taxon recognized in a classification, into which organisms are grouped because of evidence of monophyly ..., that is ranked as a species because it is the smallest ‘important’ lineage deemed worthy of formal recognition, where ‘important’ refers to the action of those processes that are dominant in producing and maintaining lineages in a particular case. Before exploring this definition further, let me briefly describe the phylogenetic approach to classification,6 on which this definition is based. The phylogenetic approach uses phylogenetic analysis to reconstruct the phylogeny7 of organisms, and produces a classificational system accurately reflecting this historical relationship. Suppose we want to know phylogenetic relationship of groups of organisms A, B and C. They are called terminal taxa. Figure 1.1 shows that we do not know what historical relationships (phylogeny) among them are like. Phylogenetic analysis is to reconstruct phylogenetic relationships among them. This analysis requires some material or data. These data are character distributions over the terminal taxa. A character distribution over taxa is a set of ordered pairs of a taxon and its character(s). For example, suppose A, B and C in Figure 1.1 are plants and the color of their flowers is red, red, and pink, respectively. Then character distribution over A-C on the color of flower provides evidence for phylogenetic analysis. Namely, one of the main jobs of the phylogenetic approach to classification is to reconstruct 6 This  approach is sometimes called cladistics. is an evolutionary history ——particularly ancester-descendant and sister-group relationships—— of organisms. A tree diagram is usually used to show phylogenetic relationships (see Figure 1.2). 7 Phylogeny  8  Figure 1.2: A phylogenetic tree and a monophyletic group. A, B, and C are terminal taxa. Lines connecting terminal taxa show historical relationships (phylogeny) among them. A closer connection in terms of the recency of the shared common ancester represents closer relationship: in this tree, A is closer to B than C, because A and B’s most recent common ancester (W , a black dot connecting A and B in the figure) is not C’s ancester. N (a group enclosed by dashed lines) is a monophyletic group, because it is a group composed of an ancestor (W ) and all of its descendents (A and B). In contrast, M (a group enclosed by dotted lines) is not a monophyletic group. M does not include all of the descendents of U (a black dot connecting W and C)——it does not contain A. M is called a paraphyletic group. Note that only a monophyletic group will be recognized as a legitimate taxonomic group in the phylogenetic approach to classification.  phylogenetic relationship among those terminal taxa from the information about which character states each unit possesses. To discuss the method of phylogenetic reconstruction is beyond the scope of this study. I may leave the details of it to textbooks of phylogenetics, such as Felsenstein (2004). So let us suppose that we successfully reconstruct the phylogenetic relationship among A, B and C as in Figure 1.2. This figure is called a phylogenetic tree. Lines connecting A, B and C show historical relationships (phylogeny) among them. A closer connection in terms of the recency of the shared common ancester represents closer relationship: in this tree, A is closer to B than C, because A and B’s most recent common ancester (W , a black dot connecting A and B in the figure) is not C’s ancester (U, a black dot connecting W and C). Here let me introduce the notion of monophyly. Monophyly is a very important kind of relationship in the phylogenetic approach, because only monophyletic groups (of taxa) will be recognized as legitimate in its classification system (since this is also beyond the scope of this study, I leave the discussion on this issue to classical textbooks of phylogenetics, such as Wiley (1981). Hereafter I simply assume that the phylogenetic approach accepts this claim). In other words, their classification system is only composed of monophyletic groups. Thus, in the phylo9  genetic classification system, non-monophyletic groups, such as reptiles (Reptilia), will never be legitimate taxonomic groups. A standard definition of monophyly is as follows (Sober 2000): A monophyletic group is a group composed of an ancestor and all of its descendants. Thus in Figure 1.2, N (a group enclosed by dashed lines) is a monophyletic group, because it is a group composed of an ancestor (W ) and all of its descendents (A and B). In contrast, M (a group enclosed by dotted lines) is not a monophyletic group. M does not include all of the descendents of U: it does not contain A. M is called a paraphyletic group. In the phylogenetic approach to classification, N is a legitimate taxonomic group, whereas M is not. Now we are ready to discuss Misher & Brandon’s definition of species. What they propose is that species are basically (some of) the smallest monophyletic groups. Suppose that A, B and C are populations of sexual organisms and we cannot apply the phylogenetic method to such populations —— we cannot analyze phylogenetic relationship within A, B and C.8 Then a group N will be the smallest monophyletic group we can observe, because A and B do not have tree-like phylogenetic relationship, and thus we cannot analyze them further. According to Misher & Brandon, a species is such an entity. This is the basic idea of PSCh. This species concept opposes the BSC in several ways. PSCh is a phylogenetic definition——it explicitly mentions phylogenetic property which a species should have, whereas BSC does not. Theoretical interests behind them are also different. The PSCh is entirely motivated by the phylogenetic approach to classification and its objective is different from explanation of the sympatric coexistence of different groups of organisms. The phylogeneticist’s program is to reconstruct the phylogenetic relationships among terminal taxa and construct a classification system reflecting this information accurately. PSCh is one way to achieve this phylogenetic goal; according to PSCh, a species is a monophyletic group just as are higher taxa like a genus and a class. Ernst Mayr, on the other 8 Within  a population of sexual organisms, phylogenetic relationship is not tree-like, as in Figure 1.2 (recall it is called a ‘phylogenetic tree’), but reticulate, because a single unit ——like an individual organism—— could have more than one ancestor. Phylogeneticists generally admit that their method of phylogenetic analysis cannot be applied to such entities.  10  Figure 1.3: A phylogenetic tree of four populations (A, B, C, and D) showing incompatibility of species classifications by different species concepts. A, B, C, and D refer to populations of sexual organisms. A thick vertical line between B and C represents reproductive isolation between (A&B) and (C&D). Thus BSC would classify populations A and B into one species and C and D into another (two rectangulars represent this). But since a group of C and D is not monophyletic ——the most recent common ancestor of C and D is also an ancestor of A and B—— it is not a species under PSCh. Thus PSCh’s and BSC’s classifications are not compatible.  hand, is highly critical of this approach. He objects to the idea that classification system should be solely based on phylogenetic relationships: a non-monophyletic group, such as reptiles, may be a legitimate taxonomic group if the members of the group share evolutionarily significant similarities.9 Hence no wonder that species classifications by BSC and PSCh are often incompatible (see Figure 1.3). In this figure, four populations (A, B, C, and D) are components of a phylogenetic tree. Suppose that A, B, C, and D refer to populations of sexual organisms. A thick vertical line between B and C represents reproductive isolation between A&B and C&D. Under the BSC, populations A and B consist of one species and C and D constitute another. On the other hand, since a group of C and D is not monophyletic ——the most recent common ancestor of C and D is also an ancestor of A and B—— it is not a species under PSCh. Thus PSCh’s and BSC’s classifications are not compatible.10 This is not just a product of philosophical imagination. It is acknowledged by many biologists that BSC and PSCh often classify the same situation differently (see, for example, de Queiroz & Donoghue 1988). 9 To discuss Mayr’s objection is beyond the scope of this thesis. I leave the details of it to Mayr & Ashlock (1991) and Simpson (1961). It is worth noting that few of the recently trained taxonomists support Mayr’s objection now. For a classic response, see, for example, Wiley (1981). 10 If there is actual gene flow between C and D in Figure 1.3, they will merge into a single branch and become a single terminal taxon. But the new taxon will still not be a monophyletic group, because the most recent common ancestor of C and D has other descendants (A and B) which are not included in the new taxon.  11  This definition also differs from the morphological species concept in important respects. As we have seen, one has to infer phylogenetic relationship among terminal taxa from character distributions over them. This indicates that when one uses PSCh, she makes some theoretical assumptions which one would not if she used the morphological species concept. Moreover, PSCh is a historical definition of species per excellence, whereas MSC defines species nonhistorically. Those points do not exhaust the main differences between the two concepts of species, but they reflect the fact that PSCh and MSC are motivated by different agendas. We have just seen that PSCh is motivated by the phylogenetic approach to classification. The morphological species concept, on the other hand, is motivated by taxonomists’ interest in the availability of morphological data and the stability of species classification. This brief review only scratches the surface of the overwhelming variety of species concepts. However it suffices to demonstrate two key points: different species concepts derive from different agendas and often oppose one other. Background 2: “Messy Situation” of the Species Concepts Facing this flood of species concepts, philosophers have repeatedly pointed out that biologists attach the label ‘species’ to biologically varied objects (I will call this the “messy situation” of the species concepts). Species Pluralism  In 1980s and 1990s, several philosophers——Phillip Kitcher  (1984), Marc Ereshefsky (1992a, 1998, 2001), Kyle Stanford (1995), and John Dupr´e (1981, 1993)——presented the pluralist account of species. Pluralism regarding species concepts recognizes this “messy situation” and makes a normative proposal based upon it. Although there are many differences in the details of such accounts, the basic idea remains roughly the same. Each species concept reviewed in the previous section claims that its criterion of grouping organisms into species is privileged; that is, each is offered as the only legitimate species concept, the only one that should be used to identifify species taxa. Pluralists reject this idea and approve the use of multiple, alternative definitions for the species category, depending on the goals of the particular biological study being carried out.  12  The pluralists’ basic argument is this.11 (i) Biologists from different areas have different reseach goals, and what they want for a species concept differs accordingly. Different species definitions naturally reflect differences in research interests. For instance, students of speciation, students of phylogenetic history, and taxonomists may have different interests in the concept of species. The former group of researchers study how new species arise in evolution and the BSC may be most useful to them; while those most interested in reconstructing the phylogenetic history of a group may find PSCh the most useful; while taxonomists may be most interested in classification based on morphological similarity when morphological data are most readily available in many cases and they want to minimize the number of assumptions on causal processes so that their classification is universally applicable. (ii) All of those goals are biologically worthwhile. For instance nobody would doubt the legitimacy of the study of speciation. (iii) The different species definitions that satisfy different goals often classify extensionally different groups as “species” (Figure 1.3). (iv) Therefore, pluralists conclude, biologists are justified in using different species concepts in different contexts: taxonomists may justifiably use, say, morphological definition, the students of speciation may rightly exploit the BSC, while the students of phylogenetic history may adopt PSCh, instead of either one of the three in all contexts. In sum, pluralists not only acknowledge that different species concepts are concerned with different biological entities12 and urge biologists to recognize this, but also recommend the use of multiple species concepts. This does not mean that all species concepts are legitimate; some species concepts may not assist any biological interests at all, and so should be expelled from scientific inquiry, even on a pluralistic view. 11 I  will offer a similar argument in Chapter 4 for a different purpose. this is not to say that pluralists accept any species concept once proposed. See, for example, Ereshefsky (1992a, 1998, 2001). 12 However,  13  1.2 1.2.1  The Persistence Question Formulation of the Persistence Question  As I have sketched above, biologists certainly recognize that no consensus has been reached on the nature or the “right” definition of species. This is a problem in its own right. Indeed, many thinkers have tried to explain the recalcitrance of the species problem, even while working out their own answer to it (see the section 1.3 for their answers and my assessments of them). This is what I call the persistence question. I formulate this question as follows: The Persistence Question of the Species Problem: Although biologists have tried long and hard to reach rational agreement on the species problem (on the nature or the “right” definition of species), they have not reached consensus and will not anytime soon. Why is that? Let me elaborate what I mean by “the persistence question” further. If biologists agreed with each other on the nature of species or the “right” definition of it, the agreement would take one of several possible forms. First, they could “collectively” accept some definition(s) of species. The species problem would then be solved when • Most biologists accept one definition of species. • Most biologists accept the same plurality of definitions of species. However, given that it is very hard to achieve consensus on the “right” definitions of species, one might wholly abandon any attempt to define it. One could give up seeking the “right” definition of species; indeed, one could forgo the concept of species altogether, instead replacing it with a different notion designed to fill the biological space that would be left open by eschwing species. There have been several attempts to do so; in this spirit, Gilmour & Gregor (1939) coined the term ‘deme’ (see also Gilmour & Heslop-Harrison 1955), and pheneticists invented the OTU (operational taxonomic unit; Sokal & Sneath 1963). Of course, very few taxonomists have embraced any such proposal, but, in any case, let’s say another solution to the species problem would be reached when 14  (1)  Why do biologists believe they need to define ‘species’?  (2)  Why does no definition command universal support?  (3)  Why do biologists keep giving one definition after another?  (4)  Why don’t biologists abandon pursuit of a correct definition?  (5)  Why don’t biologists just give up the species concept altogether?  (6)  Why aren’t we similarly bothered by analogous issues, such as the nature of life?  (7)  How could biologists do their business ——conduct their research—— about species and speciation without a unanimously accepted solution? Table 1.3: Sub-questions of the Persistence Question  • Most biologists admit that there is no “right” definition of species, and they give up using the term ‘species.’ Another possible scenario in which biologists reach consensus would be if they concluded they cannot agree on the “right” definition of species, but they concurred that using the species concept has still some merit. In this scenario, the species problem is solved when • Most biologists admit that there is no “right” definition of species, but they continue using the notion of species anyway (i.e., they only give up seeking the “right” definition of it). Presently, one can observe that none of these options commands much support among biologists. The species concept, of course, still plays an important role in contemporary biology, but no agreement has been reached on how to define it, and biologists continue proposing one definition after another. It is in this sense that the species problem persists. How can we explain this? That’s the persistence question.  1.2.2  Sub-Questions  One way of elaborating the persistence question is by breaking it down into several sub-questions. My strategy in this thesis is to answer the persistence question by asking and thoroughly addressing each sub-question. Sub-questions of the persistence question are as follows (see also Table 1.3): 15  1. Why do biologists believe they need to define ‘species’? The species controversy revolves around definitions. Tremendous effort has been made to provide an adequate definition of species. If no one needed a definition, there would be little to no controversy regarding species. Why then, is species supposed to call for definition in the first place? 2. Why does no definition command universal support? Suppose we successfully answered the first question; we are convinced that we need a definition of ‘species.’ Then we next find no definition has approached ——even come close—— to unanimous acceptance. A number of biologists have been bothered by this fact. Why is there no such definition? 3. Why do biologists keep giving one definition after another? Biologists recognize the need for species definition(s), but no definition is universally supported. Meanwhile, some biologists continue proposing definitions of species. But one may wonder why they do not give up the definitional enterprise altogether. After all, they, of anyone, know their new definition is extremely unlikely to put an end to all the disagreement, and command unanimous acceptance! 4. Why don’t biologists abandon pursuit of a correct definition? As we have seen, attempts to find the “right” definition of species have a miserable track record. One may well give up the whole enterprise altogether (or at least for a while). But as we have seen, biologists keep trying. Why don’t they just leave the practice of proposing definitions behind them? This is a flip side of the question (3). 5. Why don’t biologists just give up the species concept altogether? Given that the attempt to find the “right” definition of species has so far  16  been unsuccessful——and certainly not for want of trying——a more radical option may look attractive to biologists. If biologists have little chance of articulating the concept of species (via definition), perhaps they should abandon it and replace it with a concept that could be successfully articulated— —such as a reproductively isolated group or a deme. 6. Why aren’t we similarly bothered by analogous issues, such as the nature of life? There are various concepts in various fields of science, which also lack a clear definition. For example, there has been discussion on the nature of “life,” “drought” or “disease” (Hey 2001a,b). However, scientists have not made nearly as much of a fuss about these concepts as biologists have been about the species category; there is no “life problem” or “drought problem” in the way in there is species problem in biology. One might ask what accounts for this difference. 7. How could biologists do their business ——conduct their research—— without a unanimously accepted solution? E.g., how can we study speciation or species extinction or species diversity without a unanimously accepted definition of ‘species’? This question is closely related to questions 5 and 6. Given that ‘species’ is seen as an important concept in evolutionary biology and that biologists have reached no consensus regarding its nature, one might suspect this lack of consensus presents serious obstacles to biologists. In particular, since there is no consensus on the nature of species, and thereby the meaning of ‘species’ as well, biologists may have trouble effectively communicating with each other. But this does not at all seem to be the case. One might ask how this is possible. I do not intend this list to exhaust all the possible sub-questions that constitute the very complex persistence question. Nor do I believe that I can provide complete answers to each question listed above. If, however, one can answer some of these 17  questions, one will clearly edge closer to revealing why the species problem has not been resolved, despite the increasingly long time biologists have considered it. In this project, I will focus on Questions 1, 2, 6, and 7, although answers to these questions will offer clues to successfully addressing the others, since some questions are meaningfully interconnected.  1.2.3  A Restriction on the Answer to the Species Problem and the Persistence Question  I have articulated the persistence question. Here I would like to propose one restriction on possible answers to the species problem and the persistence question. An Answer Should Be Stand-Alone I put forward the stand-alone criterion to possible answers to the persistence question. This prohibits answering the persistence question by directly appealing to some proposed solution to the species problem. For example, suppose one believes X is the right definition of species. One could “answer” the persistence question by simply saying, “well, the species problem persists because solution X has been undiscovered until now.” Anyone can, of course, give this type of answer; one only needs a pet solution to the species problem to appeal to. For example, Mayr could say that the species problem persists, because biologists have not taken the BSC seriously enough. However, this response is clearly parasitic on one’s answer to the species problem; it presupposes that one does, in fact, have the correct answer to the species problem. In this case, the adequacy of one’s answer to the persistence question totally depends on whether one’s solution to the species problem is correct: one’s account is correct only if one’s answer to the species problem is correct. This is definitely not what we’re after. We should be able to analyze the difficulty of a certain philosophical problem before we have a solution to it; otherwise, we could not discuss, for instance, why we find it difficult to have a coherent interpretation of probability before we do have one. This does not seem to be the case. A stand-alone——not parasitic——answer is what we want.13 13 This is true even when one’s proposal is to give up our search for the right definition of ‘species’  because we cannot objectively define ‘species.’ Although several biologists have proposed this solution in the history of species controversy, many biologsts still do not give up proposing new defi-  18  1.3  Extant Answers to the Persistence Question  Though no focused treatment has been attempted previously, the persistence question has attracted some attention in the species controversy. Some biologists and philosophers have, in their own way, noted the persistence of the species problem and tried to explain it. In this section, I survey these answers and examine them.  1.3.1  Essentialism vs. Graduality —— the Vagueness Account  The most popular account emphasizes the gap between essentialistic aspects of the concept species and gradualness of evolution. Several thinkers point out that the concept of species involves some form of essentialism (or sharp boundary), but that evolution proceeds very gradually. As a result, one cannot divide the natural world neatly using the concept of species. For example, Jody Hey (2001a, p. 109f.) takes this line of argument (see also Hey 2001b): Evolutionary groups need not be distinct, and can be nested within one another, whereas categories are typically wielded with an all-or-noneness that comes with having distinct words. So finally we can sum up, in a few words, the cause of the species problem. Here it is in two sentences: (1) Biologists have been looking for a theory that explains the species they recognize and that can be used to define [the term ‘species’], and we have assumed that the theory will be about things that match up with our categories ... (2) However, the entities that are in the best theory ... need not and often do not match up with [our categories]. Mary Williams (1992) suggests a similar answer. She sees the origin of the species problem in the contrast between “T-Species” (species in taxonomic context) and “E-Species” (species in evolutionary context). Since she also suggests that TSpecies should be nonoverlapping, while E-Species are gradual, she is committed to the idea that the gap between extensional crispness implied by the taxonomic concept of species and evolutionary graduality is the cause of the persistence of nition. One can explain that biologists cannot reach the agreement on the right definition of species by saying that we cannot define it in the first place, but it does not explain why many biologists have not accepted this option. One needs to explain this.  19  the species problem. Let me call this vagueness account. One may formulate this account as follows: The Vagueness Account Taxonomic practice and the concept of species require the boundary between individual species to be sharp. However, since evolution proceeds very gradually, this boundary should be blurred, or vague. This is why we do not and cannot divide the natural world into precisely defined basic entities, or species. If it were possible for us to do so, there would be some species definition(s) which nicely draw clear boundaries which in turn do reflect the real species boundaries in the natural world. Thus the biologists could reach an agreement on the “right” species definition(s). This is not what is the case, however; there should be a number of borderline cases where we have little resource to decide what species a group of organisms belong to. This account takes it that the persistence of the species problem reflects the fact that our representation of the biological world through species classification does not match what the organic world is really like. The account also attributes the persistence problem to the fundamental gap between what our concept and/or our taxonomic practice requires and the way the organic world is. Since there is such a fundamental gap between our notion of species and the biological world, biologists will find it really hard to reach an agreement on and thus acquire the “right” definition of species that precisely reflects the way the world really is. Note that the graduality of evolution employed in this account is primarily temporal, but that the resultant vagueness is both synchronic and dischronic. Distinctions between different species are temporally vague due to evolutionary graduality. Even when one ancestral species is branching into two descendent species (and the ancestral species becomes extinct), it is hard to tell exactly when the branching occurs. This means that distinctions between different species can also be vague at one time slice; that one can hardly tell exactly when the branching occurs means there are time slices when one can hardly tell whether or not there are two——not one——species (I return to this point shortly).  20  This is a dominant account in the literature.14 A number of biologists, as well as philosophers——who otherwise support different views on species——plainly endorse it. For example, John Dupr´e begins his paper on pluralism (Dupr´e 1999) with this: Most of the philosophical difficulties that surround the concept of species can be traced to a failure to assimilate fully the Darwinian revolution. It is widely recognized that Darwin’s theory of evolution rendered untenable the classical essentialist conception of species. Perfectly sharp discontinuities between unchanging natural kinds could no longer be expected. As we will see shortly, Massimo Pigliucci (2003, 2005) and Richard Boyd (1999) propose a view, which enbraces elements of the vagueness account. Daniel Dennett (1995, p. 200) expands this form of argument beyond species and calls it “the most general form of the schema of Darwinian explanation.” Moreover, two founders of the biological species concept, Theodosius Dobzhansky (1935) and Ernst Mayr (1942, 1988), emphasize the gap between continuity in nature and the essentialistic aspect of the species category. Although Dobzhansky largely stresses synchronic, not diachronic, continuity found in the organic world, Mayr takes diachronic continuity in evolution as a cause of the problem. He compares evolutionary continuity with that in human growth. In [attempting to define ‘species’] we are confronted by the paradoxical incongruity of trying to establish a fixed stage in the evolutionary stream. If there is evolution in the true sense of the word, as against catastrophism or creation, we should find all kinds of species —— incipient species, mature species, and incipient genera, as well as all intermediate conditions. To define the middle stage of this series perfectly, so that every taxonomic unit can be certified with confidence as to whether or not it is a species, is just as impossible as to define the middle stage in the life of man, mature man, so well that every single human male can be identified as boy, mature man, or old man. It is therefore obvious that every species definition can be only an ap14 This is also the traditional “Darwinian” account. Darwin puts forward his theory of evolution to make sense of the indefinability of ‘species’ —— the fact that there has been no universally accepted definition of species in the biologists’ community. See Chapter 2 for more discussion.  21  proach and should be considered with some tolerance. (Mayr 1942, p. 114) The analogy is clear: just as one cannot tell exactly when Steve becomes middleaged, because he grows very gradually, one cannot tell exactly when a taxon becomes a new species, because this process also proceeds very gradually. George G. Simpson (1961, p. 60-61) employs a very similar analogy. He compares evolution to a piece of string with its color changing between its ends. It is blue at one end but it gradually changes color and ends up green at the other end. One cannot divide the string non-arbitrarily in terms of color (although Simpson believes that the resulting sections of a string, namely species, are perfectly real). This coincidence in analogies implies that the vagueness account looks fairly attractive to biologists and philosophers when they talk about the ontology of species. One thing to note: the vagueness account aims to draw a “big picture” of the problem and provide a fundamental explanation to the persistence of the species problem. Thus the vagueness account is compatible with various accounts of other aspects of the species problem. For instance, one would still need to offer another explanation for the existence of many different definitions. For example, Dupr´e (1999) endorses the vagueness account, but he also appeals to different research interests biologists have when he explains why there are so many different definitions of ‘species.’ Although Dupr´e may also take this as an explanation of the persistence of the species problem on its own, this and the vagueness account are not necessarily incompatible with each other.  1.3.2  Other Accounts  The vagueness account is certainly the most popular account for the persistence of the species problem. But there are other accounts in the literature. Lack of Philosophical Tools Massimo Pigliucci (2003, 2005) and Pigliucci & Kaplan (2006) follow the vagueness account, but makes an important addition. According to Pigliucci, it is a lack of philosophical consideration, not empirical findings, that is largely responsible for the problem. In particular, he continues, biologists have not found the conceptual apparatus, already available in philoso22  phy, to solve the species problem. It is family-resemblance. Pigliucci claims that species is a family-resemblance concept, like game; it is defined by a cluster of characteristics, but lacks essential properties. In other words, the species problem has long gone unsolved, because biologists approach it with the wrong conceptual tool. Essentialism is ill-suited to deal with the vague reality of the organic world. However, another type of concept——a family resemblance concept—— fares much better. If biologists would only recognize that the right tool is out there, Pigliucci bemoans, the species problem would no longer plague biology. Inflexibility of the Linnean System  Richard Boyd (1999) finds a problem in  adjusting the Linnaean classification system to a variety of explanatory needs in the biological sciences. Like classification systems in other sciences, such as the periodic table in chemistry, the Linnaean hierarchy is supposed to meet a variety of explanatory and predictive needs: a species is a unit of population genetics, ecology, phylogenetics, and so on. Generally, the Linnaean system meets them at the species level, because a taxon judged to be a species by one criterion will be judged likewise by another. The Linnaean system does not go much further, however. In particular, Boyd notices that it is difficult within the Linnaean system to modify a basic natural kind term, namely, species names, in order to meet different explanatory needs, whereas it is easy to do so in the classification systems of some other sciences. Take the periodic table as an example. Chemical elements are posited to explain a wide range of phenomena. Nevertheless, there are many phenomena that cannot be explained merely by positing elements. For example, it’s been found that different Neon atoms deflect differently through magnetic and electric fields and that different Uranium atoms have different abilities to cause fission chain reactions. Isotopes of the same chemical element, such as Neon-20 and Uranium-235, are introduced to explain these phenomena. In other words, sub-division of the basic units in a classification, such as isotopes, constitute semantic machinery for finetuning to more specific explanatory and predictive needs than the basic units of classification can accomodate. The idea is that fine-tuning like this is not available in the Linnaean system, because its semantic structure is not sufficiently flexible. As Boyd says, for example, “the Linnaean system of nomenclature does not have 23  devices, for example, to distinguish between subspecies from the point of view of ecology and subspecies from the point of view of the genetics of speciation” (p. 171). He concludes that this failure of the Linnaean nomenclature system to meet more specific explanatory and predictive needs is the cause of the species problem. ‘Species’ Is a Homonym Tom Reydon (2004, 2005) offers a totally different diagnosis of the persistence of the species problem focusing exclusively on the concept of a species. His diagnosis is that biologists and philosophers, literally, are not discussing the same thing in the species controversy. He believes different species definitions constitute altogether different notions. Take the phylogenetic species concept (history-based) (PSCh) and the phenetic species concept, for instance. These look at very different variables in species classification; the PSCh looks at the history of a population, while pheneticism looks at overall similarity (especially in morphology). They also disagree on the ontological status of species——the former takes species as historical individual, while the latter supposes they are classes of similar organisms. Moreover, the object judged to be a species by the former would evolve differently than the sort of thing judged to be a species by the latter. Reydon believes this sort of disagreement is more fundamental than it appears. In fact, the differences among definitions are so grave, he thinks, these accounts do not actually concern the same thing. Accordingly, different definitions designate different notions, on this view; ‘species’ is literally a homonym within biological discourse, meaning and referring to one sort of thing in the PSCh, but another in the BSC. If true, it is no wonder that biologists have reached no agreement on the species problem, because the supporters of different species definitions are really just talking about different things. This is Reydon’s answer to the persistence question. Degrees of Interdependency Among Parts Elliott Sober (2000) also offers an explanation of the persistence of the species problem. He compares species with organisms, and locates the origin of the species problem in differences between the functional interdependency of parts within organisms, on the one hand, and species, on the other. Sober draws attention to the fact that an organism’s parts  24  functionally depend much more on each other than the parts of a species——that is, organisms——do. For instance, if a heart is removed from a human being, he or she will not survive, even though the heart only accounts for a few percent of anyone’s total body weight. This is because other parts of a human body functionally depend on the heart——they work properly only as long as the heart functions well. One hardly sees such close functional relationships among conspecific organisms, except perhaps in some species of social insects. Sober notes this difference in functional interdependence derives from the fact that natural selection works differently at the organismal level than it does at the species level. Generally, individual organisms are subject to stronger selection pressure than groups of organisms (including a species). This means that the parts of an organism are often forced to cooperate with each other on behalf of the whole organism, whereas individual organisms do not have to cooperate so on behalf of the group. And, the more functionally integrated parts are, the easier it is to individuate the whole. Therefore, it is harder to individuate species than organisms. This is the origin of the species problem, according to Sober: Evolution may be responsible for the fact that the boundary between self and other is often clear in the case of organisms but can be more obscure when we consider species. If Darwin’s theory deserves credit for solving the problem of the origin of species, then the process that theory describes may be blamed for giving rise to the conceptual difficulty we call the species problem. (Sober 2000, p. 162, original italics)  1.4  Problems with Extant Views  We have seen various answers to the persistence question. Each answer, however, has problems of its own, and thus fails to tell the whole story behind the persistence of the species problem. I will describe these problems in this section.  1.4.1  Vagueness Account  Let us begin with the vagueness account. It faces two serious problems. First, it fails to provide a satisfactory answer to question 6 above: why does the species problem look more salient than analogous problems? After all, vague categories 25  abound. For example, can one offer necessary and sufficient conditions for being tall? The same is true of many other categories: red, drought, disease, life——the list goes on. If the vagueness account captured the whole story behind the persistence question, then scientists should be similarly preoccupied by other problems, such as the “redness” problem, the “drought” problem, the “disease” problem, the “life” problem, and the like, because the gap between essentialistic aspects of concepts and vagueness in the world holds in these——and many other cases——as well. This is not the case, however: the species problem appears to be more serious to scientists than other problems (Hey 2001a,b). Another point is this. If the gap between the essentialistic nature of the species concept, on the one hand, and the gradualness of evolution, on the other, was indeed the source of the species problem, then we could properly address the problem merely by inventing a concept whose extensional boundary is intentionally vague, thus capable of reflecting the vagueness found in nature. Although the vagueness account by itself does not imply this, it is a natural inference from the vagueness account. The problem here is that such accounts have been offered in response to the species problem, but none have not attracted much support or attention from biologists. HPC Theory Take the Homestatic Property Cluster Theory (hereafter “HPC theory”) as an account which invents a kind concept whose extensional boundary is intentionally vague, thus capable of reflecting the vagueness found in nature. This theory, originally proposed by Richard Boyd (1991) as a theory of natural kinds in general, has been applied to both species taxa and the species category by many philosophers— —such as Boyd (1999) himself, Wilson (1999), Wilson et al. (2007), and Griffiths (1999). Since this thesis is only concerned only with the species category, I will limit discussion to the HPC theory’s contribution there.15 Robert Wilson (1999, p. 197f.) nicely summarizes three basic claims of the HPC theory. First, “natural kind terms are often defined by a cluster of properties, no one or particular n-tuple of which must be possessed by any individual to which 15 See Ereshefsky & Matthen (2005) for a critique of the application of HPC theory to species taxa.  26  the term applies, but some such n-tuple of which must be possessed by all such individuals” (p. 197, my italics). For instance, a kind K will be defined by a cluster of properties f1 , . . . , fn . Each member of K must possess some of those properties, but they need not possess them all. The second point is that such properties are homeostatic: many members of the kind possess many properties in the cluster. Moreover, there are causal mechanisms behind those properties, and they cause those properties to instantiate at the same time. As a result, the possession of “any one of these properties significantly increases the probability that this individual will also possess other properties that feature in the definition” (ibid). If one member of K has f1 , this increases the probability that it has other properties f2 , . . . , fn as well. And the causal mechanisms mentioned above are responsible for this increase in probability. Finally, the HPC theory is a realistic view. Wilson says, “On the HPC view, our natural kind concepts are regulated by information about how the world is structured, not simply by conventions we have established or language games we play” (p. 198). K is a natural kind in virtue of the way nature is, not in lieu of any conventions we adopt regarding it. One example Wilson applies HPC theory to is retinal ganglion cells (RGCs). These cells are a kind of neuron found in retinas. They receive visual information from photoreceptors and transmit it to other regions, for example, in the midbrain. RGCs vary in many respects, so a couple of different classifications have been proposed for them. In one of these, various RGCs are classified into three subclasses called Y cells, X cells and W cells. Each class differs in different variables, such as receptive field centre size, axonal velocity, soma size, proportion of population, retinal distribution, central projections, and so on. Wilson asserts that each type of cell is a natural kind according to the HPC theory. He notes that not all the members of a given taxon, say, Y cells, have all the properties associated with it and that the above classification provides neither necessary nor sufficient conditions for each type of cell. Nevertheless, there is a strong tendency for those properties to be found together in individual cells. Furthermore, this tendency is not mere coincidence, because there are underlying mechanisms causally responsible for it, namely “mechanisms governing neural development and neural functioning” (p. 198). Wilson then applies the HPC theory to the species category (p. 199f.). This 27  category, he says, is defined by a cluster of properties in morphology, genetics, phylogeny, and the like. This helps us distinguish the species category from (i) non-evolutionary categories, such as diseases and predators and (ii) other taxonomic ranks, such as genus and subspecies. Predators are not defined by phylogenetic attributes and a subspecies is not reproductively isolated from other such entities. He does not specify the causal mechanisms responsible for coinstantiation of those properties, but since he refers to a variety of causal mechanisms, such as morphological development, genetics, and speciation when he discusses a species taxon, he would probably have in mind similar mechanisms when he discusses the species category. Thus described, a natural kind has indeterminate extensional boundaries. This is clear from the fact that the HPC definition usually does not specify how many f s in the cluster an object must possess to count as a member of the kind K. Suppose that the species category is defined by a cluster of properties, such as reproductive isolation, niche occupation, being monophyletic and so on, and that it qualifies as a natural kind among taxonomic categories. There should be some borderline cases where this definition is inadequate to determine whether or not a given population has reached the stage of becoming a distinctive species. Wilson recognizes this, but takes it as a virtue, rather than a defect, of his account when compared to the more traditional essentialistic account of natural kinds. Considering there is vagueness in nature and in species boundaries, in particular, a natural kind concept, such as the HPC theory, has an advantage in that it does not oversimplify the reality of nature: . . . there clearly will be cases of genuine indeterminacy with respect to both the species category and membership in particular species taxa. Yet this indeterminacy seems to me to reflect the continuities one finds in the complex biological world, whether one is investigating species, neurons, or other parts of the biological hierarchy. (p. 202) Therefore, HPC theorists should be proud of this feature of their account. From the viewpoint of the vagueness account, the HPC theory may well offer the solution to the species problem. The vagueness account proposes that the species problem comes from the gap between essentialistic nature of the concept species and evolutionary gradualism. HPC theory, however, allows the natural kind 28  concept species to leave the species boundaries vague. As a result, there should be no gap between concept and reality. Therefore, if the vagueness account of the persistence question is correct, the HPC theory may well solve the species problem. One problem with this argument is that not nearly as many biologists as philosophers have been attracted to the HPC theory as an account of the species category. As we have seen, many philosophers apply the HPC theory to the species category and discuss its implications. However, as far as I know, few biologists other than Olivier Rieppel (2007) even discuss it. Thus avoiding the gap by inventing a definition the extension of which is intentionally vague does not suffice to solve the species problem. There may be a quick response from the vagueness account to this allegation, however. The HPC theory is, after all, a philosophical theory: it was originally proposed by philosophers, and it has rarely been discussed outside philosophical circles. Moreover, one should not forget that it has only been 10 years since it was applied to the species problem (Boyd 1999, Wilson 1999, Griffiths 1999). Although it did not take long for biologists to become familiar with the species-as-individual thesis (Ghiselin 1974, Hull 1978), this may be a dramatic exception. No one should be surprised to see it takes longer for biologists to become aware of philsophical accounts like the HPC theory. This is a sensible response. No one could be certain that the HPC theory would not fare well within the biological community sometime in the future. Nevertheless, there is a reason——perhaps a good reason——to think biologists would not be interested in the HPC theory. It is that an apparently similar account to the HPC theory was proposed 50 years ago, but has failed to attract much attention from the biological community: Morton Beckner’s polytypic definition of species. Beckner’s Account Morton Beckner (1959) proposes a concept of species, in which species boundaries are intentionally vague. Beckner believes one cannot give necessary and sufficient conditions for something to be a species. Any adequate definition of species, he says, will be what he calls a polytypic concept. According to Beckner, defining a polytypic concept involves the following (p. 22). When one defines a groups K in terms of a set G of properties f1 , f2 , ..., fn :  29  1. Each [member of K] possesses a large (but unspecified) number of the properties in G 2. Each f in G is possessed by large numbers of these individuals, and 3. No f in G is possessed by every individual in the aggregate. It should be clear that the extensional boundary of such a concept will be indeterminate. The number of f s one should have, in order to count as a member of K is not specified. It should also be clear this account has the similarities to the HPC theory. Although Beckner does not give up the traditional, more essentialistic account of natural kinds——he is a pluralist about natural kinds——nor refer to causal mechanisms, he recognizes that the boundaries of different species should be vague under this definition. Beckner’s account is a predecessor of the HPC theory in this respect. Therefore, if the vagueness account is right, Beckner’s account may also attract the sympathy of biologists. But this is not the case. Over the past fifty years, biologists have simply not been interested in this idea. For example, in two anthologies (Claridge et al. 1997, Wheeler & Meier 2000) on species, most authors of which are biologists, Beckner (1959) is cited only in one paper. Although that paper (Van Regenmortel 1997) is sympathetic to Beckner’s idea, the paper itself is about viral species, which, until recently, has not attracted much attention in the species debate. Moreover, in Ereshefsky (1992b), which gathers classic papers in the species debate from both philosophers and biologists, Beckner (1959) is only once cited by Hull (1964). This casts serious doubt on Beckner’s proposal and thus the vagueness account. And this is important because 50 years are certainly not too short for biologists to become aware of Beckner’s proposal. In other words, the same response is not available to Beckner’s account as it is for the HPC theory. But if Beckner’s account and HPC theory are similar in that they allow the boundaries of different species to be indeterminate, we have good reason to expect that the HPC theory will not attract much support from biologists either, just because the HPC theory allows indeterminacy. Let me summarize here. If the vagueness account is right, that is, the gap between the essentialistic nature of the species concept, on the one hand, and the gradualness of evolution, on the other, was indeed the source of the species prob30  lem, then we could properly address the problem merely by inventing a concept whose extensional boundary is intentionally vague, thus capable of reflecting the vagueness found in nature. Under Beckner’s proposal, the gap between the essentialistic nature of the species concept, on the one hand, and the gradualness of evolution, on the other, is resolved. But somehow few biologists have endorsed it as a solution; the species problem still remains. This suggests that merely proposing a way to avoid the gap conceptually does not suffice to convince biologists that the species problem will be resolved. True, it may be the case that Beckner’s proposal actually is the solution to the species problem and it is just that the biological community does not notice it. But the fact that biologists have paid little attention to Beckner’s proposal for decades seems to carry a lot of weight against the vagueness account. This is not to say that the vagueness account is completely off the point. If evolution proceeds gradually and species definitions generally are essentialistic (possessing an “all-or-none-ness”), then there should be some borderline cases in species classification. What I have argued is that this account will leave behind some aspects of the species problem. In this sense, the vagueness account is not wholly mistaken, but rather an incomplete explanation of the persistence of the species problem.  1.4.2  Other Accounts  The last section suggests that the vagueness account does not tell the whole story behind the persistence of the species problem. Other accounts, however, do not fare much better. Pigliucci’s Account Does Not Stand Alone Pigliucci partly follows the vagueness account. But he also attributes the persistence of the species problem to a lack of conceptual resources. In the last section I provided an objection to the first part of Pigliucci’s account. The second part turns out to be no more convincing than the first. It is that the species problem persists because biologists are not fully aware of the existence of family resemblance concepts. The problem is that his account sounds like biologists are plagued by the  31  species problem, because they haven’t reached the solution to the species problem (the species-as-family-resemblance view, in his case). This violates the stand-alone criterion (p. 18): the adequacy of Pigliucci’s response to the persistence question depends on the adequacy of his response to the species problem. If the speciesas-family-resemblance view turns out to be the wrong way to treat the concept of species, explaining the persistence of the species problem via biologists’ ignorance of family resemblance concepts becomes a complete dead-end. One might object that Pigliucci’s account to the persistence question does not totally depend on the adequacy of his answer to the species problem. He does find some support for his account in the observation that the empirical study of evolution has made significant progress, even without resolution of the species problem. This observation is important, and will certainly be respected by my final account. However, it still does not logically imply family resemblance will cut the Gordian knot. Nor does it even imply that consideration of philosophical aspects of the problem should be required to answer the persistence question. This is because there may be non-philosophical and non-biological factors, which help the species problem persist, in spite of efforts biologists have made regarding empirical questions of evolution. Therefore, Pigliucci’s response to the persistence question is ultimately unsatisfactory. Semantic Inflexibility of the Linnaean System? Richard Boyd’s answer to the persistence question is that the Linnaean classification system lacks the semantic resources to accommodate finer explanatory and predictive needs. In particular, Boyd points out that under the Linnaean system, it is hard to distinguish taxa judged as subspecies under different criteria. This is an awkward answer to the persistence question, because the inflexibility of the Linnaean system, as Boyd points out, has rarely been discussed in the history of the species controversy. It is true that the Linnaean system has been criticized for its semantic inflexibility, by such thinkers as Ereshefsky (1999, 2001) and Mishler (1999). But it is not the compositional semantic inflexibility of the Linnaean nomenclature, but rather its obligation to assign a “rank” to each taxon in the taxonomic hierarchy, that has  32  been at the centre of discussion. Under the current nomenclature, a taxonomist should assign an appropriate rank to a new taxon if she wants to make the name of that taxon legitimate. One should not simply register a new taxon as a taxon; she should also pick a rank——among genus, species, subspecies and so on——and assign it to the taxon. This could cause unnecessary confusion to the taxonomic community, because a taxonomist may be forced to classify a newly found taxon as, say, a species, even when such an assignment is simply the result of guesswork, due to the lack of sufficient information about the taxon. Of course, there are other reasons to abandon this practice in the classification system, but this is not the place to examine those arguments. In any case, semantic inflexibility, as Boyd points out, has never been at the centre of discussion. Even if some taxonomists may have made the same point as Boyd, it is certainly a minor point in the species controversy when one reflects on its very long history. Moreover, Boyd’s account is in a danger of making the history of the species controversy unintelligible. After all, one could, at any time, invent semantic machinery suitable for fine-tuned explanatory and predictive needs, and then the taxonomic community could easily overcome the species problem——or so Boyd’s view seems to imply. But this is a wildly implausible account of the persistence of the species problem, because if his account is right, then one would wonder why no biologists would have put forward such a proposal. Therefore, the semantic inflexibility of the Linnaean system may play a relatively minor role in the persistence of the species problem, but it is certainly not the major culprit that Boyd suggests. ‘Species’ as a Homonym? Let us move on to Reydon’s view. Reydon’s account, recall, is that different species definitions simply correspond to different notions. Thus, no wonder that biologists have not produced a unanimously accepted definition of species; there is no such thing as species per se. Although I think this analysis sheds light on some aspects of the species controversy worth serious examination, it also leaves so many aspects of taxonomic practice unintelligible that it comes to look quite implausible. First, if Reydon is right, then biologists endorsing different definitions of species are simply not talking about species per se. But the fact they believe themselves to  33  be still remains. Biologists are likely to continue believing that their own definition is about species. Reydon’s position offers no resources for explaining why biologists and philosophers still call their own definition, the “definition of species.” Reydon’s account also indicates that biologists, as well as philosophers, are fundamentally wrong in believing they are addressing one and the same notion, namely, species. Needless to say, this is a bold position; one wants to know why seemingly all but Reydon have committed such a fundamental error. But Reydon attempts no explanation for this alleged mass confusion——none at all. Second, as we will see later, there is a general usage of ‘species,’ a usage in which biologists leave its specifics open——that is, without having any specific species concept in mind (See Chapter 3). For example, Luckow (1995) points out that taxonomists often leave the specifics of their judgments open when writing papers describing new species and making their taxon names legitimate, as though they believe there is one and the same concept, species, and feel no need to refine it into a more specific species concept.16 This indicates that biologists do have a concept of species, although they use it in an unarticulated way. This general notion of ‘species’ is also observed when biologists use an unofficial term ‘good species.’ This term refers to a taxon which satisfies most of the species criteria proposed so far ——such as reproductive isolation, niche occupation, being monophyletic, and so on——, and is thus generally judged to be a species by competent taxonomists. Biologists often study a good species to make an inference on the nature of the species category, assuming that a good species is a species, because it will be a species whichever species criterion one will eventually adopt. This notion, good species, is obviously derived from the more basic notion species. If there is no such concept as species, it is hard to make sense of this usage of good species (I will discuss this more extensively in Chapter 3). Therefore, ‘species’ is not a homonym. Biologists do not always use ‘species’ just as a substitute for a particular species definition. 16 Furthermore, the naming rules of biological taxonomy, such as the International Code of Zoological Nomenclature (Ride et al. 1999), include no definition of species. See discussion on Hugh Strickland and his contribution to the establishment of the zoological naming rules in Chapter 2.  34  Figure 1.4: Anagenesis: A species A evolves into a new species B. A extincts as a result of this process. See text for the details.  Functional Interdependence Sober ascribes the persistence of the species problem to the different degrees of functional interdependence among parts of an organism and those of a species. Parts in an organism are much more tightly tied to each other than organisms in a single species are, such that removing a heart from an organism leads to its death, whereas removing one population from a species may not mean the extinction of the entire species. This is because an organism, as a whole, is subject to stronger selection pressure than a single species. This lack of interdependence in a species partly makes the species problem hard to solve. Sober’s account has merit in that it can explain why biologists are not as generally bothered by the concept of an organism as they are of that of a species. However, one cannot extrapolate much further. Sober’s account, for example, does not explain why scientists are not bothered by similar problems, such as the “life problem” or the “drought” problem. The second, more serious concern is that Sober’s account probably shares certain crucial elements of the vagueness account. And so, his account inherits all the problems we previously identified for the vaguenessness account. Sober would deny this, because he declares that he does not accept the central idea of the vagueness account ——the idea that there is a fundamental gap between the essentialistic feature of the notion of species and the vague species boundaries in the natural world—— in the first place when he discusses species essentialism—— although he does not discuss the possible relationship between the two accounts. Sober presents an argument for the rejection of the existence of the fundamental gap when he argues that one criticism of species essentialism does not work (Sober 2000, p. 150f.). One standard criticism of essentialism is quite similar to the  35  vagueness account; that is, species essentialism does not capture the gradual nature of evolution, because an essentialist commits to the all-or-none-ness of the species boundary while there are plenty of borderline cases in species classification. But Sober does not think that this is a good objection to species essentialism. He begins his objection to this criticism with the fact that cladogenesis or branching——not anagenesis17 ——is the dominant mode of speciation; when a new species arises, it usually involves branching of a lineage. Although he does not deny that anagenesis causes the vagueness of the species boundary, in cladogenesis there is an event marking off the beginning of a branching (and thus a new species), such as the establishment of reproductive isolation. Sober illustrates this with a possible scenario of branching speciation: A small group of rabbits is isolated from its parent popultion because a river changes its course. Selection then leads this isolated population to diverge from the parental population; the isolated rabbits turn out to be the founder of a new species. We may wish to date the birth of a new species with the initial separation of the two population or the subsequent fixation of traits that prevent interbreeding between the two lineages. The point is that, whichever proposal we follow, the cutoff point is precise enough. (p. 151) Of course, a population rarely becomes isolated from another population in an instant; one cannot determine branching events at such levels of precision. However, the moment at which one lineage branches into two is determined practically and theoretically with sufficient precision. In other words, the temporal boundaries between species are not absolutely sharp, but sharp enough, all the same. Sober makes this point to argue for species essentialism, because the point of the example is that even though we cannot draw a precise species boundary in cladogenesis, the boundary is as clear as species essentialism requires.18 In other words, the gap between the essentialistic feature of the notion of species and gradual evolution is not as fundamental as critics of species essentialism (and the supporters of the vagueness account) believe. 17 In  anagenesis, one species evolves into another species without any branching of a lineage. Species A changes into species B within a single lineage. An ancestral species will extinct by definition in this case. See Figure 1.4. 18 Sober eventually objects to essentialism by arguing that a species is a historical entity. I do not address this now because it is not relevant to the current discussion.  36  Figure 1.5: A species C evolves into two new “species” A and B. See text for the details.  One may wonder how this rejection of the vagueness account and Sober’s own account of the persistence question are compatible with each other, for on reflection, they seem not to be. If Sober’s account based on weak interdependence among organisms in a single species is right, then the species boundary is, at least sometimes and perhaps frequently, indeterminate at a single time slice. Rejection of this virtually leads to the rejection of the existence of the species problem. Let us call those “species” A and B (see Figure 1.5). A and B share a most recent ancestor C and have recently branched, but there are times, say t0 , when one cannot be sure that they are different species, because each “species” exhibits only weak functional interdependence among its members.19 Put differently, there are times when the boundary between A and B is vague synchronically. However, since evolution, and speciation in particular, typically proceed gradually, let us assume that the branching process of A and B is also gradual. Therefore, the boundary between A and B is also vague when we look at them temporally from t0 −t1 , not just synchronically. Let me elaborate this by specifying a mechanism behind species branching. Suppose we take the establishment of reproductive isolation as marking the birth of a new species. (This is just an initial supposition. My argument does not depend on this choice of species criterion). Let us recall that a lesser degree of interdependence within a taxon A is either 19 Notice  that I do not specify the mechanisms behind branching of A and B and functional interdependence (or lack of thereof) within each “species.” I will consider a possible case in the next paragraph where species branching is measured by reproductive isolation.  37  relevant or irrelevant to A’s being less reproductively isolated. That is, when there is weak functional interdependence within A, it is the case either that 1. A has lesser degree of reproductive isolation, or that 2. A is sufficiently isolated from other populations at t. Either way, synchronic vagueness expressed by weak interdependence within a species, leads to temporal vagueness envisioned by the vagueness account. If Sober means (1), then the temporal boundary between species is not sharp enough even in terms of reproductive isolation at that time. Thus in A weak interdependence and synchronic vagueness of species boundary simultaneously obtain. So, synchronic vagueness can be translated into diachronic or temporal vagueness. Perhaps what Sober means is (2). Then at t, a taxon A is actually a species in spite of the weak interdependence in A; then it would be puzzling why weak functional interdependence is relevant to the species problem. Therefore, it follows from Sober’s thesis either that functional interdependence and temporal vagueness cannot be disconnected or that weak interdependence is not relevant to the species problem. But either way, one cannot maintain the conjunction of the weak interdependency thesis and the denial of the vagueness account. This can be seen as an application of the point Stamos (2002, 2003) makes regarding the relation between synchronic and diachronic dimensions of species.20 Part of his claim is that the synchronic dimension is ontologically prior to the diachronic or temporal dimension when it comes to species: if there is no species at any one time slice, then there is no species temporally or chronically, but not vice versa——a species can be real at some time slice even though there is no species across those slices. Here, I am arguing that Stamos’ claim largely holds when it comes to vagueness of the species boundary21 : if the species boundary is vague 20 Stamos uses ‘horizontal’ and ‘vertical’ dimensions to refer to synchronic and diachronic or tem-  poral dimensions, respectively. 21 Application of Stamos’ thesis to the vagueness of species boundaries does not always hold. Take the rabbit example above. Let us call the isolated population, A, and the rest of the original population, B. Let us designate t0 in Fig 1.5 as representing when the river changes its course and A is isolated from B. Now suppose that reproductive isolation is completed at t1 —— Pr(mate(A, B))=0, where mate(X,Y ) means that individuals from populations X and Y can mate each other. So, it is indeterminate whether individuals from A and B can mate (0< Pr(mate(A, B)) <1) at any one time slice t (t0 < t < t1 ). If one defines a species as a distinct lineage, then it is also indeterminate whether  38  at a single time slice (due to the weak functional interdependence), this translates into vague species boundaries in the phylogenetic dimension. My argument goes like this: (i) For any species s, if the synchronic boundary of s is vague at any one time slice t, then its boundary is also vague around t temporally or chronically. This comes from Stamos’ claim. The next premise is Sober’s thesis. (ii) Weak functional interdependence is the cause of the species problem. The third premise is an implication of Sober’s thesis. If his weak interdependency thesis is right, then there should be some species at t such that functional interdependence among its members is sufficiently weak and thereby its synchronic boundary from other species is vague. (iii) If weak functional interdependence is the cause of the species problem, then there is a species s at t such that its functional interdependence is weak and thereby its synchronic boundary between s and other species is vague. From (i), (ii), and (iii), it follows that (iv) Hence, the boundary of s is vague diachoronically. Thus, if Sober’s point on the weak interdependency entails the synchronic vagueness of the species boundary, then it will probably translate into the diachronic vagueness of species boundaries. Therefore, the problems with the vagueness account apply to Sober’s account of the persistence question as well. an organism x is a member of a distinct species A at t if one judges at t, because no one knows at t whether or not A and B will converge and become a single lineage (thus A does not persist as a distinct lineage) in the future. But when one judges at t (t1 ≤ t ) whether x is a member of A at t retrospecitively, it is determinate, because one knows at t that a lineage which x is a member of at t would persist as a distinct lineage even after t. In short, it is indeterminate whether x was a member of a distinct species A at t (t0 < t < t1 ), if one judges synchronically at t, whereas it is determinate if one judges diachronically at t (t1 ≤ t ). Thus synchronic indeterminacy of x’s membership is not always translated into diachronic indeterminacy. But this is an exceptional case where two lineages unite after temporal splitting. When the unification does not occur, the above argument is still valid.  39  Summary In this section, I examined current answers to the persistence question. It turned out that each, including the vagueness account, have problems; none fully respond to the question, in all of its nuance and subtlety. My project will answer some of those sub-questions, which previous attempts have not properly addressed.  1.5  Outline of the Project  I have formulated the persistence question ——the persistence of the species problem— — and reviewed the extant responses to it. It turns out that these are insufficient to answer the persistence question properly. This is where my project begins. In the rest of this dissertation, I will address those questions to which prior accounts have not paid sufficient attention. I outline the details of the project here. This introductory chapter will be followed by three chapters. Chapter 2 (“Sharing a Reference”) In this chapter, I will offer a part of what will be an answer to sub-question (7): how biologists could do their business without a general solution to the species problem. In particular, I will focus on the issue of communication between biologists, because the lack of an accepted solution to the species problem would naturally seem to lead to a communication breakdown. I will describe three cases where one might suspect that biologists would be unable to effectively communicate with each other because their ideas about species and speciation differ significantly: Darwin figures preeminently in the first; the second involves Grey and Strickland; and the third involves Bush, Mayr, and Coyne & Orr. Through analyzing these cases, I argue that sharing a reference would help biologists (or naturalists) communicate effectively with each other and do their business, even while they face the lack of a solution to the species problem. If they agree on which group of organisms they are discussing, in terms of extension, differences in their conceptions of species and speciation, even if significant, would not hinder them from successfully communicating with each other. Chapter 3 (“Dual Process Theory and the Concept of Species”) Several subquestions I put forward concern biologists’ ways of handling species in their study, 40  such as questions (1), (6), and (7). To help us answer those questions, I will describe the ways in which biologists work on——study, define, classify, etc.— —species and the ways in which they psychologically represent the concept of species. In particular, I show that biologists use two psychologically distinct processes in working on species and mentally representing the notion of it, and that these two processes largely correspond to the two processes proposed by the Dual Process theory in cognitive and social psychology. According to the Dual-Process theory (DPT), humans employ two psychologically distinct processes on many occasions: an implicit and unconscious process (“System 1”), and an explicit and conscious process (“System 2”). I will explain how the DPT can make sense of a variety of biologists’ practices regarding species. First, I will focus on what might be called the elusive transparency of the notion of species: biologists believe that they understand the nature of species, but they find themselves unable to adequately define ‘species’ when asked to. To explain this phenomenon, I will discuss the notion of good species. One often comes across this unofficial phrase in taxonomic papers, but it has attracted relatively little attention in the species controversy. After analyzing the meanings of ‘good species’ in taxonomic literature, I point out that ‘good species’ means a taxon judged to be a species according to multiple species criteria in many cases. Then I argue that good species involves a prototype conception of species, and as such it is processed in a more implicit, unconscious process (System 1), rather than a more explicit, conscious process (System 2). If this is the case, then biologists may well make what psychologists call attribute substitution to the species category; biologists tend to represent the species category with its prototype, good species and infer various attributes of the former from those of the latter, when they talk about the species category in a casual conversation. This explains elusive transparency. The category of good species usually looks quite homogenous in their minds in that one can easily tell a good species from other species, because a good species satisfies many different criteria of species. If biologists represent the species category by its prototype, good species, then the prototype makes it look as if they figure out the nature of species (because good species as represented in their minds looks quite homogeneous), even though biologists should be aware of the difficulty of defining ‘species.’ 41  The fact that biologists often represent the species category with its prototype, good species suggests that there are at least two ways in which biologists represent the concept of species. Psychologists have discovered that one represents a concept in different ways. One way is to represent it with its definition, just as one represents bachelor with its definition, being an unmarried man. The second way is via its prototype. An example is that one represents sport with its prototype, baseball. One can apply this distinction to understand how biologists represent the notion of species. On the one hand, they represent it with the help of various definitions of it. In other words, biologists represent the notion of species in terms of the conditions under which an object is to satisfy to become an instance of species. On the other hand, biologists often represent the notion of species by associating it with its prototypical instances, good species. Many philosophers, however, only tend to focus on the first way but pay little attention to the second way. I also discuss psychological essentialism. Psychological essentialism is a psychological inclination to assume essential properties behind superficial ones of a kind even when one is unaware of what they are. I claim that psychological essentialism is a basis of taxonomists’ practices and that psychological essentialism is arguably a product of System 1, rather than System 2 thinking. In the last section, I will describe elements of System 2 in biologists’ reasoning regarding species. I call this definition-centred reasoning about species. When biologists work on definitions of species——for instance, inventing a new definition (necessary and/or sufficient conditions for something to be a species) and defending or objecting to a certain definition——they are engaged in definition-centred reasoning. This type of reasoning, I argue, has a number of characteristics of System 2 reasoning. This concludes my attempt to describe biologists’ practices in light of the Dual Process Theory. Chapter 4 (“Answering the Persistence Question”)  In the final chapter I will  give my account of the persistence of the species problem, by answering some of the sub-questions I put forward in Chapter 1, including: (1) Why do biologists believe ‘species’ calls for definition? (2) Why does no definition command universal support? 42  (7) How could biologists do their business——conduct their research——about species and speciation without a unanimously accepted solution to the species problem? For question (1), I consider two answers. First, biologists have a practical reason to call for a definition of ‘species.’ Biologists and taxonomists need a classification system to catalog the biological world for a variety of reasons; one of them is to ensure effective communication among researchers. For this purpose, biologists believe, the terms for major classificational units should be defined so that they can be sure that they talk about the same thing with the same term. Species is a basic unit in that catalog, and thus needs to be defined. This is what many people have pointed out and few people would doubt. Psychological essentialism is another factor. Because of psychological essentialism, people, including professional biologists, believe that the species category is a natural kind: this category has essential properties causally responsible for the superficial properties which a species typically exhibits, even though they do not know what they are. This drives them to seek its causal essences. And a definition is a good way to describe them. To answer the question 2 above, I offer what I call the argument from interestrelativity, which is quite similar to the pluralist’s argument. This argument aims to show that no or very few species definitions will ever be unanimously accepted by the biological community. The argument goes as follows: if (i) the purposes or interests for which biologists employ the concept of species differ from one situation to another, (ii) there are different criteria of species invoked under different interests, and (iii) a taxon satisfying one criterion often fails to satisfy another, then (iv) there will be very few, if any, unanimously accepted definitions of ‘species.’ In the last section, I argue that good species offers a clue to explaining why biologists can still do their business regarding species without a unanimously accepted solution to the species problem (question 7). From discussions in Chapter 3, one can see that there is good reason to believe taxonomists often make attribute substitutions on the concept of species——they represent species with a prototype, good species. So long as they work on stereotypical species and do not come across any problem in studying or classifying species, they would neither have any moti43  vation to abandon this substitution nor be bothered by the species problem, because the nature of species looks sufficiently clear to them.  44  Chapter 2  Sharing a Reference In this chapter, I offer part of my answer to the question of how biologists can continue their research without a solution to the species problem, in particular how they can still effectively communicate with each other.  2.1  Introduction —— How Can Biologists Do Their Business; How Can They Communicate?  As we have seen in Chapter 1, this project focuses on the epistemological role of the species category in biological research. One can fairly summarize the current state of play regarding the species problem with the following observations: 1. Biologists have no “solution” to the species problem. There are many competing definitions of ‘species.’ None commands universal assent. 2. The species category is important in biology. 3. Biologists have made progress even in areas where the species category is important (such as speciation and biodiversity). Given the first and second observations, one may expect (though they do not strictly imply it) that biologists can make little to no progress in their research without a universally accepted species concept. However, the third proposition says otherwise, and rightly so. For evolutionary biologists frequently disagree on the nature of species, and yet we still think opponents can learn much from each other on various topics, such as speciation. This assessment is surely plausible 45  and equally commonsensical, but it does raise a philosophical puzzle. On closer reflection, it is natural to wonder how biologists could even communicate with each other——never mind learn from one another——about species, while lacking a shared theory of what a species is. For without one, it isn’t unreasonable to suppose that opposing theorists are simply talking about different things under the rubric of ‘species.’ This preliminary statement motivates the issue somewhat, but now let’s elaborate it more carefully. Possible Problem of Communication  In general, we have difficulty communi-  cating with each other when we do not talk about the same object. One of these situations is when different people attach different meanings to the same word. An easy example is this: when one person uses the word ‘bank’ for a financial institution, and the other uses it for a sloping land running along a river, confusion may well ensue. Surely a simple case like this is relatively easy to identify and handle compared to similar, but trickier, scenarios involving ‘species’ in biological contexts. In the species controversy, however, some philosophers have raised the very same concern about species pluralism (Hull 1999). Under pluralism, when different biologists mean different things (an interbreeding population, a minimal monophyletic group, etc.) by the word ‘species,’ it may prevent effective communication, because they may not be talking about the same thing, while falsely assuming they do. And basically the same type of concern could arise when we do not have a universally accepted definition of ‘species.’ Sharing a Reference How can biologists from different perspectives communicate with each other under such circumstances? Part of the answer I want to develop is supplied by the idea of sharing a reference. If biologists or naturalists can agree (to a sufficient degree) on which groups of organisms they discuss in terms of extension1 , then differences in their respective conceptions of “species” ——even if significant—— will not hamper biological discussion. This approach has been suggested several times in the history of biology. In what follows, I describe three episodes at length. They are readily listed in terms of their participants: 1 In the taxonomic literature, the term ‘circumscription’ is used to refer to the limits of a taxon as determined by an author (Lincoln et al. 1998).  46  (i) Hugh Strickland and John E. Grey, (ii) Charles Darwin and his essentialist contemporaries, and (iii) Guy Bush, on one side, and Ernst Mayr and Jerry Coyne & H. Allen Orr, on the other. The naturalists figuring in these cases were engaged in different aspects of evolutionary biology and classification. Strickland was among those who invented the first Code of Zoological Nomenclature. Grey was a curator at the British Museum, in charge of making a catalog of the organisms in 19th century England. Darwin, of course, wrote The Origin of Species without any clear definition of species at all. Bush, Mayr, and Coyne & Orr debated the plausibility of sympatric speciation. Each had to contend with the disagreement regarding the “right” definition of species in trying to make their views widely understood. More recently, Bush and others even disagreed on the priority between giving a definition of species and the study of speciation: Bush claiming that we should study speciation before giving a definition of species——others denying this. However, Grey, Strickland and Darwin all emaphasized the possibility of a shared reference to enable theoretically opposed naturalists to talk about the same group of organisms and avoid problems of miscommunication. Bush and opposition also discussed the very same taxa in their debate on the possibility of sympatric speciation and thus did not fall prey to miscommunication and unintelligibility.  2.2 2.2.1  Case Studies Strickland and Grey——Natural History in the 19th Century England  In this section, I will describe two naturalists in 19th century England who tried to ensure the possibility of communication among naturalists when facing radical disagreement about the “right” definition of species: Hugh Strickland and John Edward Grey. My account draws heavily on the work of Gordon McOuat, who has authored multiple papers on them. McOuat explains how naturalists tried to “solve” the problem of naming species without appealing to any particular species concept by looking at the founding history of the Rules for Zoological Nomenclature (McOuat 1996). He also discusses an attempt by John Edward Grey to cate-  47  gorize species in the British Museum where he worked as a curator in the natural history section (McOuat 2001). Hugh Strickland: Nomenclature and Definition of Species McOuat’s first paper focuses on naturalist Hugh Strickland’s “solution” to the species problem as a founding member of BAAS Committee on Zoological Nomenclature. In the 19 century’s England many “conservative” naturalists in the Linnaean Society largely followed the Linnaean hierarchical system of classification and the Lockean view on naming. On this view names need not represent the properties of the bearers; rather, names are directly connected to the object and do not tell us anything about the essence of the things. However, beginning in the 1830s, “reformist” naturalists began to react against the “conservative” naturalists. Some reformist naturalists such as Neville Wood and Charles T. Wood, called for the reform of the original Linnaean system of nomenclature. They argued that, contrary to Locke, names should reflect the essence of the objects named. A name should mean something with respect to its object: it should convey some information about the object in virtue of the name itself. A “name should immediately refer to some distinguishing feature (differentia)” (498, original italics). While Strickland admitted the need for reform, as a conservative he supported the Lockean account of names and replied that a name is just a tag for its object. Moreover, he wrote a proposal for “Rules for Zoological Nomenclature” in 1837 (Strickland 1837) and in 1842 proposed that the British Association for the Advancement of Science (BAAS) establish a special committee to discuss his draft in order to make a recommendation to the society with respect to its adoption as a law for zoological nomenclature. The purpose of this effort was to institutionalize his philosophy on naming species widely shared among conservative naturalists, as the official rules of nomenclature. The committee’s Report (Strickland 1843) was presented at the 12th meeting of the BAAS, but the proposal was rejected due to fierce opposition by the reformists. In spite of this, Strickland was able to print the rules in the official 1842 BAAS report, giving “the impression that it was an official document of the BAAS” (McOuat 1996, p. 509). This in turn led to acceptance of similar rules in United States and Italy. A French translation of the Report was  48  published too. This is part of the story of how conservative naturalists gained the upper hand in this controversy over nomenclature. But what is striking about this report for our purposes is that it does not specify either the ontological status or essential properties of species. For example, the first draft says that species are “tangible objects” (“Rules, Second Draught,” Strickland Papers, Cambridge University, Museum of Zoology, cited in McOuat 1996, p. 511). However, Strickland faced E. H. Bunbery’s objection that species are an abstraction and that only individuals can be tangible. As a result, he did not ultimately put any definition of species into the Rules. According to McOuat, the significance of this is that it offers a “solution” to the species problem without any definitive definition of it by mutual understanding of “competent” naturalists. “For the Rules, species were just what competent (read: institutional, published, gentlemanly, conservative) naturalists said they were.” (512). McOuat suggests that there may have been a way of understanding species without the use of definition when he says, The “Rules of Nomenclature” were rules governing behavior, of proper etiquette for membership within that elite body. Yet, there was, there could be, no agreement on exactly what a species was, definitionally. (McOuat 1996, 515. Italics added) That is, it is suggested that naturalists understood species in some non-definitional way.2 Another point is that this strategy employed by Strickland enabled naturalists to communicate smoothly with each other: “the solid core of conventional names marked something unsaid, yet something ‘agreed’ upon on trust: species are nodes in the communication of a network of ‘competent’ naturalists” (511). If particular philosophical or ontological commitments were attached to the idea of species, different naturalists would mean different things. This ontologically and philosophically stripped down characterization of species would prevent this from occurring. 2 McOuat thinks that one of the factors supporting this mutual understanding is that naturalist contemporaries of Strickland, including Darwin ——Darwin was a member of Strickland’s BAAS committee—— shared the same or similar set of examples of species (species taxa) with each other (personal communication).  49  John Edward Grey: Cataloging Species Without Species Concept McOuat’s 2001 paper discusses the same controversy between reformists and conservatives on nomenclature, in different settings: the British Museum and the parliamentary committee on the Affairs of the British Museum. The main actor here was John Edward Grey, an assistant keeper at the British Museum. After the Radicals gained more strength in the parliament, public institutions such as the British Museum came under scrutiny by the Radicals: its role, its relation to public and so on. Both reformer and conservative naturalists were called upon by this committee to voice their opinions on the current state of the Museum and its practices. Among the topics under discussion were how to name and classify organisms and how to (and whether or not to) exhibit the results to the public. Like those concerned with taxonomic practice, their concerns included whether to use the Linnaean system of classification or to introduce some new system, and what the nature of naming is (referring to essential features of species or just giving a tag to the named objects). McOuat’s point in discussing the conflict is that Grey’s method of cataloging organisms was one of the factors which made conservatives favorable in the Committee. Grey’s method for cataloging was as follows: • Assign each species a separate leaf (for example, assign leaves A, B, and C to species Xus bus, Yus cus, and Zus dus, respectively) • Assign a number to each species (for example, give ‘1’ to Xus bus, ‘2’ to Yus cus and ‘3’ to Zus dus) • Record important information about the species including the species name, any synonyms and its habitat, on its leaf (for example, one finds on the leaf A that ‘Pus fus’ is a synonym3 of ‘Xus bus’ and Xus bus lives in Madagascar, etc.) and • Make indices of synonyms and genera for the catalog (in this case the catalog has an index containing ‘Pus fus’, ‘Xus bus’, ‘Yus cus’, and ‘Zus dus’ and another index containing Xus, Yus, and Zus). In other words, using Grey’s method, the museum assigns an individual leaf to each species and make that leaf a “fact sheet” for that species so that one has only 3 A synonym is one of scientific names given to a taxon. The term ‘synonym’ often refers to those names which are not the valid name of the taxon.  50  to look at that leaf to glean important information about that species. Moreover, “the leaves could be separated and bound in any other form” (24) should a different systems of classification be adopted. This was intended to ensure the stability of species names against future possible changes in classification systems and thus render it immune to the introduction of a classification philosophy other than the Linnaean by reformists. This was effective because while they could arrange the leaves according to any new system of classification, the names themselves remain fixed. For example, suppose that Xus bus and Yus cus are originally classified into the same order and Zus dus into the other order. The catalog would arrange the first two species into one group and the third into another group. Now, imagine that a new classificational system is introduced according to which Xus bus and Zus dus, not Yus cus, constitute one order. A curator could easily deal with the change by rearranging those leaves of the species in the new system. Furthermore, by severing the process of describing a species and making a leaf for it from that of creating a classification system made up of these species, researchers in diverse fields of biology could refer to the same thing (species) regardless of the way in which those species were classified at higher levels. One could know where Xus bus lives by looking at the leaf A, regardless of where it is placed on a given classification system. Because it provided a vehicle for public reference and communication it gave an advantage to the defenders of the Linnaean system over those who advocated introducing new systems; they did not pay as much attention to communication with the public or cooperation with international community of natural history as they did to making the classification represent the natural order. One can easily see the commonalities between Strickland’s and Grey’s approaches to dealing with the conflict between reformist and conservative naturalists. For example, both of them • Invented a way of dealing with species which ensured the possibility of communication. • Gave no definition of species. • Assumed that it was enough to tie a name to a thing for the purposes of communication and research.  51  • Focused on individual species as opposed to an abstract, general definition of species. McOuat’s emphasis is on the similarities between them. He says that “Gray’s glowing success was Strickland’s goal writ institutionally” (McOuat 2001, p. 27).  2.2.2  Darwin’s Strategy for ‘Species’  We have seen how Strickland and Grey dealt with the conflict between reformist and conservative naturalists without seeking the “right” definition of species or the “right” classification philosophy. According to John Beatty (1982, 1986), Charles Darwin also dealt the problem of communication with his fellow naturalists on the concept of species. Furthermore, Darwin offered a similar solution to the ones Strickland and Grey adopted. He did not give any definition of ‘species’ and did not delve into the controversy of what a species really is while appreciating which taxa competent naturalists would recognize as a species. Beatty begins both papers by noting the fact that when Darwin proposed his theory of divergent evolution, he faced a problem of communication with respect to the concept of species. However, his “communication problem” is not the same as the one Strickland and Grey had. Part of the problem which Strickland and Grey had to solve in order for naturalists to communicate was to ensure that all naturalists could refer to the same taxon by a single scientific name regardless of whether they subscribed to the same philosophy of classification or changed their positions over time. Darwin had a different task. He had to make his transmutation theory of species intelligible in spite of the fact that, on the most common definitions at the time species were immutable. In fact, many naturalists accepted this as something close to a conceptual truth: If a species were real, then it would be constant and could not change into a different species by definition. Beatty emphasizes this by saying that ‘species’ was a theory-laden concept. One could not use the concept of species without committing to the immutability thesis of species. Charles Lyell’s remark in Principles of Geology exemplifies this understanding. ... the majority of naturalists agree with Linnaeus in supposing that all the individuals propagated from one stock have certain distinguishing characters in common, which will never vary, and which have re52  mained the same since the creation of each species. (Lyell 1837, p. 407, quoted in Beatty 1982) Against this background belief, “evolving species” is simply a conceptual contradiction, and so Darwin’s theory of evolution by natural selection was unintelligible to naturalists, to whom he addressed the Origin (Beatty 1986, p. 266). An obvious solution was to give another definition of species which did not imply the immutability of a species. If he could use such a definition, then his fellow naturalists could find the theory of evolution intelligible, even though they might not agree with him. However, this was not a viable option to Darwin, because he accepted as a fact to be explained that there had been no universally accepted definition of species in the naturalists’ community, or that nobody could give a satisfactory definition of species. Darwin believed that his theory of divergent evolution could make sense of this indefinability of species (see also our discussion on the “vagueness account” in Chapter 1). His theory asserts that some varieties are actually incipient species that eventually turn into a new species after a long period of transition. If a variety must go through this process in order to become a species and different taxa are at different stages of this transition at a given timeslice, we should expect to see a gradation in nature from a single “distinct” variety to a borderline case (a taxon which naturalists can hardly agree to be a variety or a species with confidence) to multiple “distinct” species. Thus for Darwin, it is no wonder we have difficulty in finding out the “right” definition of species, because nature does not provide us with any resource to allow us to give a clear-cut definition of it. According to Beatty, Darwin’s solution to this “communication problem” was to adopt what is called the “referential use” of the term ‘species.’ Instead of accepting the conventional definition of species or putting forward a new definition of his own, Darwin chose to talk about what competent naturalists call species, without accepting the implication of immutability attached to the conventional definition. Beatty cites the following formulation of the problem of evolution by Darwin in his manuscript Natural Selection: we have to discuss in this work whether forms called by all naturalists distinct species are not lineal descendants of other forms. (Darwin 1987, p. 97, quoted in Beatty 1986, Beatty’s italics) 53  By talking about what naturalists call “species,” and not species per se, Darwin succeeded in speaking of “species” but making his theory of evolution intelligible to other naturalists at the same time. Darwin’s presentation of his thesis is now that those groups that are called “species” evolve. This does not involve any contradiction, because we can use “what naturalists call species” to identify the referent of the term, without accepting the characterization implied by the definition of species as immutable. Notice that if Beatty is right in his interpretation of Darwin’s usage of ‘species,’ then Darwin’s strategy assumed that there had been general agreement as to whether a taxon was to be judged to be a species at a single time slice ——possibly in a given flora or fauna—— between himself and other competent naturalists as well as among those same competent naturalists themselves. Otherwise the referent of the “so-called species” would radically differ from one naturalists to another (including Darwin), and the phrase “what is called species by naturalists” would not help naturalists communicate with each other whether or not Darwin’s theory of evolution were discussed. This is shown in Darwin’s remark on Agassiz’s critique. In response to Agassiz, who wrote in his review of the Origin that if a species is not real, as Darwin claims, then it cannot vary, he wrote I am surprised that Agassiz did not succeed in writing something better. How absurd that logical quibble —— “if species do not exist, how can they vary? As if any one doubted their temporary existence. [Darwin to Asa Gray, August 11, 1860] (Darwin 1887, vol.2, p. 333) Darwin does not doubt that a species exists at a single time slice. This is not surprising if Beatty’s reading is right, because the problem of communication occurs to Darwin because there is a fundamental disagreement about a diachronic property of a species (its immutability), not about any synchronic property. In contrast, disagreement over synchronic properties of “species,” if ever, was not so large that it brought about any intelligibility problem for Darwin. Thus Darwin found his solution to the problem of communication without offering a new definition of species, just as Strickland and Grey did. For Darwin, the problem of communication was the intelligibility of his theory. The referential use of the term ‘species’ rendered his theory intelligible without recourse to 54  any explicit definition of species. Another noteworthy commonality among Strickland, Grey, and Darwin is that they all tried to work on individual taxa generally recognized as species by competent naturalists.4 In other words, a key to ensuring communication, for all of them, was a general consensus about the extensions of ‘species’ among naturalists. Strickland and Grey’s focus was to tie a scientific name to a single species taxon (or more precisely speaking, a taxon generally recognized to be a species) and keep the linkage stable. Darwin’s attention was focused on what is called a species: what is commonly recognized as a species by naturalists. Strickland and Grey tied a label to such a species taxon and made this connection a commodity for the taxonomic community, and Darwin relied on this commodity to make his theory intelligible to other naturalists.5  2.3  Guy Bush and Sympatric Speciation  In the preceding sections I discussed the “taxonomic” side of the concept of species: how biologists or taxonomists talk about it in taxonomic journals, and how to make nomenclature or the reference of the scientific names of species taxa stable across changes in taxonomic philosophy. In this section I will discuss the study of speciation: the study of the process by which species are produced, rather than that of classifying and naming species. I will focus in particular on two conflicting views about whether priority should be given to the definition of species or to the study of speciation. My objective is to understand how it is possible that the two camps 4 In this sense, some may take Strickland and Grey as partially adopting the referential use of the term species. 5 It is possible to offer a different, but not incompatible interpretation of Darwin’s strategy based on Beatty’s description. By speaking about what naturalists called species and not species per se, Darwin was able to jettison unwarranted assumptions about species and speak of his theory without lapsing into unintelligibility. But this strategy may have had yet another effect of leaving the “vague” idea of species alone —— talking about species without delving into the endless controversy on the nature of species. This would have been clear if Darwin had emphasize the point of his strategy by adding “whatever the nature of species is” or “whatever the right definition of species is.” If one remembers some characteristics of the unarticulated concept of species ——that biologists generally believe that they understand species and that they share this understanding of species and yet they are drawn into an endless controversy when they try to make the details of their understanding or the nature of species explicit (see Chapter 3)—— the referential use of ‘species’ provides a detour which allows them to discuss species without falling into this futile controversy. This relative indifference to the “true nature” or the right definition of species also characterizes his strategy on the notion of species.  55  of biologists engaged in this debate can communicate with each other without experiencing any serious incomensurability, while genuinely and fundamentally disagreeing about the issue. The question at issue is this: is it desirable, or even possible to arrive at a (preferably “correct”) definition of species before studying speciation or after — — or does it really matter? Biologists in one camp such as Ernst Mayr, Jerry Coyne, and Allen Orr argue that we should have a definition of species before we begin to study speciation, or at least that deciding on a definition of species fosters our study of speciation. While biologists in the other camp, including Guy L. Bush argue that we should study speciation first in order to arrive at the right definition of species, or, arriving at a decision about which definition of species to adopt is not a necessary condition for commencing with research on speciation. For instance, Bush says: It is widely assumed by most evolutionary biologists that this concept [‘species’] is so fundamental that it must be established before the process of speciation can be investigated, let alone understood or discussed.... This perspective strikes me as wrong minded, as quite the opposite is true. It is unlikely that species can be defined unless we first understand how populations actually diverge and establish separate breeding systems in nature. (Bush 1993, p. 242) Trying to explain speciation within the context of a preconceived species concept places the cart before the horse. It is an understanding of the factors that result in the reduction and eventual elimination of gene flow between sister populations ——the very process of speciation itself—— that is necessary before a clear species definition is possible. (Bush 1994, p. 286) Thus, such biologists opt not to pick any particular species concept to guide their study of speciation. Although the people in this camp are not always unanimous in their positions ——some of them suggest they are sympathetic to a particular species concept, while others are explicity pluralistic about the concept of species or agnostic with respect to the status of a particular population—— they all agree that choosing a species concept is not necessary to the study of speciation. One may suspect that this lack of agreement regarding whether the priority 56  goes to defining species or studying speciation would hinder communication between these camps. However, this is not the case. As will be seen in section 2.3.4, the two camps just mentioned also happen to disagree on the frequency with which sympatric speciation6 occurs in the biological world and whether this mode of speciation actually occurs in populations called host races. Host races are defined, at the moment, as genetically differentiated, sympatric populations of parasites that use different hosts, and between which there is appreciable gene flow (Dr`es & Mallet 2002). Bush and others believe that sympatric speciation occurs in nature more frequently than the conventional wisdom, which Coyne & Orr and Mayr subscribe to, suggests and think that some host races are in the process of sympatric diversification. However, when we take a close look at one case of host races, the dispute does not arise from a lack of agreement regarding the priority of the speciation study but hinges on empirical points —— whether or not there is sufficient evidence to rule out the possibility of explaining the properties of host races in terms of allopatric diversification. In other words, disagreement about the priority of the speciation study does not hinder communication between biologists of the two camps.  2.3.1  Mayr on the Priority of Defining Species and Studying Speciation  Several biologists have discussed whether priority should be given to defining species or studying speciation. One such biologist is Ernst Mayr (1942, 1957, 1970). Mayr took a strong stance on this issue. He argued that ‘species’ should be defined before starting research on speciation. For example, in Systematics and the Origin of Species he says, A concise definition of species is, for [a student of speciation], a necessity, because his interpretation of the speciation process depends largely on what he considers to be the final stage of the process, the species. (Mayr 1942, p. 114) In 1957, he suggested this position when he commented on Darwin’s failure to solve the problem of multiplication of species (i.e., speciation) in the Origin: 6 Sympatric speciation is the diversification of one species into multiple species (usually two) in a single geographic location.  57  Having thus eliminated the species as a concrete unit of nature, Darwin had also neatly eliminated the problem of the multiplication of species. This explains why he made no effort in his classical work to solve the problem of speciation. (Mayr 1957, p. 4) Here Mayr explains that the title of Darwin’s book, On the Origin of Species, does not accurately reflect the contents thereof by citing the fact that Darwin emphasized the subjectivity of species in it. In Populations, Species, and Evolution, Mayr seems to provide a synthesis of his two earlier arguments, which I have just discussed: “Descent with modification,” true biological evolution, could be proved only by demonstrating that one species could originate from another.... Darwin failed to solve the problem indicated by the title of his work. ... he never seriously attempted a rigorous analysis of the problem of the multiplication of species, of the splitting of one species into two. I have examined the reasons for this failure ... and found that foremost among them was Darwin’s uncertainty about the nature of species. ... An understanding of the nature of species, then, is an indispensable prerequisite for the understanding of the evolutionary process. (Mayr 1970, p. 10) Mayr’s Two Arguments In these quotations, Mayr seems to make two arguments for his position. One may be called the logical argument and the other may be called the historical argument. The logical argument is that since speciation is, after all, the process of producing new species, we will not be able to understand its nature unless we understand the nature of species.7 The historical argument illustrates this point with the historical case of Darwin, in which, according to Mayr, Darwin could not solve the question of speciation because he saw ‘species’ as an arbitrary concept. While the main purpose of this section is not the critical analysis of each view on whether priority should be given to defining species or to studying speciation, nonetheless it would be useful to point out some possible problems with Mayr’s 7A  similar argument to this is given with regard to language evolution. For example, Robert Berwick, professor of computational linguistics at MIT, suggests that the definition of language is necessary for the evolutionary study of its origin by saying, “If you can’t define what it [language] is, why study it from an evolutionary point of view?” (MIT 2008)  58  “logical” argument. His logical argument states that since speciation is the process of species formation, the study of speciation requires a definition of species so that researchers can know what they are studying (since they cannot study something if they do not know what it is). However, I believe that there is reason to believe that this requirement is too strict. There are two possible interpretations of Mayr’s logical argument. First, in the 1970 book Mayr said that we need to understand “the nature of species” before embarking upon the study of speciation. This suggests that we need the “right” definition (not just a definition) of a species before commencing the study of speciation, because only the “right” definition could tell us about the nature of what is defined. However, in 1957 Mayr also said that “a concise definition” is necessary. This seems to imply that what we need is “a definition” of species. If Mayr meant that we need the right definition of species before the speciation study, his argument should be formulated as follows. (1) Speciation is the process of species formation. (2) For all X, we cannot begin to study the process of forming X until we have the right definition of ‘X.’ (C) Therefore, we cannot begin to study speciation until we have the right definition of ‘species.’ In this argument the second premise (2) is clearly false. One can begin to study the process of forming X without the right definition of ‘X.’ For example, scientists disagree on the “right” definition of ‘life’ (Hey et al. 2003), but this does not prevent those scientists from studying life. The second interpretation revises this point. According to this interpretation, what is necessary for the study of speciation is a definition, not the right one. (1) Speciation is the process of species formation. (2) For all X, we cannot begin to study the process of forming X until we have a definition of ‘X.’ (C) Therefore, we cannot begin to study speciation until we have a definition of ‘species.’  59  I believe that this is not a sound argument if a ‘definition’ is intended to distinguish all X’s from all non-X’s. If we have good cases in which the formation process of X did occur, we can begin to research the process by investigating that case, even if we do not have any definition of X. This is clear in cases where a discovery involves serendipity. Suppose the legend about discovery of cheese is true: a merchant in the Middle East found that milk stored in a container made from an animal’s stomach had turned into cheese. The merchant was suprised to find cheese in his container and tried to find out how milk turned into cheese. However, when he was asked of what cheese is, he probably could not give a definition of it. But he could begin to study how cheese is formed, however crude his “research” was. Of course, if we had absolutely no concept of X, we would not be able to study X’s formation, because we would not know where to start our research. It follows that one needs some characterization of X before starting a study of X. However, characterization is not identical with a definition. We can characterize something without fully defining it. For example, water is a chemical substance; this is a characterization of water, but not a definition of it. In addition, as we shall see shortly, Bush’s study can be seen as a good counterexample to Mayr’s argument; he offers no definition of species, but no serious biologist would doubt that his study of, for example, Rhagoletis pomonella is a study of speciation. Of course, Bush assumes that species have some characteristic properties (he says that the reduction of gene flow to zero occurs when speciation completes, for example). However, this is not a definition of species for him.  2.3.2  Coyne and Orr on the Priority of Defining Species and Studying Speciation  With regard to the priority issue, Jerry Coyne and Allen Orr are in the same camp as Mayr, but their position is significantly more modest. Instead of claiming that an appropriate definition of species is required for the study of speciation, they argue that defining species fosters the study of speciation. Although they do not explicitly state their reasons for adopting this more modest position (Coyne & Orr 2004), their argument goes as follows. They admit that the species problem has an element of interest-relativity; there are, as a matter of 60  fact, many aspects from which one can define ‘species’ and work on the species problem, and one’s interest ——what one wants to do with the notion of species and what one wants to understand—— affects which aspect one focuses on. For example, “Systematicists, whose task is unraveling the history of life, often prefer species concepts different from those used by evolutionists more interested in evolutionary processes” (p. 10). Phylogeneticists, who study phylogenetic relationships among taxa would be motivated to define ‘species’ in terms of its phylogenetic relation as in a version of phylogenetic species concept, which defines a species as a least inclusive monophyletic taxon in the tree of life (see section 1.1.2 for details), whereas the students of speciation do not have such a motivation and thus prefer to a definition of species in terms of the process by which distinctive species maintain its identity (for example, the biological species concept). Given this interest-relativity in species studies, the authors believe, biologists should be aware of their objective in the research and should select (or invent) a species concept which is in accord with their research objective (p. 26). This choice of a species concept in turn forms their research program, because a selected definition partly determines which characteristics of organisms they should pay attention to in the study of the origin of species. “[D]eriving a species concept is important because it frames one’s entire research program on the origin of species” (p. 10). Let us take a quick look at why Coyne and Orr select the biological species concept as their concept and how this selection affects their research program on speciation. Their primary interest in the species problem is to understand the origin of discontinuity found in nature. The authors point out that conspecific organisms are largely distinctive in their phenotype and genetics, and they view this as a fundamental biological fact. Their next task is to see which definition of species is the best for understanding the origin of discontinuity (p. 26). The authors conclude that the biological species concept is the best, because reproductive isolation and thereby the significant reduction of gene flow between incipient species is necessary to allow their divergent evolution (p. 31). This choice led them to concentrate exclusively on the establishment of reproductive isolation in the subsequent studies; their book is rightly summarized as a study of how populations acquire reproductive isolation.8 In their words, 8A  review of this book represents this in its title: “On the origin of reproductive isola-  61  If one accepts some version of the biological species concept, then the central problem of speciation is understanding the origin of those isolating barriers that actually or potentially prevent gene flow in sympatry. (p. 57). Other factors are mentioned only if they can shed a light on the establishment of reproductive isolation. One can see that although Coyne and Orr are largely in the same camp as Mayr, their position is more modest than Mayr in several ways and their argument is different from his. They do not say that defining ‘species’ is necessary for the speciation study; defining it only facilitates the study of its origin. They are more ready to admit the plurality of possible approaches toward the study of species and speciation than Mayr was. Their argument places the role of defining a species as an object of their study within the research program as a whole. A definition is characterized as a “working” definition in their study, as it were, not something distinguishing every example of species from that of nonspecies once and for all. However, this modest view would be still unsatisfactory for Bush. So let us go on to see Bush’s position on the priority issue.  2.3.3  Bush on the Priority of Defining Species and Studying Speciation  Guy L. Bush has developed his idea on the priority issue and the species problem in general in a series of papers (Bush 1993, 1994, 1995, Bush & Smith 1998, Bush & Butlin 2004). In these papers, he objects to the biological species concept and also denies the Mayrian idea that the definition of species should be given prior to the study of speciation. Before 1993, however, he did not go against the Mayrian idea on the priority issue and accepted the biological species concept. For example, in his 1969 paper (Bush 1969), in which he studied sympatric formation of host races in Rhagoletis pomonella, Bush adopted the biological species concept as a research tool citing Mayr’s book (Mayr 1963). Host races represent a continuum between forms which freely interbreed to those that rarely exchange genes. The latter may approach tion.” (Schilthuizen 2005)  62  the status of a species generally regarded as an interbreeding population reproductively isolated from all other such populations (Mayr, 1963.) (Bush 1969, p. 237, n.1) His attitude toward Mayr changed around 1993. Since then, Bush has raised a number of objections to the biological species concept and argued that priority should be given to the study of speciation, rather than to the definition of species. Since his argument that the prior selection of a species concept may hinder subsequent research, and his argument against the biological species concept are interdependent, I will discuss his general objection to the biological species concept, before discussing his objection to the prior selection of a species concept. In his 1993 paper, Bush offers an extensive argument on priority, particularly with respect to the biological species concept. He first points out that there are populations which are not recognized as species taxa according to the biological species concept, but are recognized as a “good species” by taxonomists (The notion of ‘good species’ will be discussed shortly). “[T]here is mounting evidence that many closely related taxa, long accepted as “good species”, may exchange genes at a low level” (242). He repeats this point in later papers (Bush 1995, Bush & Smith 1998). Bush also notes that speciation proceeds gradually with few exceptions and that many causal factors can affect whether two populations will eventually merge or diverge —— their genetic structure, their environment, the pattern and degree of geographical and ecological isolation, to name a few. As a result, no one can “pinpoint the precise time, place, or circumstance when two or more sister populations pass through this diffuse yet critical threshold of genome change and become irrevocably committed to different evolutionary paths” (243). Furthermore, Bush asserts that different speciation events have different “thresholds” and conditions. For example, populations in different speciation events reach the “threshold” of becoming distinct species with different level of gene flow. Therefore, Bush supposes, the supporters of the biological species concept will require that we focus on the end result of the speciation process, and not the process itself (or the cause of it). This is because complete genetic isolation occurs only at the end of the speciation process. This means that the biological species concept does not help us to understand the process of speciation itself or its causes. Because the BSC [biological species concept] focuses on a level of 63  divergence between populations that has nothing to do with the actual phase in which populations become committed to one lineage or another, it is not a particularly useful concept to apply in speciation studies. (243) Thus, Bush concludes that the biological species concept does not help us understand what occurs when a population diverges into two distinct species.9 Bush’s Point on Priority Issue  Let us turn to Bush’s point on the priority issue.  His starting point is close to Coyne & Orr’s: A researcher’s prior commitment to a species concept affects the way that he studies speciation. In particular, since different species concepts appeal to different aspects of the biological world, selecting a single species concept leads us to focus a particular aspect of it. Bush expresses this by saying “how one defines species directly influences one’s views of how they arise in nature” (242). We have just seen an example in which he pointed out that the biological species concept only focuses on the end result of the speciation process, and not the process itself. However, Bush parts way with Coyne & Orr in the next step of the argument. From the first premise, Bush infers that there is no universal species concept, which 9 I do not believe that all of his objections to the biological species concept just described are right. For example, I do not think that reproductive isolation is a merely end result of the speciation process. On Bush’s view, the speciation process comes to an end when reproductive isolation is established. However, as Robert O’Hara (1993) points out, one cannot recognize that a particular speciation event ends at t synchronically (i.e., if the observer is at t). This is because if two lineages merged just after t (say t ), then the divergence at t would not be recognized as a genuine speciation. In other words, even if a single lineage does in fact diverge into two distinct lineages at t, these lineages may maintain the option of merging into a single lineage again in the near future. We would not count the divergence at t as a genuine instance of speciation until this possibility disappears. One may defend Bush in the following way: Since Bush believed that two lineages cannot merge again once reproductive isolation is established between them when he adopted the biological species concept (Bush 1975), he may still hold this view even though he no longer endorses the biological species concept itself. However, different isolation barriers serve different roles in preventing isolated populations from merging together (Coyne & Orr 2004), and some barriers are not effective forever. For example, when ecological factors serve a reproductive barrier, the reproductive isolation provided by this barrier is no longer effective when the ecological condition no longer obtains. In the case of species of cichlid in an African lake (Seehausen et al. 1997), individuals could recognize their conspecifics (their potential mates) by their color. However, it is reported that since transparency is lost in the lake because of environmental problems, they cannot tell their conspecifics from others, and the reproctive isolation has been collapsing. Therefore, the establishment of reproductive isolation is not merely the end result of the speciation process. It is also a cause of speciation in that it maintains two lineages apart for a sufficiently long time.  64  would cover every aspect of all organisms. “[I]t is doubtful whether a single, unequivocal definition of the species category can satisfy all the biological and philosophical criteria deemed essential by authorities of different taxa” (Bush 1994, p. 286). This point can also be derived from some of his observations about the speciation process. Bush believes that different speciation events have different “thresholds” and conditions. If so, then no species concept focusing on a single causal factor could cover all the speciation processes. A species concept citing a causal factor X would not work when the condition for the “threshold” in a given speciation process pertains exclusively to another causal factor Y . Therefore, if one adopts a single species concept, there will still be speciation events in which something remains to be understood. For instance, the biological species concept does not make sense of the fact that some “good” species ——those taxa which are not reproductively isolated yet occupy distinctive ecological niches and differentiate morphologically, for example—— are actually species, and so cannot explain their speciation process. Then Bush concludes, “Strict adherence to any one particular species concept may actually impede our understanding of the speciation process” (Bush 1995, p. 58).  2.3.4  Coyne & Orr and Bush on Rhagoletis pomonella  We have seen Coyne & Orr and Bush’s arguments regarding the priority issue in the previous section. From what we have seen, it is clear that their disagrement lies primarily in what they see as the significance of the fact that selecting a species concept restricts a researcher’s perspective regarding their study of speciation. They agree that selecting a particular species concept leads a scientist to focus on one aspect of the living world rather than others. For example, adopting the biological species concept helps us explain or understand a particular biological fact —— whether it is sympatric coexistence of morphologically distinct populations (Coyne & Orr) or what happens at the end of the speciation process (Bush). They differ in how they evaluate the relevance of one’s research program to the whole project of the speciation study once it has been restricted by the selection of a species concept. For Coyne & Orr, the scope of the research program is set once a researcher becomes aware of his research objective (what he wants to understand concerning  65  the phenomenon of speciation) so the adoption of a species concept is supposed to facilitate the overall study of speciation. On the other hand, for Bush, selecting a species concept will unnecessarily restrict the scope of the research project and be even harmful to the project to study speciation as a whole. There will always be something in the phenomena of speciation wrongly excluded by such a perspective. The reason why I have explored the disagreement on the priority issue is that Coyne & Orr and Bush disagree on another issue——the plausibility of sympatric speciation. The main question I would like to ask is whether the disagreement on the former issue affects that on the latter issue, in particular whether or not the disagreement on the priority issue makes each other’s claim on the plausibility of sympatric speciation sound unintelligible to others. When Coyne & Orr and Bush discuss the second issue, they have focused on whether host races of Rhagoletis pomonella, a fly found in North America, are undergoing the process of sympatric speciation. I will address this debate in this section. R. pomonella’s Host Races  The two host races occur on hawthorn and apple  trees respectively. They currently live sympatrically in the northeastern and midwestern United States where hawthorn and apple trees coexist, while the hawthorn race also occurs in apple-free regions in the southern United States. Moreover, an isolated group of the hawthorn race inhabits the hawthorns of central Mexico. In the mid 19th century, R. pomonella was found breeding on introduced apple trees in the Hudson Valley. Over the next 50 years, the apple race spread throughout the eastern United States, becoming a serious agricultural pest. This spread and the genetic evidence for monophyly of the apple race implies that it arose only once, probably in the last 150 years. Coyne & Orr (2004, p. 159) concisely summarized the life histories of the two races: The life histories of the races are similar. Both have one generation per year. Adult flies congregate for mating on the plants, and females oviposit on ripening fruit. Larvae complete their development in the ripe fallen fruits, pupate in the soil, and undergo a facultative diapause until spring. Adults spend one to two weeks moving about and feeding on diverse food before congregating again on the host plant for mating. 66  Sympatric Speciation?  Because flies mate on or very near the host fruit, host  and mate choice is tightly linked. This is frequently observed in the mating behaviors of many parasites and other habitat specific organisms (Bush 1992). Coyne & Orr (2004) cite Feder (1998) to convey a hypothesis supported by Bush and others: Because apples and hawthorns are sympatric, Feder (1998) suggests that individuals with a genetic preference for apples gained a selective advantage, perhaps involving the use of an empty niche or escape from parasitism. This change in preference, coupled with a tendency to mate on the host plant, could have promoted the evolution of host races. (p. 160) What might have happened to the flies, according to Bush and others, is this: Suppose some flies of the hawthorn race moved to apple trees by accident, which could have occurred at any time because hawthorn and apple trees are sympatric. If the life on apple trees happens to be adaptively advantageous, natural selection would be in favor of those with a preference for such a habitat. And if those flies tended to mate more frequently with those on the apple trees than on the hawthorn trees, then the gene flow between the hawthorn and the apple populations would have been significantly reduced. If this scenario is true, then the apple race is evolving independently and can be expected to diverge further from the hawthorn race. Possible Non-sympatric Origins  Coyne & Orr agree with Bush that the two  races live sympatrically and that the origin of the apple race is evolutionarily recent. However, Coyne and Orr contend that there are several possible scenarios other than the sympatric origin that Bush does not consider: the two races may not be sister groups10 , or, even if they are sister groups, they may not have been formed in sympatry. Coyne & Orr cite three possibilities. I address two of them here. The first possibility concerns the genetic similarity between the hawthorn and the apple races. No race or species in Rhagoletis is genetically more similar to the apple race than the hawthorn race, but there are some other species ——besides R. pomonella—— which are more similar to some populations of the hawthorn 10 Sister groups are groups where each is the nearest relative of the other (Lincoln et al. 1998). They have a single most recent common ancestor which neither of them shares with any other group.  67  race than the apple flies. Moreover, there are a couple of species that are sufficiently genetically similar to the apple race to be possible ancestors not shared by the hawthorn race. Given that there are factors that could be disturbing the phylogenetic analysis based on genetic data ——for example, allozymes used to infer the phylogenetic tree vary geographically—— the phylogenetic analysis might not accurately reflect the phylogeny among Rhagoletis populations, and the genetic similarity between the two races might come from hybridization between them, not their common ancestor. The second possibility that Coyne & Orr consider is that the apple race of R. pomonella may not have formed sympatrically, but rather descended from another hawthorn race living in a different area. In other words, the apple race may have arised as a result of colonization by some population of the hawthorn race living in a distant habitat. Coyne & Orr (2004) cite Carson (1989) as providing some support for this hypothesis. Carson points out that the hawthorn genus (hawthorn tree, a host of the fly) is a very diversified one (for example, there are 40 recognized species in the Michigan-Illinois region alone) and that the hawthorn flies have accordingly frequently speciated through colonization. He also points out that there are a couple of races of R. pomonella which could have been preadapted to the introduced apple trees. Thus, Carson suggests that some of those hawthorn flies had a reasonable possibility of colonizing the apple trees from their habitats. If this is true, the apple race did not come from the hawthorn race which live in the same geographic area but descended from another hawthorn race travelling from some different area. Bush and others reply to the second point (Bush et al. 1989). One of their responses is that R. pomonella has not speciated through colonization as frequently as Carson (and Coyne & Orr) suggests. Carson’s point is that since R. pomonella is adapted to diverse species of the hawthorn tree, the fly must have often shifted their host by colonization, not sympatric speciation. The argument by Bush’s camp is that the taxonomic status of many “species” in the hawthorn genus is dubious and the R. pomonella hosts only a restricted number of those hawthorn “species.” This means that diversity found in the classification of the hawthorn tree may be artificial rather than real and that this fly may not have experienced as many colonizations as Carson expected. Another consideration is that there is not the amount of genetic 68  divergence in R. pomonella which could be observed if Carson’s scenario were right. The authors note that electrophoretic analysis of different hawthorn-hosted populations gives no evidence of genetic differentiation; this supports their claim that there is no significant differentiation between the hawthorn-hosted popultions. Still, Coyne & Orr reply that there is no systematic research for these races and that there is evidence for “substantial genetic and temporal differentiation between neighboring R. pomonella population that live on different hawthorns” (Coyne & Orr 2004, p. 161).  2.3.5  Why Does Their Disagreement on the Priority Issue Not Affect the R. pomonella Case?  So much for the debate on sympatric speciation of R. pomonella. Our purpose is not to take sides, but to see where disagreement between the two camps lies and whether this disagreement is relevant to another disagreement with respect to the priority issue. Now it is clear that their disagreement about R. pomonella is by and large empirical, not conceptual. Their disagreement is about whether or not there is sufficient evidence to rule out the possibility that the apple race of R. pomonella did not arise sympatrically from the hawthorn race. Those in Coyne & Orr’s camp offer hypotheses for the possible origin of the sympatric coexistence of the apple and hawthorn races, while Bush and others give evidence against them. Though the debate may not end any time soon, this is due to the lack of empirical evidence to show that the genetic similarity between apple and hawthorn races does not come from hybridization between the two races. Thus there is little incommensurability in this debate when it comes to the definition of species. Disagreements on Definition and Priority: Irrelevant Disagreement between the two camps on the right definition of species makes little difference to this debate. Although Coyne & Orr endorse the biological species concept and Bush has no species definition to support, there is no evidence that this prevents them from understanding each other. Although Coyne (1994) objects to Bush by noting that the apple race is not a full species even if it did arise sympatrically, this is not at the centre of their attempt to downplay the possibility of sympatric speciation in  69  R. pomonella. Moreover, Bush’s camp does not seem to care much about the fact that the apple race is not officially recognized as a species among taxonomists; he is indifferent to whether or not the race of interest is indeed a species. Furthermore, their disagreement about whether priority should be given to studying speciation or defining a species is not relevant to the disagreement about the host races of R. pomonella. As far as I know, neither camp ever discusses the priority issue in the debate. Sharing a Reference  How do their disagreements regarding the priority issue  and the right species concept not cause them to fall into incommensurability? Recall where the disagreement occurs between Coyne & Orr’s and Bush’s camps: both camps agree that the selection of a species concept makes a difference to the way in which the study of speciation proceeds. However, Coyne & Orr see this positively because it allows researchers to focus on a particular aspect of the speciation process, which will help their study. Bush, on the other hand, sees it as an obstacle to the pursuit of a universal theory of speciation. Adopting a species concept may motivate biologists locally, but hinder their study globally. This is because the way in which the process of speciation varies from one species to another, as does the threshold for becoming a species. The focus of research is restricted by the selection of a species concept. It seems that selecting a species concept before studying a possible case of speciation helps us restrict the scope of our research in two ways. First, selecting a species concept helps us focus on some population(s) but not others. If a species is defined by the property G and a researcher accepts this definition, then she does not have to pay attention to cases where a population of interest does not have or is not expected to have the property G. Secondly, selecting a species concept helps us focus on some aspects of the population(s) or the speciation process but not others. For example, if a biologist accepts a definition of ‘species’ in terms of the property G, she would examine how a given population acquires the property G when she studies a particular case of speciation, but not how the population will acquire another property F if having F is not related to having G. Therefore, if the general disagreement on the priority issue is about the influence that the selection of a species concept has on the way in which a biologist studies the speciation process, 70  then both camps possibly disagree in these two regards. When we consider the R. pomonella case, Coyne & Orr and Bush both assume that the apple race could be a case of speciation. Coyne may contend that the apple race is still not a full species, but this does not discourage him from discussing the R. pomonella case. Bush may not believe that the speciation process is now complete (because there may be gene flow between the two races), this also does not stop him. Both camps in the debate may restrict their focus in the first way — —selecting a particular species definition helps them focus on some population but not others—— but this does not make them unintelligible, because both of them discuss the same apple host race. The first way of restricting the focus does not prevent both camps from participating in the debate meaningfully. Reproductive Isolation and Speciation This leaves open whether or not the second way of restricting the focus in the study of speciation ——selecting one species definition helps them focus on some aspects of the population or the speciation process but not others—— keep their disagreement empirical. However, the second way of restriction does not prevent both camps from engaging in the debate intelligibly. Coyne & Orr see this debate as meaningful only if the apple race may eventually establish reproductive isolation. Bush would not take this condition seriously; the apple race may be a full species even if it does not establish full isolation from other races. However, in this case the difference between Coyne & Orr and Bush lies only in emphasis. True, reproductive isolation is a crucial definition for speciation study for Coyne & Orr, but not for Bush. Nevertheless the establishment of reproductive isolation is important for Bush too, because this is still a part of speciation process. For example, when Bush describes the process of sympatric speciation (Bush & Butlin 2004), he places the completion of reproductive isolation at the end of the speciation process. Furthermore, the process of forming distinct species through natural selection could also be important for both Coyne & Orr and Bush (Coyne & Orr (2004) admit the theoretical possibility of sympatric speciation through natural selection); the disagreement only lies in the reasons why selection could be important for their study —— it is because it either leads to the establishment of reproductive isolation (Coyne & Orr), or because it contributes to the diversification of lineages (but not necessarily reproductive isolation; Bush). 71  In other words, both agree that selection leads to reproductive isolation. The only difference is whether or not reproductive isolation is conceptually relevant to being a species. Therefore the definitional connection between reproductive isolation and specieshood (or the lack thereof) is largely irrelevant when considering the R. pomonella case. This is why the disagreement over the priority issue does not prevent both camps from meaningfully discussing whether or not the apple race of R. pomonella is in the process of sympatric speciation. Lessons From the R. pomonella Case Our analysis shows how the disagreement about the role of a definition of species in the study of speciation does not necessarily lead to miscommunication or incommensurability between the parties involved when they are engaged in the study of a particular case of speciation. An important factor for this is that Coyne & Orr and Bush all talk about the same population (the apple race of R. pomonella) to a reasonable degree and believe that this population is at least a possible instance of sympatric speciation ——that is, a new species may come into existence as a result of this process, whatever definition of species one may adopt—— in spite of their disagreement over the right definition of species. This disagreement may affect the taxonomic status of the race, as Coyne suggests, but did not prevent both camps from engaging in the debate. Another factor is that both camps still believe that reproductive isolation and natural selection play significant roles during speciation, even though they disagree on the roles they play in the definition of ‘species.’ Thus although their emphasis on the causal roles of reproductive isolation and natural selection differs, it does not so much that it prevents both camps from being unintelligible from the others’ side. The first insight is in accord with those of Strickland, Grey and Darwin. Strickland and Grey invented new ways (a nomenclauture and an information-storing system) to ensure that the same taxonomic name is tagged to the same taxon so that naturalists could talk about the same taxonomic object, even when they have different ideas of what the nature of species is. Even when a taxon named as Xus bus turns out to be a non-species, its referent is largely stable thanks to the nomenclature or Grey’s storing system. Darwin did not invent either a nomenclature or an information-storing system, but he came up with a way of talking about species while eschewing its immediate connotation, i.e., its immutability. This can be taken 72  to be Darwin’s strategy to talk about the same thing with his fellow naturalists even though there is a grave difference in their conception of ‘species.’ In this sense, all those three people tried to overcome the problem of communication among naturalists regarding the concept of species by inventing ways of sharing a reference with other naturalists. And this is true of Coyne & Orr and Bush. This is partly due to the naming system in zoology ——because of this, they could agree on what Rhagoletic pomonella refers to to a sufficient degree—— and the fact that the referent of “the apple race” can be fixed by some description of the race (for example, description of its geographical or ecological habitat). To summarize, in all cases which we have discussed in this chapter, sharing an extension makes it possible for the parties in disagreement on the theoretical issue of species to communicate meaningfully with each other without falling prey to incommensurability.  2.4  Conclusions  We have seen various biologists (or naturalists) in this chapter who faced—— consciously or not——the problem of communication. What can they teach us about ensuring successful communication between scientists who disagree—— even radically——on the “right” definition of species or the priority between defining species and the study of speciation? At a detailed level, slightly different strategies were employed in each of our successful case studies: Strickland avoided including any detailed definition of species in the “Nomenclature”; Gray invented a new method of cataloging organisms; Darwin made only referential use of ‘species’; Bush and others did not take any specific measure, but talked about the same host race. At a more abstract level, however, those naturalists employed the very same strategy for handling potential problems of communication, namely, sharing a reference and ensuring the same group of organisms is under discussion. Indeed, the “Rules of Zoological Nomenclature” were invented for this very purpose by instituting the rule of tagging a name to a taxon. Gray’s cataloguing method demands that a single leaf be assigned to a distinct taxon whether or not it is seen as a single “species” by any particular criterion. Darwin managed to intelligibly convey his idea of transmutation of species to his contemporaries by making only referential use of ‘species’——discussing those judged as species by  73  competent naturalists. Darwin and his critics could thus talk about the same taxa, regardless of whether they were immutable. Bush and Mayr’s camps still have radically different ideas on the “right” definition of species and the priority between defining species and the study of speciation, but this did not hinder their debate on sympatric speciation of the R. pomonella case, because they implicitly agree on the referent of the group of organisms being addressed. It is this sharing that helps biologists communicate efficiently, regardless of one’s preferred definition of ‘species,’ the classificatory status of a taxon in question, or one’s view on research priorities involving species and speciation.  2.4.1  The Incommensurability Problem and the Communication Breakdown  The Incommensurability Problem This analysis of the case studies confirms a leading account on the way in which possible communication breakdown among the scientific community is avoided. In philosophy of science, the problem of communication breakdown has been discussed in relation to the incommensurability problem. According to those who proposed this problem, such as Thomas Kuhn (1962) and Paul Feyerabend (1975), when a revolutionary change occurs in science (e.g., when the Newtonian mechanics was replaced with the relativistic theory), the two theories may not be able to be directly compared or incommensurable, because some vocabularies employed in the two theories may have different meanings. In a familiar example, ‘mass’ in the Newtonian theory refers to proper mass and relativistic mass but ‘mass’ in Einstein’s theory only refers to the latter kind as the content of the concept of mass changes in the transition from the Newtonian to the Einsteinian theory. Kuhn describes this by saying that ... the physical referents of these Einsteinian concepts [space, time and mass] are by no means identical with those of the Newtonian concepts that bear the same name. (Newtonian mass is conserved; Einsteinian is convertible with energy. Only at low relative velocities may the two be measured in the same way, and even then they must not be conceived to be the same.) (Kuhn 1962, p. 102)  74  Newton and Einstein talk about different things by ‘mass.’ Perhaps Newton talks about the Newtonian mass while Einstein talks about the Einstein mass, but the concepts of the Newtonian and the Einsteinian mass are not identical. Different theories can be compared only if they discuss the same object. Therefore, the Newtonian and Einsteinian theories are incommensurable; they cannot be directly compared. This is how Kuhn & Feyerabend reach their conclusion. A problem with this incommensurability thesis is an assumption on the meaning of a theoretical term (Field 1973, Kitcher 1978, 1993, Newton-Smith 1982, Sankey 1994). Kuhn & Feyerabend’s argument is based on the holistic account of the meaning of a term: the meaning of a theoretical term is determined by the roles it plays in scientific theory. The reason why Kuhn and Feyerabend believe that Newton and Einstein mean something different by the same word ‘mass’ is that that term plays different roles in their theories. The Newtonian mass and the Einsteinian mass are supposed to possess different properties, as the above quote indicates. This means that the concept of mass has different relationships with other concepts in each theory; a theory is a network of concepts and the concept ‘mass’ is located differently in the Newtonian and the Einsteinian theories. Critics have called this assumption into question. They agree with Kuhn & Feyerabend that the Newtonian mass and the Einsteinian mass possess different properties. Nevertheless, critics argue, Newton and Einstain could talk about the same object by the term ‘mass.’ Kitcher’s Account Philip Kitcher (1978, 1993) proposes a different account on the meaning of a theoretical term. His account starts with the distinction between type and token of a term. A token of a term is an individual instance of the term as a type. When I utter “Rose is a rose,” the first and second ‘rose’ are different tokens of the word ‘rose’ but of the same type; they are two instances of the same word. Some tokens of the word ‘rose’ are italicized (like ‘rose’), but this does not mean that the word ‘rose’ as a type is italicized. When Kuhn and Feyerabend discuss a reference of a theoretical concept, they implicitly assume that every token of the same term refers to the same object. In contrast, Kitcher’s idea is that different tokens of a term may have different modes of reference (a medium through which one refers to objects with a word, such as descriptions and baptism) and thereby 75  refer to different objects. If we call the Newtonian mass “massN ,” different tokens of ‘massN ’ may have different references. From the way Newton used one token of the term, for instance, that token of ‘massN ’ refers to an object to which ‘the Einsteinian mass’ (or ‘massE ’) also refers, while other tokens of the two terms do not correspond in their referents. Kitcher calls the compendium of modes of references “reference potential.” Each concept has its reference potential, and when the reference potential changes, that concept also changes. Kitcher’s main example is ‘dephlogistoned air.’ One mode of reference for this term is a description “dephlogistoned air is a kind of air from which phlogiston is removed.” If you understand the term this way, however, this term refers to nothing because there is no phlogiston and thus every sentence containing ‘dephlogistoned air’ is false, which means that Priestley contributed nothing to the history of chemistry as long as he used the term ‘dephlogistoned air.’ This is counterintuitive. According to Kitcher’s account, Priestley was able to talk about oxygen at least in some contexts when he used ‘dephlogistoned air,’ because some tokens of this term have modes of reference which makes them refer to oxygen. Kitcher believes that this account helps us overcome the incommensurability charge. For if theoretical terms in two theories could at least partially have the same reference, then we can compare the theories by seeing what they say about the same objects. Thus we can say that ‘massN ’ sometimes refers to proper mass in some places and relativistic mass in others, and ‘massE ’ only refers to the latter kind of mass. We can also say that Priestley did talk about oxygen by “dephlogistoned air” and compare what Priestley said of oxygen with Lavoisier’s conceptions of it. Many philosophers approve of Kitcher’s account: although some have pointed out that Kitcher’s account does not explain how rational a particular theoretical change is (Brigandt 2006, forthcoming)11 , other philosophers’ proposals are in line with his account (Field 1973, Newton-Smith 1982, Weber 2004). Case Studies and Kitcher’s Account The analysis of the cases studies in this chapter largely fit with Kitcher’s account. The lesson from Kitcher’s solution to the incommensurability problem is that theoretical terms from different theories 11 An  examination of this topic is beyond this paper’s scope.  76  could share the same reference in some contexts and thus there could be meaningful discussion between proponents of two different theories. Although Kitcher discusses a theoretical term and ‘species’ and a taxon’s name may not be purely theoretical terms, it is clear that naturalists, such as Strickland and Grey, have repeatedly tried to preserve the stability of reference for a species name——with the help of a new method of cataloguing, for example—— when they face the possibility of miscommunication within the biological community. Kitcher’s account is particularly useful here because it also makes sense of Darwin’s referential use of the term species. From his account, what Darwin ——and other naturalists partially adopting this use of the term—— did is that he added a new mode of reference to the term species (or emphasized an extant one for his particular purpose).12 The reference potential of the term species was enriched by this and ‘species’ thus could refer to the taxa which competent naturalists would recognize as a species. This enrichment made it possible that in some contexts Darwin talks about species while eschewing the connotation that it is immutable, while his contemporaries could still support the immutability of species as a definitional truth. In the debate between Coyne & Orr and Bush, as we have seen in the previous section, the same kind of reference-sharing occurs on the species Rhagoletis pomonella and its apple host race. Due to the zoological naming rules, the name Rhagoletis pomonella refers to the same object (a particular taxon) between the two camps and so does ‘the apple race of Rhagoletis pomonella’ with the help of additional description. There is more to successful communication between the camps than simply sharing a reference of the species name, however. Not only do both camps agree on which race is at issue in the debate, but they also agree, at least to some extent, on relevant causal factors working in the possible case of speciation. Selection of a species definition determines which causal relations researchers draw attention to in a given case. Since Coyne & Orr endorse the biological species concept and Bush does not, one may think that they differ on their views on important causal factors in sympatric speciation. But they do agree that reproductive isolation and natural selection play a significant role in sympatric speciation, although they do not concur on the degree of causal relevance of those 12 See  LaPorte (2003) for a similar analysis.  77  factors. This is why robust reference of the species name is not the only thing that makes the claims from one side look intelligible from the other. Both sides also do agree on some conceptions of ‘sympatric speciation,’ and thereby share a part of extension of the term——the case of the apple race could be an instance of sympatric speciation. If two groups of biologists endorsed different species concepts and claimed that the factors mentioned in their opponent’s definition are irrelevant to speciation because it is not mentioned by their own definition, they would find themselves in communication breakdown. Therefore, we may conclude that our analysis of possible cases of communication breakdown regarding species and speciation largely resonates with Kitcher’s solution (thus the fairly standard view in philosophy of science) to the incommensurability charge. Kitcher introduces the concept of reference potential and described how two scientists can communicate with each other even when they are in different theoretical frameworks or paradigms. This is because tokens of different concepts, such as ‘dephlogistoned air’ and ‘oxygen,’ could have the same reference through their modes of reference even when the conceptions of them radically differ. In the case studies of this chapter, we have repeatedly observed that naturalists managed to preserve the stability of reference of taxonomic names and the concept of species when they face the danger of communication breakdown (Strickland, Grey, and Darwin). In the debate between Coyne & Orr and Bush regarding the apple race of R. pomonella, they agree on the possible extension of ‘sympatric speciation’ as well as the reference of the species name and the apple race. Sharing a reference of a taxon name and other key terms is a substantial factor to ensure successful communication among those biologists who may have different ideas with regard to the nature and the right definition of species.  78  Chapter 3  Dual-Process Theory and the Concept of Species 3.1  Introduction  The history of the recent debate on the so-called “species problem” is the history of competing species concepts. If one were to write a book on the subject, it would start with the advent of the New Systematics (Huxley 1940) and the Biological Species Concept (Mayr 1942, Beurton 2002), and continue with the many other species concepts that followed it. But this perspective on the history of the recent species problem would only illuminate one aspect of the problem, because biologists often deploy the notion of species without advocating a particular species concept. This fact about biologists’ use of species was remarked upon by naturalists as early as Charles Darwin (1859, p. 44). No one definition [of ‘species’] has as yet satisfied all naturalists; yet every naturalist knows vaguely what he means when he speaks of a species. (italics added) Darwin also highlights the other problem faced by biologists using the notion of species in their work. He explains that although biologists (naturalists) generally have some idea of what a species is, they cannot articulate either the nature of this 79  understanding or what is understood by species clearly enough to reach a consensus on the proper definition of ‘species.’ Put another way, Darwin’s remark suggests there are two different modes of understanding the species concept: 1. Biologists may understand species through definition. 2. Biologists may understand species, while leaving its content unarticulated (Darwin calls this a “vague” way of understanding). This chapter aims to show that Darwin’s view of the matter in the nineteenth century is still by and large right about biologists’ thinking about species. To put it differently, biologists think about species in two psychologically different ways. I am going to argue that this fact is especially relevant to determining why the species problem persists and how biologists can conduct their business——undertake their research——without a “solution” to the species problem. My argument will proceed in two stages. This chapter concerns the first. I show that biologists work on ——study, define, classify, etc.—— species by two psychologically distinct processes. I also argue that these two processes largely correspond to the two processes proposed by the Dual Process theory in cognitive and social psychology. Chapter Outline The structure of this chapter is as follows. The next section describes the Dual-Process theory. According to the Dual-Process theory, human beings employ two psychologically distinct cognitive processes on many occasions. The Dual-Process theory describes probable features of those processes. The first process (System 1) is considered to be implicit and automatic and the second process (System 2) is considered to be an explicit and controlled process. The following sections will explain how the Dual-Process theory can make sense of biologists’ practices involving species. The third section introduces a phenomenon I call elusive transparency. Many naturalists believe that they understand the nature of species, but find themselves unable to adequately define it. This does not mean that biologists know nothing about species; rather, they have an implicit understanding of the concept. A couple of studies (Luckow 1995, McDade 1995) also suggest that biologists’ implicit understanding of species may differ in content 80  from their explicit one. For example, many taxonomists do not make their choice of species definition explicit when they describe a new species in their paper. The broad aim of this chapter is to argue that these facts fit the characterization of two processes in the Dual-Process theory. To make this argument, I introduce two possible components of System 1 process regarding species. The first component is the concept called “good species.” I survey a variety of usages of this phrase and show that in many cases ‘good species’ means (i) a taxon judged to be a species according to multiple species criteria or (ii) a taxon generally judged to be a species by the taxonomic community. I then analyze this notion in the light of the Dual-Process theory. The idea is that this notion may well be processed in System 1 (an implicit, automatic process), rather than System 2 (an explicit controlled process) and, with the help of the notion, one can make sense of elusive transparency. Another component of System 1 reasoning is so-called “psychological essentialism,” namely the inclination to assume there are essential properties responsible for the superfical features of an object. I will argue that psychological essentialism may well be associated with System 1 reasoning about species. In the last section, I will describe System 2 processing in taxonomists’ reasoning about species. I conclude that two psychologically distinct processes are at work in taxonomists’ minds, as it were, and describe how each taxonomic practice should be classified in terms of the two systems posited by the Dual-Process theory. Let me acknowledge one thing. As should be clear, significant parts of the arguments I will put forward in this chapter depend on the Dual-Process theory. Although this theory is adopted by a number of cognitive and social psychologists, it is still incomplete, and has much room for refinement (see Appendix A). This chapter does not aim to prove my thesis, because Dual-Process theory itself has not been proved yet. Instead, the objective of this chapter is to offer a promising picture of the different modes of understanding species, and how these are manifested in biological practice. I concede that the picture remains unproven, in an everyday sense, but I believe it is worthy of serious consideration.  81  3.2 3.2.1  Overview of the Dual-Process Theory The Basic Claim of Dual-Process Theory  Dual-Process accounts have been exploited in various subfields of psychology, such as cognitive psychology and social psychology, to explain a variety of experimental results. Although the details differ from one account to another, the basic tenet is similar: people have two different information-processing systems, and each has a cluster of distinctive characteristics. People employ either process (or, occasionally, both at the same time) in problem-solving tasks. Psychologist Jonathan Evans gives a brief description of such accounts: What dual-process theories [of human behavior] have in common is the idea that there two different modes of processing . . . Almost all authors agree on a distinction between processes that are unconscious, rapid, automatic, and high capacity, and those that are conscious, slow, and deliberative. (Evans 2008) Similar accounts appealing to two distinctive mental processes have been proposed in various areas of psychology, including reasoning (belief bias, matching bias,Wason selection task, hypothetical thinking, etc.), judgment and decision making (judgment of probability, decision making under risk, social judgment theory), and social cognition (processing of social information: person perception, stereotyping, and attitude change; for various applications, see Sloman 1996, Chaiken 1980, Fazio 1990, Saunders 2008, Ohtomo & Hirose 2007, Brewer 1988, and others). Several different names have been given to each process, including “associative system /rule-based system” (Sloman 1996), “heuristic/analytic processing” (Evans 1984), “tacit/explicit thought processes” (Evans & Over 1996), “experiential/rational system” (Epstein 1994), “quick and inflexible modules/intellection” (Pollock 1989), “intuitive/analytical cognition” (Hammond 1996), “Autonomous Set of Systems (TASS)/Analytic System” (Stanovich 2004). However, to avoid the impression that the property after which each process is named is essential to it, let us adopt the following rather neutral terminology: ‘System 1’ for the first processing system in the above quotes, and ‘System 2’ for the second. On our terminology 82  then, Dual-Process theorists assert that there are two processing systems (Systems 1 & 2) and that we often employ either or both of them in problem-soving tasks. An Example of the Dual-Process Account: The Belief-Bias Effect in Deductive Reasoning Let us take deductive reasoning as an illustration of how the Dual-Process account is invoked to explain experimental results (Evans 2003). In this experiment, subjects are asked to evaluate deductive arguments and answer whether or not a given argument is logically valid. Arguments shown to the subjects in the experiment have two variables. The first one is validity of the argument, and the second variable is believability of the conclusion. Some arguments are logically valid and others are not. Some arguments have a conclusion that one easily believes, while others do not. For example, the following argument is logically valid, and moreover one could easily believe its conclusion: (a) 1. No police dogs are vicious. 2. Some highly trained dogs are vicious. 3. Therefore, some highly trained dogs are not police dogs. Compare this argument (a) with the following syllogism (b). (b) 1. No nutritional things are inexpensive. 2. Some vitamin tablets are inexpensive. 3. Therefore, some vitamin tablets are not nutritional. Argument (b) is logically valid too, but its conclusion is hard to believe, due to the commonsense view that all vitamin tablets are nutritional, which (b)’s conclusion denies. Researchers have found that subjects are more likely to see this type of argument as invalid than arguments like (a); indeed, as Evans reports, 90% of subjects see arguments like (a) as valid, while less than 60% think arguments like (b) are valid. Likewise, subjects are far more likely to see invalid arguments with a believable conclusion as valid than invalid arguments with an unbelievable conclusion (70% vs. 10%). Evans suspects that subjects’ judgments are affected by both the logical validity of the argument and the believability of its conclusion. If 83  an argument has a believable conclusion, then subjects tend to take the inference to be valid, even if the argument is invalid; if an argument has an unbelievable conclusion, then subjects tend to take the inference to be invalid, even if the argument is valid. This is called belief-bias. To explain this result, Evans proposes the hypothesis that there are two different information-processing systems working in the subjects’ minds, which are responsible for conflicting responses. One system checks validity of an argument, while the other determines whether the conclusion of an argument is easy to believe and then uses this as a cue to the logical validity of an argument. These processes yield the same results in cases like argument (a), but offer different, incompatible results in cases like argument (b).  3.2.2  Characterization of the Two Systems  Properties of the Two Systems We have seen how the Dual-Process account is supposed to account for our deductive reasoning. However, Dual-Process accounts are not only put forward for deductive reasoning, but rather for a variety of phenomena, including: • probabilistic reasoning (representative heuristics) (Kahneman & Frederick • • • •  2002), impression formation (Brewer 1988), persuasion and attitude change (Chaiken 1980), attitude-guided behavior (Fazio 1990, Ohtomo & Hirose 2007) and moral judgments (Haidt 2001, Saunders 2008).  Dual-process theorists propose that two distinct systems are working and exhibiting a similar cluster of properties across these phenomena. Characterization of the two systems varies from one author to another, but there is significant overlap in what authors list as properties of each respective system. Evans (2008) summarizes them in a table. One finds similar tables in many papers on the topic (see Table 3.1). Each system has a distinct cluster of properties. For instance, System 1 processing is unconscious, fast, automatic, and supposedly evolutionarily old. It also takes low effort, has a high capacity, and so on. System 2 processing has another 84  Table 3.1: Characteristics of Two Processes (Shortened and modified from Evans 2008) System 1 (S1)  System 2 (S2) Consciousness  Unconscious  Conscious1  Implicit  Explicit  Automatic2  Controlled  Low Effort  High Effort  Rapid  Slow  High Capacity  Low Capacity  Default Process  Inhibitory  Holistic, perceptual  Analytic, reflective Evolution  Evolutionarily old  Evolutionarily recent  Evolutionary rationality  Individual rationality  Shared with [other] animals  Uniquely human  Nonverbal  Linked to language  Modular cognition  Fluid intelligence Functional characteristics  Associative  Rule-based  Domain-specific  Domain-general  Contextualized  Abstract  Pragmatic  Logical  Parallel  Sequential  Stereotypical  Egalitarian  1 Related properties: being volitional, responsiveness to verbal in-  structions (Evans 2003), thinking aloud (Evans & Over 1996) 2 As a result of this, only the final product of processing comes to  consciousness.  85  cluster of properties: conscious, slow, controlled, and supposedly evolutionarily novel (being presently unique to human beings). It also takes high effort, has a low capacity, and so on. How to Find Two Systems Conflicting Responses  As in the above example, Dual-Process theorists take  conflicting responses from subjects to indicate that there are two psychological processes working. This is because Dual-Process theorists believe that conflicting responses probably originate from different psychological processes. In the experiment above, subjects’ judgments regarding the validity of a given syllogism are clearly divided. When the hard-to-believe conclusion does follow from the premises, only 60% of the subjects think that the argument is valid, while the remainder see it as invalid. When the easy-to-believe conclusion does not follow from the premises, 30% of the subjects correctly see it as invalid, while the remainder sees it as valid. These kinds of conflicts in responses are frequently found in psychological experiments (Stanovich 1999), and some of them have been called “fallacies” or “biases,” because one response deviates from the “correct” laws of logic or mathematics. This is about diversity in collective response. But researchers also observe that conflicting responses often arise and remain within a single subject (Sloman 1996). One example is the Linda problem. The Linda problem is a psychological experiment in which subjects are asked to read the profile of a woman called Linda and choose her likely occupation. Her profile strongly suggests that she is a feminist. In one version of the experiment, subjects are asked whether Linda is more likely to be (a) a bank teller or (b) a feminist bank teller. Many subjects choose (b), which is an “intuitive” answer. This happens to be false, however, as it involves the conjunction fallacy: whenever Linda is a feminist bank teller she should be a bank teller at the same time, but she could be a bank teller without being a feminist bank teller.1 However, psychologists report that even when a subject is told what the 1 Psychologists think that this is an instance of representative heuristics: subjects’ choice is based on a judgment of similarity between the profile and the alternatives ((a) and (b)). For more detail on the Linda problem, see, for example, Kahneman & Tversky (1983). For arguments on Dual-Process  86  right answer of the question is and why, strong positive feelings for an “intuitive” but wrong answer (b) often persist in her mind. Stephen Jay Gould depicts this feeling vividly: I know [the right answer], yet a little homunculus in my head continues to jump up and down, shouting at me — ‘but she can’t just be a bank teller; read the description.’ (Gould 1991, p. 491; quoted from Sloman 1996) Kahneman & Tversky (1983, p. 300) also report that when one subject was told the right answer to the same problem, she replied unapologetically, “I thought you only asked for my opinion.” Steven Sloman thinks that this persistence of an “intuitive” answer is quite suggestive; for a psychological process responsible for the “wrong” answer is still active and effective even after the subject is somehow convinced about the “right” answer. Based on observations like this, Sloman (1996) puts forward a criterion for the existence of dual processes (for reasoning tasks) and calls it “Criterion S”: A reasoning problem satisfies Criterion S if it causes people to simultaneously believe two contradictory responses. (p. 11) Psychologists believe that these kinds of conflicts occur when the two different processes are working in our minds and feeding contradictory answers to our consciousness (Evans 2008). This criterion will be helpful in determining when biologists seem to be deploying different information-processing systems. How to Classify Individual Processes Dual-Process theorists generally use clusters of properties like those listed in Table 3.1 to tell to which system a given process belongs. However, they take no property in the table as a necessary or sufficient condition for either system. For instance, few psychologists may see the disposition to look for believability as particularly holistic (as opposed to analytic), but this is not seen as a problem. Moreover, there is some crossover in properties. Some properties are not uniquely attributed to one system (Samuels 2008). The evolutionary origin of each system constitutes one example. System 1 is supposed theory explanations, see Kahneman & Frederick (2002).  87  to have a longer evolutionary history than System 2. Indeed, System 2 is supposed to be unique to human beings. However, some System 1 characteristics, such as belief-bias, are also believed to be unique to human beings, because belief-bias requires an explicit belief system (see section 3.6.2 for further discussion). Although one does not find crossover in all the properties listed, one should be careful in classifying an individual process into either system on the basis of such properties. This also illustrates the difference in emphasis on properties among the DualProcess theorists. Although Evans has discussed the issue of evolutionary origin for over a decade (Evans & Over 1996, Evans 2003, 2006b, 2008), for example, many Dual-Process theorists are not interested in it. In spite of all this, those properties in the table are homeostatic (in the sense described in Chapter 1, section 1.4.1): if an individual process has one property in one cluster, then it tends to have other properties in the same cluster. This is why Dual-Process theorists classify a given process into either system when it has many features of one system and few, if any, features of the other. I followed the same classificatory procedure in the belief-bias case and will in the rest of this chapter. Homeostasis is also the reason why there are few disagreements among the Dual-Process theorists regarding the classification of individual processes.  3.2.3  Summary  In this section we outlined the Dual-process theory in cognitive and social psychology. According to the Dual-Process theory, human beings employ two psychologically distinct cognitive processes on many occasions. The theory proposes that those processes have a distinct cluster of homeostatic properties. For example, the first process (System 1) is considered to be an implicit and unconscious process and the second process (System 2) is considered to be explicit and conscious (see Table 3.1 for other probable properties). The phenomenon called belief-bias illustrates the point: when subjects are asked to judge the validity of a syllogism, some subjects appear to reach their judgement via shortcut ——they use some “cue,” such as the believability of the conclusion of an argument—— others try to judge the logical validity of the argument directly. Psychologists suspect that distinct psychological mechanisms are responsible for this variation in their responses. Other  88  experiments like the Linda problem provide support for this interpretation, because a single subject often exhibits conflicting responses to one and the same question. Cognitive psychologists view this conflict in subjects’ response as evidence for the Dual-Process theory. They also have discovered similar phenomena in a variety of experiments and suppose that the same or similar kinds of processes are activated in the subjects accross those experiments. So much for the basics of the Dual-Process theory. I concede that the theory has qualifications and limitations, but I leave them to Appendix (on p. 199). I also describe further details of the theory there. This Dual-Process theory is a broad framework for this chapter and I employ it to explain how biologists work on the notion of species. However, we need to articulate the phenomena to be explained. This is what I do in the next section.  3.3  Elusive Transparency  Dual-Process Theory claims that there are two distinctive psychological systems. Each has a cluster of different properties; one is unconscious, fast, autonomous, implicit, and takes low effort, while the other is conscious, relatively slow, deliberative, explicit, and takes high effort. In the following sections I will argue that this theory can make sense of biologists’ practices involving the species concept. For this purpose, in this section I describe rather puzzling phenomenon I call elusive transparency. One major objective of this chapter is to offer a plausible account for this phenomenon with the help of the Dual-Process theory. One thing to note. The arguments I put forward in the rest of the chapter are based on psychological experiments and taxonomic literature. However, as far as I know, no psychological experiments have been performed to determine the way in which biologists think about species. One may thus find some of my inferences speculative and spot the lack of decisive evidence for my arguments. Although I will try to construct a coherent picture of what biologists do with species, as clearly as I can, to prove decisively that my picture is true certainly eclipses the scope of this paper. Nevertheless, given that there has been relatively little attention paid to the questions I am pursuing, constructing a coherent, albeit tentative, picture of what biologists do with species is an essential and necessary prelude to more  89  advanced research. After all, we must begin somewhere.  3.3.1  Elusive Transparency  Several researchers have observed that individual biologists and taxonomists believe they understand the nature of species, but subsequently find themselves at a loss when asked to define it. This is what might be called elusive transparency about species: biologists tend to believe that the nature of species is transparent, but it turns out to be extremely elusive and difficult to articulate in a way that commands wide, much less universal, assent. Take the quote from Darwin (p. 79) “No one definition [of species] has as yet satisfied all naturalists; yet every naturalist knows vaguely what he means2 when he speaks of a species.” One can make sense of this quote, given elusive transparency regarding species. Since naturalists believe that they know the nature of species, they try provide an adequate definition, one that all can accept. It turns out that they cannot, yet they still believe they know something about species. And they are right. Their understanding is just a “vague,” or unarticulated one. Note that elusive transparency should not be taken to imply that everything biologists believe they understand about species is illusory. Naturalists in the quote from Darwin are not completely ignorant of the nature of species. For otherwise it would be puzzling that they can classify and conduct fruitful research with species. They know something about species——it is just not in the form of a definition. Hey’s Observation  Elusive transparency about the species concept has been ob-  served by biologists themselves on several occasions. Population geneticist Jody Hey provides an example. In his book on the species problem, Hey (2001a) reports that biologists, including himself, often find themselves casually using the word ‘species’ in conversation with colleagues, as if they fully understand its common meaning. And this is despite the fact that they know, as intimately as anyone, all 2 One might think that Darwin adopts the referential use of ‘species’ here (when one adopt this, she refers to the taxa that competent taxonomists would classify as species. See Chapter 2 for details). This is not the case, because this quote is followed by this: “Generally the term [‘species’] includes the unknown element of a distinct act of creation.” Here, Darwin simply describes the content of the general understanding of species.  90  the difficulties that have attended every attempt to define the notion (see also Hey 2006). In his own words, Hey confesses, It has been my experience ——and I am guessing that it is a typical one—— that when talking with biologists, one hears [the term ‘species’] tossed about regularly in a manner that supposes there is one single common meaning. If pressed on that common meaning, biologists are stuck, but they persist in using the word in a casual way much as laypersons do, as if it has a well-known meaning. (Hey 2001a, p. 11) This strikes much the same chord as Darwin’s passage, written nearly 150 years earlier.3 Biologists still claim to know what ‘species’ means, and take this understanding for granted in casual conversation. And, equally, biologists still find it hard to articulate their understanding to others, much less to give the single correct definition of ‘species.’ Notice that elusive transparency is not only concerned with individual biologists but also with the biologists’ community as a whole (or, more precisely, what they believe the majority of biologists believe). When one feels that the nature of species is transparent to him, he would naturally think that it is also transparent to other biologists. One can clearly read this from Hey’s observation above. When Hey assumed that the term ‘species’ has the “common” meaning, he also assumed that the meaning is shared in the biological community. Darwin’s words, by themselves, do not talk about this as explicitly as Hey does, but he does intend to say that an individual naturalist believes that his understanding has something common with others’ understanding. Indeed, after the quote, Darwin points out “a distinct act of creation” as a common element of the notion of species (Origin, 44). Two Modes of Understanding Species  Darwin and Hey’s observations of elu-  sive transparency seem to suggest that there are two possible “modes” of thinking about species in biologists’ minds. One can describe them by using the terms Darwin and Hey use: 3 Cronquist  (1978) and Pigliucci & Kaplan (2006) also point to this apparent transparency of the notion of species to biologists.  91  1. an unarticulated or implicit mode of understanding ——understanding “one single common meaning” of ‘species,’ or “what biologists means by ‘species”’— — and, 2. an explicit mode of understanding, as represented by giving a definition of species or stating the “common meaning” clearly. It is not hard to see that these two modes have a number of the properties assigned to the two systems in the Dual-Process Theory (see Table 3.1 on p. 85). The first unarticulated mode is implicit (biologists fail to articulate their understanding), not symbolically represented (nonverbal), unconscious, and perhaps the default mode (Hey’s biologists are in the casual mode of understanding before asked to articulate it), whereas the second mode is explicit, symbolically represented (linked to language; defining something is a linguistic activity par excellence), and conscious (giving a definition is a conscious process; see section 3.7.2 for further discussion.) Furthermore, elusive transparency suggests that each biologist gives somewhat conflicting responses to the species problem, in accordance with these two modes of understanding. This does not necessarily mean a biologist will have different ideas about the nature of species when in the unarticulated mode, than when in the explicit mode of understanding. The difference between the two modes may not lie as much in distinct species conceptions, as in different recognitions of the degree to which one understands the nature of species. In the unarticulated mode, a biologist simply believes she can see the nature of species, while in the explicit mode, she admits her understanding is less than complete, since she cannot easily or adequately define it.  3.3.2  Luckow and McDade’s Observation  It has just been described how the Dual-Process theory makes sense of the elusive transparency witnessed by Darwin and Jody Hey. A conceptual component of elusive transparency is the apparent transparency of the species concept to biologists. However, because Hey is a geneticist, some may wonder if his anecdote correctly captures the way in which taxonomists understand the notion of species, in particular, the apparent transparency of species. But it is also reported that taxonomists tend not to articulate which species concept they use in their papers (Luckow 1995, 92  McDade 1995). Melissa Luckow (1995) surveyed 130 papers on species (not all of which focus on the species problem) in two professional journals for systematists, Systematic Botany and Systematic Zoology (now called Systematic Biology) from 1989-1993. In doing so, she observed that the authors of many papers do not specify which species concept they are using, despite an apparent expectation that they do so.4 Luckow writes, . . . most of the papers were not explicit about which species concept was being used. (598) Only 20% of the papers in Systematic Botany were explicit as to the criteria being used; the default criterion was usually gaps in qualitative variation. An even smaller subset (8%) of papers specified which species concept was being used. (600) This observation conforms with Hey’s anecdotal report that biologists tend to take their notion of species for granted and not feel the need to make it explicit in practice——at least not until they are asked to do so. One might wonder if it is because of lack of space, not lack of interest or attention, that taxonomists do not articulate their choice of species concept. This may be true of some papers, but certainly not most. Declaring one’s choice would not take much space at all, if one does not discuss what is the right concept of species. A few sentences or couple of paragraphs would suffice, and many papers describing new species are 3-5 pages long. Thus, adding a few sentences to indicate their preferred definition would not much reduce——if at all——the possibility of acceptance. Therefore, I do not believe that they fail to articulate their choice because they fear this will cost them acceptance of their paper.5 4 Luckow  also points out that some authors even sometimes use different criteria within the very same paper with no explanation as to whether this reflects their pluralistic position or a possible inconsistency in the application of species criteria (598). However, I found no evidence of such an abuse of different criteria in the papers she is referring to. 5 Another possible explanation is that a taxonomist refrains from specifying her choice of species definition because she may believe that participating in a controversial issue, such as the species problem, does not help her paper accepted. But this is hardly the case. As far as I know, there is no evidence that systematists have such a concern. After all, Luckow would not recommend any measure for systematists if she believes that most of them could not implement it.  93  Lucinda McDade (1995) surveyed 104 monographs6 in three journals of plant taxonomy (Systematic Botany Monographs, Opera Botanica, and Systematic Botany) published over a 10 year period (for the first 2 journals from 1984-1993, for Systematic Botany from 1988-1993). She found that few monographs discuss the species controversy. She suggests that the reason for this is that the authors believe that the controversy is of little practical importance to them: The monographic work of systematics thus appears to have been largely unaffected by recent contributions to the species controversy. . . . This suggests that many monographers find the species controversy to have little practical bearing on their work. (613) This indicates that many researchers shrug off “theoretical” discussion in their work. McDade’s observation does not directly prove the point that taxonomists or biologists have a shared unarticulated idea of species (or they believe they have one) as Darwin and Hey suggest, but it does mesh nicely with it. First, McDade’s observation does imply that taxonomists often can and do deal with species in their work without delving into the details of the species controversy. And the substantial part of the controversy involves discussions of the costs and benefits of different species definitions (see, for example, Wheeler & Meier 2000). In other words, there is something in their business on species classification other than discussing what the “right” definition is. The controversy concerning the single “correct” or “right” species concept is often seen as irrelevant to taxonomic practice. This also fits with Luckow’s observation, since if taxonomists did feel the need to discuss the notion of species, they would not avoid declaring which species concept they employed in their papers. Second, even if monographers avoid “theoretical” discussion of species definitions, they should have some idea of species; otherwise, they would not know how to use morphological data in writing monographs. That idea should be somewhat “vague,” or unarticulated.7 Perhaps it is something like “A species is 6 A monograph is a comprehensive taxonomic work on a single taxonomic group (usually at a higher level than species). It typically summarizes and revises the information on all the extant species and adds description of newly discovered species in the group. 7 This does not preclude the possibility that a systematist has more articulate idea, such as some version of phylogenetic species concept or the biological species concept. However, if the authors of most monographs have such an articulated idea, then they would more frequently express their idea  94  a group of organisms morphologically distinguished from other such groups” (i.e., close to the morphological species concept; see Chapter 1). But every taxonomist knows this is not sufficiently articulated, because one can find morphologically distinguishable groups at higher or lower levels than species. Summary In this section we have paid attention to the elusive transparency of the species concept to biologists. Elusive transparency suggests that biologists follow two different modes of understanding —— (i) the unarticulated or implicit mode and (ii) the explicit mode. Those modes have several properties of the two systems as explicated by the Dual-Process theory. Luckow and McDade’s observations show that taxonomists often do not bother to articulate the notion of species they use, nor to delve into the species controversy in their papers. Those observations fit with Hey’s report that biologists tend to take their unarticulated notion of species for granted before they are asked to articulate it. McDade’s remark also suggests that there is something else to taxonomists’ business on species other than merely giving and discussing definitions of species. McDade’s observation also implies that taxonomists have some idea of species. It is somewhat “vague” in that it has room for articulation. The aim of this chapter is, as we have seen, to explain how this elusive transparency occurs with the help of the Dual-Process theory. I outlined the DualProcess theory in section 3.2 and introduced elusive transparency in this section. I also suggested that those two “modes” of understanding the notion of species correspond to two Systems as explicated by Dual-Process theory. But we have not seen how the Dual-Process theory could be applied to biologists’ reasoning on species. This is what I do in the following sections.  3.4  Notion of ‘Good Species’  This section and the next focus on the phrase and concept “good species” and how the Dual-Process Theory can make sense of it. ‘Good species’ is a rather unofficial technical term used in systematics and more broadly in biology. One often sees this phrase used in taxonomic description papers and scientific papers of species. This is what McDade reports is not the case (see the beginning of her quote).  95  on topics related to species, such as speciation. For the moment, ‘good species’ is defined as a “species which is distinctive or well-defined by a single or multiple criteria” (further explication will be given shortly). The notion of good species seems quite relevant to the themes discussed in the previous section. For example, ‘good species’ is used to refer to individual species as in “Xus bus [a name for a taxon] is a good species.” Once a taxon is considered to be a good species, the fact that it is a species is generally taken for granted; one can stop pursuing the correct definition of species as long as one is only studying that taxon Xus bus. This is not only because the criterion according to which one judges the taxon to be a “good species” is often implicit in the paper, but also because being judged to be a good species sometimes implies that it will be judged to be a species under many other criteria——as if it is supposed that Xus bus would be a species no matter what definition one adopted for a species. In this section I shall survey biologists’ usages of ‘good species.’ I have gathered usages of ‘good species’ from two main sources: (i) a mailing list for taxonomists, called Taxacom8 , and (ii) abstracts which include the term ‘good species.’ For those who want to know the summary of these surveys, please go to the summary section (3.4.3) on p. 110. Following that, I discuss how the Dual-Process Theory would make sense of this notion. From the vantage of the Dual-Process Theory, good species can be taken to be prototypical of the concept ‘species.’ Many versions of the DualProcess theory agree that judgments involving prototypes are generally processed by System 1, not System 2. I conclude that the System-1 processing is involved in biologists’ judgments concerning good species. This conclusion also will give content to the System 1 processing in the context of the notion of species, since we have only described so far System 1 as an unarticulated mode of understanding species.  3.4.1  Survey from Taxacom  Taxacom is a mailing list mainly for systematists. The history of Taxacom goes back to an Internet bulletin board set up in 1987 by Richard Zander. Later, two 8 http://mailman.nhm.ku.edu/mailman/listinfo/taxacom  96  mailing lists, “taxacoma” and “taxacomb,” began at Michigan State University in 1990. The list was moved to Harvard in 1992 (the archive covers messages dating back to this year), and then to University of California Berkeley in 1994. It finally relocated itself to the University of Kansas Natural History Museum in 1999. The list has been about taxonomy from the beginning, but the objective of the list was originally to foster communication among taxonomists on computer-related topics such as software applications for systematics research. However, the topics have grown to include a variety of subjects related to taxonomy, whether is computerrelated or not.9 Although anyone can join, as long as they have their own email account, most participants are expected to be taxonomists or somehow engaged in taxonomy. Otherwise there is little benefit to joining the list. Typical messages include various advertisements for positions, fellowships, conferences, as well as conversations on diverse topics related to systematics and taxonomy. For example, one recent message which received many replies asked participants to identify the single paper describing the greatest number of species. Thus this is a good place to collect data on the actual uses of the phrase ‘good species’ by professional taxonomists. There is another advantage to surveying this mailing list in order to understand how taxonomists think about species. Recall that one of the goals of this chapter is to determine the nature of the rather unarticulated or implict understanding of species by biologists. However, in more official venues for expressing their thoughts such as professional journals, precision in thought and writing is valued. In contrast, more informal conversations take place on mailing lists than in refereed papers. Since ‘good species’ is not a formal term but rather taxonomic jargon,10 examining how professional taxonomists use this phrase on a mailing list may yield valuable insight in the way into which biologists think about species. 9 This  description of the history of Taxacom is based on Christine Hine’s book (Hine 2008) and description at Taxacom info page (URL: http://mailman.nhm.ku.edu/mailman/listinfo/taxacom) [Retrieved on June 15, 2008]. 10 One may cite as an evidence for this the fact that scare quotes are often added to “good” as in ““good” species.” A further discussion will be given on it in the section 3.5.3.  97  Method Taxacom has an archive on the Internet.11 To gather usages of ‘good species,’ I downloaded past conversations (from November 1992 to May 2008) to search for messages which include the phrase “good species.” I surveyed how the phrase was used in those messages, excluding those in which ‘good species’ appears only in a quote from another entry. I also excluded from the survey any messages irrelevant to the current purposes. For one, I did not count the messages where ‘good species’ appears only as a part of a longer phrase and does not refer to good species. For instance, when a message contains “X is a good species indicator,” this means that X is good as a species indicator, or that X provides good evidence that a taxon at issue is a species. In other words, the phrase “good species” here is simply a part of a longer phrase “a good species indicator” and does not refer to good species. I did not count those messages as a legitimate usage of ‘good species.’ When an author provides their affiliation in the message, I added it to the name of the sender. However, since not all authors do this, even when they are affiliated with an institution, there are more affiliated authors than there appear to be. The number of the messages I reviewed is 35. I have included a list of the messages that I surveyed as an appendix B.1 (p. 205). Result It turns out that taxonomists use ‘good species’ in several different, but related ways. (a) An Alleged Species That Satisfies One Species Criterion In some messages ‘a good species’ is used to mean “a species which is well-defined or distinctive according to one criterion.” The following message contains one example of this usage: Dear All, The Electronic Atlas of the Plants of British Columbia shows Sax11 http://mailman.nhm.ku.edu/pipermail/taxacom/  98  ifraga tischii12 as a synonym13 of Saxifraga rufidula [...]. However, most databases still show S. tischii as a good species based on morphological differences (unfortunately NCBI shows no molecular sequences for S. tischii). ... (Ken Kinman, March/20/2007, “Saxifraga tischii a good species?”) Kinman asks whether Saxifraga tischii is a well-defined species on a morphological criterion of species. But there is more than one criterion that can be used to delimit “a good species.” Some consider a species to be “good” because it is distinct by morphological criterion; others rely on the biological species concept, according to which a species is reproductively well isolated from other similar groups. For example, . . . Here’s something to consider: There are cases in which two sister species look perfectly alike, which is termed “twin species” or “cryptic species”. Even though they look identical, they do not interbreed and are hence good species. . . . (Susanne Schulmeister [Institute of Zoology and Anthropology University of G¨ottingen], Feb/08/2002, “Comments to Cladistics/Eclecticism, part 3”) The following example is particularly interesting because Ken Kinman, whom we have just seen using ‘good species’ when talking about a morphologically distinct group, also calls a reproductively isolated, but not morphologically distinct, group “a good species.” Not surprisingly, I follow Ernst Mayr’s version of semi-species [i.e., incipient species ——YA]. Although many cladists14 probably prefer a more morphological species concept, Mayr’s biological species concept is a little more rigorous, and a semi-species is therefore not regarded as a good species (but on the way to getting there). . . . (Ken Kinman, Oct/12/1999, “Mayr on semi-species & speciation”) 12 In  taxonomic literature species names are italicized, like “Saxifraga tischii.” But in the mailing list, they are not, partly because it is informal communication. I will leave this intact when I cite messages from Taxacom. 13 For the meaning of ‘synonym’ in systematics, see footnote 3 of Chapter 2 (on p. 50) 14 Cladistics is a school of taxnomy which believes classification should be entirely based on phylogenetic relationships among terminal taxa. See footnote 6 of Chapter 1 (p. 8).  99  (b) An Alleged Species That Satisfies More Than One Species Criterion The fact that the same author has employed different criteria for ‘good species’ on different occasions suggests another usage of the term. That is, a taxonomist sometimes implies that multiple alternative species criteria are satisfied by “good species.”15 For example, Thomas DiBenedetto refers to a group judged to be a species based on reproductive isolation and morphological criterion as a “good species.” He imagines a situation in which A and B are sister species at one time, but later a new population C buds off from A by autopolyploidy and becomes reproductively isolated and morphologically distinct. In my example above, the isolated population (C) is the result of autopolyploidy with a single propagule getting established and founding the new population. C has both morphological apomorphies (polyploidy often yields these) and is a “good” biological species. (Thomas DiBenedetto, Feb/06/2002, ‘Cladistics and “Eclecticism”’) Larissa Vasilyeva hints at the idea that a good species is identified based on multiple alternative species criteria by pointing out that many species classified by Linnaeus were “good” species and are still considered to be so. Although Linnaeus may have generally depended on morphology in his identification of a species (see, for example, Larson 1968), modern biologists may well use other criteria even though they still classify them as species. Linnaeus was a genius, he guessed a lot of ‘good’ species and genera persisting today. (Larissa Vasilyeva, Jan/28/2002, “natural”)  (c) A Taxon Generally Recognized as a Species  According to a third use, bi-  ologists sometimes use ‘good species’ to refer to a group generally regarded as a species. For example, of the level of confidence in a particular species designation, Lawrence Kirkendall says 15 Note that the first use (a) does not contradict this second use (b). Nothing in the first use precludes the possibility that a species recognized by a criterion A may be recognized by a criterion B.  100  With respect to the identification, I wish to point out two aspects that may have been overlooked [...]: confidence that (1) the sample matches the specific species concept (which usually has been derived from one or more descriptions, +/- illustrations, and type specimens), and (2) the specific species concept being used represents a “good” species. Both affect confidence in the “result”, a statement that the sample represents species xxx of genus Nnn. (Lawrence Kirkendall [University of Bergen, Zoological Institute], Jun/14/1995, “Confidence”) In this quotation, Kirkendall points out that when one assesses the level of confidence one has in a particular classification, there are two elements to consider: (1) the degree to which the samples used in constructing the classification adhere to the species concept the taxonomist employs, and (2) the degree to which taxa generally classified as species by that species concept (other than those taxa in the classification in question——i.e., those instances not included in the sample) are actually species. This leads to evaluation of the species concept employed itself. This second element is an instance of the usage presently under consideration: will the species concept employed properly recognize a “good species” ——a taxon generally judged to be a species by taxonomists—— as species? Kirkendall seems to presuppose that a “good species” is generally judged by competent taxonomists to be a species regardless of which species concept they adhere to, if any; otherwise he would take that species concept (which he uses to judge a group to be a “good” species) to be the “right” one, and so beg the question.16 In this context, I must also point out that when biologists use ‘good species,’ they do so with the expectation that their judgment with respect to species classification will be shared by others. When considering whether or not the Citizendium17 , a wiki-based free encyclopedia (like Wikipedia) with more oversight by experts, could play the role of a public information storage facility for biodiversity, Richard Zander says, The Wikipedia has a new rival, the Citizendium, that is supposed to provide more expert oversight and better ombudsing, but I can’t see 16 I am inclined to think that Kirkendall here proposes the same kind of test for the adequacy of a species criterion as James Mallet does. See my discussion on his account of species on p. 107. 17 http://en.citizendium.org/wiki/Main Page  101  even this system helping when two experts see or contribute different things (maybe because they live in different areas of the world or examined different specimens or their ocular micrometers are badly calibrated). “Black Maples have fuzzy lower leaf surfaces, and are good species. This is MY page.” “Not always! and they are not good species! and it is not YOUR page!” “Always! or they are not Black Maples! which is a good species! Get your OWN page!” (Richard Zander [Missouri Botanical Garden], May/14/2007, “encyclopedia of life”) From this, one can see that if Xus bus is considered to be a good species by a taxonomist Mr. A, then he should be able to reasonably expect that his judgment will be shared by other taxonomists. This usage is consistent with all the other usages discussed in this section; if, for example, a group is distinctive morphologically and thus considered to be a good species by a taxonomist, then she should be able to expect others to make the same judgment. The same is true of the case in which multiple criteria are used for the judgment; the more criteria that agree with the classification, the stronger the expectation that the classification will be agreed upon by others will be. This usage fits most naturally with the third usage. If Xus bus is judged to be a species independent of any particular criteria, one can assume that almost all taxonomists will consider it a species. So much for results of the survey on Taxacom (see Table 3.2 (p. 110) for the summary of the results). Let move on to the next survey on professional journals.  3.4.2  Survey from Abstracts of the Papers in Professional Journals  I have surveyed the abstracts of papers published in professional journals as well as the messages posted to the mailing list Taxacom. I had two reasons for conducting surveys of the academic journals in addition to the mailing list. First, taxonomists are expected to have conversations that are less formal on the mailing list than they do in professional journals. I discussed various usages of ‘good species’ gleaned from the messages in the Taxacom archive. Since academic journals are a more formal venue in which scientists can share their ideas, they may use ‘good species’ differently there. Second, surveying academic journals can help us see how the term ‘good 102  species’ is used by a wider variety of authors. As with most other mailing lists, the members who post messages to Taxacom may not be representative of the large body of subscribers. For example, not every member of the list is as eager to post messages as others. Rather, only a small portion of subscribers post messages regularly while the majority of the community rarely post messages, if ever. For example, according to Christine Hine (2008), in 1995 there were approximately 2500 messages posted to Taxacom by roughly 500 message senders.18 Since many messages are various kinds of advertisements, there may well be even less subscribers who engage in serious discussions. Moreover, although Taxacom has over one thousand members, this is only a small fraction of the entire community of taxonomists; many taxonomists do not subscribe to Taxacom for various reasons. For example, Hine (2008) shows that the number of taxonomists registered with the World Taxonomist Database19 is significantly larger than that of Taxacom subscribers.20 By contrast, all taxonomists are expected to write and read journal papers (including monographs) periodically, so there must be a wider range of authors and readers of academic journals than there is for the mailing list. Method I searched for papers that had abstracts containing the term “good species” at (1) Web of Science21 and (2) Zoological Record22 in Cambridge Scientific Abstracts from the University of British Columbia Library Web page.23 Both of them are standard search engines for academic purposes. Web of Science is a leading online academic research database. It provides access to five databases: the Science Citation Index (1900-present), Social Sciences Citation Index (1956-present), Arts & Humanities Citation Index (1975-present), Index Chemicus (1993-present), and Current Chemical Reactions (1986-present) and contains 8700 academic journals 18 Since some may use more than one email accounts when sending messages to the list, the number  of actual senders may be less than 500. 19 http://www.eti.uva.nl/tools/wtd.php 20 From  my survey conducted on June 9, 2008, Taxacom has 1782 subscribers while the World Taxonomist Database has 4651 registered members. 21 http://apps.isiknowledge.com/ 22 http://www-ca6.csa.com/ids70/advanced search.php?SID=u5tc426crl4rvsr4bv1tith3b4 23 http://www.library.ubc.ca/  103  from humanities to natural science.24 Zoological Record contains papers in various fields including conservation and environmental science, evolution, genetics, geographical and fossil records, marine and freshwater biology, parasitology and disease, reproduction, taxonomy and systematics dating back as far as 1978. One can search for papers according to several parameters such as author, title, journal and abstract. I searched for papers by specifying the parameter “good species” in the abstract and title fields, and the search engines found the papers which had the phrase “good species” in their abstracts and titles. Since the string “good species” does not always refer to good species, I excluded all irrelevant papers from the review in the same way as I did in the search of Taxacom. For example, I did not count the abstracts where a string of words “good species” does not refer to good species, as in “a good species indicator.” While some papers appear as a result at both engines, they were each reviewed only once. My search of Web of Science yielded 112 results and that of Zoological Record yielded 82 results. Since there are 43 papers appearing twice, the number of the papers I reviewed was eventually 151. I have also included a list of the reviewed papers as an appendix B.2 (on p. 207). Result I mentioned three different usages when I discussed how the authors on Taxacom use “good species”: (a) a group classified as a species by a single criterion, (b) a group classified as a species by multiple (mainly two) criteria, and (c) a group which competent taxonomists would be expected to classify as a species independently of any species criteria. While all of those usages are found in the abstracts, there are some interesting differences in the details. Furthermore, there is another usage which occurs regularly in the abstracts that one does not often find in the messages on Taxacom. (α) An Alleged Species That Satisfies One Species Criterion The authors of these papers still cite a single criterion, e.g., a morphological one or a reproductive one, when they call a given species “good.” This paper mentions morphological 24 Retrieved  from http://scientific.thomson.com/products/wos/ on June 12, 2008.  104  difference before calling Paracercion barbanan a good species. Eight species occurring in Japan and continental East Asia are separated by the morphology of their male terminalia and by the structure of the female pronotum and adjacent laminae mesostigmales. Paracercion barbanan is confirmed as a good species, probably restricted to China, where it co-occurs with P. impar and other spp. (Dumont 2004) There is also a paper citing the interbreeding criterion in the judgment of “good species” (Edelaar 2008). This is the same criterion used as in the messages on Taxacom regarding reproductive isolation. (β ) An Alleged Species That Satisfies More Than One Species Criterion As with the messages on Taxacom, there is a usage in which multiple criteria are cited for the judgment of “good species.” The papers in which this usage occurs tend to make the fact that their judgment is based on multiple criteria explicit, while the authors of the messages in the mailing list tend not to emphasize that they are using multiple criteria. One example of this occurs in the title of the paper (Saito et al. 2007): Polistes formosanus Sonan, 1927 (Hymenoptera : Vespidae), a good species supported by both morphological and molecular phylogenetic analyses, and a key social wasp in understanding the historical biogeography of the Nansei Islands. In this case the authors cite two alternative species criteria (morphology and phylogeny) when they call the taxon “good species.” Another example in which the authors make reference to more than two criteria when calling a taxon a good species (this is also an example of using ‘good species’ as a way to emphasize a taxon’s specieshood——see below) is the following: Polytene chromosomes of four members of the Simulium perflavum species group in Brazil are described,. . . Chromosomal, morphological and ecological evidence indicates that S. maroniense Floch & Abonnenc, previously considered synonymous with S. rorotaense, is a good species. (Hamada & Adler 1999)  105  Note that this usage (β ) does not necessarily conflict with the usage (α). Mentioning one criterion when one judges a taxon to be a good species does not necessarily deny that the taxon may satisfy another criterion of species. This is the same as in the Taxacom messages. (γ) A Taxon Generally Recognized as a Species As another usage, which corresponds to the third one we have seen on the mailing list, one can point to a case in which ‘good species’ is used to refer to a taxon that an author assumes is generally classified as a species without citing any particular criterion of species classification. Reference Point When biologists use the term good species, they sometimes view good species as a reference point. That is, the authors assume that “good species” are species and compare the taxa under consideration with those recognized as good species in order to infer something significant about the taxa. For example, this usage appears when biologists aim at measuring the effectiveness of a novel method of classification by using a species as “reference point,” as in this paper: In just over a decade, the use of molecular approaches for the recognition of parasites has become commonplace. For trematodes, the internal transcribed spacer region of ribosomal DNA (ITS rDNA) has become the default region of choice. Here, we review the findings of 63 studies that report ITS rDNA sequence data for about 155 digenean species from 19 families, and then review the levels of variation that have been reported and how the variation has been interpreted. ... Closely related species may have few base differences and in at least one convincing case the ITS2 sequences of two “good” species are identical. (Nolan & Cribb 2005) The authors seem to suggest that those groups between which no difference in the ITS2 sequences were observed are generally accepted as species by taxonomists by referring to them as “good species.” They infer that other alleged species might not be distinguishable in this area of sequence. The next paper uses ‘good species’ in a similar way. 106  The genetic divergence between the eastern European, southern European, and Asian chromosome forms of the pygmy wood mouse Sylvaemus uralensis, whose karyotypes differ from one another in the amount of centromeric heterochromatin, has been reevaluated using allozyme analysis. . . . However, the allozyme differences between all chromosome forms of the pygmy wood mouse are comparable with the interpopulation differences within each form and are an order of magnitude smaller than those between “good” species of the genus Sylvaemus. Thus, the chromosome forms of S. uralensis cannot be considered to be separate species. (Bogdanov 2004) In these cases, the authors do not cite any paper to support that those “good species” are actually recognized as species. But it is implied that they are recognized so and fellow taxonomists and biologists would not find it difficult to see that implication. Chan and Levin also seem to use ‘good species’ in this way: Accurate phylogenies are crucial for understanding evolutionary processes, especially species diversification. It is commonly assumed that “good” species are sufficiently isolated genetically that gene genealogies represent accurate phylogenies. However, it is increasingly clear that good species may continue to exchange genetic material through hybridization (introgression). . . (Chan & Levin 2005) The authors do not explicitly discuss species definitions. But this usage is observed whether or not species definitions are discussed. James Mallet (1995a,b, 1996) uses ‘good species’ in a similar way, although he does this in his discussion of species concepts (Mallet is not alone for this usage of ‘good species’ in discussion of species concepts. See Bush 1995). Dobzhansky discovered that certain ‘good species,’ characterized by strong hybrid inviability when crossed, were morphologically inseparable. (Mallet 1995b, p. 295) From this quotation, it may seem that Mallet sees a taxon separated by a reproductive barrier as a “good” species. However, the next quotation suggests that he uses this phrase in a more general sense. Second, many good species seemed to have little in the way of sterility barriers (e.g., dogs, pheasants and Crinum lilies). The explosion of 107  data we have today confirms this: for instance, intraspecific sterility barriers caused in insects by the endosymbiont Wolbachia have little to do with speciation. (ibid.) Here he implies that a “good” species may have a reproductive barrier within itself. Thus he seems to use ‘good species’ to make a reference to a taxon which is generally recognized as a species by taxonomists. Indeed, he makes it explicit what he means by ‘good species’ in his reply to Kerry Shaw’s criticism (Shaw 1996) to his species concept. I used the term ‘good’ species several times meaning that people generally agree that ‘the blue whale’ and ‘the fin whale,’ for example, are species, . . . . Unless taxonomists are mad, there is something reasonable about such species. . . . (Mallet 1996, p. 174) Note that he does not refer to any particular criterion when he says that “the blue whale” and “the fin whale” are recognized as a species. This is why I judge Mallet uses this phrase independent of any particular criterion.25 Notice also that he finds a normative element in taxonomists’ judgments. One should probably respect taxonomists’ (presumably quite solid) judgments on species classification. Despite the above discussion, it may be hasty to assume that if a taxon is recognized as a species by a taxonomist (or even the taxonomic community) it is a good species. This abstract shows this is not always the case. Whereas previously all populations of Pleuroxus known from the subantarctic islands and southernmost South America were considered to belong to a subspecies of P. aduncus (described from France), now there are five distinct species, only one of which [P. varidentatus] resembles P. aduncus to any significant extent, but even it is a good species. (Frey 1993) Here although one population of Pleuroxus is not previously recognized formally as a species, it is now called a “good species.” So a taxon may not be called a “good species” merely because it is recognized by a taxonomist to be a species. The quotation from Frey also shows that a taxon does not need to be recognized as a species by taxonomists to be a good species. If Frey is right, then P. varidentatus is a good species but has not yet generally recognized as a species by taxonomists. 25  Shaw (1996) also take it that Mallet uses ‘good species’ this way.  108  Quality of Description In a related usage, ‘good species’ is sometimes used in a contrast to poorly described species. . . . Recognition of numerous and poorly circumscribed orchid taxa is a serious obstacle to their conservation because rare, poorly defined species may be prioritized for conservation over taxonomically “good” species. (Pillon & Chase 2007) The authors contrast poorly described species with good species.26 The adverb ‘taxonomically’ is suggestive here, because this seems to mean that such a species is good in terms of taxonomic practice. Thus the authors give us an impression that a good species is well described in a taxonomic paper.27 (δ ) Specieshood  Lastly, there is a use of the term ‘good species’ that one does  not often find on Taxacom. It is when a taxonomist calls a taxon a good species in order to emphasize the fact that the taxon is not a subspecies, synonym, or species complex——entities which are often confused with a species. This can be seen in the following quotation (See also Hamada & Adler 1999): Polistes formosanus Sonan, 1927 is closely related to P. japonicus de Saussure, 1858, and has been treated variously as a good species or subspecies or synonym of P. japonicus. . . (Saito et al. 2007)28 26 In  the quotation above, the authors write “poorly defined species,” not “poorly described species.” In taxonomic literature, ‘definition’ often means “description which allows us to distinguish one taxon from another.” For instance, in the International Code of Zoological Nomenclature (Ride et al. 1999) ‘definition’ is use to refer to “A statement in words that purports to give those characters which, in combination, uniquely distinguish a taxon.” 27 I am inclined to offer an instance of this usage of ‘good species’ from a source different from a paper abstract. We have already seen that Guy Bush uses ‘good species’ this way in Chapter 2 (p. 62). Arthur Cronquist (1978), a plant taxonomist, quotes, as “an old joke” among taxonomists, “a good species is what a good taxonomist says it is” and compares this with a statement on pornography made by Potter Stewart, a judge of the Supreme Court of the Unites States, that ‘hard-core pornography’ was hard to define, but “I know it when I see it.” The same comparison is made by Pigliucci & Kaplan (2006) (without mentioning Cronquist), although they simply use the term ‘species,’ not ‘good species.’ 28 Taxonomists often do not make a distinction between use and mention with regard to a species name. If Polistes formosanus is a good species or a subspecies, “Polistes formosanus” refers to a taxon, not to a taxon’s name; if Polistes formosanus is a synonym of P. japonicus, “Polistes formosanus” refers to a name of a taxon, not to a taxon itself, since a synonym is a name, not a taxon. Since this ambiguity is common in the taxonomic literature, I will only note it if necessary.  109  (a) (b) (c)  Usages from the Taxacom mailing list An alleged species that satisfies one species criterion An alleged species that satisfies more than one species criterion A taxon generally recognized as a species  (α) (β ) (γ) (δ )  Usages from professional journals An alleged Species that satisfies one species criterion An alleged species that satisfies more than one species criterion A taxon generally recognized as a species Specieshood  Table 3.2: Summary of the usages of ‘good species’ observed. (a)-(c) are found in the messages to the Taxacom mailing list. (α)-(δ ) are from the papers in biology journals.  As in the quotation above, a typical example of this usage might be “As a result of research we did, Xus bus turns out to be a good species, not synonymous with Xus cus.” Here ‘good species’ is used to stress the fact that the taxon is a new species and not equivalent to a part of a previously recognized species.  3.4.3  ‘Good Species’: The Variety of Usages  Let me summarize the usages observed in the messages to the Taxacom mailing list, and those found in biology journal articles. See Table 3.2. There are no major differences between the kinds of categories found in the two sources, despite our expectation that one would find significant difference between them. This coincidence indicates that the usage of ‘good species’ is stable across official and unofficial contexts. The summary above also clearly shows that ‘good species’ does not have a single definite meaning. This is not surprising, because ‘good species’ is rather informal terminology in the taxonomic community. Nevertheless, different usages overlap significantly. Meanings common to both surveys are the following: 1. an alleged species that satisfies one species criterion, 2. an alleged species that satisfies more than one species criterion, and 3. a taxon generally recognized as a species. Hereafter I will focus on the second and third meanings for further analysis of ‘good species.’ This is because those meanings are closer to the “unarticulated” or “indeterminate” notion of species, which is the subject matter of this chapter, than 110  the first meaning. Although calling a taxon “good species” because it satisfies one species criterion does not preclude that it also satisfies other criteria, the speaker may well focus on that particular criterion she appeals to, and thus that token of ‘good species’ has a clear meaning when compared to the second and third usages. On the other hand, the third meaning is more unarticulated or indeterminate than (1), because it by itself says little about biological characteristics which its referent (a good species) has. The Distinctness of Good Species It is also worth noting that good species are probably considered to be distinct from other species. Good species are often supposed to meet multiple criteria of specieshood. In such a case, the taxon probably looks distinct from other species to many taxonomists, because taxonomists from opposing perspectives would, nonetheless, agree that it is a species.29 This would encourage many taxonomists to accept it as a legitimate species taxon. From this point of view, one can make sense of the fact that quality of description matters in judging whether a given taxon is a good species. Quality of description partly depends on distinctness of that taxon; if a taxon is not distinct, it would be hard to describe30 it so that one could easily distinguish it from others. * In this section, I have surveyed a variety of usages of the unofficial taxonomical terminology, ‘good species.’ I continue to analyze this notion in the next section but in the light of the Dual-Process theory, which I introduced in section 3.2. One main point in question is which psychological process (System 1 or System 2) is activated when biologists reason about species with the help of the concept good species. Answering this question will also illuminate how the Dual-Process theory could be applied to biologists’ reasoning on species. 29 One may find some support from cognitive psychology in this regard. Cognitive psychologists point out that human beings are particularly keen observers of multiple interrelational associations among objects. Humans tend more quickly to learn how to tell one category from another when two categories are different in multiple properties, than a single property (Kornblith 1993). See Murphy (2002) for survey of the psychological literature. 30 See note 26 for the meaning of ‘description’ in taxonomy.  111  3.5  Good Species and Dual-Process Theory  We have seen how the phrase ‘good species’ is used by biologists and what it means to them. Here I will argue that good species is a prototype of species and that System 1 processing is activated when biologists reason using good species. First, I will overview a general account of prototypes. I sketch prototype effects, which motivate cognitive psychologists to pay attention to prototypes. I describe how cognitive psychologists characterize prototypes. I then clarify possible senses in which good species is a prototype of species. Next I apply this framework to describe biologists’ dealings with the concept of species. In particular I show that the notion of “good species” shares a number of features with other prototypes and conclude that it is a prototype of species. Finally, I argue that since prototypes or stereotypes are generally processed in System 1, the notion of “good species” may well be processed in System 1 (an implicit, unconscious process), rather than System 2 (an explicit conscious process).  3.5.1  Prototype Effects  Here I give a succinct account of the nature of prototypes in cognitive psychology. Cognitive psychologists began to pay serious attention to prototypes when they discovered what are called prototype effects on concepts (Rosch 1978). Prototype effects have been seen as a critical blow to the traditional theory of concepts, aptly dubbed the classical theory. According to the classical theory, a concept is represented by necessary and sufficient conditions for its application: a bachelor is an unmarried man (Laurence & Margolis 1999). In this picture, every instance of a concept is treated the same way in our mind, as long as it satisfies those conditions: if we believe Tom and George are both bachelors, then our mind represents them in the same way in this regard. Prototype effects provide strong evidence against this picture. Different instances of a concept may be represented differently in our mind. When subjects undertake various tasks involving concepts (e.g., naming instances of a concept or judging the membership of a particular instance of a concept), their responses often differ. For instance, subjects require less time to identify a typical member of a category (e.g., a dog for pet), than an atypical member (e.g., a snake). 112  Consider an experiment performed by Armstrong et al. (1983) on prototype effects.31 In this experiment, subjects are instructed to evaluate how typical an instance of a category is. The instructions read: This study has to do with what we have in mind when we use words which refer to categories . . . Think of dogs. You all have some notion of what a ‘real dog’, a ‘doggy dog’ is. To me a retriever or a German Shepherd is a very doggy dog while a Pekinese is a less doggy dog. Notice that this kind of judgment has nothing to do with how well you like the thing . . . You may prefer to own a Pekinese without thinking that it is the breed that best represents what people mean by dogginess. On this form you are asked to judge how good an example of a category various instances of the category are . . . You are to rate how good an example of the category each member is on a 7-point scale. A 1 means that you feel the member is a very good example of your idea of what the category is . . . For the category fruit, for instance, they are asked to judge how good apple is as a fruit on a scale of 1 (very good) to 7 (very poor). The same question is asked for other instances of fruit, such as strawberry, plum, pineapple, fig, and olive. The result is that subjects tend to judge that different objects are good to different degrees as instances of fruit (see Table 3.3). For fruit, the best example is apple, while strawberry follows in second place. The worst is olive. The same result is obtained for many other categories, such as sport, vegetable, vehicle, even number (2, 4, 34, 106,. . . ), female (mother, sister, housewife, and so on), and plane geometry figure (square, circle, triangle etc.). This indicates that different items of a category may be represented differently in our minds, even though each belongs to the same category. Psychologists have found other effects involving prototypes. Here are some examples: • Response Time: When subjects are asked to respond true/false to “X is a member of category Y ,” “responses of true are invariably faster” when X is “the items that have been rated more prototypical [in a preliminary experi31 The  authors note that this is a follow-up experiment to the one reported in Rosch (1975).  113  Table 3.3: Typicality ratings of instances of categories (Shortened and modified from Armstrong et al. 1983)  Fruit  Even number  apple  1.3  4  1.0  strawberry  2.1  8  1.5  plum  2.5  10  1.7  pineapple  2.7  18  2.6  fig  5.2  34  3.4  olive  6.4  106  3.9  Sport  Female  football  1.4  mother  1.7  hockey  1.8  housewife  2.4  gymnastics  2.8  princess  3.0  wrestling  3.1  waitress  3.2  archery  4.8  policewomen  3.9  weight-lifting  5.1  comedienne  4.5  Table 3.4: Response time for prototypical and non-prototypical members of categories (Shortened and modified from Armstrong et al. 1983. Items used in the experiment are shown in parentheses. Numbers are msec.)  Categories  Prototypical examplars  Non-prototypical examplars  fruit  903 (orange and banana)  1125 (fig and coconut)  sport  892 (baseball and hockey)  941 (fishing and archery)  even number  1073 (8 and 22)  1132 (30 and 18)  female  1032 (aunt and ballerina)  1156 (widow and waitress)  ment]” (Rosch 1978, p. 198).32 See results from a similar experiment done by Armstrong et al. (1983) in Table 3.4. • Speed of Learning: “[Y]oung children learn category membership [for artificial categories] of good examples of categories before that of poor examples.” (Rosch 1978, p. 198) 32 Rosch  (1978) is reprinted in “Concepts: Core Readings” (Margolis & Laurence 1999). Pagination comes from this reprinted version.  114  • Availability: “[T]he most prototypical items were the first and most frequently produced items when subjects were asked to list the members of the category.” (p. 199) These three points indicate that subjects give graded responses to various instances of a concept. Both a robin and a penguin are instances of the concept bird, but subject responses to a robin differ from those regarding a penguin. In other words, concepts are mentally represented in such a way that not every instance of them elicits the same response. These various facts about prototype effects do not imply that the membership relation is itself graded, but rather that subjects respond in grades to various instances of a concept —— put another way, prototype effects concern subjects’ representation of a given concept, not the concept itself. For example, subjects show prototype effects with regard to the concept odd number, even though no odd number (e.g., 3) is odder than any other instance (Armstrong et al. 1983). Cognitive psychologists have found other prototype effects, including: • Agreement on membership: “Perception of typicality differences is, in the first place, an empirical fact of people’s judgments about category membership. . . . subjects overwhelmingly agree in their judgments of how good an example or clear a case members are of a category, even for categories about whose boundaries they disagree. . . ” (Rosch 1978, p. 197) • Hedges: Although “A robin is a bird” and “A penguin is a bird” are both true, adding some hedges (qualifying terms such as “almost,” “virtually,” and “strictly speaking”) could change their truth values: “A penguin is technically a bird” is judged to be true while “A robin is technically a bird” is not (Lakoff 1973, Rosch 1978). • Inferences: Subjects tend to infer features from more to less representative members. For example, given the information that a robin (a more representative member of bird) has a feature F, subjects more readily infer that a penguin (a less prototypical bird) has F than otherwise (from a penguin’s feature to a robin’s feature) (Lakoff 1989).  115  3.5.2  Prototype —— Exemplar or Clustered Properties?  What Is a Prototype? Prototype effects show that our representations of categories have a graded structure, and exhibit gradients of membership. Eleanor Rosch says, “all categories show gradients of membership; that is, subjects easily, rapidly, and meaningfully rate how well a particular item fits their idea or image of the category to which the item belongs” (Rosch 2000, p. 66). Then what is a prototype? Laurence & Margolis (1999, p. 28, n. 35) point out that cognitive psychologists use the concept of prototype in one of the two senses below: (a) A prototype is a highly exemplary instance of a concept. (b) A prototype is a cluster of properties shared by many instances of a concept. In (a), a prototype can be (i) exemplary in-and-of itself, as apples are judged by experimental subjects to be prototypical, exemplary fruits. Or a prototype can be (ii) exemplary by virtue of possessing a sufficient number of properties that are exemplary of the concept, as when experimental subjects judge apples to be prototypical fruits because they have so many properties considered exemplary of fruits, like being sweet and brightly colored. Now one can see that according to (a. ii), a prototype could be taken to be similar to a natural kind as explicated by the homeostatic property cluster view (Boyd 1997, 1999, see also discussion in Chapter 1). Both a prototype and a HPC kind are associated with a property cluster which tend to be instantiated by a member of a kind (category). Both have vague boundaries. One typically cannot draw a strict line between prototypical and nonprototypical members of a category. According to sense (b), prototypes are property clusters. One crucial difference from (a. ii) is that while (a. ii) says that a prototype is a category associated with a property cluster, (b) asserts that a prototype is not merely associated with a property cluster; it is a property cluster. A prototype of fruit may be represented as a cluster of properties shared by many instances of the category, such as sweet, having the right size (a watermelon is too big, nuts are too small), edible, along with the structure they possess. Smith et al. (1988) take this position. If one adopts this option and analyzes the concept fruit, she would give different weight to each of those properties according to the diagnosticity of each attribute (sweetness, size, 116  . . . ); thus sweetness may be more important than size in evaluating how typical an apple is of fruit. According to these authors, this whole structure (a list of properties, the importance of each properties and so on) and how they are represented in one’s mind are identical to a prototype. A prototype is neither a particular thing or a kind. However, this should not be taken to imply that these two views are in some sort of competition for the “correct” meaning of ‘prototype.’ According to Laurence & Margolis (1999), they are not in such a relation. This is a purely semantic choice for cognitive psychologists. And so, the choice is up to us. Then, What Could a Prototype of Species Be?  So much for psychologists’  conceptions of prototype. According to the two views, there are three alternatives for a prototype of species, if it is a good species. (a. i) A good species is a prototype of species, in the sense that it is not only an alleged species, but is considered to be a highly exemplary member of the species category. (a. ii) A good species is a prototype of species in the sense that it is not only an alleged species, but also satisfies sufficiently many of a cluster of properties that are considered exemplary criteria of the species category. (b) The prototype of species is the cluster of properties shared by taxa called “good species.” The labels like “(a. i)” indicate that they are application of corresponding conceptions of a prototype. According to the first option, a good species is a highly representative instance of species, such as Homo sapiens and the American robin, Turdus migratorius. In the second option, good species is a category associated with a property cluster, and it typically contains properties indicated as species criteria by various species definitions, such as reproductive isolation, occupation of a distinctive niche, phylogenetic features, and so on. According to this option, good species as a prototype of species is similar to the natural kind according to the HPC theory. The third option would say that the prototype of species is a property cluster and its representational structure (this option is not similar to the HPC view in  117  this regard. For the HPC view, a kind is defined by, but not identical to, a property cluster). Those properties in the cluster are shared by the taxa called “good species.” Perhaps one criterion, such as reproductive isolation, counts more than being monophyletic for being a good species. According to this option, this whole representational structure is called a prototype. Here the first option will be ignored. The reason is this: for this option, a good species is represented as a particular species, such as Homo sapiens. But it is not clear that all or most biologists share the same individual species as a prototype of species, because many biologists have their own “focal” organisms in their research; botanists mainly study plants (or subareas of plants), while zoologists study animals (or subareas ——even units as small as a genus or species—— of animals). Thus, it is not likely that biologists studying such different areas would conceive of the same particular species when they think of good species or a prototype of species. Rather, they would probably have in mind some species relevant to their study——their focal organisms——when they think about species. Indeed, this tendency has been seen in the history of the species controversy; biologists tend to invent a species definition particularly suited to the organisms they themselves study. For instance, it is commonly thought that the biological species concept is well-suited for someone studying birds like Mayr, but not at all for practicing botanists. Botanists often have a different view of species than ornithologists (for example, Cronquist 1978). If we adopt the first option (a. i) for good species, then it follows that different biologists have different prototypes of species. Although there may be a sense in which one understands a prototype of species this way, this chapter so far has focused the conceptions of species common in the biologists’ community. Therefore, we will not hold the first option for further consideration. But this argument does not entirely preclude opting for Sense (a). If one sees good species as a kind of species (a. ii), just as apple is a prototype of fruit, then one can choose Sense (a) and avoid the above difficulty. Moreover, this option is compatible with the two usages of ‘good species’ we have observed on p. 110: (2) an alleged species that satisfies more than one species criterion and (3) a taxon recognized generally as a species by the biological community. It is obvious that (a. ii) and (2) are compatible. The option (a. ii) and the usage (3) are compatible because, as I have suggested earlier, a taxon which satisfies more than one 118  species criterion is likely to be recognized as a legitimate species by the biological community. I am also inclined to opt out of the option (b). According to this option, a prototype is not a particular physical thing or a kind, because a prototype is a property cluster. However, the usages of ‘good species’ as described in the previous sections suggest that good species is a kind, even though it may not be a natural kind. Then the option (b), by definition, implies that good species is not a prototype of species. However, I will discuss what psychologists call attribute substitution ——a psychological tendency to infer what attributes a kind has from the attributes its prototype has (see section 3.5.4)—— to account for elusive transparency, and this presupposes that good species is a prototype of species. The option (b) forbids us to take this explanatory strategy. Therefore, we will not hold the option (b) for further consideration. To summarize, if good species is a prototype of species, then it will be an alleged member of the species category, which is associated with a property cluster. These properties are typically various species criteria indicated by species definitions, such as reproductive isolation and occupation of a distinctive ecological niche. It could be taken to be close to a natural kind as explicated by the HPC view.  3.5.3  Good Species Is a Prototype of Species  Now we can see whether good species is a prototype of species. Good species as presented in the section 3.4 has several features of a prototype of species. ‘Good X’  A couple of linguistic features of good species make it seem like a  prototype of species. First, the phrase ‘good X’ is quite often used by psychologists to refer to prototypical instances of a concept. When psychologists attempt to find prototypes of a concept operationally, they almost always ask subjects to pick “good” instances of a concept. Recall that Armstrong et al. use ‘good X’ to refer to a prototypical member of a category in the instructions of their experiment (see p. 113). There are many other examples like this. In an experiment by Smith et al. (1988), subjects are asked to rate each item “for how good an example it is of the  119  category” (502). Psychologists themselves also commonly refer to a prototype by ‘good X’ or derived phrases, like “best example of X”. “. . . the social stereotype of a bachelor will characterize the best examples, and those undisputed bachelors who do not fit the social stereotype will be less good examples.” (Lakoff 1989, p. 402)33 “subjects overwhelmingly agree in their judgments of how good an example or clear a case members are of a category . . . ” (Rosch 1978, p.197, see also p. 198) “Lastly, we want to examine the data for conjunction effects. Table 17.3 presents the relevant data and predictions for vegetable concepts . . . The top half of the table contains the data for instances that were “good” members of their corresponding conjunctions . . . ” (Smith et al. 1988, p. 504).34 Notice that Smith et al. add quotation marks to “good members.” The use of quotation marks even coincides with taxonomists’ use of them. If you reread quotes from section 3.4, you might be surprised to see how often they add quotation marks to “good” in “good species” (see also note 10). The use of quotation marks by taxonomists reflects that ‘good species’ is an informal term, rather than official or technical terminology, and that the judgment that a taxon is a good species is also unofficial. The use of scare quotes by psychologists probably reflects that prototypes have the same unofficial character. This coincidence indicates that taxonomists and psychologists do not just use the same phrase; ‘good X’ has a similar linguistic function for both psychologists and taxonomists. Psychologists generally use the phrase ‘good X’ and phrases derived from it to refer to a prototype of X. This gives strong support to the claim that good species is a prototype of species. Hedges More linguistic evidence comes from hedges. “A penguin is technically a bird” is true, but “A robin is technically a bird” is not true, because a robin is 33 Lakoff  (1989) is reprinted in “Concepts: Core Readings” (Margolis & Laurence 1999). Pagination comes from this reprinted version. 34 Smith & Medin (1999) also name one of the models they examine, “the best example model” (p. 210).  120  a prototypical member of a bird, while a penguin is not (see section 3.5.1). The same thing is true of “good species.” Sentences such as, “Technically, good species Xus bus is a species”, sound false. In contrast, “Xus bus is not a good species, but technically a species.” sounds true, just as “A penguin is technically a bird” sounds true. In other words, a good species and a prototype will never be borderline cases. Indeed, taxonomists use “Xus bus is a good species” to claim that that the taxon is not a borderline case (see the discussion on specieshood on p. 109). Agreement On Specieshood Related but nonlinguistic support for our claim is agreement among biologists on specieshood. Rosch (1978) points out that subjects make similar judgments on how good a given object is, relative to a certain category. For example, much more subjects judge that apple is a good instance of fruit than that olive is. This implies that a prototype of a category is generally considered to be a member of that category. This is what is observed about good species. We have seen that once a biologist judges some taxon to be a good species she will form a strong expectation that other competent biologists will concur. And this should be the case if she is right about the pieces of evidence supporting her judgment: if, for example, a taxon Xus bus is clearly demarcated from other taxa in terms of phylogeny (i.e., if Xus bus is a monophyletic group and others are not), then other natualists will follow her judgment, as long as her phylogenetic analysis has no problems and they accept the same species concept she employs. Furthermore, biologists often use ‘good species’ to refer to species taxa that almost all biologists would judge to be a species, as Mallet (1995b) does. Thus, good species has a feature of a prototype of species in agreement on membership. Response Speed  We have seen that subjects take less time to identify a prototype  of a category as a member of that category than nontypical members. We have good reason to believe the same is true of good species and species, although, strictly speaking, there is no experimental data to confirm it. Recall that a good species is often clearly demarcated from other such taxa (see p. 111). If a good species is distinguished clearly and taxonomists can spend less to identify clearly distinguished taxa than otherwise, then they could identify a good species as a  121  species more quickly.35 It is probably true that taxonomists can identify clearly distinguished taxon (according to some adequate criterion) quickly as a species. Therefore, taxonomists arguably would take less time to identify a good species as a member of the species category than a non-good species. Thus, good species has another important feature of a prototype of species. Inference From Good Species Biologists sometimes infer that properties of a good species are characteristic of many or all species, including borderline “bad” species. For example, Chan & Levin (2005) point out that some good species hybridize with each other and then suggest that species in general could engage in the same degree of hybridization (see the quote from them on p. 107). Psychologists found that subjects tend to infer properties exhibited in a prototypic member of a category to non-typical members: if a robin has a property F, then a penguin probably has it (Lakoff 1989). It is worth noting that Lakoff’s remark also shows that this inference would be made for all members of the category, not just atypical members, because a category generally consists of “good” (prototypic) and “bad” (non-prototypical) members. Subjects tend to believe that if a robin has a property F, then any bird (including a “bad” member of bird like penguin) would have it. If it is sensible to read Lakoff’s remark this way ——although Lakoff does not particularly emphasize this reading—— then people arguably make similar inferences involving good species. Availability and Order of Learning  Good species has several features of a pro-  totype of species. However, this does not mean that good species is necessarily a prototype of species, because we have no evidence about other features of a prototype, such as availability and the order of learning (see section 3.5.1 for details). But this is simply due to the lack of experimental evidence. It is not easy to say 35 Note  that this response speed does not take into account the time they spend to gather data on character of a taxon. Suppose two species differ in the strucuture of a very tiny organ and it takes time to find its structure. But once researchers acquire appropriate data on it, they may take little time to judge a taxon as a good species. In such a case, then one can say that their judgment speed is sufficiently fast.  122  anything meaningful about these issues without experimantal data.36 So, it remains to be seen whether good species has those features of species, but this should not be taken to undermine the claim that good species is a prototype of species. Summary Good species has a number of features of a prototype: two linguistic features (‘good X’ and hedges), agreement on membership, response speed, and inferences. Although we do not know whether good species has other features of a prototype, such as availability and the order of leaning, this is solely due to a lack of experimental data. If we have a chance to perform relevant experiments, we can see the results. However, even with this limitation, good species is arguably a prototype of species, because many features of a prototype are observed in good species.37  3.5.4  How Biologists Reason With the Help of Good Species——Attribute Substitution  I argue that good species is a prototype of species. If this is true, what is the implication? In particular, how do biologists reason with good species? One possibility is that biologists often represent the species category as a whole by its prototypes in their minds. Psychologists call this an instance of attribute substitution (Kahne36 For  order of learning, however, there is an anecdotal evidence. Pigliucci & Kaplan (2006, p. 223) point that when biologists teach about species, they give examples to students first. If this is because students can learn more efficiently this way, this may suggest that the students can learn faster with good examples, i.e., good species. 37 The use of prototypes in taxonomic practice may be more widespread than this section suggests. Jody Hey (2001a) points out that when taxonomists engage in species classification, they often rely on a prototype of that species and treat each species as if it is a natural kind, rather than an individual, as Ghiselin (1974) and Hull (1976, 1978) suggest: Most biologists are working in contexts where others have gone before and have devised categories on the basis of their samples. When another investigator comes along, new samples are taken with the aid of a series of search images, such as field guides, or a series of criteria, such as identification keys. In the case of search images [something like prototype], organisms are identified to kind in a typological way [i.e., similarity to the prototype], and in the case of distinct criteria, organisms are identified to kind as if kinds are classical categories. When an organism is found that fits neither search images, nor catalogs of criteria, then the typical reaction is to consider it as a representative of an unknown species. All of these behaviors treat species as categories, and not as entities in nature. (p.162)  123  man & Frederick 2002, see also section 3.7.1). When attribute substitution occurs, a subject represents one attribute of an object with another somewhat relevant, not identical, attribute. Kahneman and Frederick call the first kind of attribute “target attribute” and the second “heuristic attribute.” In their words, ‘attribute substitution’ means that an individual assesses a specified target attribute of a judgment object by substituting another property of that object ——the heuristic attribute—— which comes more readily to mind (p. 53, original italics). Take an example from the Linda problem (see p. 86 for description of the problem). In this case, subjects represent a category (feminist bank tellers) by its prototype and judge the probability that Linda belongs to that category by the degree to which she resembles the prototype. Kahneman and Frederick note that it is because Linda’s profile resembles the prototype of feminist bank teller more than that of bank teller that subjects judges that Linda is more likely to be a feminist bank teller than a bank teller. This is partly because the heuristic attribute (the degree of resemblance) is more accessible to subjects than the target attribute (the probability that Linda is a bank teller). If attribute substitution is one of the things subjects do with a prototype of a category, and good species is a prototype of the notion of species, then it is no wonder that biologists do the same thing with good species. Indeed, biologists do seem to represent the species category by its prototype, good species, in their minds and make an inference about the attributes a species could have (target attributes) from the attributes a good species has (heuristic attributes). Let me formulate this here: Attribute Substitution of Species with Good Species: Biologists, often implicitly, represent the species category by its prototype, good species, in their minds and infer what attributes a species has (target attributes) from the attributes a good species has (heuristic attributes). Attribute substitution makes sense of observations made by biologists. According to Hey’s observation (section 3.3.1), for example, when he (and he assumes other biologists) participates in casual conversation, he sees the term ‘species’ as 124  having one common meaning, although he should be fully aware of the fact that there is no universally accepted definition of ‘species.’ This makes sense if he substitutes a prototype for the species category in his mind and infers the meanings of the term ‘species’ from the way in which a good species is, because a good species usually looks quite homogeneous in that it tends to satisfy many species criteria and one can easily tell that it is a species——We have seen that humans beings tend more quickly to learn how to tell one category from another when two categories are different in multiple properties, than a single property (see footnote 29). Even though there is no clear and common meaning in the term ‘species’ (a target attribute), the prototype makes it look as if there is such a meaning (a heuristic attribute), and a speaker easily believes that she grasps it. This also explains why they find it hard to precisely define species; when one substitute species with the prototype, her understanding of the notion is not mediated by words and a prototype may not easily translate into a strict definition——the category has the gradient of membership.38 Good Species Makes Sense of the Observations of Luckow and McDade Substitution of the species category with its prototype also makes sense of observations made by Luckow and McDade. As we have seen (section 3.3.2), they note that taxonomists are not interested in explicitly declaring their preferred species concept (Luckow) or discussing the species controversy in general (McDade) in their description papers. Now we can see how Kahneman and Frederick’s insight can account for the observations. If the substitution occurs in their mind, then naturally they would feel little need to spell out their conception of species and get involved in the controversy, because substitution may automatically lead them to believe, by the help of the prototype, that they understand the nature of species. A cate38 Some might wonder whether Hey substitutes the species category by good species as a category or individual instance(s) of good species, such as Homo sapiens. I am inclined to think that he substitutes the species category by the category of good species, not a particular instance of good species, because Hey seems to talk exclusively about the species category and the term ‘species,’ but not individual species. Yet this does not exclude the possibility that a biologist substitutes the species category with instance(s) of good species, not good species as a category. To take an example from the Linda problem, a subject may represent the category bank teller by a prototypical instance of bank teller (e.g., a teller the subject saw in the bank the day before the experiment) while she could have a prototype image of feminist bank teller even if she has not met such a bank teller in her life.  125  gory mediated by its prototype looks transparent to a subject (a heuristic attribute), even though the nature and the right definition of species has been the subject of a considerable controversy (a target attribute). If the substitution occurs, we can make sense of the observations made by Luckow and McDade. And there is a piece of evidence that it does occur. Taxonomic practice helps the substitution occur in the minds of taxonomists when they describe a new species or write a monograph. There is a recommendation in the taxonomic community that a (new) species should be described clearly, so that it can be easily distinguished from other species (Winston 1999). Described this way, individual species are likely to be represented as a good species in the minds of taxonomists, because distinctness is a feature of a good species (see section 3.4.3). And if a species is judged to be a good species, then a taxonomist would feel little need to spell out their conception of species or to be engaged in the species controversy, because a taxon at their hand is a good species and thus a species according to different species criteria anyway. Substitution also nicely meshes with Hey’s remark that taxonomists are essentialists and use prototypes in describing individual species (see footnote 37 for the quote from him). If taxonomists make use of prototypes in their minds when they classify organisms into individual species, then they may well use them at the category level. This is not to say that attribute substitution is always the cause of the phenomena observed by Luckow and McDade; a biologist may not make explicit her choice of species definition because of her deliberate decision. Yet the attribute substitution is still a possible, but often neglected explanation of many such cases.  3.5.5  Good Species Reasoning Involves System 1 Processing  Good species is reasonably seen as a prototype of biological species. This helps us see how good species is more likely to be processed in System 1 than System 2. In many psychological models a prototype is supposed to be activated and utilized in System 1, not System 2. Steven Sloman offers a good example. Sloman calls System 1 the similarity- or association-based reasoning process and System 2 the rule-based reasoning process. Similarity- or association-based reasoning processes information statistically (or quasi-statistically) in terms of similarities. It  126  “encodes and processes statistical regularities of its environment, frequencies and correlations amongst the various features of the world” (Sloman 1996, p. 4). Belief bias is an example of this process: in this example, subjects tend to see an invalid syllogism with a true conclusion as valid (see p. 83 in this chapter). Sloman says that reasoning is affected by a subject’s prior beliefs and content of a syllogism, and attributes the above tendency to associative memory. On the other hand, when the rule-based system is activated, one consciously follows rules which are “abstractions that apply to any and all statements that have a certain well-specified, symbolic structure... they have both a logical structure and a set of variables” (Sloman 1996, p. 4). This system is exemplified by the conjunction rule of probability: Pr(A) ≥ Pr(A&B). This rule can be applied to whatever event A and B stand for. Applying this distinction, categorization by similarity (particularly prototype) is taken as an example of the associative system (System 1). For instance, in categorization through prototype, one classifies an item X as belonging to a category Y by estimating the degree of X’s similarity to Y ’s prototype. Sloman contrasts this processing with rule-based categorization (System 2). Kahneman & Frederick (2002) also place representativeness heuristics (one of the prototype heuristics) in System 1. This is in contrast to extensional reasoning. Take the Linda problem as an example. Their idea is that subjects often represent a category (bank teller) by its prototype and take the extent to which Linda’s profile is similar to this prototype as the probability that she belongs to that category. They call this substitution prototype heuristics, and assign them to System 1. Extensional attributes, on the other hand, involve “an aggregated property of a set or category for which an extension is specified.” One example is the probability that Linda is a bank teller, because it is about whether or not a set of bank tellers contains Linda as an element. The normative principle of extensional attributes is controlled by (conditional or simple) adding —— for example, the probability that Linda is a bank teller is the sum of two probabilities, namely, the probability that Linda is a feminist and a bank teller at the same time, plus the probability that Linda is not a feminist and a bank teller (Pr(B) = Pr(F&B) + Pr(¬F&B)). Principles like this are the very thing prototype heuristics ignores. Furthermore, in Dual-Process models of social psychology, prototypes (or stereotypes) are also generally classifed as System 1 processing. One example is Russel 127  Fazio’s paper on how an attitude toward an object guides one’s behavior——how an attitude toward a presidential candidate affects one’s voting behavior, for instance (Fazio 1990). Following Kruglanski & Freund (1983), Fazio notes that subjects refrain from using an ethnic stereotype in their decision-making when the stakes are high, which is one condition for activating a deliberative processing (System 2) in his model of the attitude-behavior relation. Ohtomo & Hirose (2007) draw on Fazio’s framework when they construct a model for how one’s attitude affects eco-friendly or eco-unfriendly behavior. In their model, prototype image (“a mental image of the type of person who engages in the undesirable behavior.” (119)) is a determining factor of one’s environmental behavior. If one has a less negative prototype image toward littering behavior, for example, a reactive decision process (System 1), as opposed to an intentional decision process (deliberative process, System 2), leads her to do the same thing herself. Prototypes are taken to be a component of System 1 in many psychological models. Therefore, if good species is a prototype of the species category, then we have a good reason to believe that good species tends to be activated in System 1, rather than System 2.  3.5.6  Summary  Let me summarize the arguments I put forward in this section. Roughly I made two points about the way in which biologists reason about species. In the last section, I surveyed a variety of usages of the unofficial taxonomic phrase ‘good species.’ In this section, I analyzed the notion in the light of the prototype theory and the Dual-Process theory. After briefly overviewing the phenomenon called prototype effect in cognitive psychology, I made the first point that since the notion of good species has a number of features of a prototype as explicated in cognitive psychology, we have a good reason to view good species as a prototype of the concept of species. I also outlined the meanings of the term ‘prototype’ and articulated possible ways in which a good species is a prototype of species. I made explicit what I mean by ‘good species’ hereafter. Then I discussed how biologists reason with the help of good species in section 3.5.4. I point to what Kahneman & Frederick call attribute sub-  128  stitution: Biologists represent the species category by good species in their minds and infer what a species’ attributes are from the attributes of a good species. Then I showed that we can make sense of the phenomenon called elusive transparency, as described in the section 3.3.1, in the light of attribute substitution. For example, if one represents species with good species, then one tends to take its boundary to be clear even when she should be fully aware of the fact that it is not and the term ‘species’ does not have a clearcut defintion, because its prototype, a good species, looks so homogenous that the term ‘species’ could admit of a clearcut definition. I then turned to the second point: since good species is likely to be a prototype of species, and a prototype is supposed to be activated and utilized in System 1, not System 2 in many psychological models, we can infer that good species tends to be activated in System 1, rather than System 2. A number of Dual-Process theorists generally view reasoning with a prototype as System 1 reasoning, whether it is in cognitive psychology or social psychology. Therefore, if we can accept the first point of this section, it leads to the conclusion that prototype reasoning in general is a System 1 reasoning, which includes the reasoning with good species, a prototype of species. In this section I addressed one way in which biologists reason about species when they may activate System 1 reasoning. This indicates how the Dual-Process theory may be applied to illuminate biologists’ reasoning about species. I continue this project in the next section: I will explore another way in which biologists reason about species with System 1 reasoning, i.e., via what is called psychological essentialism.  3.6  Psychological Essentialism  In this section I shall discuss another possible component of System 1 processing in taxonomic reasoning, namely, psychological essentialism. Psychological essentialism about a given object is the psychological disposition to assume that the relevant object has an essence ——hidden properties causally responsible for the way an object is (in particular, its superficial properties)—— even when one is not aware of what the essence is. The first part of the section elaborates the phenomenon of psychological essentialism and demonstrates its presence in science, particularly  129  in taxonomy. In the second part, I discuss the status of psychological essentialism in Dual-Process theory. Some psychologists classify psychological essentialism as a System 2 process. I shall argue that the grounds for this classification are not strong enough and that psychological essentialism has several important features of System 1 processing. Pychological essentialism thus joins good species as a component of System 1 processing in taxonomic practice.  3.6.1  Psychological Essentialism and its Widespread Use in Taxonomy  Psychological Essentialism Psychological essentialism is widely observed in categorization. When one categorizes objects (particularly natural objects), one usually pays attention to two kinds of properties: overt or “superficial” properties, and “deep” or hidden structure. Objects generally have both kinds of properties. Gold looks yellow and has a particular density (superficial properties), but also has the atomic number 79 and an atomic structure (deep structure). Psychological essentialism is the psychological tendency to assume there is an intrinsic “deep” essence behind the superficial properties of an object, and to see this as relevant to the kind-membership of the object. Laurence & Margolis (1999) note that people take the deep structure of an object to be more important to its category membership: [P]eople are apt to view category membership for some kinds as being less a matter of an instance’s exhibiting certain observable properties than the item’s having an appropriate internal structure or some other hidden property. (p. 45) This does not mean that one has to have a detailed picture of the essence of the relevant kind when one thinks like an essentialist. Indeed, one may believe that all instances of gold have intrinsic properties in common and that they are causally responsible for their superficial properties, but still have little idea of what that essence is. This sort of representation of essence is called essence placeholder. On this kind of psychological essentialism, one has some idea of what function the essence will play in an instance of a kind, but little idea of how exactly it will do it. Medin (1989) makes this point when he says, 130  People act as if things (e.g., objects) have essences or underlying natures that make them the thing that they are. (p. 1476) . . . people adopt an essentialist heuristic, namely, the hypothesis that things that look alike tend to share deeper properties (similarities). Our perceptual and conceptual systems appear to have evolved such that the essentialist heuristic is very often correct. . . (p. 1477)  Supporting Experiments  So how do we know when people have the disposition  to think of a subject in essentialist terms? Gelman (2004) suggests that we think of an object along essentialist lines when we assume a category has, among other indicators, what she calls “inductive potential” and “underlying structure.” • Inductive Potential: when one makes induction about what properties an object has, one should place more weight on its essence than its appearance. • Underlying Structure: Internal structure is of more importance than appearance for predicting the properties of an organism. Underlying structure is causally responsible for superficial properties. One experiment for inductive potential goes like this (Gelman & Markman 1987). Pictures of three items are shown to preschool children: (a) a leaf, (b) a insect, and (c) an insect that looks like a leaf (the authors call it “leaf-insect.”). They are asked what properties each item has. Even though a leaf and a leaf-insect share many perceptual properties (being green, having striped markings, etc.), when children are told of their category (leaf, insect, and insect respectively), they make more inferences about novel properties from one organism to another of the same category with different appearances, than to those of a different category but with a similar appearance. Even if a leaf has a property F, children do not infer that a leaf-bug would also have it, but they infer that a leaf with a different shape would. What is remarkable is that even 3 and 4 year old children follow this pattern. Part of the findings on representation of underlying structure are based on the following experiment (Gelman & Wellman 1991). In this experiment, 4 and 5 year-old children are asked about an object’s identity and function when its insides or outside is removed. Objects include both living things and artefacts, such as  131  turtles, dogs, eggs, bananas, cars, books, and pencils. Among the questions asked to children are • Removal of insides, and identity: “What if you take out the stuffy inside of the dog, you know, the blood and bones and things like that, and got rid of it and all you have left are the outsides? Is it still a dog?” • Removal of outsides, and function: “What if you take off the stuff outside of the dog, you know, the fur, and got rid of it and all you have left are the insides? Can it still bark and eat dog food?” Gelman and Wellman found that more children say removal of insides makes a difference to the identity and function of an object than removal of outsides. From experiments like this, one of the authors concludes that, for children, insides are “privileged with respect to identify, functioning, and word extension” (Gelman 2003, p. 83). To summarize: when one categorizes (natural) objects, one often assumes it has a deep essence behind its superficial properties, and that this essence is causally responsible for those superficial properties. Thus, knowledge of essence helps us predict and explain the superficial properties of objects, and deep structure is more relevant to determining its kind membership. But one need not possess a view of what the essence is before one endorses this sort of essentialism. Rather, one may just have an “essence placeholder” in one’s mind when one catagorizes objects this way. Widespread Use of Psychological Essentialism in Science and Taxonomy One remarkable thing about psychological essentialism is that many preschoolers, as well as adults, exhibit this disposition in a variety of experiments. From this fact, one might reasonably suspect that most people, including scientists, take this stance. Indeed, psychological essentialism, as described above is exhibited by a wide variety of scientists, especially taxonomists. Several facts indicate this. First, biologists adopt this attitude when engaged in the search for the “right” definition of species (the species category). The search for the “right” definition plainly assumes 132  that the species category is a natural kind, and that it possesses an essence awaiting discovery. Most definitions of species presuppose that species is not a merely conventional rank, nor an individual (recall that Ghiselin and Hull talk about a species taxon, not the species category). If we do know what its essence is, there is no species problem to begin with. These assumptions are part and parcel of psychological essentialism. Furthermore, most of the definitions of species on offer appeal to deep or hidden structure. The distinction between superficial and hidden properties is somewhat blurred, but one can make a rough distinction. Most morphological characters are manifest properties, while most genetic properties, like DNA sequences, are deep or hidden. Interbreeding, ecological niche, and phylogeny are probably hidden structures, in that they are causally responsible for morphological characters and are not as easy to observe. So, the biological, ecological, and phylogenetic species concept (history-based, PSCh) appeal to hidden properties, and perhaps issue from essentialist leanings. Now it is true that some species concepts, like the taxonomic species concept (see Chapter 1 for details), the phenetic species concept (Sokal 1973), and some versions of the phylogenetic species concepts (especially the phylogenetic species concept (character-based); see for example Nixon & Wheeler 1990), are supposed to appeal exclusively to superficial characters. In addition, even among definitions that do not appeal to superficial properties, many still put emphasis on pattern, rather than process (compare genotypic cluster species concept (Mallet 1995b) with biological species concept). But definitions appealing to superficial properties do not constitute the majority of recent species concepts. Moreover, although Mallet focuses on genotypic pattern, his focus is still on genotypic pattern, which is not a superficial or manifest property. Second, an attitude quite similar to psychological essentialism is found in the history of systematics, although it is not about the species category per se. Ronald Amundson (2005) calls it cautious realism. He describes it as follows: In periods during which new theories are being developed, scientists are often willing to commit themselves to a phenomenal law, and to the claim that the law points to an important underlying causal explanation, but they are not be ready to commit themselves to the nature of the underlying explanation itself. I call this position cautious realism. 133  Cautious realists believe in a reality underlying a phenomenal law, but they are not yet ready to name it. (p. 15) A cautious realist believes that there are phenomenal laws and that there are common cause(s) for them. But she is cautious enough to refrain from pointing to what the cause(s) are. It is easy to see that, conceptually, this bears a remarkable resemblance to psychological essentialism. According to Amundson, several naturalists took this position about species and higher taxa before Darwin. Hugh Strickland (discussed in Chapter 2, section 2.2.1) is one such example (Amundson 2005, p. 46ff.). Amundson describes him as an empirical essentialist and realist regarding the higher groups. Strickland believed in the objective reality of higher taxa like birds by elaborating the distinction made by William MacLeay between two kinds of similarity——affinity and analogy. According to Strickland, affinities indicate a real and natural relationship between taxonomic groups, while analogies (like adaptation for swimming) are merely accidental similarities and fail to indicate real relationships between taxa. Affinities exhibit an agreement in essential characters among taxa and thus imply that those taxa form a natural group placed in the Natural System (the objective classificatory organization of organisms). Thus Strickland does believe that affinities inform us of the arrangements of groups in the Natural System and is fully committed to essentialism regarding the higher taxonomic groups. Nevertheless, he is cautious enough not to spell out what the essence is. In particular Strickland is skeptical about explanatory power of affinities on the arrangements of groups. Although affinities tell us where a given group is located in the Natural System, Strickland does not believe that our knowledge of affinities provide any deeper, causal explanation for those arrangements. Amundson says, Strickland “gave no indication that the discovery of taxonomic essences would enable new causal explanations. ... [Taxonomic essences] are the true basis of the Natural System, the organization of life. However, Strickland gives us no hints of what he would consider a deeper explanation of the Natural System” (Amundson 2005, p. 51). Moreover, Strickland was aware of the possibility that one is wrong about what character is essential character to a taxonomic group. He is a fallibilist on essence and did not believe that one can a priori know what they are. In this sense Strickland was an essentialist on species and higher groups, but he did not go far beyond it. 134  The third reason for supposing that psychological essentialism is widespread in taxonomy, is that it is so widespread in science in general. Amundson (2005) himself notes that cautious realism is fairly common in the history of science when new theories arise. Mendeleev illustrates this point (p. 33). Mendeleev discovered phenomenal regularities among chemical substances, as organized in the periodic table, and believed that there was something behind them, but did not venture to say what it was. Hilary Kornblith (1993) joins Amundson in this regard, albeit from a different perspective. Kornblith’s argument begins with the idea that induction is at the heart of scientific methodology and that understanding how various kinds of induction work is the key to figuring out why natural science has been so successful. He then suggests that we must have adopted the form of induction assuming hidden structure, because induction from hidden structure (if successful) is more efficient than that based solely on superficial properties. Roughly, this is because the former has a more solid basis than the latter. Induction from the atomic structure of gold, for example, offers more inferential potential than a yellow-colored appearance. Even if simple induction gives us a reason to expect observable properties to come together, he says, “the success of a theory explaining why such properties are inevitably united in nature provides us with a much firmer basis for continuing to expect the observable properties to be found together” (p. 43). Kornblith thinks that psychological studies of essentialism reinforce this point. After citing psychological findings that adult subjects adopt psychological essentialism, he suggests that this essentialistic attitude is inherited from their childfood.39 Furthermore, psychologists have emphasized the programmatic character of psychological essentialism (recall the discussion on essence placeholder). It is simply not the case that children will only make an induction from one property to another in a kind after they know the internal essence of a kind. Rather, at least in certain cases, they adopt the essentialistic attitude in the first place, before they are aware of what that essence is. Psychological findings about the essentialistic attitude thus strongly suggest that human beings in general ——let alone scientists—— adopt essentialism as a methodological (perhaps default) strategy in 39 He even argues that psychological essentialism is an innate characteristic of human beings. I won’t delve into this, because it is unnecessary for the claim I want to derive here from Kornblith.  135  exploring their environment. Kornblith describes this by saying We presuppose that kinds have such unobservable essences even when we are unaware what the essence of a kind is. As a result, we are not at any time inclined to classify objects solely in virtue of their observable features, but instead take for granted that the observable features of objects are only an imperfect guide to their true natures. (p. 81)40 One may wonder whether Kornblith is too quick to conclude that scientists adopt psychological essentialism just because children or general adult do; scientists might be more careful to avoid the essentialistic assumption, just as they are careful to avoid the conjunction fallcy. This concern is legitimate. But given that Kornblith already argued that the success of modern science partly comes from essentialistic ideas, what he tries to do here is to trace the origin of essentialistic attitude in one’s development. In this sense, he still gives some support for the idea that scientists adopt psychological essentialism in their research.  3.6.2  Is Psychological Essentialism Processed in System 2?  Psychological essentialism is prevalent in science and in taxonomy, in particular. If the Dual-Process theory is on the right track, which system does psychological essentialism belong to? Not all Dual-Process theorists discuss psychological essentialism, and not all students of psychological essentialism discuss Dual-Process theory. However, those theorists who do, generally claim that psychological essentialism belongs to System 2, not System 1. I disagree. Their claim is based on the fact that psychological essentialism has a couple of characteristic features of System 2, such as rule-based reasoning and uniqueness to human beings. However, it 40 John Dupr´e (1993) also describes this aspect of the essentialist view of natural kinds, an aspect of which he is quite critical. He says that supposition of essence is a “promissory note” for the existence of lawlike correlations among instances of a kind:  . . . what makes a kind explanatorily useful is that its instances share the same properties or dispositions and are susceptible to the same forces. But since we have no way of deciding how much such concomitance to expect in any particular kind, the discovery of a kind adds little, if anything, to the discovery of whatever correlations may turn out to characterize it. An essence can be seen a promissory note on the existence of such correlations. (p. 80)  136  turns out that some of those features are not found exclusively in System 2. Moreover, psychological essentialism is not rule-based reasoning; this view stems from psychologists missing an important distinction regarding rule-following. Psychological essentialism also has several important——indeed central——features of System 1. Therefore, I argue that psychological essentialism may well be a component of System 1. Of course, it is not always an easy job to decide which system a particular processing belongs to. I believe, however, that the discussion in this section will be good enough to give the Dual-Process theorists serious motivation to reconsider their classification of psychological essentialism. This section consists of two parts. In the first part, I survey considerations given by Dual-Process theorists for the idea that psychological essentialism belongs to System 2. Then, in the second part (from p. 139), I argue that those considerations are not firm enough to reach a final conclusion, and that psychological essentialism has several features essential to System 1. For those who want to skip this section, please go directly to the summary section of this chapter on p. 145. Alleged System 2 Characteristics of Psychological Essentialism Some Dual-Process theorists argue that psychological essentialism plays a role in System 2, rather than System 1 (Sloman 1996, 2002, Smith & DeCoster 2000). Their position stems from the idea that the key difference between the two systems lies in the distinction between a similarity- or association-based and a rule-based reasoning, and the view that essentialistic categorization represents a rule-based reasoning (see p. 126 for a brief description of Sloman’s characterization of two systems). Since Smith & DeCoster (2000) do not discuss this issue extensively, I will mainly focus on Sloman’s characterization of psychological essentialism in the rest of this section. Rule-Based Reasoning  Sloman draws our attention to the fact that essentialists  hold certain beliefs about the regularities in nature when engaged in categorization. When children think that removing inside stuff (blood and bones, for instance) from an animal changes its functioning substantially, but removing the outside stuff (e.g., fur) does not, they seem to make an assumption about how the world works,  137  namely, that the stuff inside an organism plays a causal role for its behaviors. In other words, essentialists seem to have some system of beliefs or rules, which they employ in categorization; therefore, they argue, essentialistic classification is a theory- or rule-based categorization. In contrast, similarity-based categorization, as exhibited in psychology experiments, does not involve rule-based thinking. If one categorizes an organism only on its superficial similarity to a prototype of a category, then a leaf and leaf-insect would be in the same category. Being Uniquely Human Susan Gelman (2003) points out another feature of psychological essentialism in favor of this position, although, incidentally, she does not offer it to support Sloman’s view (she does not discuss Dual-Process theory). Gelman discusses the origin of psychological essentialism and suspects that other animals do not have essentialistic attitudes. She cites Michael Tomasello’s study on tool-use (Tomasello 1999). In this experiment, human children and chimps use a rakelike tool to get a treat from a cage. The trick is that one needs to hold the tool in a certain way to track it successfully, which requires a grasp of the underlying causal mechanism in the situation. Tomasello found that human children, after some trials, figure out the right way to hold it, whereas chimps keep holding the tool the wrong way and thus fail to get the treat. From this experiment, Gelman suggests that chimps lack one component of psychological essentialism, that is, grasping underlying causes, which leads her to suspect that chimps lack psychological essentialism. She also cites Daniel Povinelli’s view that chimps may not have the capacity to reason about unobservable phenomena (Povinelli 2000).41 If her speculation is right, this may be another reason to believe that psychological essentialism is a System 2 process: some Dual-Process theorists suggest that System 1 processing is not unique to human beings, but rather shared with other animals, while only human beings possess System 2 processing (Stanovich 1999, Evans & Over 1996). For example, Evans & Over (1996) follow Reber (1993) in pointing out that implicit reasoning (System 1) is invariant among individuals, robust (e.g. to neurological damage), and independent to ages and IQ. According to their inference, these facts suggest that System 1 evolved earlier than System 2, 41 However,  Gelman admits that it is too early to conclude that this matter is settled. She thinks we need more empirical evidence to decide the issue (p. 284).  138  and is thus shared with other animals. On the other hand, System 2 is unique to human beings, because System 2 reasoning often involves explicit use of language, and possession of complex and developed language is a unique feature of human beings. Therefore, Gelman’s suggestion that an essentialistic attitude is unique to human beings seems to confirm that psychological essentialism is a System 2 processing. Objections to these Characterizations However, these points should not be taken as the final word on the issue. Implicitness of Psychological Essentialism  First, Sloman (1996, 2002) and oth-  ers seem to overlook an important distinction regarding rules. One possible concern regarding the concept of “rule-based reasoning” is that many reasoning processes in human beings, as well as other animals, conform to some rule or another. For example, when one exhibits belief bias, one’s reasoning can no doubt be described as conforming to some rule (something like: “If you believe a conclusion to be true, take an entire argument to be valid”). At this point, Dual-Process theorists typically reply by appealing to the distinction between consciously following a rule and merely conforming to one (see, for example, Evans & Over 1996). When System 2 is involved, subjects recognize the rules they use, often by verbalizing them; when System 1 is activated, on the other hand, subjects do not have to be aware of the “rules” they are conforming to. For instance, when one uses a Venn diagram to see if a syllogism is valid, one is probably aware of the rules one uses and the way in which one reaches an answer. In contrast, when one merely uses the truth value of a conclusion as a cue, one might not consciously decide to use any rule. One’s reasoning may be described as conforming to a rule, but this does not mean one is aware of that rule. This is a common response from Dual-Process theorists to this concern, and Sloman (1996) follows this too. What he seems to overlook, however, is that when subjects engage in essentialist thinking, they do not seem to do so consciously. Rather, when children employ essentialistic reasoning, they usually do it implicitly. Gelman (2003) suggests this point (without mentioning any implication re-  139  garding Dual-Process theory). After citing an epigram from Stephen Jay Gould (“Classifications are theories about the basis of natural order, not dull catalogs complied only to avoid chaos.” from Gould (1989)), she says, Preschool children are not scientists, but they would (implicitly) agree with Gould’s assessment of categories, at least in certain domains. (p. 26, italics added) On another occasion, she also says that people hold essentialistic beliefs implicitly: . . . people implicitly believe that the causal essence is shared by all and only members of the category. . . . (p. 301, italics added) This characterization of psychological essentialism is shared by Kornblith (1993). In the following quote, he takes issue with the Quinean view on cognitive development that “children’s conceptual categories should correspond to distinctions drawn on the basis of superficial observable properties” (p. 63). He clarifies where the adult and children’s conceptual structures would differ with regards to natural kinds if the Quinean view is right, given that adults exhibit psychological essentialism. On the Quinean view, the child’s conceptual structure is radically different from that of the adult . . . because there is no implicit awareness that it is the unknown internal structure of objects which determines their kind. (p. 68, italics added) In other words, Kornblith says that if the Quinean view is right, then children would have no implicit awareness regarding internal structure, whereas adults would have such an awareness, due to their possession of psychological essentialism. Now Kornblith indicates that when one adopts the essentialistic attitude, one does so implicitly. When one employs psychological essentialism, one merely conforms to the rules rather than following them consciously. Psychological essentialism is thus not best described as a rule-based system. Even according to Sloman’s criterion, psychological essentialism is not taken to be a rule-based reasoning. Uniqueness to Humans is not Sufficient for a Process to Count as System 2 The second point is concerned with the point that essentialist reasoning may be evolutionary novel. Recall that Gelman suspects psychological essentialism is unique 140  to human beings. One might appeal to this to argue that psychological essentialism should be taken as a System 2 process. However, uniqueness to human beings turns out not to be the strong criterion for System 2 that one would expect; it is not true that all System 1 processes are shared with higher animals. Historically, the distinction is much more blurred. In a series of papers, Evans (2003, 2006a, 2008) discusses this issue. Citing Toates (2006), Evans first points out that higher cognition may not be unique to human beings. Many higher animals have stimulus-bound and higherorder cognitive processes. Examples of stimulus-bound processes are conditioning and instinctively programmed behavior. On the other hand, when the higher-order cognitive processes control organisms they do not merely follow direct stimulusresponse links. An organism in such a state is generally seen to have a representation of its environment. In human beings, these higher-order processes are linked to consciousness, while there is disagreement whether (and how much) this is true of animals. Nonetheless, it is more appropriate to say that this higher cognition is not unique to us, but highly developed in us, Evans says. His second point is that some System 1 processings have a uniquely human component to them. Here, Evans cites Goel (2005). From neuroimaging studies, Goel found that when subjects exhibit belief bias, the prefrontal cortex ——a brain region believed to have evolved in human evolution—— is activated. This is not surprising, because belief bias occurs with the help of an explicit belief system that is not shared with other animals in its fully developed form. Since belief bias is a paradigmatic System 1 phenomenon, this implies that some System 1 phenomena will be observed only in human beings; uniqueness to human beings does not imply that that process will be classified as System 1. A related point is that System 1 is probably not a single system, all components of which have one and the same evolutionary origin, because there are different types of implicit cognitive processes, such as associative learning (evolutionarily old) and modular cognition (evolutionarily recent), and their evolutionary history is probably different. Although Evans follows Mithen (1996) in thinking System 2 evolved recently and uniquely in human beings, this does not mean that no System 1 phenomena are unique to human beings.  141  System 1 Characteristics of Psychological Essentialism It should now be clear that the standard reasons for counting psychological essentialism as System 2 processing are not as strong as some have hoped. In contrast, there are good reasons to believe that the essentialistic attitude is a System 1 cognitive process. Implicitness and Unconsciousness  Some objections to Sloman’s characteriza-  tion can be directly applied to an argument that psychological essentialism has System 1 features. Implicit processing is widely seen as a feature of System 1 processing (see Table 3.1 on p. 85). As I suggested above, Evans & Over (1996) take this distinction between implicit/explicit processing as a central feature of the dual processes. Their emphasis on the distinction is generally accepted by various Dual-Process theorists in the same or slightly different forms, such as unconsciousness/consciousness or unawareness/awareness distinctions (Smith & DeCoster 2000, Kahneman & Frederick 2002, Stanovich 1999, Stanovich & West 2000), including Sloman himself (Sloman 1996).42 Implicitness or unconsciousness of psychological essentialism is also suggested by the fact that young children exhibit the essentialistic framework. It is unlikely that such small children use the framework as explicitly or consciously as one uses a Venn diagram to judge the validity of a syllogism. Being a Heuristic Another System 1 feature of psychological essentialism is being a heuristic. Gelman (2003) describes the essentialistic attitude as a kind of heuristic in that it is a generally (but not always) true assumption about the world and guides our attention to particular aspects of the world: . . . for biological species, essences are heuristically useful much of the time. The basic idea that underlying features cause clusters of surface features seems to be quite right with respect to animal and plant kinds. In this sense, essentialism is akin to statistical reasoning biases. . . : generally useful but not fully right. (p. 302) 42 However,  Sloman notes that awareness is taken to be neither a necessary or sufficient condition to identify systems (p. 6).  142  Essentialism is a reasoning heuristic that allows us to make fairly good predictions much of the time, but it should not be confused with the structure of reality. (p. 324) Being a heuristic is also widely accepted as a System 1 feature, although it may not be as strongly emphasized as implicitness or unconsciousness. Chaiken (1980), Chen & Chaiken (1999), Evans (2006b) name System 1, the heuristic process. Their use of ‘heuristic’ is more or less similar to Gelman’s. For instance, Chaiken (1980) says that when one is in a heuristic process, one uses more accessible and generally reliable cues, such as the source’s identity to decide to accept some opinion. Stanovich (1999), Stanovich & West (2000), Smith & DeCoster (2000), Kahneman & Frederick (2002) also characterize System 1 as “heuristic.” Stereotypes and Essentialism Social psychologists have found that essentialistic beliefs reinforce one’s social stereotypes. Applying Medin’s insights in cognitive psychology (Medin 1989) to social psychology, Yzerbyt et al. (1997) discuss this linkage. The conventional view is that social stereotypes are merely a label with a list of features attributed to a social group, like a race or a gender. Yzerbyt et al. (1997) propose the view that a stereotype contains rich content and plays an explanatory role for a social group or its members. When people build a stereotype, they not only receive information on what a certain social group is, but also make a “gestalt” for that group——a coherent scenario or story with rich content to make sense of their behavior. This is where essentialistic ideas come in. The essentialist attitude (they call it “subjective essentialism”) provides a foundation to build a “gestalt” with a causal story, and makes it easier to use stereotypes for explanatory purposes (Keller 2005). If one believes that members of a race X have “inherent X-ness” (e.g., biological characteristics —— like DNA sequences) that make them what they are, it will make one’s stereotype-explanation for what X’s do look stronger and more plausible. “[S]tereotypes allow people to provide an account for why things are the way they are. . . . rationalization is best served by an essentialistic approach to social categories” (Yzerbyt et al. 1997, p. 39). Many Dual-Process theorists in social psychology categorize the use of stereotypes in System 1 (Table 3.1 on p. 85). Among them are Epstein (1994), Smith 143  & DeCoster (2000), and Sloman (1996). Evans (2008) cites studies showing that stereotype-based reasoning works in implicit memory and takes a pessimistic view on whether conscious thinking can overcome implicit racial and social stereotypes. If the essentialistic attitude is as strongly linked to stereotypes as we have seen, and there is a good reason to assign stereotypical thinking to System 1, it would be natural, pace Sloman and others, to assign the essentialistic attitude to System 1, as well.  3.6.3  Summary  In the last section I addressed one way in which biologists reason about species when they may activate System 1 reasoning —— i.e., prototype reasoning with good species. In this section I continued this project and explored another way in which biologists reason about species with System 1 reasoning: psychological essentialism. Psychological essentialism is the psychological disposition to assume that the relevant object has an essence even when one is not aware of what the essence is. In the case of species, if one adopts psychological essentialism, she would tend to assume that species, as a single natural kind, has essential properties causally responsible for the various phenomena (such as phenotypical cohesion in a single species) even when she is unaware of what those essential properties are. After briefly reviewing the notion of psychological essentialism and its supporting experiments, we saw that psychological essentialism prevails in science and in particular biology and taxonomy, whether it is in the history or the current states of those sciences. Lastly, I discussed the issue of which system psychological essentialism belongs to in the Dual-Process framework. Whereas some psychologists, such as Steven Sloman (1996, 2002), claim that psychological essentialism belongs to Systems 2 reasoning, I made a point that psychological essentialism actually belongs to System 1 reasoning. I argued that Sloman’s arguments are not as strong as he hopes, and that we have good reasons to believe that psychological essentialism has a number of features attributed to System 1 reasoning, such as implicitness and unconsciousness. In the last two sections I have explored how the Dual-Process theory is applied to biologists’ reasoning about species. I pointed to two elements of System  144  1 reasoning: good species as a prototype of species and psychological essentialism. Then I described how biologists use these elements in their reasoning about species. This also indicated how System 1 reasoning works in biologists’ minds and explained some puzzling phenomenon. In the next section, I will summarize the results and draw a coherent picture of System 1 and 2 processing in biologists’ reasoning about species.  3.7  Dual Processes in Inferences Concerning Species  In this chapter, I have discussed possible elements of System 1 processing in biologists’ thinking involving species. I began with Hey’s observation (elusive transparency) and then discussed Luckow and McDade’s observations. I then argued that good species and psychological essentialism are components of System 1 processing in biologists’ reasoning on species. In this section, I attempt to sketch a coherent picture of System 1 and 2 processing in biologists’ reasoning about species. First, I argue that good species and psychological essentialism can account for elusive transparency, as well as the observations of Luckow and McDade. Then I describe possible System 2 processing regarding species, since my focus has so far concerned System 1 components. One such process is what I will call definition-centred reasoning on species. In this kind of reasoning, people are, in one way or another, engaged in definition: proposing a definition, objecting to it, evaluating it, revising it, comparing it with others, and so on. I describe this processing and argue that this is System 2 reasoning, because it has a number of features which other System 2 reasonings have. This will complete the picture of System 1/System 2 reasonings of biologists involving species.  3.7.1  Good Species, Psychological Essentialism, and Elusive Transparency  Elusive Transparency  Let us begin with elusive transparency. I have suggested  that when the elusive transparency of species confronts biologists, two different modes of understanding species are involved (see section 3.3.1): 1. An unarticulated or implicit mode of understanding (understanding “one sin145  gle common meaning” of ‘species,’ or “what biologists mean by ‘species’ ”) 2. An explicit mode of understanding, as represented by giving a definition of ‘species.’ Good species and psychological essentialism give us good clues to explaining the first mode. I described the way in which biologists reason about species by the help of its prototype, good species (see section 3.5.4). My suggestion was that biologists often represent the species category as a whole by its prototypes and that this “attribute substitution” can account for the observations made by Hey, Luckow, and McDade. When a biologist represents the species category by the prototype (good species), she tends to believe that the term ‘species’ has one common meaning. She would also feel little need to articulate her choice of species definition and engage in the species controversy, because a good species is a species under most species definitions. Now we can summarize this by saying that attribute substitution makes sense of the first mode of understanding species. Kahneman and Frederick provide another piece of evidence that when a biologist reasons about species in the first mode above, she activates System 1 reasoning. They note that one condition for the occurrence of attribute substitution is that the substitution is not intervened by a reflective process (System 2) (p. 54). For example, it is found that one tends to fall prey to the conjunction fallacy when one does not attend to the conjunction rule; subjects tend to commit to the fallacy less when subjects have a chance to be cognizant of relevant statistical rules. There is good reason to suspect that biologists, in Hey’s observation, make the same kind of substitution when engaged in casual conversation (recall the quote from Hey), because they are probably not attending to details about species. Psychological essentialism also provides an explanation for elusive transparency. If one takes an essentialist attitude toward the species category, then one would believe that ‘species’ has a common meaning, because one implicitly believes that the species category is a natural kind with a common essence. But biologists still find it hard to precisely define ‘species,’ because one can take this attitude without detailed knowledge of what that common essence is. These explanations are compatible with each other. Representing a category by a prototype does not have to come together with having an essentialist attitude 146  Table 3.5: Features of System 1 in biologists’ reasoning regarding species (I intentionally leave System 2 features blank in this table)  System 1  System 2  Paying attention to prototype by substitution of the species category by good species (prototypes) Assuming that species category is a natural kind with a simple causal essence (psychological essentialism) Assuming that ‘species’ has a single (succinct) meaning Implicit understanding of species Assuming essence placeholder  about that category, but they are not incompatible with each other either. Representing apple by its prototypical instance does not prevent us from believing that apples have some causal essence in common behind their surface properties. Now it is clear that substitution of a category by its prototype and essentialism can jointly make sense of the relevant observations made by Hey, Luckow, and McDade. Here is a summary table of features of System 1 in biologists’ reasoning regarding species (Table 3.5). Of course, I do not intend to say that this table exhausts every component of System 1; other components will be included as we come to better understand biologists’ reasoning.  3.7.2  System 2 Reasoning: Definition-Centered Reasoning on Species  So far, I have listed components of System 1 reasoning. However, our picture will be incomplete if we do not say what System 2 reasoning biologists engage in with respect to species. Quotes from Darwin and Hey give us a hint. Both contrast System 1 understanding with understanding through giving a definition. Darwin says that there is no definition of species unanimously accepted. Hey notices that biologists often stumble when asked to explicitly state the nature of species. In the analysis of elusive transparency, I also suggested that the second mode of understanding is through definition. I call this definition-centred reasoning about 147  species. This reasoning involves the following activities. In defnition-centred reasoning on species, one • Seeks necessary or sufficient conditions (often causal essence) which all or almost all and only members of a category satisfy for membership. • Considers possible exceptions and counterexamples in the course of seeking conditions or rejecting other definitions. • Fills in specific details of the causal essence which are left open by psychological essentialism. • Evaluates different definitions from different points of view. • Identifies a group of organisms as a species by appealing to a specific definition. To put it simply, this is reasoning about the species category in terms of its definition. Its aim is to find and justify an adequate definition of species. Definition-centred reasoning is at the heart of intellectual activities driving the species controversy. Those activities just listed are, for instance, seen in discussions of the biological species concept. Mayr (1942) and others propose a definition of species appealing to causal essence of the species category (gene flow). Some point out cases where the definition diverges from our intuitive judgments (e.g., Ehrlich & Raven 1969), while people in Mayr’s camp defend the original definition or revise it accordingly (e.g., Mayr 1963, 1992). Meanwhile, other definitions are proposed, and some compare the biological species concept with them and judge which is best (Hull 1997, Mayden 1997). Things are not very different for other species concepts (Wheeler & Meier 2000); in one way or another, all the major species concepts have undergone this process. There are many reasons to take this reasoning as System 2 processing. In cognitive psychology, understanding of a category through definition is in contrast to that through a prototype (Landau 1982). Moreover, in general, definition-centred reasoning has a number of the conventional features of System 2 (see also Table 3.1 on p. 85): • Controlled: In automatic processes, nothing but the results of information processing pop up in one’s mind. This is not true of the definition-centred process. One can control one’s thinking. (It is rarely the case that one knows 148  that a case at hand is a counterexample to the biological species concept, but is unaware of how one reached that conclusion.) • Taking High-effort: It usually takes considerable mental effort to build a definition ——especially those carefully crafted by biologists—— just as it takes concentration to use a Venn-diagram correctly. • Conscious: One is obviously conscious when engaged in the above activitites. For example, to find a counterexample to a definition, one is usually aware of what the definition says. • Explicit: To discuss a definition of species, one needs to spell it out. Landau (1982) notes that people are more likely to represent a category through a definition when they are instructed to give a justification for their categorization than when they are instructed only to categorize an object. • Relatively Slow: One rarely comes up with the right definition instantly. Furthermore, the definition-centred reasoning possesses features which some claim belong to System 2. • Looking for Exceptions: According to Lieberman (2003), System 2 evolved so that it could manage the situations that System 1 does not. Thus, System 2 represents “situations that are exceptions to the rules” (p. 54). The tendency to look at differences among individual situations in System 2 processing is also pointed out by Brewer (1988). • Cultural Learning (Sloman 1996): Many System 2 processings involve cultural knowledge, such as probability theory and class-inclusion logic. This kind of knowledge is transmitted in a culture through verbal communication. Species concepts have this feature: one learns phylogenetic species concepts from other biologists. • Domain-general: One uses the same kind of reasonings not only for species, but for many other domains. These features collectively provide good reason to assign the definition-centred reasoning to System 2. This does not mean that the definition-centred reasoning is 149  Table 3.6: Features of System 1 and System 2 in biologists’ reasoning regarding species  System 1  System 2 Definition-centred reasoning  Paying attention to prototype by substitu-  Paying attention to borderline cases  tion of species with good species (prototype) Assuming that species category is a natural  Allowing the possibility that there are mul-  kind with a simple causal essence (psycho-  tiple definitions of species: Possible plural-  logical essentialism)  ism or disjunctive meaning or no definition  Assuming that ‘species’ has a single (succinct) meaning Implicit understanding of species  Explicit understanding of species (through precise definitions)  Assuming essence placeholder  Filling in placeholder with detailed causal story  the only kind of System 2 reasoning for biologists, but it is certainly one of them. To summarize, I fill in the remaining blanks from Table 3.5 to contrast the two kinds of reasonings, see Table 3.6. Let me discuss here the point that the definition-centred reasoning regarding species allows the possibility that there may be multiple definitions of species, because it is included in Table 3.6, but I have yet to discuss it. The definitioncentred reasoning has this feature because, under discussion, one may abandon psychological essentialism regarding the object in question, as some have done with respect to species. Psychological essentialism assumes that the category has a single causal essence. Definition-centred reasoning may keep this assumption, but it does not have to when it turns out to be difficult to do so. Recall that pluralism is proposed for philosophical concepts such as probability, because it is better to employ different definitions (or interpretations) in different contexts (Gillies 2000). In such a case, the definition-centred reasoning may allow the possibility that there may be multiple adequate definitions of the category, because it may assign priority to having multiple definitions of the category for different contexts over having nothing at all. Of course, one may choose to give up searching for any definition 150  of the category and thereby give up the definition-centred reasoning altogether. If one is convinced that probability is a hopelessly hodgepodge category and thus does not have an essence, then one would not try to find any definition for it. This amounts to abandon (psychological) essentialism for probability. Summary of the Chapter In this chapter, I argued that two psychologically distinct systems are at work in biologists’ reasonings about species. Following the Dual-Process theory in cognitive and social psychology, I argued that one of the systems is unconscious, implicit, fast, automatic, and takes low effort (System 1). The other one is conscious, explicit, slow, controlled, and takes high effort (System 2). After I described the notion of good species, I proposed that the System 1 is activated in biologists’ minds when they use the notion of good species and exhibit the essentialist attitude. This interpretation can make sense of several phenomena about biological thinking. For example, biologists often confront the elusive transparency of species —— they believe they understand the nature of species in everyday contexts, but find it very hard to precisely define if called upon to do so. I suggest that when engaged in the casual mode, as it were, of understanding species, they implicitly substitute the category species with its prototype, i.e., good species. This is why they believe that ‘species’ has a single simple meaning when they talk about it casually. In the last section of the chapter, I proposed definition-centred reasoning regarding species as a possible System 2 processing and that this is at the core of the species controversy. This completes the project of modelling how biologists think about species. In the next chapter, I describe how these insights help us explain the persistence of the species problem.  151  Chapter 4  Answering the Persistence Question 4.1  Introduction  I have argued that sharing a reference would help biologists/naturalists communicate effectively with each other and conduct their business without any solution to the species problem (Chapter 2), that biologists employ two kinds of informationprocessing (System 1 and System 2) when they work on species, and that prototype reasoning involving good species and psychological essentialism are components of System 1 processing (Chapter 3). In this chapter, I will make use of these results to explain the persistence of the species problem. This chapter proceeds as follows. In Chapter 1, I divided the persistence question into several sub-questions. Of those questions, I select four principal questions: (1) Why do biologists believe they need to define ‘species’? (2) Why does no definition command universal support? (6) Why aren’t we similarly bothered by analogous issues, such as the nature of life? (7) How could biologists do their business ——conduct their research—— about species and speciation without a unanimously accepted solution? 152  In this chapter I will answer those questions. For the first question, I appeal to psychological essentialism, as well as practical considerations. Naturalists tend to think that they need a definition of ‘species,’ because they require a classification system, and that in turn motivates them to seek for such a definition. However, psychological essentialism also plays a significant role here. Psychological essentialism leads biologists to believe the species category is a natural kind and that it has some causal essence, even though they don’t know what it really is. This provides biologists with a motivation to seek the essence of species via definition. This sheds light on sub-question (6), because people adopt the essentialistic attitude more strongly toward some kinds of things than others, and groups of living things are among these objects. Continuing, I argue that no (or very few) definitions of ‘species’ will be accepted unanimously in taxonomical community, because biologists have many different interests, but there is little consilience among these interests——a taxon judged to be a species under some interests often will not count as one under others. The vagueness account (see Chapter 1) does not give proper emphasis on the role that divergent interests play in the species problem. That is why this account will not work. For question (7), good species offers us a clue. In Chapter 3, I argued that taxonomists often make attribute substitution on the concept of species when, for example, they write a description paper of a new species; they represent the species category with its prototype, good species, just as subjects tend to represent the category bank teller by its prototype image. Part of my answer to question (7) is that this substitution makes it easier for biologists to work on species when they face a “good” species——a taxon which will be judged to be a species either according to more than one criterion or by the general taxonomic community. For example, when they work on good species and the psychological substitution occurs, they could set aside the complications of the species problem, because the species problem is serious when biologists work on borderline, “bad,” species. In other words, although it is hard to find the nature of species and the right definition of it, biologists do not always face the difficulties concerning species when they study species thanks to the existence of good species. At the end of the chapter, I briefly summarize the entire project and argue that 153  Nonobvious properties  Cause  Boundary intensif.  Innate potential  high  Stability over outward transformations high  high  high  high  high  high  high  high  high  ??  n/a  high  high  high  high  high  high  high  medium  high  high  ??  low  medium  low  n/a  ??  noninherent ??  low  high  noninherent high  low  n/a  low  low  low  low  low  n/a  Inductive potential  Living kinds (ex.: cats) Nonliving natural kinds (ex.: gold) Ethnicities (ex.: Jews) Social kinds (ex.: doctors) Simple artefacts (ex.: pencils) Machines (ex.: computers) Noncoherent (ex.: tchotchkes)  Table 4.1: Evidence for essentialism across different types of concepts. “??” indicates that there is not sufficient evidence for judgment. The first row is for various components of psychological essentialism. Boundary intensitivity is the belief that the extensional boundary between inside and outside the category is clear and crisp——an instance must be either a member of a category X or a nonmenber. There are no borderline cases. See Chapter 3 for characterization of other components. Source: Gelman (2003).  the distinction between System 1 and System 2 in general (see Chapter 3) helps to make sense of the way in which biologists work on species.  4.2  Why Do Biologists Believe They Need To Define ‘Species’?  Naturalists try to define ‘species’ for two long-held reasons. First, they tend to believe that they need a definition for classification. Taxonomists agree that it is important to build a classification system and that species is the basic unit of it. If biologists agree on the “right” definition of ‘species,’ it will no doubt help taxonomists build a classification system, although it may not be necessary, because as we have seen in Chapter 2, Strickland’s committee managed to settle on taxonomic naming rules without including any definition of ‘species’ and the current ones also inherit this feature. This is widely accepted among taxonomists and philosophers, so we need not delve into it any further. 154  The second reason is that biologists take an essentialistic attitude toward species, even though they do not fully understand its essence. As we have seen, when one adopts psychological essentialism about a kind, one believes its instances share some causal essence, even though one is not aware of what it is (essence placeholder; Gelman 2003, Medin & Ortony 1989).1 If one adopts an essentialistic framework, one naturally wants to articulate what the essence is, and giving a definition is the most natural way to do so. Notice that one would probably prefer a simple definition to a complex and disjunctive one (although one would not automatically reject the latter).  4.2.1  Question 6: Why Aren’t We Similarly Bothered By Analogous Issues, Such As the Nature of Life?  It is also worth noting this helps explain why the species problem looks more serious to biologists than other possible problems, such as defining ‘life’ or ‘drought’ which do not bother scientists. This is due to the following fact: people adopt the essentialistic attitude toward a variety of kinds of things, but the one concerning forms of life is stronger than those regarding other objects, such as artifacts, social kinds and other forms of natural kinds (e.g., “doctors” or “drought”; Gelman 2003, see also Table 4.1). Thus, psychological essentialism is domain-specific, at least in application. People are psychological essentialists more strongly toward forms of life than other kinds of thing. Generic phrases illustrate this point. Sentences including generic noun phrases, such as ‘an elephant,’ ‘elephants,’ and ‘the elephant,’ often refer to kinds. For instance, “A tiger is a four-legged animal” is true, even if an individual tiger happens to lose one leg, whereas “All tigers are four-legged animals” is false. From points like this, a psychologist Susan Gelman (2003) believes that the use of generic noun phrases in conversation implicitly conveys an element of essentialistic ideas, since essentialism presupposes the concept kind. The use of generic noun phrases such 1 Note that one might also be philosophically motivated to adopt the essentialistic attitude at the same time (one might call this philosophical essentialism). If a taxonomist is convinced of the truth of Kripkean essentialism about a natural kind, for example, he would not only be psychologically but philosophically motivated to make an essentialistic assumption about the species category. In other words, when one has a philosophical reason to believe that there are necessary and sufficient conditions for being a species, then she would not want to give up the search for them prematurely.  155  as ‘the tiger’ implicitly convey the idea that the tiger is a kind and the essentialistic ideas about it. Gelman also revealed that people use generic phrases more frequently for a particular kind of objects than others. In one experiment where conversation between a mother and her child was analyzed, generics were used more than five times more often for animal kinds than for artifacts: sentences like “Bats are awake all night” are uttered more often than sentences like, “The battery makes the clock go.” This result indicates that the essentialistic tendency one has toward animal kinds is among the strongest ones. Although Gelman did not perform the same experiment on species, if a biologist, among other people, holds a stronger essentialistic attitude toward animal kinds including individual species than to other kinds of objects including artifacts, she may well hold a strong attitude toward the species category as well. This is not only because biologists, among other people, assume that individual organic forms belong to the same category, but also because the species category is the immediate higher category to which individual species belong. Therefore, biologists will feel a stronger motivation to define species than other scientific concepts, and they will take the lack of a unanimously accepted solution to the species problem more seriously. This point is absent in the vagueness account, and that is why it does not answer the persistence question. One might point out that these results may not imply that we have the stronger essentialistic attitude toward species than life, because both species and life are presumably natural and living kinds. However, Gelman provides a reason to believe that we do have the essentialistic attitude towards the species category more strongly than life, although she does not address this concern by herself. Gelman cites several psychologists such as Coley, Medin, Proffitt, Lynch & Atran (1999) in this regard. Coley et al. make an experiment about inductive potential of biological categories across different ranks. In this experiment, the researchers give to the subjects a fact that members of a category at a certain level have a property (for example, “All trout have enzyme X”), and then ask them whether the members of a category at the next higher rank have that property (“If all trout have enzyme X, how likely is it that all fish have enzyme X?”). By asking this question, they try to see how readily one makes an inductive inference from a category at a given rank to another category at the next higher rank. The researchers did this 156  experiment to two different populations: undergrad students in a Western university (Northwestern University) and the Itzaj Mayan adult population in the rainforests of Guatemala. They revealed that the subjects of both populations exhibit the stronger attitude to make induction to a taxon at the specific or generic level than to one at the kingdom level. For instance, “U.S. students readily generalize from brown-backed gray squirrels to gray squirrels (specific), and from gray squirrels to squirrels (generic), but not ... from mammals to animals (kingdom). ... The Itzaj show the same patterns, though they were tested with categories indigenous to their environment (e.g., agouti, spider monkey)” (Gelman 2003, p. 53f.). When U.S. students believe that brown-backed gray squirrel has a property F1 and mammals have another property F2 , they are more willing to believe that gray squirrels have F1 than that animals in general have F2 , and the same pattern is observed in the Mayan people. Thus, as far as inductive potential is concerned, people in two very different populations (university students in the U.S. and the Itzaj people in Guatemala) believe that a taxon at a rank lower than genus has stronger inductive potential than the one at the “higher” level. This provides reason to believe that one may well have an essentialistic attitude more strongly toward the species category than life, because life includes all the taxa at the kingdom level and is even at a higher rank than kingdom in the taxonomic hierarchy. If, as Coley et al. suggest, the essentialistic attitude we take toward a taxon at a certain level gets weaker as the taxon is at a higher rank, then one adopts the essentialistic framework more strongly toward the species category than life, given that she would hold a strong attitude toward the species category as well as individual species taxa. Thus it seems plausible that biologists will feel a stronger motivation to define ‘species’ than ‘life,’ although we do not have conclusive evidence because no experimentation has been carried out on the issue.  4.3  Why Does No Definition Command Universal Support?  This question is an important part of the persistence question. This is illustrated by the fact that attempts to account for the persistence of the species problem almost exclusively focus on this point. If biologists could reach an agreement on some 157  possible solution to the species problem, then no one would raise the persistence question in the first place. I objected to several extant accounts of this issue (see Chapter 1). Now it is time to give my own account. To answer this question, I will propose an argument. This argument assumes that a taxon judged as species under one criterion will often not be a species under another. It concludes that biologists lack a unanimously accepted definition of ‘species,’ because the interests biologists have in it are so diverse. I call this the argument from interest-relativity.  4.3.1  Argument from Interest-Relativity  This argument is basically the same argument as species pluralists propose for their position. It aims to show that no or very few species definitions are accepted by the majority of biologists, because there are different interests that biologists have in species, and different definitions largely reflect different interests. Interests This argument has three premises. The first premise is that (I): Biologists have many different interests in species. Let me explain what I mean by ‘interests.’ Here, an interest is a motivation for species definition(s). For example, Coyne & Orr (2004) claim that the biological species concept has an explanation of the coexistence of different species in a single biota as its goal. Some versions of the phylogenetic species concept (PSCc, Nixon & Wheeler 1990) are motivated by the search for basic unit (or an “atom”) in phylogenetic analysis. Species definitions are supposed to serve this interest(s). There are a variety of interests in species concepts. Some interests are shared by multiple definitions, while others are not. The biological species concept and the recognition species concept (RSC; Paterson 1985) have the explanation of the coexistence of different species in a single habitat as a common interest. According to the recognition species concept, a species is the “most inclusive population of individual biparental organisms which share a common fertilization system” (p. 149, see also Chapter 1).2 The fertilization system is a suite of mechanisms, which 2 Paterson’s  paper is reprinted in Ereshefsky (1992b). Pagination comes from this reprinted ver-  sion.  158  enable individual organisms to mate with each other, such as a perceptual mechanism that helps an organism to recognize potential mates. Since reproduction occurs among those which share a common fertilization system, both definitions may be used to determine whether or not two populations in a given habitat belong to the same species; BSC and RSC share an interest in explaining the coexistence of different species.3 Many species concepts, on the other hand, take applicability to asexual organisms to be desirable, but this is not true of BSC and RSC. Interests may also be theoretical. The phylogenetic species concept (historybased) may be taken as aiming to apply the phylogenetic principle of monophyly to species, as well as higher categories: a species should be monophyletic, just as other higher taxonomic groups are. They could be practical, however. The evolutionary species concept (Simpson 1961, Wiley 1978) is sometimes accused of lacking practicality, because it is hard to see when a taxon satisfies the conditions specified by the concept, in particular, “historical tendency.” Those who reject the evolutionary and other species concepts would require a definition to be easily applicable in research. Thus, few would disagree with (I). So much for the first premise. The second premise is (R): Under some interest In or combination of interests I1 . . . , In , along with factual beliefs B1 , . . . , Bn , biologists erect a set of criteria C1 , . . . ,Cn for any group of organisms to count as a species. When biologists have In or I1 . . . , In in conjunction with B1 , . . . , Bn , they will be moti3 The recognition concept, however, has an interest which the BSC does not have. Paterson believes that the BSC smacks of teleology, especially when Dobzhansky endorses the idea of reinforcement of reproductive isolation. Reproductive isolation occurs when two populations are geographically isolated, but Dobzhansky notes that the isolation thus produced may not be complete if those populations remain merely geographically isolated. Reproductive isolation will be complete as an adaptation, he argues, when two partially isolated populations reencounter after the initial isolation. When the reencounter occurs, individuals from the two populations may be able to mate with each other and have hybrid offsprings, because reproductive isolation may be only partial. However, as the two populations may have occupied different niches during the initial isolation, those hybrids may not be as well adapted to either niche as their parents. If this is the case, the hybrids will be adaptively inferior to organisms from both populations. This means that it will be evolutionarily advantageous for individuals to avoid mating with individuals from the other population. Paterson sees this account of speciation as teleological, because it sounds as if reproductive isolation is formed in order to protect the identity of both “species.” In this sense, Paterson has an interest in excluding any element of teleology from the definition of species. However, not all advocates of the BSC are committed to the idea of reinforcement. See Chandler & Gromko (1989) for an objection to Paterson’s argument.  159  vated to accept only those definitions that best (or nearly best) satisfy C1 , . . . ,Cn . The idea behind this premise is this: a species definition D specifies conditions C1 , . . . ,Cn for a taxon to be an instance of species. The biological species concept, for instance, sets a condition of being reproductively isolated from other such populations. This condition issues from its interest ——explaining the coexistence of different species, for example—— and beliefs about nature. The latter in this case is that lack of gene flow is responsible for separating different species. It is because of this particular combination of interest and belief that the BSC sets reproductive isolation as a condition. The ecological species concept provides another example. Leigh Van Valen, who proposes it (Van Valen 1976), shares the same interest as Mayr ——namely explaining the coexistence of different species in a single habitat—— but does not hold the factual belief that only reproductive isolation is responsible for the coexistence of different species. This is why he rejects the BSC and emphasizes niche differentiation in his definition of ‘species’ (see Chapter 1 for the wording of his definition). This example illustrates that opposing species concepts can be motivated by the same interest. The BSC and the ecological species concept, for example, share the same interest; the difference between them lay in opposing factual beliefs. Non-Consilience The third premise is about non-consilience among criteria: (N): A taxon which satisfies a set of criteria C j1 , . . . ,C jm often does not, or is not expected to, satisfy another set of criteria Ck1 , . . . ,Ckn . (N) describes a well-known empirical fact in evolutionary biology. Take the biological species concept and the phylogenetic species concept (history-based). The latter concept claims a species is the least inclusive monophyletic group.4 A taxon is a species under the former definition if it is reproductively isolated from other such entities. However, it is found that such a taxon is often non-monophyletic (see Chapter 1 for discussion). Thus, such a taxon would not be judged to be a species 4A  monophyletic group is a group composed of an ancestor and all of its descendants.  160  Figure 4.1: This represents the process where one species diverges into two species as time progresses from bottom to top of the figure. ‘SC’ stands for a ‘species criterion.’ (Source: de Queiroz 2005)  under the latter definition; a taxon judged to be a species under one definition will not be so judged under another. This non-consilience is due to gradualism and the contingency of evolution. Contingency, in this case, refers to the fact that there is no fixed order in which the speciation process proceeds. Let us look at Figure 4.1 from de Queiroz (2005). It represents the process whereby one species diverges into two species as time progresses from the bottom to the top of the figure. ‘SC’ stands for a ‘species criterion.’ Eight criteria are represented in the figure, but that is not to say that there are not more or less legitimate criteria. When two populations reach the point where they satisfy some criterion in the figure, they will be judged as distinct species according to that criterion. Imagine ‘SC4’ stands for reproductive isolation. When two populations reach that point, they are reproductively isolated from each other. Once two populations go through all the criteria, they are seen as two different species by all criteria and thereby become “good” species. This diagram illustrates 161  how a borderline species occurs. Before two populations satisfy all the criteria, they are not yet two species according to some criteria: even if two populations reach SC3 (which is, for instance, acquisition of different niches), no supporters of the BSC would classify them as two different species, because they have yet to reach the level of SC4 (reproductive isolation). Likewise, no supporters of the phylogenetic species concept (history-based) would see two populations at SC3 as different species when they have not yet reached SC6 (for example, becoming a monophyletic group). Thus populations reproductively isolated from other groups are at “the reproductively isolated stage” of speciation, but perhaps not individual species simpliciter. de Queiroz takes “borderlines cases” of species to be a situation in which two populations reach some stages, but not all. Since the speciation process proceeds slowly in most cases, there will be many borderline cases in nature. If the speciation does not proceed gradually, (N) may not be true: if two populations satisfy all the criteria at once, then a taxon that satisfies C j1 , . . . ,C jm will probably satisfy Ck1 , . . . ,Ckn ; thus that taxon will instantly be a “good” species. The point of (N) is not merely that a taxon judged to be a species under a definition D1 will often not be judged to be a species under another definition D2 . It is also that non-consilience obtains even among important and shared interests or criteria, such as selecting a basic unit in phylogenetic analysis and explaining the coexistence of species in a single habitat. For instance, occupation of a distinct niche does not necessarily lead to being a monophyletic group or being reproductively isolated. In other words, a species concept serving Im may not serve In , when Im is not identical to In . One thing to note. Some might wonder whether (N) contradicts what I have argued on the notion of good species. On the one hand, according to one meaning of the phrase ‘good species,’ a good species is a taxon which will be judged to be a species according to more than one species criterion, such as reproductive isolation and monophyly. On the other hand, (N) claims that a taxon satisfying one species criterion often fails to satisfy another. But there is no contradiction between them. First, what I have argued about good species does not require that “good” species overwhelm “bad” or borderline species in number. The existence of good species does not deny that there are a number of borderline cases. Secondly, (N) does not imply that borderline species greatly exceed good species in number, either. (N) 162  does claim that there are a number of borderline cases, but it says little about what proportion of the entire species category borderline species account for. Although I shall argue shortly that biologists can set aside the complications of the species problem when they face good species (see section 4.4), and this explanation would be weakened if borderline species greatly exceed good species in number (because this means that biologists could hardly set aside the complexities of the species problem in their research) (N) does not imply that. Non-Existence of a Unanimously Accepted Definition From these premises, we can infer: (U): Very few, or no species concepts will be accepted unanimously. From (I) and (R), it follows that only those definitions that will do best under most interests will be accepted by different strains of biologists unanimously. However, (N) implies that there are very few or no definitions which satisfy all of the conditions derived from interests. Therefore, few to no species definitions will be accepted unanimously. It is worth noting that the denial of each premise is close to a necessary condition for the possibility of having an unanimously accepted definition. If (I) does not hold, a definition which has the highest marks under an interest In may be accepted unanimously, because In would be the only interest biologists have. If (R) is not true ——that is, if biologists accept those definitions that fail to satisfy conditions Ck1 —— there may be a unanimously accepted definition: even though a definition D gets lower grades than others under an interest In , biologists may see it as a legitimate enough definition. If there is consilience among conditions C1 ,C2 , . . . ,Cn , ((N) is not true), then some definitions may satisfy all of them.5 5 Some  may wonder whether the argument from interest-relativity works if a biologist has more than one interest (this amounts to revising (R)). If a biologist has more than one interest (at one time or different interests at different times), this (with (I) and (N)) implies that she would be a species pluralist; she would support different definitions of species at one and the same time or over different times. But this is not incompatible with the conclusion of the argument from interest-relativity, which is that there is no single definition of ‘species’ universally accepted in the biological community.  163  4.3.2  Argument from Interest-Relativity and the Vagueness Account  In Chapter 1, we saw that many thinkers adopt what I call the Vagueness Account as an answer to the persistence question. According to this account, the species problem persists because there is a gap between the concept species and vagueness found in the organic world. The argument from interest-relativity makes it clear that this account overlooks important components in the persistence question. Vagueness Does Not Imply the Lack of a Universally Accepted Definition of ‘Species’ In the first place, vagueness regarding species classification does not imply that one has a problem with the definition of species in the way biologists do now. Even if biologists were to reach consensus on which species definition to use and thus solve the species problem in its present form, there would still be some vagueness regarding species classification. Suppose that biologists come to favor the biological species concept over all others at least for sexual organisms. Nevertheless, application of the BSC would exhibit some vagueness. For instance, when two populations are partially isolated ——if 30% or 50% of the populations cannot reproduce with each other—— are they reproductively isolated? This vagueness may cause biologists to disagree on how to apply the BSC, but this has nothing to do with disagreement on which definition to apply. This example shows that the existence of the gap between the essentialistic feature of the notion of species and gradual evolution does not imply the lack of a universally accepted definition of ‘species.’ This indicates that the vagueness account does not capture the point of the persistence question. True, it is still possible to interpret the above argument (the argument from interest-relativity) in terms of vagueness: a taxon x is a species when it satisfies all the criteria and is not so when it satisfies no criterion. The more criteria it satisfies, the more specieshood it will have, but there is no clear demarcation between a non-species taxon and a species taxon. In this sense, vagueness may well be necessary for the lack of a unanimously accepted definition of ‘species.’ However, this formulation does not capture interest-relativity: each set of criteria is provided from a certain point of view (or a certain interest), and that interest motivates biologists to propose and defend a definition of ‘species.’  164  In other words, vagueness could come from many sources and interest relativity is merely one of them. Vagueness could also come from other sources, including application of a definition, as we have seen in the above example. The vagueness account does not distinguish those sources. This is why the vagueness account does not capture the heart of the species problem.  4.3.3  Argument from Interest-Relativity and the Species Pluralism  One may notice that the argument from interest-relativity implies that pluralism is a viable position. Indeed several notable philosophers of biology support this position by an argument similar to this (Kitcher 1984, Ereshefsky 1992a, 1998, 2001, Stanford 1995, Dupr´e 1981, 1993). However, it is not clear how much support pluralism receives from biologists. As far as I know, not many biologists explicitly endorse species pluralism as described above (Dieckmann et al. 2004, is one exception). Compared to other movements in the taxonomic community, such as the one for a new naming code (PhyloCode; see Cantino & de Queiroz 2000, for details), the supporters of species pluralism are not as conspicuous. Then one might ask why species pluralism does not enjoy as much support from the biological community as from the philosophical community. We do not know enough to be able to answer this question fully now, but we can point to some possible factors from the findings we have seen. Psychological essentialism is one of those factors. When one adopts psychological essentialism toward a certain kind of object, then she would tend to apply the essentialistic framework to them, such as believing that there is some hidden “essence” behind surface phenomena (see Chapter 3 for details). If a taxonomist applies this to the species category, then it would discourage her from adopting species pluralism; she would still believe that there is a hidden causal factor responsible for a variety of phenomena which partly makes biologists use different species definitions. In other words, psychological essentialism leads biologists to assume that the species category is a single natural category and there is a unique single causal factor behind it, even though we have not yet fully defined it. This aspect of psychological essentialism makes it hard for biologists to endorse species pluralism wholeheartedly, because pluralists say that there are equally legitimate ways to define ‘species.’  165  Prototypical thinking about species also does not fit well with pluralism. When the attribute substitution occurs, one would tend to represent the species category with its prototype, good species. This makes a biologist believe that the species category is more homogeneous than it actually is, because a good species would be more homogeneous than other “borderline” species when a good species satisfies multiple species criteria.6 Then biologists won’t feel motivated to introduce pluralism. Psychological essentialism and prototypical thinking lead one to believe that there is one and the same kind of entity called ‘species’ in the nature. If one holds this belief, pluralism is not a preferable option. Even facing the difficulties in finding the “right” definition of ‘species,’ she would still believe that biologists should pursue it. Although this is not to say that they are all the factors which discourage biologists to adopt pluralism, I believe they may well be behind biologists’ minds if they are hesitant to adopt species pluralism.  4.4  Study of Species and Speciation Without a Unanimously Accepted Solution  Now I will turn to the last sub-question concerning the persistence question: how having two systems sheds light on another question in this project——how biologists do their business with species, despite lacking a unanimously accepted solution to the species problem. I already addressed this question in Chapter 2. I offered an explanation of how biologists do not fall prey to communication breakdown even when they do not have the same conceptions or definitions of species. Namely, biologists of various camps generally agree on what individual species names refer to, and this enables them to restrict their disagreements only to empirical matters. We also saw that naturalists have invented a variety of methods to ensure that species names and the term ‘species’ itself refers to the same groups of organisms among different schools of biologists. In this section I will offer another explanation based on the 6 Hilary Kornblith (1993) cites a finding made by cognitive psychologists that human beings are particularly keen observers of multiple interrelational associations among objects. Humans tend more quickly to learn how to tell one category from another when two categories are different in multiple properties, than a single property. See also footnote 29 in Chapter 3 (p. 111).  166  results we observed in Chapter 3.  4.4.1  Substitution and the Species Problem  In Chapter 3, I claimed that biologists often make what psychologists call attribute substitution when they work on the notion of species: they tend to represent species with its prototype (good species)7 and infer the attributes a species could have (target attributes) from those a good species has (heuristic attributes), just as subjects tend to represent some category (bank teller) with its prototypical image and infer the attributes a bank teller has from those a prototypical bank teller may have. When this substitution occurs, the concept of species looks sufficiently clear in their minds and they do not pay serious attention to the difficulties in defining the concept. The idea here is that as long as biologists do not come across any problem in studying or classifying species, they would neither have any motivation to abandon this substitution, nor be bothered by the species problem, because the nature of species looks sufficiently transparent to them. Take describing a new species as an example. It is observed that many taxonomists fail to specify their choice of species definition when they describe a new species or revise an extant classification (Luckow 1995, see also Chapter 3). I suspect that in many such cases taxonomists substitute species in general by its prototype, good species, and feel no need to spell out their idea on the nature of species or even have no specific definition in mind. If a taxon which they describe as a new species is so clearly demarcated from others that taxonomists are allowed to hold the substitution, they would then go about their business without being bothered by differences among various species definitions. However, this is not always the case. The new taxon may turn out not to be a prototypical species. That is, one may come across borderline cases. Recall that sibling species play an important role when Mayr argues in favor of the BSC over the morphological species concept (Mayr 1942). Two phenotypically very similar, but distinctive reproductive communities are not prototypical species to those holding a morphological concept of species. Then one should think about 7 I think that a biologist substitutes the species category either by the concept or category of good species, or individual good species, such as Drosophila melanogaster. See footnote 38 in Chapter 3 (p. 125) for discussion.  167  the nature of species and which concept of species to adopt. However, sibling species are an exception, not the rule. In many cases, taxonomists do not face such a situation. Then they do not have to give up the substitution of the species category by its prototype and keep doing their business. Let me note one thing. I suspect that taxonomists often commit attribute substitution when, for instance, they describe a new species. This description is true of many taxonomists, but not all biologists, especially species theorists engaging in the species controversy. Biologists such as Ernst Mayr would be sensitive to species definitions, especially when engaged in the species controversy. Recall that Luckow’s observations are about taxonomists in general, not participants in the species controversy. As we have seen, those participants are especially sensitive to competing definitions of species. But where average taxonomists are concerned, they undertake their research relatively unbothered by the species controversy. Good Species Overcome Non-Consilience One can make a similar point with the argument from interest-relativity. One premise of this argument is non-consilience (N): A taxon which satisfies a set of criteria C j1 , . . . ,C jm often does not, or is not expected to, satisfy another set of criteria Ck1 , . . . ,Ckn . We can argue that the idea of good species makes communication between biologists feasible by overcoming non-consilience among species definitions. According to one usage of ‘good species,’ a good species is nearly neutral with respect to any interest that a biologist may happen to have, in that a good speices will be judged to be a species according to many definitions of species.8 Each definition of species is specified by a cluster of criteria C1 , ...,Cn , which is based on a combination of interests and factual beliefs. Since a good species satisfies most of those clusters and each cluster is based on research interest(s) and factual beliefs, a good species will be judged to be a species under most interests. Thus, when a biologist finds a new form and it turns out to be a good species, either (i) she might not be particularly motivated to use a concept fine-tuned to 8 One usage of ‘good species’ does refer to a taxon judged to be an instance of species by a single species criterion. But I rule out this usage for further consideration in the current study (Chapter 3, section 4.3). Here ‘good species’ refers to a taxon judged to be a species by more than one species criteria or by the taxonomic community.  168  an interest at hand, in the first place, or (ii) even when different biologists have different interests and thus adopt different definitions, they can share what that species refers to, i.e., its extension. In the first case, if a biologist does not adopt a species definition particularly fine-tuned to an interest, but is still to use the notion of species, then she would employ a somewhat unarticulated notion of species. We have already seen, in Chapter 3, that this kind of general notion of species is akin to good species, a prototype of the notion species. In such a case, one does tend to substitute the concept species with its prototype and can communicate with other biologists even with an unarticulated notion of species, due to the fact that a taxon in question is a species according to different definitions. In the second case, the attribute substitution does not occur, and biologists have a clear sense of what they mean when they use the term ‘species’: they can explain what they mean by ‘species’ when they are asked to articulate it, like “Oh, I use ‘species’ as Mayr would.” Thus, when two biologists talk about species, it might be the case that two of them use ‘species’ with different meanings —— Tom may follow Mayr but George may use the term in the way in which it implies that a species is a minimum monophyletic group. However, even if such is the case, the extension of the term ‘species’ is shared by biologists with different definitions, because a good species is neutral to one’s choice of species definition. As a result, possible differences in the meaning of ‘species’ may not prevent them from having effective communication. Therefore, in both cases, as long as biologists deal with good species, interest-relativity does not prevent them from communicating with each other. This is expressed in terms of the argument from interest-relativity. Recall that interest-relativity ((I) and (R)) combined with non-consilience (N) imply that no unanimously acceptable definition is yet available. And if none of the definitions are unanimously accepted and many taxa are judged differently under different definitions, one might doubt that biologists could do their business. For biologists have trouble successfully communicating when talking about different groups under the term, ‘species.’ Suppose that Tom and George happen to study the same genus and Tom follows Mayr (BSC) and George uses the term ‘species’ to refer to the least inclusive monophyletic group (PSCh). If a reproductively isolated group does not overlap the least inclusive monophyletic group to a significant de169  gree, then Tom’s “species” (a reproductively isolated group) may not be George’s “species” (a minimum monophyletic group) . This could be a source of confusion and communication breakdown. However, the concept of good species brings about partial consilience among relevant parties. In a good species, groups judged to be a species by different criteria coincide or overlap considerably. Tom’s “species” happens to be George’s “species” and both of them refer to the same group by the term ‘species.’ In such a case, whatever interest or whichever definition a biologist happens to have, the same species name denotes the same (or nearly the same) group of organisms. We have seen in Chapter 2 that this coincidence greatly helps naturalists endorsing different conceptions of species communicate sucessfully. In the argument from interest-relativity, this amounts to denial of (N): a taxon judged to be a species by a cluster of criteria will be judged so by another cluster of criteria. This means that biologists may have different interests and thereby adopt different definitions of species, but those different definitions do not make much difference in terms of their references——they may simply be different ways of denoting the same group of organisms. And we have seen that this reference-sharing makes it possible that the biological community denotes the same taxon even when different members of the community endorse different definition of ‘species.’ Thus we can connect insights from our psychological analysis of good species to the results from the argument we have put forward in this chapter, the argument from interest-relativity.  4.4.2  Reference-Sharing and Prototypical Reasoning  I proposed two answers to the same sub-question (7): How could biologists conduct their research about species without a unanimously accepted solution? In Chapter 2, I argued that the fact that a species name refers to the same group of organisms, despite having different connotations to different biologists, helps different camps of biologists avoid possible communication breakdown, and that naturalists have made a considerable effort to ensure the stability of reference in the history of biology. In Chapter 3, I described how prototypical reasoning about the notion of species is activated in biologists’ minds. Based on this result, I argued in this chapter that this helps biologists conduct their research without being  170  bothered by the complexities of the species problem. Then some might wonder whether there is any relation between the two answers I proposed, because they are the answers to the same question after all. I am inclined to think that biologists’ reliance on reference-sharing and their prototype reasoning regarding species has a weak connection. First, if Xus bus is a good species, then it is likely that the reference of the species name ‘Xus bus’ is fairly stable in the biological community. Recall that the phrase ‘good species’ has three meanings and we focus on two of them in this project (Chapter 3, p. 110): a good species is either (i) a taxon which will be judged to be a species according to more than one species criterion or (ii) a taxon which will be generally judged to be a species by the taxonomic community. When a taxon is a good species in either sense, its species name ‘Xus bus’ successfully would refer to the same group of organisms even when different biologists have different conceptions about that group or the species category in general. When a taxon is a good species in the first sense, it will be judged to be a species according to many species criteria. This raises the probability that biologists who have different conceptions on the taxon or the nature of species would view the taxon as a species, which means that the same species name would probably refer to the same group of organism as a species. If a taxon is a good species in the second sense, the species name ‘Xus bus’ would also have the same reference among biologists, because there would be an agreement in the biological community that Xus bus (a taxon, not a name of the taxon) is a species, and thus it is natural to assume that there is not a serious disagreement on the extension of the species name in question. Of course, this argument does not mean that every taxonomist would agree whether or not a given organism belongs to Xus bus if it is a good species. Yet we can still conclude that if Xus bus is a good species, then it is likely that the reference of the species name ‘Xus bus’ is fairly stable in the biological community. Nevertheless, the two proposals are largely independent as answers to the above question (7). Although both proposals are answers to the same question, a close look at them reveals that they aim to explain different phenomena. My point in Chapter 2 is that biologists do not frequently experience communication breakdown regarding species (either as the species category or individual species) because the term ‘species’ or a species name refer to the same object across different 171  biologists. In other words, reference-sharing resolves the concern about possible incommensurability between biologists of different camps. On the other hand, as we saw in the last chapter, prototype reasoning is evoked to account for elusive transparency, i.e., the fact that biologists often appear to ignore the complexities of the species problem and behave as if there is little problem about the notion of species. Reference-sharing and prototype reasoning are proposed to explain different phenomena. A striking difference between the two phenomena is that reference-sharing could occur when biologists explicitly reason about the species problem whereas elusive transparency occurs when biologists implicitly reason about species or even ignore the complexities of the species problem. The apple race of Rhagoletis pomonella nicely illustrates this point (Chapter 2). In the debate regarding this race, both camps of biologists ——Coyne & Orr on the one hand, and Bush on the other hand—— agree that there is an interesting issue and disagreement, and that the debate is relavant to the species problem in general. This is a quite different attitude from those we have seen in the cases of elusive transparency (Chapter 3): when biologists fall prey to elusive transparency, they tend to overlook the complexities of the notion of species rather than deal with them. Therefore we can say that reference-sharing could occur whether one engages either in an explicit or implicit thought process, while elusive transparency is a phenomenon which could occur even when a biologist implicitly ignores the complexities of the species problem. Thus reference-sharing and prototype reasoning are largely independent accounts for different phenomena, even though both accounts could be applied to the same taxon, because if Xus bus is a good species, then it is likely that the reference of the species name ‘Xus bus’ is fairly stable in the biological community.  4.5  Conclusions and Summary  In this thesis, I have explored why the species problem has persisted for such a long time. Since this question is multi-faceted, and I have divided it into different sub-questions to address it, it may be helpful to now see the big picture. In this section, I summarize the accounts I have offered so far. In Chapter 1, after I outlined the current state of the species problem, I for-  172  mulated what I call the persistence question in terms of consensus on the “right” solution to the species problem: The Persistence Question of the Species Problem: Although biologists have tried long and hard to reach rational agreement on the species problem (on the nature or the “right” definition of species), they have not reached consensus and will not anytime soon. Why is that? I overviewed possible forms in which the species problem is “resolved” ——for example, biologists may agree on one species definition and decide to give up others, or most biologists may admit that there is no “right” definition of ‘species,’ and they give up using the notion ‘species’—— and observed that none of them is realized in the biological community. Since a variety of factors are at work behind the persistence of the species problem, I divided this question into several sub-questions. I focused on the following four in this project:  (1) Why do biologists believe they need to define ‘species’? (2) Why does no definition command universal support? (6) Why aren’t we similarly bothered by analogous issues, such as the nature of life? (7) How could biologists do their business ——conduct their research—— concerning the origin, evolution, and ecological combination of species without a unanimously accepted solution? In the rest of Chapter 1, I reviewed various answers to the persistence question which are proposed in literature. One notable answer to the persistence question is what I call the vagueness account, according to which the species problem persists because there is a fundamental gap between the essentialistic feature which the notion of species requires of individual species ——that is, the species boundaries should be so sharp that a taxonomist could classify every organism into one species or another—— and the gradual nature of biological evolution, which implies the existence of many borderline cases in species classification. I argued that this account does not answer fully the persistence question. I made two points. First, the vagueness account does not explain the saliency of the species 173  Questions  (1)  Explanations • Naturalists have a practical need for classification • Naturalists take the essentialistic attitude toward living kinds (psychological essentialism): they (tacitly) believe that the species category has a causal essence.  (2) (6) (7)  • Argument from interest-relativity • Psychological essentialism works more strongly toward living kinds. • Good species and attribute substitution save biologists from going into the species controversy. • Sharing references makes it possible for biologists with different ideas on the nature of species to avoid communication breakdown Table 4.2: Answers to sub-questions  problem over other apparently similar problems. Objects which have apparently vague boundaries abound even in the biological world. Yet the species problem is far more salient in the biological community than other such problems, such as the concept of life. Mere existence of vague boundaries does not explain fully the way in which the species is in the current form. Secondly, if the gap between the essentialistic nature of the species concept, on the one hand, and the gradualness of evolution, on the other, was indeed the source of the species problem, then we could properly address the problem merely by inventing a concept whose extensional boundary is intentionally vague, thus capable of reflecting the vagueness found in nature. The problem here is that such accounts ——such as Beckner’s polytypic definition of ‘species,’ and the homeostatic property cluster theory—— have been offered in response to the species problem, but none has attracted much support or attention from biologists. Thus inventing such a definition is not sufficient to solve the species problem. This is not to say that the above gap does not exist or is irrelevant to the persistence of the species problem. Yet there is more than that in the persistence question, and I explore other possible factors contributing to the persistence of the species problem. Factors Contributing to the Persistence of the Species Problem  My plan in  this thesis is that I point out various factors that should be considered in giving answers to these sub-questions above. This is what I did in Chapters 2-4. For the first question, I noted that since biologists have a practical need for classification 174  and species is the basic unit of the classification system, they are motivated to have a definition of ‘species.’ Another factor is that biologists tend to adopt the essentialistic attitude toward living kinds. This disposition to assume a relevant object has essential properties, even when one is not aware of what the essence is (psychological essentialism), is prevalent among species theorists, as well as other scientists, whether or not they have philosophical arguments for the position, such as Kripkean essentialism of natural kinds (Kripke 1980). For example, most of the definitions of ‘species’ on offer appeal to deep or hidden structure, such as gene pool or phylogeny. If one takes this attitude toward living kinds, including species, one naturally assumes the existence of essential properties causally responsible for the superficial properties exhibited by instances of the kind. This motivates biologists to look for the correct description of those essential properties ——the “right” definition of species—— because defining a kind of object is a good way to convey information about its essential properties. Psychological essentialism also offers a clue to explaining why the species problem is more prominent among the scientific community than similar tasks, such as defining life (question (6) above). The essentialist attitude is not equally strong toward all kinds of things; it is domain-specific. Subjects adopt the essentialistic framework toward living kinds (like cats) more strongly than other objects, such as artifacts. Thus a biologist may well hold a strong essentialistic attitude toward the species category as well. This is not only because biologists assume that certain organic forms belong to the same category, but also because the species category is the immediate higher category to which those forms belong. This difference in the strength of psychological essentialism accounts for the relative prominence of the species problem. This point is absent in the vagueness account, and that is why it does not answer the persistence question. For question (2), I put forward the argument from interest-relativity. The argument has three premises: (I) biologists have very different interests in species, (R) given some interest (or combination of interests) along with factual beliefs, biologists erect a set of criteria for any group of organisms to count as a species and accept only the definition(s) that best satisfy those criteria, and (N) a taxon satisfying one species criterion often fails to satisfy another. From these premises, 175  it follows that there will be very few, if any, unanimously accepted definitions of species, because a definition of it under one interest will fail to satisfy criteria provoked by other interests. This argument indicates that the vagueness account misses important aspects of the persistence question, the fact that biologists have different interests in species and the lack of consilence among them brings about the problem. In this sense the vagueness account is incomplete. It is true that if the argument from interest-relativity is sound, then the vagueness between the notion of species and gradual evolution still obtains: if these premises (I), (R), and (N) are true, then the vagueness still obtains. However, the opposite relation does not hold: the existence of the vagueness does not imply that those premises are true. In other words, while vagueness between the essentialistic feature of the notion of species and gradual evolution may come from various sources, the vagueness account does not discriminate different sources of the vagueness: the vagueness may or may not come from a source relevant to the persistence of the species problem, but the account does not offer a way to tell the kind of vagueness relevant to the persistence of the species problem. Moreover, vagueness regarding species classification does not imply that one has a problem with the concept of species in the way biologists do now. Even if biologists reach consensus on which species concept to use and thus the species problem in its present form disappears, there may still be some vagueness regarding species classification, as long as there is vagueness regarding how to apply the species definition, not which concept to apply. We saw this by taking an example from the biological species concept. This is why the gap itself is not enough to account for the impact the species problem has on biologists. As we have seen, biologists have done their business (such as studying speciation and constructing classifications) without a universally accepted solution to the species problem (question (7)). I suggested two factors that make this possible. First, biologists often believe that they agree regarding the reference of ‘species’ and individual species names (e.g., Homo sapiens), and this enables them to avoid communication breakdown. If biologists or naturalists can agree (to a sufficient degree) on which groups of organisms they discuss in terms of extension, then differences in their respective conceptions of ‘species’ and ‘speciation’ ——even if significant—— will not hamper biological discussion. In Chapter 2, I described three cases in which naturalists faced this worry and took the above strategy in dif176  ferent ways. When Charles Darwin faced the incommensurability threat about his theory of evolution, he used ‘species’ to refer to a taxon judged to be a “species” by his fellow naturalists (John Beatty calls this the referential use of the term species), so that any naturalist, regardless of her belief in the immutability of species, could figure out what the reference of ‘species’ (or individual species names) is in his theory. John Edward Grey invented an information-storing system in which information of an individual taxon could be securely stored no matter where the taxon is placed in a classification system. This helps ensure that the same species name could refer to the same taxon regardless of whichever rank it is assigned in a taxonomic system. Jerry Coyne & Allen Orr and Guy Bush radically disagree in their conceptions of species and the priority between the study of speciation and a definition of ‘species,’ but they can disagree on whether a host race of Rhagoletis pomonella is going through sympatric speciation process without falling prey to incommensurability. Their disagreement is empirical and both camps do make sense of their opponent’s claim. One reason for this is that Coyne & Orr and Bush agree on what the name for the host race in question refers to, i.e., what the group of organisms at issue is. This analysis gets along with the idea supported by leading philosophers of science, that is, the one that sharing a reference is a key to overcoming potential incommensurability (Kitcher 1978, 1993). Second, the way in which biologists represent the notion of species is also relevant to their success in preventing the species problem from getting in the way of their business. As we saw in Chapter 3, biologists often use the term ‘good species’ in formal and informal venues. The term has slightly different usages from one context to another, but in many cases ‘good species’ refers to (a) a taxon which will be judged to be a species according to many species criteria, such as reproductive isolation and monophyly, or (b) a taxon which be generally judged to be a species by the biological community (and, the two usages are continuous in that a good species in the sense (a) is likely to be a good species in the sense (b)). Then it turns out that the notion of good species has a number of features of a prototype; we have a reason to believe that good species is a prototype of species in biologists’ representations. If this is the case, then biologists may well make what Kahneman and Frederick call attribute substitution to the species category; biologists tend to represent the species category with its prototype, good species, and 177  infer various attributes of the former from those of the latter, when they talk about species (the species category) in a casual conversation. This explains a variety of apparently puzzling observations about the way in which biologists work on the notion of species, including the observation that in casual conversation, biologists act as though ‘species’ has a common meaning when they should be aware of the fact that there is no universally accepted definition of ‘species’ in the biological community. The idea is that a good species usually looks quite homogeneous in their minds in that it tends to satisfy many species criteria and one can easily tell that it is a species. If biologists represent the species category by its prototype, good species, in a conversation, then ——even though there is no clear and common meaning in the term ‘species’—— the prototype makes it look as if there is such a meaning (because good species as represented in their minds looks quite homogeneous), and a speaker easily believes that she grasps it clearly. This leads biologists to believe that ‘species’ has one common meaning and to have effective communication about species when there are many species definitions on offer in the biological community. Division between System 1- and System 2-Reasonings In Chapter 3, I adopted the framework of the Dual-Process theory to explore the mental machinery biologists employ in the study of species. However, I have not utilized the division between Systems 1 and 2 explicitly to answer the persistence question. I briefly describe the way in which adopting the framework of the Dual-Process theory as a whole is more helpful than using only some components of it, such as theories on prototypes and psychological essentialism, in order to answer the persistence question. There are two reasons why we should adopt the entire framework of the DualProcess theory. First, we have seen several aspects of the way in which biologists work on species——good species, psychological essentialism, elusive transparency, and definition-centred reasoning. We have also found that some aspects nicely mesh with each other, with the help of conceptual resources from the DualProcess theory. Take elusive transparency. As we have seen in Chapter 3, Jody Hey (2001a) observes that biologists often find themselves casually using the word ‘species’ in conversation with colleagues as if they fully understand its common 178  meaning, despite the fact that they know all the difficulties that have attended every attempt to define the notion. Here biologists give conflicting responses to the nature of species, which indicates that different processes are working in their minds. Good species and psychological essentialism, both of which are componenets of System 1, explain how elusive transparency of the nature of species occurs——for example, when a biologist holds an essentialistic framework toward the species category, then this framework will strengthen her assumption that species is a single coherent category in that there is a causal essence for observable phenomena about species. This in turn leads her to the belief that the term ‘species’ has “one common meaning.” The idea of good species explains other observations relevant to elusive transparency, such as that taxonomists are often not involved in the species debate when it is natural to think they would be——if a taxon at hand will be judged to be a species according to many species criteria, then a taxonomist will be less motivated to be involved in the debate, say, on which species definition is the right one. This satisfying coherence would be lost if one neglects the Dual-Process theory as an explanatory framework. Second, the DPT tells us what, among various kinds of reasoning employed by biologists on species, species theorists have somehow ignored. The species theorists’ focus has been, by and large, given to various definitions of ‘species.’ For example, the vagueness account seems to put an emphasis on definitions, because a common assumption behind many definitions of species is that a definition divides a species and a non-species crisply. This focus corresponds to System 2 reasoning on species (definition-centred reasoning on species). On the other hand, components of System 1 have not attracted as much attention in the philosophical study of the species problem. This contrast indicates that a particular kind of reasoning regarding species has been somehow neglected by many species theorists. Although I have tried to point out some such aspects in this project, I do not believe that this exhausts the the sort of reasoning biologists exercise with respect to species. Adopting the DPT as a whole sheds light on what we could do to understand how biologists deal with the notion of species and the species problem.  179  Bibliography Amundson, R. (2005), The Changing Role of the Embryo in Evolutionary Thought: Roots of Evo-Devo, Cambridge University Press, Cambridge. Armstrong, S., Gleitman, L. & Gleitman, H. (1983), ‘What some concepts might not be’, Cognition 13, 263–308. Baum, D. & Donoghue, M. (1995), ‘Choosing among alternative ”phylogenetic” species concepts’, Systematic Botany 20, 560–573. Beatty, J. (1982), ‘What’s in a word? Coming to terms in the Darwinian revolution’, Journal of the History of Biology 15, 215–239. Beatty, J. (1986), Speaking of species: Darwin’s strategy, in D. Kohn, ed., ‘The Darwinian Heritage’, Princeton University Press, Princeton, pp. 265–281. Beckner, M. (1959), The Biological Way of Thought, University of California Press, Berkeley, CA. Beurton, P. (2002), ‘Ernst Mayr through time on the biological species concept: A conceptual analysis’, Theory in Biosciences 121, 81–98. Blackwelder, R. (1967), Taxonomy: A Text and Reference Book, Wiley, New York. Bogdanov, A. (2004), ‘Allozyme variation of the pygmy wood mouse sylvaemus uralensis (rodentia, muridae) and estimation of the divergence of its chromosome forms’, Russian Journal of Genetics 40, 897–909. Boyd, R. (1991), ‘Realism, anti-foundationalis, and the the enthusiansm for natural kinds’, Philosophical Studies 61, 127–48. 180  Boyd, R. (1997), How to be a moral realist, in G. Sayre-McCord, ed., ‘Essays on Moral Realism’, Cornell University Press,, Ithaca, NY, pp. 181–228. Boyd, R. (1999), Homeostasis, species, and higher taxa, in R. Wilson, ed., ‘Species: New Interdisciplinary Essays’, The MIT Press, Cambridge MA, pp. 141–185. Brewer, M. (1988), ‘A dual process model of impression formation’, Advances in social cognition 1, 1–36. Brigandt, I. (2006), A Theory of Conceptual Advance: Explaining Conceptual Change in Evolutionary, Molecular, and Evolutionary Developmental Biology, PhD thesis, University of Pittsburgh. Brigandt, I. (2009), ‘Natural kinds in evolution and systematics: Metaphysical and epistemological considerations’, Acta Biotheoretica 57, 77–97. Brigandt, I. (forthcoming), ‘The epistemic goal of a concept: Accounting for the rationality of semantic change and variation’, Synthese. Bush, G. (1969), ‘Sympatric host race formation and speciation in frugivorous flies of the genus Rhagoletis (Diptera, Tephritidae)’, Evolution 23, 237–251. Bush, G. (1975), ‘Modes of animal speciation’, Annual Review of Ecology and Systematics 6, 339–364. Bush, G. (1992), ‘Host race formation and sympatric speciation in Rhagoletis fruit flies (Diptera: Tephritidae)’, Psyche: A Journal of Entomology 99, 335–357. Bush, G. (1993), A reaffirmation of Santa Rosalia, or why are there so many kinds of small animals?, in D. Lees, D.R.; Edwards, ed., ‘Evolutionary Patterns and Processes’, Academic Press, London, pp. 229–249. Bush, G. (1994), ‘Sympatric speciation in animals: new wine in old bottles’, Trends in Ecology and Evolution 9, 285–288. Bush, G. (1995), ‘Reply from G.L. Bush’, Trends in Ecology and Evolution 10, 38.  181  Bush, G. & Butlin, R. (2004), Sympatric speciation in insects, in M. L. R. Dieckmann, U.; Doebeli, ed., ‘Adaptive Speciation’, Cambridge University Press, Cambridge. Bush, G. & Smith, J. (1998), ‘The genetics and ecology of sympatric speciation: A case study’, Population Ecology 40, 175–187. Bush, G., Feder, J., Berlocher, S., McPheron, B., Smith, D. & Chilcote, C. (1989), ‘Sympatric origins of R. pomonella’, Nature 339, 346–346. Cantino, P. & de Queiroz, K. (2000), ‘PhyloCode: A phylogenetic code of biological nomenclature’, Website. http://www.ohiou.edu/phylocode/. Carruthers, P. (2006), The Architecture of the Mind, Oxford University Press, Oxford. Carson, H. (1989), ‘Sympatric pest’, Nature 338, 304. Chaiken, S. (1980), ‘Heuristic versus systematic information processing and the use of source versus message cues in persuasion’, Journal of Personality and Social Psychology 39, 752–766. Chan, K. & Levin, S. (2005), ‘Leaky prezygotic isolation and porous genomes: Rapid introgression of maternally inherited DNA’, Evolution 59, 720–729. Chandler, C. & Gromko, M. (1989), ‘On the relationship between species concepts and speciation processes’, Systematic Zoology 38, 116–125. Chen, S. & Chaiken, S. (1999), The heuristic-systematic model in its broader context, in S. Chaiken & Y. Trope, eds, ‘Dual-Process Theories in Social Psychology’, Guiford Press, New York, pp. 73–96. Claridge, M., Dawah, H. & Wilson, M., eds (1997), Species — The Units of Biodiversity, Chapman and Hall, London. Coley, J., Medin, D., Proffitt, J., Lynch, E. & Atran, S. (1999), Inductive reasoning in folkbiological thought, in D. Medin & S. Atran, eds, ‘Folkbiology’, The MIT Press, Cambridge, MA, pp. 205–232. 182  Coyne, J. (1994), ‘Ernst Mayr and the origin of species’, Evolution 48, 19–30. Coyne, J. & Orr, H. (2004), Speciation, Sinauer Association Inc, Sunderland MA. Crane, J. (2004), ‘On the metaphysics of species’, Philosophy of Science 71, 156– 173. Cronquist, A. (1978), Once again, what is a species?, in ‘Biosystematics in Agriculture’, Rombergered Allanheld & Osmun, Montclair NJ, pp. 3–20. Darwin, C. (1859), On the Origin of Species by Means of Natural Selection, John Murray, London. Darwin, C. (1887), The Life and Letters of Charles Darwin, Including an Autobiographical Chapter, John Murray, London. Darwin, C. (1987), Charles Darwin’s Natural Selection: Being the Second Part of His Big Species Book Written from 1856 to 1858, Cambridge University Press, Cambridge. de Queiroz, K. (2005), Ernst Mayr and the modern species concept, in J. Hey, W. Fitch & F. Ayala, eds, ‘Systematics And the Origin of Species: On Ernst Mayr’s 100th Anniversary’, National Academic Press, Washington DC, pp. 243– 263. de Queiroz, K. & Donoghue, M. (1988), ‘Phylogenetic systematics and the species problem’, Cladistics 4, 317–38. Dennett, D. (1995), Darwin’s Dangerous Idea: Evolution and the Meanings of Life, Touchstone Book, New York. Devitt, M. (2008), ‘Resurrecting biological essentialism’, Philosophy of Science 75, 344–382. Dieckmann, U., Metz, J., Doebeli, M. & Tautz, D. (2004), Introduction, in U. Dieckmann, M. Doebeli, J. Metz & D. Tautz, eds, ‘Adaptive Speciation’, Cambridge University Press, Cambridge, UK, pp. 1–16.  183  Dobzhansky, T. (1935), ‘A critique of the species concept in biology’, Philosophy of Science 2, 344–55. Dobzhansky, T. (1937), Genetics and the Origin of Species, New York: Columbia University Press. Dr`es, M. & Mallet, J. (2002), ‘Host races in plant-feeding insects and their importance in sympatric speciation’, Philosophical Transactions - Royal Society of London Series B 357, 471–492. Dumont, H. (2004), ‘Distinguishing between the East-Asiatic representatives of paracercion weekers and dumont (Zygoptera: Coenagrionidae)’, Odonatologica 33, 361–370. Dupr´e, J. (1981), ‘Natural kinds and biological taxa’, Philosophical Review 90, 66– 90. Dupr´e, J. (1993), Disorder of Things: Metaphysical Foundations of the Disunity of Science, Harvard University Press, Cambridge MA. Dupr´e, J. (1999), On the impossibility of a monistic account of species, in R. Wilson, ed., ‘Species: New Interdisciplinary Essays’, The MIT Press, Cambridge MA, pp. 3–22. Edelaar, P. (2008), ‘Assortative mating also indicates that common crossbill Loxia curvirostra vocal types are species’, Journal of Avian Biology 39, 9–12. Ehrlich, P. & Raven, P. (1969), ‘Differentiation of populations’, Science 165, 1228– 32. Eldredge, N. & Cracraft, J. (1980), Phylogenetic Patterns and the Evolutionary Precess: Method and theory in comparative biology, Columbia University Press: New York. Endler, J. (1973), ‘Gene flow and population differentiation’, Science 179, 243–50. Epstein, S. (1994), ‘Integration of the cognitive and the psychodynamic unconscious’, American Psychologist 49, 709–724. 184  Ereshefsky, M. (1992a), ‘Eliminative pluralism’, Philosophy of Science 59, 671– 90. Ereshefsky, M. (1998), ‘Species pluralism and anti-realism’, Philosophy of Science 65, 103–120. Ereshefsky, M. (1999), Species and the Linnaean hierarchy, in R. Wilson, ed., ‘Species: New Interdisciplinary Essays’, The MIT Press, Cambridge, MA, pp. 285–305. Ereshefsky, M. (2001), The Poverty of the Linnaean Hierarchy: A Philosophical Study of Biological Taxonomy, Cambridge University Press: Cambridge. Ereshefsky, M. (2010), ‘Microbiology and the species problem’, Biology and Philosophy 25, 553–568. Ereshefsky, M. (forthcoming), ‘What’s wrong with the new biological essentialism’, Philosophy of Science. Ereshefsky, M. & Matthen, M. (2005), ‘Taxonomy, polymorphism, and history: An introduction to population structure theory’, Philosophy of Science 72, 1–21. Ereshefsky, M., ed. (1992b), The Units of Evolution: Essays on the Nature of Species, The MIT Press, Cambridge, MA. Evans, J. (1984), ‘Heuristic and analytic processes in reasoning’, British Journal of Psychology 75, 541–568. Evans, J. (2003), ‘In two minds: Dual-process accounts of reasoning’, Trends in Cognitive Sciences 7, 454–459. Evans, J. (2006a), Dual system theories of cognition: Some issues, in ‘Proceedings of the Cognitive Science Society’, Vancouver. Evans, J. (2006b), ‘The heuristic-analytic theory of reasoning: Extension and evaluation’, Psychonomic Bulletin and Review 13, 378–395. Evans, J. (2008), ‘Dual-processing accounts of reasoning, judgment, and social cognition’, Annual Review of Psychology 59, 255–278. 185  Evans, J. & Over, D. (1996), Rationality and Reasoning, Psychology Press, Hove, England. Fazio, R. (1990), ‘Multiple processes by which attitudes guide behavior: the mode model as an integrative framework’, Advances in experimental social psychology 23, 75–109. Feder, J. (1998), The apple maggot fly, Rhagoletis pomonella, in S. Howard, D.J.; Berlocher, ed., ‘Endless Forms: Species and Speciation’, Oxford University Press, Oxford. Felsenstein, J. (2004), Inferring Phylogenies, Sinauer Associates, Sunderland, Massachusetts. Feyerabend, P. (1975), Against Method, 3rd edn, Verso. Field, H. (1973), ‘Theory change and the indterminacy of reference’, The Journal of Philosophy 70, 462–81. Franklin, L. (2007), ‘Bacteria, sex and systematics’, Philosophy of Science 74, 69– 95. Frey, D. (1993), ‘Species of Pleuroxus (Anomopoda, Chydoridae) from the subantarctic islands and southernmost South America: a partial unravelling of the Pleuroxus aduncus problem’, Hydrobiologia 262, 145–188. Gelman, S. (2003), The Essential Child: Origins of Essentialism in Everyday Thought, Oxford UP, Oxford. Gelman, S. (2004), ‘Psychological essentialism in children’, Trends in Cognitive Sciences 8, 404–409. Gelman, S. & Markman, E. (1987), ‘Young children’s inductions from natural kinds: the role of categories and appearances’, Child Development 58, 1532– 1541. Gelman, S. & Wellman, H. (1991), ‘Insides and essences early understandings of the non-obvious’, Cognition 38, 213–244. Reprinted in Margolis and Laurence (eds.) Concepts; Core Readings, MIT Press, 1999. 186  Ghiselin, M. (1966), ‘On psychologism in the logic of taxonomic controversies’, Systematic Zoology 15, 207–15. Ghiselin, M. (1974), ‘A radical solution to the species problem’, Systematic Zoology 23, 536–544. Gillies, D. (2000), Philosophical Theories of Probability, Routeledge, London. Gilmour, J. & Gregor, J. (1939), ‘Demes: a suggested new terminology’, Nature 144, 333. Gilmour, J. & Heslop-Harrison, J. (1955), ‘The deme terminology and the units of micro-evolutionary change’, Genetica 27, 147–161. Goel, V. (2005), Cognitive neuroscience of deductive reasoning, in K. Holyoak & R. Morrison, eds, ‘The Cambridge Handbook of Thinking and Reasoning’, Cambridge University Press, Cambridge, pp. 475–492. Gould, S. (1989), Wonderful Life: the Burgess Shale and the Nature of History, W W Norton, New York. Gould, S. (1991), Bully for Brontosaurus. Reflections in Natural History, Norton, New York. Grant, V. (1981), Plant Speciation, 2nd edn, Columbia University Press, New York. Griffiths, P. (1999), Squaring the circle: natural kinds with historical essences, in R. Wilson, ed., ‘Species: New Interdisciplinary Essays’, The MIT Press, Cambridge MA. Haidt, J. (2001), ‘The emotional dog and its rational tail: A social intuitionist approach to moral judgment’, Psychological Review 108, 814–834. Hamada, N. & Adler, P. (1999), ‘Cytotaxonomy of four species in the Simulium perflavum species group (Diptera: Simuliidae) from Brazilian Amazonia’, Systematic Entomology 24, 273–288. Hammond, K. (1996), Human Judgment and Social Policy: Irreducible Uncertainty, Inevitable Error, Unavoidable Injustice, Oxford University Press, Oxford. 187  Hey, J. (2001a), Genes, Categories, and Species, Oxford University Press, Oxford. Hey, J. (2001b), ‘The mind of the species problem’, Trends in Ecology & Evolution 16, 326–329. Hey, J. (2006), ‘On the failure of modern species concepts’, Trends in Ecology and Evolution 21, 447–450. Hey, J., Waples, R., Arnold, M., Butlins, R. & Harrison, R. (2003), ‘Understanding and confronting species uncertainty in biology and conservation’, Trends in Ecology and Evolution 18, 597–603. Hine, C. (2008), Systematics as Cyberscience: Computers, Change and Continuity in Science, The MIT Press, Cambridge, MA. Howard, D. & Berlocher, S., eds (1998), Endless Forms: Species and Speciation, Oxford University Press, New York. Hull, D. (1964), ‘The effect of essentialism on taxonomy - two thousand years of stasis I’, British Journal for Philosophy of Science 15, 314–326. Hull, D. (1976), ‘Are species really individuals?’, Systematic Zoology 25, 174–91. Hull, D. (1978), ‘A matter of individuality’, Philosophy of Science 45, 335–360. Hull, D. (1997), The ideal species concept — and why we can’t get it, in M. Claridge, H. Dewah & M. Wilson, eds, ‘Species: the Units of Biodiversity’, Chapman and Hall, London, pp. 357–380. Hull, D. (1998), Introduction to section IV, in D. Hull & M. Ruse, eds, ‘Philosophy of Biology’, Oxford University Press, Oxford. Hull, D. (1999), On the plurality of species: Questioning the party line, in R. Wilson, ed., ‘Species New Interdisciplinary Essays’, The MIT Press, Cambridge MA. Huxley, J., ed. (1940), The New Systematics, Clarendon, Oxford.  188  Kahneman, D. & Frederick, S. (2002), Representativeness revisited: Attribute substitution in intuitive judgment, in D. Gilovich T, Griffin & D. Kahneman, eds, ‘Heuristics and Biases’, Cambridge University Press, Cambridge, pp. 49–81. Kahneman, D. & Tversky, A. (1983), ‘Extensional versus intuitive reasoning: The conjunction fallacy in probability judgment’, Psychological Review 90, 293– 315. Keller, J. (2005), ‘In genes we trust: The biological component of psychological essentialism and its relationship to mechanisms of motivated social cognition’, Journal of Personality and Social Psychology 88, 686–702. Kitcher, P. (1978), ‘Theories, theorists, and theoretical change’, The Philosophical Review 87, 519–547. Kitcher, P. (1984), ‘Species’, Philosophy of Science 51, 308–335. Kitcher, P. (1993), The Advancement of Science: Science without Legend, Objectivity without Illusions, Oxford University Press, Oxford. Kornblith, H. (1993), Inductive Inference and its Natural Ground, The MIT Press, Cambridge MA. Kripke, S. (1980), Naming and Necessity, Harvard University Press, Cambridge. Kruglanski, A. & Freund, T. (1983), ‘The freezing and unfreezing of layinferences: effects on impressional primacy, ethnic stereotyping, and numerical anchoring’, Journal of experimental social psychology 19, 448–468. Kuhn, T. (1962), Structure of Scientific Revolution, 3rd edn, The University of Chicago Press, Chicago. Lakoff, G. (1973), ‘Hedges: A study in meaning criteria and the logic of fuzzy concepts’, Journal of Philosophical Logic 2, 458–508. Lakoff, G. (1989), Cognitive models and prototype theory, in U. Neisse, ed., ‘Concepts and Conceptual Development’, Cambridge University Press, Cambridge, pp. 63–100. 189  Landau, B. (1982), ‘Will the real grandmother please stand up? The psychological reality of dual meaning representations’, Journal of Psycholinguistic Research 11, 47–62. LaPorte, J. (2003), Natural Kinds and Conceptual Change, Cambridge University Press, Cambridge. Larson, J. (1968), ‘The species concept of Linnaeus’, Isis 59, 291–299. Laurence, S. & Margolis, E. (1999), Concepts and cognitive science, in E. Margolis & S. Laurence, eds, ‘Concepts: Core Readings’, The MIT Press, Cambridge MA, pp. 3–81. Levin, D. (1979), ‘The nature of plant species’, Science 204, 381–384. Lieberman, M. (2003), Reflexive and reflective judgment processes: A social cognitive neuroscience approach, in K. Forgas, J.; Williams & W. von Hippel, eds, ‘Social Judgments: Implicit and Explicit Processes’, Cambridge University Press, Cambridge, pp. 44–67. Lieberman, M. (2008), What zombies can’t do: A social cognitive neuroscience approach to the irreducibility of reflective consciousness, in J. Evans, ed., ‘In Two Minds: Dual Processes and Beyond’, Oxford University Press, Oxford. Lincoln, R., Boxshall, G. & Clark, P. (1998), A Dictionary of Ecology, Evolution, and Systematics, 2nd edn, Cambridge University Press, Cambridge. Luckow, M. (1995), ‘Species concepts: Assumptions, methods, and applications’, Systematic Botany 20, 589–605. Lyell, C. (1837), Principles of Geology, Murray, London. Maclaurin, J. & Sterelny, K. (2008), What Is Biodiversity?, University of Chicago Press, Chicago. Mallet, J. (1995a), ‘Reply from Mallet’, Trends in Ecology and Evolution 10, 490– 491.  190  Mallet, J. (1995b), ‘A species definition for the modern synthesis’, Trends in Ecology and Evolution 10, 294–299. Mallet, J. (1996), ‘Reply from Mallet’, Trends in Ecology and Evolution 11, 174– 174. Margolis, E. & Laurence, S., eds (1999), Concepts: Core Readings, The MIT Press, Cambridge MA. Mayden, R. (1997), A hierarchy of species concepts the document in the saga of the species problem, in M. Claridge, H. Dewah & M. Wilson, eds, ‘Species: the Units of Biodiversity’, Chapman and Hall, London, pp. 381–424. Mayr, E. (1942), Systematics and the Origin of Species from the Viewpoint of a Zoologist, Columbia University Press, New York. Mayr, E. (1957), Species concepts and definitions, in E. Mayr, ed., ‘The Species Problem’, American Association for the Advancement of Science, pp. 1–22. Mayr, E. (1963), Animal Species and Evolution, Harvard University Press, Cambridge MA. Mayr, E. (1969), Principles of Systematic Zoology, McGraw-Hill, New York. Mayr, E. (1970), Populations, Species and Evolution: An Abridgement of Aminal Species and Evolution, Harvard University Press, Cambridge MA. Mayr, E. (1988), Toward a new Philosophy of Biology, Harvard University Press, Cambridge. Mayr, E. (1992), ‘A local flora and the biological species concept’, American Journal of Botany 79, 228–38. Mayr, E. & Ashlock, P. (1991), Principles of Systematic Zoology, McGraw-hill, New York. McDade, L. (1995), ‘Species concepts and problems in practice: insight from botanical monographs’, Systematic Botany 20, 606–22.  191  McOuat, G. (1996), ‘Species, rules and meaning: The politics of language and the ends of definitions in 19th century natural history’, Studies in History and Philosophy of Science Part A 27, 473–519. McOuat, G. (2001), ‘Cataloguing power: delineating ‘competent naturalists’ and the meaning of species in the british museum’, The British Journal for the History of Science 34, 1–28. Medin, D. (1989), ‘Concepts and conceptual structure’, American Psychologist 44, 1469–1481. Medin, D. & Ortony, A. (1989), Comments on part I: psychological essentialism, in O. A. Vosniadou S, ed., ‘Similarity and Analogical Reasoning’, Cambridge University Press, Cambridge UK, pp. 179–195. Mishler, B. (1999), Getting rid of species?, in R. Wilson, ed., ‘Species: New Interdisciplinary Essays’, The MIT Press, Cambridge, MA, pp. 307–315. Mishler, B. & Brandon, R. (1987), ‘Individuality, pluralism, and the phylogenetic species concept’, Biology and Philosophy 2, 397–414. Mishler, B. & Theriot, E. (1999), Monophyly, apomorphy, and phylogenetic species concepts, in Q. Wheeler & R. Meier, eds, ‘Species Concepts and Phylogenetic Theory: A Debate’, Columbia University Press, New York, pp. 44–54. MIT (2008), ‘No easy answers in evolution of human language’, ScienceDaily. Retrieved July 24, 2008, from http://www.sciencedaily.com/releases/2008/02/ 080217102131.htm.  Mithen, S. (1996), The Prehistory of the Mind, Thames and Hudson, London. Murphy, G. (2002), The Big Book of Concepts, The MIT Press, Cambridge MA. Newton-Smith, W. (1982), The Rationality of Science, Routledge Kegan & Paul, London. Nixon, K. & Wheeler, Q. (1990), ‘An amplification of the phylogenetic species concept’, Cladistics 6, 211–223. 192  Nolan, M. & Cribb, T. (2005), ‘The use and implications of ribosomal DNA sequencing for the discrimination of digenean species’, Adv Parasitol 60, 101– 163. O’Hara, R. (1993), ‘Systematic generalizations, historical fate, and the species problem’, Systematic Zoology 42, 231–246. Ohtomo, S. & Hirose, Y. (2007), ‘The dual-process of reactive and intentional decision-making involved in eco-friendly behavior’, Journal of Environmental Psychology 27, 117–125. Okasha, S. (2002), ‘Darwinian metaphysics: Species and the question of essentialism’, Synthese 131, 191–213. Otte, D. & Endler, J., eds (1989), Speciation and its Consequences, Sinauer Association Inc, Sunderland MA. Paterson, H. (1985), The recogniton concept of species, in E. Vrba, ed., ‘Species and Specieation’, Transvaal Museum Monograph No4, Pretoria. Pigliucci, M. (2003), ‘Species as family resemblance concepts: the (dis-)solution of the species problem?’, Bioessays 25(6), 596–602. Pigliucci, M. (2005), ‘Wittgenstein solves (posthumously) the species problem’, Philosophy Now 50, 51. [http://www.philosophynow.org/issue50/50pigliucci.htm (Jun. 11, 2005 retrieved)]. Pigliucci, M. & Kaplan, J. (2006), Making Sense of Evolution, University of Chicago Press, Chicago. Pillon, Y. & Chase, M. (2007), ‘Taxonomic exaggeration and its effects on orchid conservation’, Conservation Biology 21, 263–265. Pollock, J. (1989), ‘OSCAR: a general theory of rationality’, Journal of Experimental & Theoretical Artificial Intelligence 1, 209–226. Povinelli, D. (2000), Folk Physics For Apes, Oxford University Press, Oxford.  193  Reber, A. (1993), Implicit Learning and Tacit Knowledge, Oxford University Press, Oxford. Reydon, T. (2004), ‘Why does the species prolbem still persist?’, Bioessays 26(3), 300–305. Reydon, T. (2005), ‘On the nature of the species problem and the four meanings of ‘species”, Studies in History and Philosophy of Biological and Biomedical Sciences 36, 135–158. Ride, W., Cogger, H., Dupuis, C., Kraus, O., Minelli, A., Thompson, F. & Tubbs, P. (1999), International Code of Zoological Nomenclature, 4th edn, London: The International Trust for Zoological Nomenclature. Ridley, M. (1989), ‘The cladistic solution to the species problem’, Biology and Philosophy 4, 1–16. Rieppel, O. (2007), ‘Species: kinds of individuals or individuals of a kind’, Cladistics 23, 373–384. Rosch, E. (1975), ‘Cognitive representations of semantic categories’, Journal of Experimental Psychology: General 104, 192–233. Rosch, E. (1978), Principles of categorization, in B. Rosch, E.; Lloyd, ed., ‘Cognition and Categorization’, Erlbaum, Hillsdale NJ, pp. 27–48. Rosch, E. (2000), Reclaiming concepts, in W. Nunez, R.; Freeman, ed., ‘Reclaiming Cognition: The Primacy of Action, Intention and Emotion’, Imprint Academic, Thorverton, England, pp. 61–77. Ruse, M. (1987), ‘Biological species: Natural kinds, individuals, or what?’, The British Journal for the Philosophy of Science 38, 225–42. Sahlin, N., Wallin, A. & Persson, J. (forthcoming), ‘Decision science: From Ramsey to Dual Process Theories’, Synthese. Saito, F., Kojima, J., Nguyen, L. & Kanuka, M. (2007), ‘Polistes formosanus (Hymenoptera: Vespidae), a good species supported by both morphological and 194  molecular phylogenetic analyses, and a key social wasp in understanding the historical biogeography of the Nansei Islands’, Zoological Science 24, 927–939. Samuels, R. (2008), The magical number two, plus or minus: Dual process theory as a theory of cognitive kinds, in J. Evans, ed., ‘In Two Minds: Dual Processes and Beyond’, Oxford University Press, Oxford. Sankey, H. (1994), The Incommensurability Thesis, Avebury, Aldershot, England. Saunders, L. (2008), Reason and intuition in the moral life: A dual process account of moral justification, in J. Evans, ed., ‘In Two Minds: Dual Process and Beyond’, Oxford UP, Oxford. Schilthuizen, M. (2005), ‘On the origin of reproductive isolation’, Bioessays 27, 669–670. Seehausen, O., Alphen, J. & Witte, F. (1997), ‘Cichlid fish diversity threatened by eutrophication that curbs sexual selection’, Science 277, 1808. Shaw, K. (1996), ‘What are ‘Good’ species?’, Trends in Ecology and Evolution 11, 174–174. Simpson, G. (1961), Principles of Animal Taxonomy, Columbia University Press, New York. Sloman, S. (1996), ‘The empirical case for two systems of reasoning’, Psychological Bulletin 119, 3–22. Sloman, S. (2002), Two systems of reasoning, in D. Gilovich, T.; Griffin & D. Kahneman, eds, ‘Heuristics and Biases: The Psychology of Intuitive Judgment’, Cambridge University Press, Cambridge, pp. 379–396. Smith, E. & DeCoster, J. (2000), ‘Dual-process models in social and cognitive psychology: Conceptual integration and links to underlying memory systems’, Personality and Social Psychology Review 4, 108–131. Smith, E. & Medin, D. (1999), The exemplar view, in E. Margolis & S. Lawrence, eds, ‘Concepts: Core Readings’, The MIT Press, Cambridge MA, pp. 207–221. 195  Smith, E., Osherson, D., Rips, L. & Keane, M. (1988), ‘Combining prototypes: A selective modification model’, Cognitive Science 12, 485–527. Sneath, P. (1976), ‘Phenetic taxonomy at the species level and above’, Taxon 25, 437–50. Sober, E. (2000), Philosophy of Biology, 2nd edn, Westview Press, Boulder. Sokal, R. (1973), ‘The species problem reconsidered’, Systematic Zoology 22, 360–74. Sokal, R. & Sneath, P. (1963), Principles of Numerical Taxonomy, Freeman, San Francisco. Stamos, D. (2002), ‘Species, languages, and the horizontal/vertical distinction’, Biology and Philosophy 17, 171–198. Stamos, D. (2003), The Species Problem: Biological Species, Ontology, and the Metaphysics of Biology, Lexinton Books, Lanham. Stanford, P. (1995), ‘For pluralism and against realism about species’, Philogophy of Science 62, 70–91. Stanovich, K. (1999), Who Is Rational?: Studies of Individual Differences in Reasoning, Lawrence Erlbaum Associates, Mahwah, NJ. Stanovich, K. (2004), The Robot’s Rebellion: Finding Meaning in the Age of Darwin, The University of Chicago Press, Chicago. Stanovich, K. & West, R. (2000), ‘Individual differences in reasoning: Implications for the rationality debate?’, Behavioral and Brain Sciences 23, 645–665. Sterelny, K. & Griffith, P. (1999), Sex and Death, Chicago: University of Chicago Press. Stevens, P. (1992), Species: historical perspectives, in E. Keller & E. Lloyd, eds, ‘Keywords in Evolutionary Biology’, Harvard University Press, Cambridge, pp. 302–311.  196  Strickland, H. (1837), ‘Rules for zoological nomenclature’, Magazine of Natural History 1, 173–176. Strickland, H. (1843), ‘Report of a committee appointed “to consider of the rules by which the nomenclature of zoology may be established on a uniform and permanent basis”’, Report of the 12th Meeting of the British Association for the Advancement of Science pp. 105–121. Templeton, A. (1989), The meaning of species and speciation: A genetic perspective, in D. Otte & J. Endler, eds, ‘Speciation and its Consequences’, Sinauer Associates, Inc., Massachusetts. Toates, F. (2006), ‘A model of the hierarchy of behaviour, cognition, and consciousness’, Consciousness and Cognition 15, 75–118. Tomasello, M. (1999), The Cultural Origins of Human Cognition, Harvard University Press, Cambridge MA. Van Regenmortel, M. (1997), Viral species, in M. Claridge, H. Dawah & R. Wilson, eds, ‘Species — The Units of Biodiversity’, Chapman and Hall, London, pp. 17– 24. Van Valen, L. M. (1976), ‘Ecological species, multispecies, and oaks’, Taxon 25, 233–9. Weber, M. (2004), Philosophy of Experimental Biology, Cambridge UP, Cambridge. Wheeler, Q. & Meier, R., eds (2000), Species Concepts and Phylogenetic Theory: A Debate, Columbia University Press, New York. Wiley, E. (1978), ‘The evolutionary species concept reconsidered’, Systematic Zoology 27, 17–26. Wiley, E. (1981), Phylogenetics. The Theory and Practice of Phylogenetic Systematics, J Wiley and Sons, New York. Wilkins, J. (2006a), ‘The concept and causes of microbial species’, History and Philosophy of the Life Sciences 28, 389–408. 197  Wilkins, J. (2006b), ‘Species, kinds, and evolution’, Reports of the National Center for Science Education 26, 36–45. Wilkins, J. (2009), Species: A History of the Idea, University of California Press, Berkeley, CA. Williams, M. (1992), Species: current usages, in E. Keller & E. Lloyd, eds, ‘Keywords in Evolutionary Biology’, Harvard University Press, Cambridge, pp. 318– 323. Wilson, R. (1999), Realism, essence, and kind: Resuscitating species essentialism?, in R. Wilson, ed., ‘Species: New Interdisciplinary Essays’, The MIT Press, Cambridge MA, pp. 187–207. Wilson, R., Barker, M. & Brigandt, I. (2007), ‘When traditional essentialism fails: Biological natural kinds’, Philosophical Topics 35(2), 189–215. Winston, J. (1999), Describing Species: Practical Taxonomic Procedure for Biologists, Columbia University Press, New York. Yzerbyt, V., Rocher, S. & Schadron, G. (1997), Stereotypes as explanations: A subjective essentialistic view of group perception, in P. E. N. Spears, R; Oakes & S. Haslam, eds, ‘The Social Psychology of Stereotyping and Group Life’, Blackwell, Oxford, pp. 20–50.  198  Appendix A  Further Description of the Dual-Process Theory General characteristics of the two systems by the Dual-Process theory were presented in section 3.2.2 (p. 84). Although psychologists largely agree that DualProcess theory is well-motivated empirically, elaboration is needed with regard to its theoretical foundation. I will discuss several points in this Appendix.  A.1  Elaboration of “System”: Token or Type?  First, we should further clarify what Dual-Process theorists mean by ‘system.’ For there are two different ways of understanding the Dual-Process theory, as Richard Samuels (2008) discusses, related to the type-token distinction. According to the token thesis, the systems described in Dual-Process theory are the same across a variety of areas; System 1 postulated in the belief-bias experiment is numerically the same system as the one postulated in the Linda problem, and so on. The type thesis would say that ‘system’ refers to a type or kind; each system is merely a collection of information-processing processes postulated over different phenomena which share a cluster of properties. Dual-Process theory is about two types of processes, on this view. Samuels rejects the token thesis: if Dual-Process theory is, indeed, a plausible theory, then it has to be about types of processes. He discusses various versions  199  of the token thesis as well, but argues that none of them will work either. For instance, he emphasizes that it is hopeless to believe there are exactly two systems in the mind. This is because each system would have many different functions, but coginitive psychologists generally believe that a complex system is decomposed into subsystems corresponding to specific functions. The human vision system powerfully illustrates this. Each subsystem of it has a specific function such as depth perception, color identification, and categorization. If one counts those subsystems as “systems,” the number of systems will never be two. One may reply that there are two systems at a more abstract level. However, even at a quite abstract level, there are a number of things our mind is supposed to do: perceive, reason, remember, emote, and communicate through language, to name just a few. So, it is not plausible to think that there are only two systems which do all these things. The second reason to reject the token thesis is that Dual-Process theorists characterize systems somewhat differently, depending on the phenomenon for which they are postulating them. As we have already noted, not all psychologists attribute exactly the same cluster of properties to each system. Sloman (1996, 2002) associates creativity and fantasy with System 1, but this association is not common among other Dual-Process theorists. Furthermore, Dual-Process theory covers a variety of subareas of psychology —— planning, decision-making, deductive inference, social cognition, etc. It is unlikely that one and the same system works across such diverse areas of cognition. For those reasons the token thesis should be rejected. Ultimately, Samuels concludes, we should pursue the only alternative: the type thesis.  A.2  Relationship Between Systems 1 and 2  Now we by and large know what the two systems are. But what is the relationship between them? Researchers have discussed multiple models purporting to capture the relationship between the two systems. I focus on the following two: 1. Parallel model 2. Default/intervention model In the parallel model, System 1 and System 2 work simultaneously. When outputs from the systems do not coincide, there will be competition between them for the 200  final output. Sloman (1996) thinks this models the interaction of both systems: “When a person is given a problem, however, both systems may try to solve it: each may compute a response, and those responses may not agree” (p. 6). He takes conflicting responses as evidence of both systems working simultaneously. In the syllogism example, this model would suppose that reasoning through the Venn diagram and that driven by the belief-bias occur simultaneously (p. 13). In the default/intervention model (for example, see Kahneman & Frederick 2002), a default process is assumed (usually System 1). When one faces a task, the default process activates first. Then, under certain conditions, the other process will “check” the result from the first process. When outputs from both processes coincide, then that will be the final output. If they do not coincide, the second process will intervene before the output from System 1 becomes the final output. Thus the final result may come from the intervening process when conditions allow it to work. In the syllogism case, at first one may think that argument (b) is invalid just because it is hard to believe that vitamin tablets are not nutritious. The subject may then “think twice” and visualize or draw a Venn diagram to confirm that the conclusion does, in fact, follow from the premises. When the logical process is not allowed to perform its function well (under time-pressure, for instance), the second process fails to intervene the first and the answer from the first process will be the final one. Researchers do not agree on which model to take generally. In reasoning tasks, many scholars assume the default/intervention model, while researchers like Sloman (1996) endorse the parallel model. The same author is somtimes sympathetic to both models in the same paper (Smith & DeCoster 2000, p. 112, and p. 122). Matters are complicated by the fact that speed of processing is often taken to be crucial in distinguishing System 1 from System 2; even when Systems 1 and 2 begin to work simultaneously, outputs from System 1 tend to come earlier than the one from System 2, because System 1 processing is faster than System 2 processing. So, receiving output from System 2 later than that from System 1 does not discriminate one model from another, as in the belief-bias experiment. Furthermore, no evidence or argument has so far been given to suggest that only one model should be applied to all the cases in which dual processes are at work——the appropriate model may differ from one case to the next. In any case, for the purposes of this 201  project, we need not decide which, if either, model is superior, since my account does not require it.  A.3  The Unity of Systems and Causal Mechanism Behind It  “System,” in Dual-Process theory, is taken to be a type——a collection of mental processes which have a cluster of properties in common. One concern here is whether this “system” has only a phenomenal basis or is based on deeper causal mechanisms. In other words, one may ask how deep the unity of the System is. This concern is raised by several thinkers (Samuels 2008, Evans 2008, Sahlin et al. forthcoming). We cannot reach any decisive conclusion on this issue now. For, as these authors all agree, Dual-Process theory is still in progress. However, there is some support for the idea that systems have some causal mechanism behind them.  A.3.1  The Unity of System 2  There are a couple of proposed causal mechanisms behind the systems, but Evans’ hypothesis for System 2 now looks the most promising. Evans (2008) proposes that the use of working memory (a system temporarily storing information that is currently processed) may well be the key component of System 2. If working memory involves System 2 processing, then one can render intelligible several common features of System 2. For instance, working memory is linked to consciousness, because an item one is conscious of is typically represented in one’s working memory. Another point is individual difference. It is found that some perform far better than others in the tasks involving System 2 (Stanovich & West 2000). Although many subjects say that Linda is more likely to be a feminist bank teller than a bank teller, some still do reach the correct answer. Individual differences like this are known to correlate with one’s general intelligence and working memory capacity. Working memory also makes sense of the fact that System 2 processing is slow, sequential, and capacity-limited. Storing information and accessing working memory takes time. Other features, such as voluntariness and higher-order control, are also accounted for. 202  However, Evans’ account is incomplete in some regards (Samuels 2008). First, this account says nothing about System 1. Even if his account is on the right track for System 2, it remains an open question whether System 1 possesses causal mechanisms. Second, Evans’ proposal does not provide an account for every property of System 2. Evans himself concedes that System 2 features, such as uniqueness to human beings and association with decontextualized thought, may not be accounted for by reference to working memory.  A.3.2  Neurological Basis  Evans’ proposal leaves open how deep the causal unity of System 1 goes. Nevertheless, System 1 processings, as well System 2 processings, may have some common neurological foundation. Smith & DeCoster (2000), for example, link System 1 and 2 to different learning systems (slow and fast systems). Slow learning does not involve consciousness and awareness. In this mode, one records “information slowly and incrementally so that the total configuration in memory reflects a large sample of experiences” (p. 109). On the other hand, fast learning is conscious and involves episodic memory: one remembers an event after that event occurred (no need to perceive many events of the same kind). Patients whose hippocampal region of the brain is impaired show deficiency in tasks related to the fast learning system (System 2) ——such as rapid learning of new associations among objects and intentional retrieval of information—— but not those related to the slow learning system (System 1). Matthew Lieberman (2003, 2008) also suggests that the two systems use different regions of the brain. Liebermann calls the two systems, “X-system” (System 1) and “C-system” (System 2), respectively. Each system has a standard set of features as characterized in the Dual-Process theory. From neural imaging studies, Liberman found that each system is associated with particular regions of a brain. According to Lieberman, when one uses X-system, regions such as the amygdala, basal ganglia, and lateral temporal cortex are found to be active, while regions such as the anterior cingulate cortex, lateral prefontal cortex, and the medial temporal (the hippocampus, etc.) are active when one uses the C-system. If this is true, the distinction between System 1 and System 2 has some neurological basis.  203  We have seen a couple of attempts to provide a causal basis to the Dual-Process theory. It is not true at present that the theory has an adequate theoretical account in this regard. However, neither is it true to say that the Dual-Process theory will prove unable to provide one in the future. It remains to be seen how much both systems have a physiological and neurological basis. So, it is too early to reject Dual-Process theory for this reason.  A.4  Transfer (Consolidation)  We should also bear in mind that some processes are not intrinsically classified into System 1 or 2. What psychologists call transfer or consolidation illustrates this point: one processes one type of information with one system at one time, but will do it with the other system later (Smith & DeCoster 2000). In the syllogism example, as one becomes accustomed to use the Venn diagram, one may be able to do it unconsciously and implicitly, and take less time to get to the answer. Chess playing offers a good analogy. A chess player can consider more moves implicitly and unconsciously as he becomes a better player. The same process could acquire properties linked to System 1 as one becomes more accustomed to deploying it, even though it had System-2 properties when employed the first time.1  1 Another possible criticism toward the Dual-Process theory comes from evolutionary psychology. Evolutionary psychologists propose that our mind is composed entirely of cognitive modules. Each module has a specific function. It is adapted to a particular cognitive task in that it is to give a specific kind of output to a specific kind of input. Modules are each domain-specific, in this sense, and somewhat independent from each other. Some evolutionary psychologists claim that this massive modularity hypothesis (MMH) would be incompatible with the Dual-Process theory, because System 2 (in the Dual-Process theory) is close to what MMH claims does not exist: a domain-general information-processing mechanism. However, one wonders whether MMH and Dual-Process theory are truly incompatible, because Peter Carruthers (2006) tries to account for System 2 in terms of massive modularity.  204  Appendix B  Sources Surveyed for Usages of ‘Good Species’ In Chapter 3, I outlined various usages of the phrase ‘good species’ in a mailing list called “Taxacom” and professional journals. This appendix provides the sources I surveyed for the usages of ‘good species.’  B.1  List of Entries Surveyed from the Taxacom Mailing List  See Table B.1. Table B.1: List of Entries from the Taxacom Mailing List  From  Date  Subject  Arthur Chapman  18 Nov 1993  Botanical Author Names  Richard Jensen  3 Apr 1995  nominal characters  Lawrence Kirkendall  14 Jun 1995  Confidence  Richard Jensen  9 Jul 1996  Exactly what is a species?  Timothy S. Ross  4 Jan 1996  Saintpaulia taxonomy  Mary E. Petersen  20 Sep 1996  Peter Davis - Pieris  Daniel L. Geiger  17 Apr 1998  Hybrid Speciation  (Continued on Next Page. . . ) 205  List of Entries from Taxacom (Continued)  From  Date  Subject  Tim Holtsford  8 Jan 1999  Nicotiana mutabilis  Andreas Gminder  16 Feb 1999  This should start a thread ...  Neil Snow  18 Feb 1999  Names for Money  Ken Kinman  23 May 1999  Rhipidomys & other mammals  Curtis Clark  11 Oct 1999  semi-species  Ken Kinman  12 Oct 1999  Mayr on semi-species & speciation  Ron Kaneen  12 Oct 1999  semi-species  Alexey Solodovnikov  11 Nov 1999  subspecies  John Grehan  15 Nov 1999  The political meaning of species  Geoff Read  16 Nov 1999  The political meaning of species  Ken Kinman  6 Feb 2000  paraphyletic evolution  Richard Jensen  2 Feb 2001  Culex molestus  Don Colless  2 Feb 2001  Culex molestus  Larissa Vasilyeva  28 Jan 2002  natural  Richard Jensen  5 Feb 2002  Cladistics and “Eclecticism”  Susanne Schulmeister  8 Feb 2002  Comments  to  Cladis-  tics/Eclecticism, part 3 Richard Jensen  3 Apr 2002  Botanical  nomenclatural  query Ron Gatrelle  21 May 2002  Undescribed species and the internet  Diana Horton  5 Nov 2002  Taxonomic ‘Jargon’  Ron Gatrelle  18 Apr 2003  Protecting  Doug Yanega  11 Nov 2003  BSC/PSC  Richard Pyle  20 Apr 2004  Real species and ideology  (Continued on Next Page. . . )  206  List of Entries from Taxacom (Continued)  From  Date  Subject  Erast Parmasto  1 Sep 2004  Non type specimens  Mark Egger  10 May 2006  typification knot  Ken Kinman  20 Mar 2007  Saxifraga  tischii  a  good  species? Richard Zander  14 May 2007  encyclopedia of life  Anthony Pigott  6 Apr 2008  Emendation / species splitting  Paul van Rijckevorsel  10 May 2008  “deprecation” (was Taxonomy... in the 21st century)  B.2  List of Papers Surveyed on ‘Good Species’  • Alderweireldt, M. (1999) Journal of arachnology 27, 449–457. • Artemiev, M. (1982) Zoologicheskii Zhurnal 61, 1359–1371. • Asada, N., Fujiwara, K., Ikeda, H., and Hihara, F. (1992) Zoological Science 9, 397–404. • Azaizeh, H., Salhani, N., Sebesvari, Z., Shardendu, S., and Emons, H. (2006) International Journal of Phytoremediation 8, 187–198. • Baixeras, J. and Dominguez, M. (1994) Annales de la Societe Entomologique de France 30, 345–359. • Baran, T. and Buszko, J. (2005) Entomologica Fennica 16, 9–18. • Beaver, R. and Gebhardt, H. (2006) Deutsche Entomologische Zeitschrift 53, 155–178. • Bedalov, M. and Fischer, M. (1995) Phyton Annales Rei Botanicae 35, 103– 113. • Benick, G. (1981) Revue Suisse de Zoologie 88, 561