UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Materialism : metaphysics and methodology Constabaris, Adam John 1995

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

Item Metadata


831-ubc_1995-0467.pdf [ 3.17MB ]
JSON: 831-1.0086957.json
JSON-LD: 831-1.0086957-ld.json
RDF/XML (Pretty): 831-1.0086957-rdf.xml
RDF/JSON: 831-1.0086957-rdf.json
Turtle: 831-1.0086957-turtle.txt
N-Triples: 831-1.0086957-rdf-ntriples.txt
Original Record: 831-1.0086957-source.json
Full Text

Full Text

MATERIALISM: METAPHYSICS AND METHODOLOGY by A D A M JOHN CONSTABARIS B.A., The University of British Columbia, 1993 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF T H E REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in T H E F A C U L T Y OF G R A D U A T E STUDIES (Department of Philosophy) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA August 1995 © Adam John Constabaris, 1995 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of T^'toSe pK^ The University of British Columbia Vancouver, Canada Date DE-6 (2/88) 11 A b s t r a c t In contemporary discussions of materialism, the term "reductionism" is used in several senses. First, I distinguish materialism, the metaphysical claim that "everything is physical" from physicalism, which is, broadly speaking, a research strategy that gives some sort of privilege to physics. I discuss two theses commonly associated with materialism, the view that the natural world is divided into levels corresponding to the various special sciences, and the claim that physics is "causally complete". In the second chapter, I discuss reductionism in the sense in which that term applies to empiricists who espoused a doctrine known as "the Unity of Science". Ametaphysical empiricists were less concerned with the classical materialist goal of ontological economy than they were with conceptual economy, establishing a single theory stated in the language of physics which would be adequate for all scientific purposes. According to various empiricists, such unity was to be achieved by explaining higher level theories in terms of lower level theories, a phenomenon known as intertheoretic reduction. I then show how the most widely discussed account of intertheoretic reduction as a kind of derivation is related to empiricistic views on the nature of theories and of explanation. I discuss the reasons that the empiricist's linguistic account of reductionism was re-formulated by metaphysically minded philosophers as the claim that every property is a physical property (property reductionism). Then I discuss the "multiple realizability" arguments, which are widely thought to establish both the methodological autonomy of the special sciences from physics and the metaphysical thesis that there are non-physical properties. I argue that the conclusion of such arguments is better stated as a U l conclusion about the representational power of physical language. Next, I state the recent arguments that 'non-reductive' forms of materialism based on supervenience (conceived as a relation between physical and non-physical properties) appear to be compatible with non-standard emergentist versions of materialism, and suggest that a more linguistic construal of supervenience ought to be employed in the formulation of materialism. In the final chapter, I attempt to show the independence of reductionist methodological claims from the metaphysical claim of property reductionism. iv ABSTRACT ii ACKNOWLEDGEMENTS vi CHAPTER ONE: MATERIALISM 1 Definitions and (some) Presuppositions 3 Materialism: The Nuts and Bolts 6 Levels of Nature 9 The Closure of Physics 14 CHAPTER TWO: REDUCTIONISM 19 Introduction 19 Intertheoretic Reduction and Logical Empiricism 22 A Formal Model of Reduction 25 Theory and Ontology 28 The Functionalist Challenge 31 CHAPTER THREE: SUPERVENIENCE AND EMERGENCE 37 Supervenience Materialism: The Consensus View 37 Supervenience • 40 Emergence 44 Why Supervenience? 48 A proposal 50 V CHAPTER FOUR: REDUCTIONISM REVISITED 52 A Brief Science Lesson 53 Property Reduction Again 55 Methodological Reductionism 58 Conclusion 61 BIBLIOGRAPHY 62 vi A c k n o w l e d g e m e n t s I would like to thank Steven Savitt for making me worry about many of the ideas I intially wanted to put into this thesis (as well as a lot of ideas I did put into it), Alan Richardson for illuminating and entertaining discussions of logical empiricism, and Leslie Burkholder for bibliographical assistance. 1 C h a p t e r O n e : M a t e r i a l i s m The history of philosophy is littered with examples where ontology and epistemology have been stirred together into a confused and confusing brew (Earman 1986, p.7) Physicalism, goes the standard story, comes in two basic flavors: reductive, and non-reductive. Reductive physicalism, or "reductionism" for short, is often associated with logical positivism and is widely regarded as hopeless. Non-reductivists typically argue that it is possible to retain the ontological and metaphysical portions of physicalism while rejecting the positivists' epistemological cum methodological doctrine of "the Unity of Science". Curiously, although their central goal is to establish the methodological autonomy of the special sciences from physics, anti-"reductionists" usually attempt to refute a metaphysical thesis. This curious situation is compounded further by the charge that the most popular form of non-reductive physicalism — supervenience physicalism ~ bears a close relation to emergentism, a doctrine which is at odds with orthodox physicalism. My aim in this thesis is to disentangle the methodological thesis of reductionism from the metaphysical thesis of reductionism. I take non-reductive physicalism to be based on the view that the metaphysical monism of physicalism does not imply the sort of linguistic or epistemological monism associated with the classical doctrine of "the Unity of Science" (Carnap 1933, Oppenheim and Putnam 1958). I am in substantial agreement with non-reductivists that materialism (the metaphysical thesis of physicalism) has no such implications. I do think, however, that non-reductivists have failed to distinguish clearly between ontology on the one hand and epistemology on the other, leading to a number of confusions about reductionism. The objects, properties, and processes described in the 2 vocabularies of the various special sciences may all be "physical" objects, properties, and processes, but this does not mean that the special sciences are or ought to be physics, as the classical reductionist seems to tell us. We can, in a phrase, be metaphysical reductionists without being epistemological reductionists. However, the failure of classical reductionism does not, contra Fodor (1974) imply the disunity of science (Smith 1992). Anti-reductionist arguments are, at least nominally, usually directed at the largely static account of intertheoretic reduction given by Ernest Nagel (1961, ch. 11). More recent dynamic accounts of reduction (Wimsatt 1976, Hooker 1981) which respect (to a certain degree) the autonomy of the special sciences have been by and large ignored. Such accounts of reduction support a modified view of the unity of science. I will not argue directly for such unity, but only show how it is possible to defend 'weak' unity against some popular antireductionist arguments. Non-reductive physicalists often employ the notion of supervenience in their formulations of physicalism; roughly, the idea behind supervenience physicalism is "no difference without a physical difference". Supervenience is supposed to capture the idea that the physical facts determine all of the facts without implying that the special sciences are reducible to physics. While supervenience seems to be on the right track (if taken as a metaphysical claim, which it usually is), several philosophers (Kim 1992, 1993, Horgan 1993a) have noticed the resemblance supervenience physicalism bears to a doctrinal tradition Brian McLaughlin (1992) has called "British Emergentism" (e.g: Broad 1925). Emergentists accept that everything is composed out of microscopic physical particles and that no two things can differ in any respect unless they differ in some physical respect, but 3 deny the key physicalist claim that physics is 'causally complete' . In light of the resemblance between emergent materialism and supervenience materialism (and other historically embarassing ones), Horgan argues, the would-be non-reductivist must require that the supervenience relation itself be given a 'materialistically acceptable' explanation (1993a). This seems to represent a step back towards reductionism, since the core notion of reduction is the explanation of one theory by another. Definitions and (some) Presuppositions In this thesis, I will be looking at physicalism from the point of view of a philosopher of science. The basic problem to which physicalism is a response is quite simple (Melnyk 1993): physics is not the only successful science; whatever reasons we have for accepting the claims of physicists as true (or empirically adequate, or reliable predictive instruments) about what the world is like apply to at least the more successful "special sciences", including large chunks of chemistry, biology, geology, and psychology. Nevertheless, it seems clear that the various sciences do fit together in some way — total "incommensurability" between (for example) cytology and physics is out of the question1. Physicalism is a story about the way some things 'hang together', to borrow Sellars' (1963) famous phrase. Sellars himself noted that there is no single scientific image (and I have my doubts about the uniqueness of "the" manifest image), although he thought that integrating the various scientific images was a fairly simple task in comparison to the larger project. In 1 Two theories are "totally incommensurable" if they cannot be seen as dealing with (in some sense) same subject-matter. In my example, cells have positions, sizes, and momenta, for example, and these terms seem to be used in much the same sense by biologists as they are by physicists; so there is some measure of 'agreement' between the two disciplines. 4 addition to the microphysical image, there are macrophysical images, biological images, and so forth. In Sellarsian terms then, physicalism is one way of getting a scientific image. In the final sections of this chapter, I will try to lay out the basic metaphysical 'image' common to most physicalists. I will reserve the name "materialism" for the ontological/metaphysical component of physicalism. Very roughly, materialism is the thesis that the special sciences require no more, ontologically speaking, than does microphysics. This rough formulation might be cashed out as the claim that every concrete thing that exists is either a microphysical 'particle' or is a merological sum of microphysical 'particles' , or as the claim that the special sciences deal with subsets of the domain of microphysics. I will also construe materialism as involving some sort of claim that the physical facts determine the non-physical facts (although the invocation of 'facts', which seem suspiciously like linguistic entities, troubles me). To lay some cards on the table here, I think that the doctrine Hellman and Thompson present under the name "Physicalist Materialism" (Hellman and Thompson 1975, 1977; see also Post 1987) is one of the best attempts at expressing a "non-reductive" materialism. However, in order to understand the appeal of reductionism (in both its L By 'concrete', I mean to exclude such problematic 'objects' as holes, centers of gravity, and national economies. Facts about such 'objects' fall into the scope of the second thesis of materialism (i.e it is facts about concrete objects and their properties which determine the facts about holes, centers of gravity, and national economies. 3 By "microphysical particles", I mean whatever it is that physics tells us are the smallest identifiable things. Current quantum theory presents many difficulties in this respect, e.g. the apparent non-locality of entangled states. Apart from non-locality, however, quantum theory is concerned with what happens in very small regions of space-time. For the purposes of this thesis, I will, in general, ignore these complications to focus on what I take to be the broader issues surrounding materialism and physicalism. 5 metaphysical and epistemological forms), I will have to use a somewhat looser and more standard characterization of materialism in terms of 'levels' and 'causal completeness'. In contrast, "physicalism" has been used by some philosophers to.refer to a methodological program (Field 1991, Poland 1994) distinct from the thesis of materialism. Physicalism is distinct from classical reductionism a la Ernest Nagel (1961), Kemeny and Oppenheim (1956), and Oppenheim and Putnam (1958), but like classical reductionists, physicalists attribute some kind of privileged methodological or epistemological status to physical theory. In Field's case, the privilege of physics consists in rejecting theories which postulate processes for which physical explanations cannot be found (1991, pp.271-272). In matters epistemological, I am a fallibilist; many of our beliefs, including some of our most cherished scientific theories, may very well be false. All we can hope to do, at any stage of the game, is to formulate the best theories we can and take our lumps when those theories fail to yield correct predictions. I am interested primarily in the issue of what sort of "global" research methodology is defensible, and whether any such research methodology follows from the ontological claims physicalists are prone to make. Like William Wimsatt, I think that "what we need is a philosophy of science for real, fallible people"4. As such, I will try to limit appeals to what we might discover "at the end of inquiry", or to what God, Laplace's demon, or Broad's (1925) "mathematical archangel" could discover given complete microphysical knowledge of the state of the universe at a particular time. One good reason for avoiding 'confused and confusing brews' of epistemology and ontology is falliblism — the proper methodology for beings as 4 Quoted in Callebaut (1993, p. 154) 6 epistemically 'frail' as humans often are may bear little resemblance to the methodology best suited to something with the calculational abilities of Laplace's demon. I'm also going to presuppose "realism" in a couple of areas. First, a sort of bare-bones realism about theories: at least some theories are to be interpreted literally, i.e. as making claims about the world (even the unobservable parts of it), rather than as "mere" prediction-making instruments5. At any rate, I don't see why a constructive empiricist (van Fraassen 1980) would be terribly interested in the metaphysical components of physicalism (i.e. materialism). I'm also going to assume, for the most part, realism about properties; at least, I'm not always going to try to find ways of putting my claims about properties that would be acceptable to nominalists (cf. Horgan 1993a, p. 557); to do so would complicate the discussion unnecessarily. However, this realist assumption about properties will play an important role in the argument of the third chapter. Materialism: The Nuts and Bolts According to the materialist, contemporary physical theory provides us with the best guide to a fundamental ontological domain from which, metaphorically speaking, everything is "built". Depending on how we cash out the "building" metaphor6, this sort of claim is held to be plausible, or even relatively trivial, by many7. Materialism, thus 5 Rosenberg (1994) defends the view that biological and psychological theories ought to be interpreted instrumentalistically, since the phenomena with which those theories deal are too complex to be dealt with by humans. 6 And, perhaps, the term "everything": we might construe materialism as applying only to the contents of the natural world, allowing that mathematics deals with abstract objects which exist but are not to be found in the natural world. 7 Even some anti-physicalists as Dupre (1993) and Crane and Mellor (1990) concede that something like this is probably correct; indeed, it seems to amount to little more than the denial of substance pluralism (e.g. vitalism, Cartesian dualism), views these authors do not support. 7 construed, gains much of its plausibility from the recognition that one branch of physics, usually referred to as "microphysics", deals with the smallest known entities. For the concrete items of everyday experience (e.g. tables, rocks, and trees) and many 'scientific' entities (e.g. molecules, tectonic plates, and cells) it seems as if we can take the "building" metaphor almost literally. Materialism faces the same problem all versions of monism have faced: the apparent diversity of kinds of things in the world. Not only are there microphysical things, but also sealing wax, canned meat products, epithelial cells, computers, brains, the occasional syphilitic mayor, oxidizing agents, filbert nuts, and maybe, just maybe, snide comments. In addition, we attribute many different kinds of properties to things, not just 'physical' ones like mass and spin: snideness, hardness, having a temperature, and epitheliality (?) are just some of them. The raw materialist 'intuition' is that the concrete things we interact with on a daily basis are composed entirely of scads of microscopic entities (quarks and leptons?), and that the differences between these objects are all due ultimately to differences in the relations between the microphysical entities. Orthodox materialists usually add or assume that microphysics is (at least potentially) "self contained" (Field 1991, p. 283), "causally closed", or "causally complete" (Horgan 1993a, p. 560), in the sense that the chance of any given microphysical event occurring is fixed by other microphysical events in conjunction with the laws of microphysics (Papineau 1993, p. 16). 8 Hellman and Thompson (1975) claim that physicalist doctrine consists of two sorts of claims8. The first claim is straightforwardly ontological: "everything is physical", or, to be slightly more precise, "everything is composed completely out of microphysical parts", I will call this somewhat vague claim the composition principle, or C P for short. Such a thesis is widely regarded as relatively trivial, even by anti-materialists (Crane and Mellor 1990, Dupre 1993). Indeed, C P amounts to little more than the denial of various forms of substance pluralism (e.g. Cartesian dualism, vitalism). Contra Crane and Mellor (1990, p. 187), however, Hellman and Thompson's "Principle of Physical Exhaustion" is a somewhat more comprehensive thesis than C P : the former states that every entity (including, e.g. properties) is a physical entity, in the sense that everything belongs to a set-theoretic hierarchy based on microphysical urelements. Materialists have traditionally been unwilling to allow for "non-physical" properties (more on this in chapter 2). C P , as it stands, is compatible with traditional forms of property pluralism, which Crane and Mellor defend (Crane 1994). The second sort of claim is related (but by no means equivalent) to causal completeness and concerns the primacy of the physical ~ the way the world is depends (in some sense) on the way its fundamental physical constitutents are arranged. I will call this somewhat vague claim the determination principle, or D P for short. Antireductionists like to cash out D P in terms of supervenience -- there can be no difference without a physical difference. The concept of supervenience has received a lot of attention in the philosophical Other materialists who have made similar proposals are Horgan (1987) and Pettit (1994). Hellman and Thompson's approach has led to what Poland (1994) refers to as "the consensus view"; Post (1987) explicitly acknowledges the influence of Hellman and Thompson. 9 literature, and there are numerous ways of making it more explicit (no difference in whafl). The chief function of supervenience in formulations of non-reductive physicalism is to capture the notion that all the facts are determined by the physical facts without implying certain kinds of reductionist claims. Levels of Nature At the heart of most physicalistic accounts of the special sciences is the claim that nature has a hierarchical structure (Oppenheim and Putnam 1958, Simon 1969, Wimsatt 1976, Lycan 1987, Collier 1988). At the (unique) lowest level, there are the fundamental microphysical particles. At higher levels are things like molecules, cells, multicellular organisms, planets, and galaxies; higher level entities are composed of lower-level entities. In general, an object from level n+\ will have objects from level n as its highest-level parts9. The special sciences, on this view, are distinguished from physics by the fact that they deal with different levels of nature. In Oppenheim and Putnam's classic account (1958), the hierarchy of ontological levels is reflected in a hierarchy of sciences: (micro-) physics, is the fundamental science, while chemistry, cellular biology, and sociology all deal with higher levels of nature (cells, certain kinds of multicelled organisms, and groups of psychological individuals, respectively). Reductionism is the thesis that the theories at higher levels are reducible to (explainable in terms of) lower-level theories, and, via the transitivity of the reducibility relation, to microphysics. It is usually added that no object at a given level has objects from higher levels as parts. 10 Additionally, objects from higher levels will often have properties .which lower-level objects lack: only things from the chemical level have a valence, or are reducing agents, while only things from the biological level are alive, and so forth. Thus, many properties are 'emergent' in some sense, but this is not a sense which threatens materialism. By "properties", I mean to include also relations — for example, can successfully interbreed is a relation which 'emerges' at some biological level. In general, I will call such properties "higher-level" properties, since they only apply to (or are only instantiated by) objects at some level higher than the microphysical. Supervenience theses are often formulated as relations between higher level-properties and microphysical properties. Although Oppenheim and Putnam's account of levels is simplistic, the core of it has survived intact. Oppenheim and Putnam (O & P) distinguish only six levels, the microphysical, the atomic, the molecular, the celluar, the organismic, and the sociological. Further, their 'layer cake' model seems to miss out whole branches of science, such as the /wacroscopic /^ organic sciences of geology and astrophysics. Wimsatt (1976) argues that we can expect the structure of the sciences to be more like a branching tree (p. 253), where objects at one level (say, one of the molecular levels) can compose objects which are studied by different sciences (e.g. molecular biology vs. geology). Things from higher levels have things from lower levels as parts; this alone, however, does not tell us enough about the nature of levels. As Oppenheim and Putnam (1958) point out, we can specify the part-whole relation in a number of ways10. One of their suggestions is that we take "x is part of y" to mean something like "x is spatially contained 1 0 Or, if you prefer, there are a number of part-whole relations. 11 in y"11. As Wimsatt (1976) points out, the objects at a given level will all be of roughly the same size, so this specification seems like a good start. It breaks down in some cases, however. Wimsatt, for example, takes it that black holes are from a higher level than are bacteria, yet theory allows for the existence of black holes smaller than many bacteria. However, the physicalist places the further requirement that the set of objects at a given level be 'natural' in some sense. Cells, while they may have somewhat 'fuzzy' spatial boundaries (from the perspective of quantum mechanics), seem to be genuine individuals, while the object O composed of seventeen particular cells from Jean Chretien's liver, four square miles of the Skeleton Coast, and the stuffed body of Jeremy Bentham does not. By the rough account of the part-whole relation sketched above, both cells and objects like O are perfectly respectable objects, since they are both wholes composed of microphysical entities and have (tolerably clear) spatiotemporal boundaries. But what is it that distinguishes "natural" wholes from "unnatural" ones? The 'levels'view seems to lead us right into one part of "the" problem of "natural kinds", viz. distinguishing natural merological sums from unnatural ones. First, as Simon (1969, p. 89-90) notes, higher-level objects have parts which interact in characteristic ways. For instance, there are relatively strong intermolecular forces between the molecules which comprise tables (and all macroscopic 'physical' objects, for that matter); between celestial objects (e.g. galaxies), the primary form of interaction is gravitation, while nations are composed of parts which are tied together by legal and social 1 1 Four-dimensionalists can read this, if they like, as "x is spatiotemporally contained in / ' , as this modification makes little difference to the discussion which follows. 12 relations. We are reluctant to regard mereological sums as objects (individuals) without relatively tight relations between their constituents, O being a case in point. Wimsatt (1976) employs the metaphorical account of the aims of science as "carving the beast of nature at its joints": With a joint, if you cut to the left or right of it, the going is a lot harder. Similarly, it seems at least roughly true of a level that if one, starting with that size "mesh", makes the mesh continuously bigger or smaller and tries to construct theories with the entities that result (many of which we would not consider to be entities at all), the going will be a lot harder ... I assume that this means either that the theories are more complex, or the phenomena are less regular, or usually both, for an imaginary theory at the slightly larger or smaller size scale.(Wimatt 1976, p. 238) Wimsatt suggests that we regard a level as "a local maximum of predictability and regularity" (1976, p. 238, italics original). Roughly, the idea is that within a certain range of sizes, certain mereological sums of microphysical particles can be more usefully and economically dealt with than others because the behavior of those mereological sums is more regular than that of other mereological sums of the same size. In a sense, then, theory is a guide to what objects there are at a given level. The more predictively useful a given theory is, the more natural the objects discriminated by that theory are. The idea that the levels are 'natural' is bolstered by the consideration that as products of natural selection, we should expect that we are fairly reliable detectors of regularities (at least those important for our survival) (Wimsatt 1976, p. 238). Wimsatt's account of levels seems to make them dependent on which way of 'carving things up' gives the best theories ~ but 'best' by whose lights? Wimsatt, I think, is quite happy to say that they're our theories, and we judge them by our criteria of simplicity, elegance, explanatory and predictive power, etc.. Note, however, that Wimsatt's account 13 also depends on whether the phenomena at a proposed level are regular, and this seems to call for some account of what constitutes a genuine regularity in nature. For philosophers of realistic persuasions (which includes, as far as I can tell, all contemporary physicalists), the problem of "natural kinds" is still a live, and very important issue. We take it that some collections (in the mathematical sense) of objects delineate natural kinds, while others do not ~ but what is it that distinguishes such natural collections from the unnatural ones? I don't pretend to have a satisfactory solution to the problem; indeed, I find it so intractable as to think that we're better off without the notion of "natural kinds" in the classic sense. We could take a modified Goodmanian (1954) line here and just say that 'natural' kinds are the ones which best suit our parochial purposes. I do think that some such doctrine is the most defensible one here, but I will not pause to defend it, only to flag this as a difficulty faced by philosophers who are altogether too 'realistic' about natural kinds12. There is much more to be said about 'levels', but I have restricted myself here to what is common in various writings. What matters, for my purposes, is that: 1) There is more than one level of nature, but only finitely many (Oppenheim and Putnam 1958) 2) Microphysical entities are at the unique lowest level 3) Objects at higher levels are composed completely of objects from lower levels 4) The purpose of the special sciences is to find and describe the regularities holding between higher-level objects. I am unsure as to whether to include Wimsatt in this class; much of what he says does seem to commit him to a fairly strong version of realism about natural kinds, but in other places, he refers to his position as "Kantian", which suggests a view closer to the Goodman/Quine account. 14 The Closure of Physics A materialist position should certainly assert... that physics is causally complete (i.e. all fundamental causal forces are physical forces, and the laws of physics are never violated) (Horgan 1993a, p. 560) "Casual completeness", or something like it, is by many accounts a central thesis of materialism. Other philosophers who have supported its centrality are Kim (1992a, 1993b), Papineau (1993), and Field (1991). Indeed, the growing literature on "mental causation" revolves around the theme of causal completeness (see, e.g. Fodor 1989, Kim 1989, Marras 1994) and the problem of accounting for the causal efficacy of mentality it apparently raises. My purpose in this section is not to discuss that literature (although I will have something to say about it later), but rather to get a bit clearer about what the principle of causal completeness states. I think Horgan's brisk account (which is fairly representative) suffers from ambiguities, which I will try to rectify. I cannot claim full originality for my efforts here; my attempts at clarification are inspired, in the main by Brian McLaughlin's 13 (1992) excellent paper on "British emergentism" . I think that causal completeness is best understood by contrasting it with metaphysical doctrines which oppose it, i.e. Cartesian dualism and emergentism. Cartesian dualism, however, is ruled out by the bare-bones compositional principle CP stated above, and so an account of causal closure need not rule out causation between mental and physical substances. The contemporary and not wholly imaginary opponent of causal completeness is the emergentist, who agrees with CP but adds that there are fundamental, 'non-physical' 1 3 Well, let me toot my own hom a bit here: McLaughlin does not discuss "causal completeness" in so many words, although he does mention "downward causation". 15 forces which are exerted only by complex structures of microphysical particles (e.g. Sperry 1986, Legget 1987). A full contrasting of orthodox materialism with emergent materialism will have to wait for a subsequent chapter, but I want to give the reader an idea of the background concerns in the present discussion, (i) ... all fundamental causal forces are physical forces... One problem here is what a "physical force" is. If it means 'force which acts on microphysical particles', then the thesis is plausible, but unhelpful. Suppose, for a moment, that emergentism is true, and there is a law to the effect that thinking about Vienna (a non-physical property which emerges from certain arrangements of microphysical particles) creates a gentle eddy in the brain, which is sufficient to move free electrons several nanometers before they are stopped by electrostatic forces. The law tells us of a force which acts on microphysical particles, so the force is on the present account a physical force. This is, apparently, not what Horgan has in mind. We need some non-trivial way of picking out the physical forces from the others. As McLaughlin (1992) points out, Newton's three laws and Schrodinger's equation only tell us how a body will act under the influence of forces ~ they do not tell us what forces influence the behavior of bodies. Newton's laws and the law of gravity are not necessarily violated if two massive objects in an otherwise empty universe fail to be accelerated toward one another — for example, suppose that the masses, distance, and charges on the objects are such that a mutually repulsive electrostatic force equal in magnitude to the gravitational force also acts on the objects. It is at least conceivable that structured collections of microphysical particles could exert forces which act on individual 16 microphysical particles (and some contemporary physicists have suggested that there are such forces). I think the best way of cashing out "physical force" here turns on something of a bet on what form physical laws will take: the fundamental forces of physics are all forces that can be exerted by one microphysical particle on another. Thus, the net force on a particle is the vector sum of the forces exerted on that particle by all the individual particles in the universe. The restriction to forces exerted by single microphysical particles is crucial here, because it is what rules out the possibility of 'configurational forces' possessed by structures of microphysical parts. However, it is worth discussing weakening the claim on which this bet is based; Leggett (1987) has suggested that the problem of accounting for the apparent determinacy (classicality?) of 'manifest' and other non-quantum phenomena in terms of quantum mechanics with its indeterminacies might require the postulation of something resembling McLaughlin's "configurational forces" — determinacy may 'emerge' from complex systems of quantum mechanical entities14. Even if Legget's (tentative) theory is correct, however, there still might be some (critical) level (well below, say, lower biological levels) above which no new configurational forces or the like emerge. We could reformulate a doctrine resembling orthodox materialism in terms of the processes and laws of the critical level. Even if microphysics is not "causally complete", some higher (physical) level might be. Such a 1 4 Leggett's thesis is not exactly that there are configurational forces, in the sense in which the electroweak force is a "force"; he argues that complex systems of fundamental particles may possess some property which renders the behavior of those complexes determinate in a way that behavior of individual particles are not. The non-committal but vaguer term "causal power" seems more appropriate in this context. 17 doctrine is still of considerable interest, as it does not attribute emergent causal powers to mental or biological entities. (ii) .. the laws of physics are never violated... This claim suffers from the same problem vis a vis "physical law" as (i) does with "physical force". There is, however, another problem here. We are all familiar with the notion of laws with restricted domains of applicability: "the second law of thermodynamics" applies only to closed systems, the Wiedmann-Franz law applies only to certain metals, and so forth. We do not regard it as a violation of the second law of thermodynamics if entropy spontaneously decreases in some region of space-time (this may well have happened on Earth with the evolution of complex forms of life). Similarly, the laws of physics are not necessarily violated if the behavior of some microphysical particles in some situation fails to conform to those laws, if those laws are only applicable to a restricted domain. It is, I think, part of the materialist conception of physical laws that they apply everywhere and always. On a generous reading of "never", we could read (ii) as "the laws of physics are not violated in any region of space-time", and I do think that this is what Horgan had in mind. Is it not, after all, the microphysicist's job to come up with accurate models of the behavior of microphysical particles? The emergentist is saying that, as a result of 'emergent' forces, the behavior of microphysical particles in (and around?) certain structured collections will be different from the behavior predicted by microphysical theories which only allow for forces exerted between pairs of particles. Insofar as the emergentist is trying to carve out a home for special-scientific properties, he or she is stepping on the microphysicist's toes. On the emergentist's view, what actually happens may not be something that is possible, given microphysical theory. It is this feature which, I 18 think, disturbs the orthodox materialist. Philosophers are not forbidden from framing empirical hypotheses, but as it goes, the claim that microphysics is in this sense uncompletable is a rather strong one, and strong hypotheses require a great deal of support15. 1 5 Of course, this also applies to materialism. 19 C h a p t e r T w o : R e d u c t i o n i s m ... the word "reductionism" seems by now to have acquired a negative, faintly disreputable flavor ... [b]eing a reductionist is a bit like being a logical positivist or member of the Old Left ~ an aura of doctrinaire naivete hangs over him (Kim 1989, p. 266) Introduction "Reductionism" refers to any number of views, but in this chapter I wish to discuss some versions of reductionism associated with "The Unity of Science". Said doctrine is usually associated with logical empiricism, although there are several versions of it (for two, see Carnap 1934, Oppenheim and Putnam 1958). Even with such restrictions in mind, "reductionism" stands variously for epistemological, methodological, and metaphysical theses. The first two are closely related; the epistemological claim is that every true scientific theory can be reduced to, or explained in terms of, true microphysical theory. Reductionist methodology is aimed at finding such explanations. Metaphysical reductionism, as I shall use the term, refers to the ontology both of particulars and of properties. CP, which was introduced in the previous chapter, captures much of the force of 'particular' reduction; property reductionism is the view that every property is a physical property. Unfortunately, these three views have become entangled in contemporary discussions of reductionism. 20 Most contemporary writers on materialism usually have property reductionism in mind; consequently, "non-reductive materialism" refers to views which puport to be materialist, yet admit the existence of properties which are in some sense not physical. However, the arguments directed at "reductionism" are supposed to defeat not only this metaphysical version of reductionism, but also the epistemological and methodological versions as well16. Like Wilson (1985), I think that this is a mistake; the issue of property reductionism is largely independent both of methodological and epistemological reductionist theses. The current confusion stems, I think, from a failure to appreciate certain facts about the empiricist tradition, in which the best known (but certainly not only) accounts of intertheoretic reduction were proposed. It is not often remarked that the logical empiricists were not, as a matter of principle, directly concerned with ontological matters. Logical empiricists distinguished themselves by their concerns with matters of formal logic (hence "logical") and observation (hence "empiricism"). Insofar as ontological issues such as that of "materialism vs. dualism" could not be decided by applying logical techniques to statements about observations, empiricists rejected them as "pseudo"-issues not worthy of discussion. As a result, reduction was usually defined as a relation between theories and only derivatively a relation between entities. Within the empiricist tradition, ontological unification emerges as more of a byproduct of successful reductions rather than the aim of reduction, which is to unify scientific knowledge and language. It is thus . "Particular reductionism", or CP, is accepted by nearly all philosophers who consider themselves physicalists (but see Haugeland 1982, 1984). 21 not all that surprising that empiricist accounts of reduction fail to satisfy philosophers of a more 'realistic' bent. Successful or completed reductions represent scientific progress. Reductionists are divided on the benefits of successful reductions, but for a start we can distinguish the following kinds of progress which have been suggested: 1) A more unified and deeper understanding of the world (this more or less follows from the conception of reduction as explanation). 2) Increased conceptual economy; reductions show us that the conceptual apparatus of the reduced theory is (at least in principle) eliminable in favor of the conceptual apparatus of the reducing theory (e.g. Kemeny and Oppenheim 1956, Hempel 1965b, 1969). 3) Ontological simplicity; the particulars and properties postulated by the reducing theory are shown to be responsible for the phenomena which the reducing theory is able to predict and explain. Thus, we need not commit ourselves to the existence of higher-level entities that exist "over and above" the lower-level entities in order to account for those phenomena. 4) Instrumental gains; finding out which processes underlie certain phenomena may give us increased power to control the occurrences of those phenomena. Empiricists tend to emphasize (1) and (2) above the others, although (3) plays some role. In contrast, contemporary reductionists tend to take (3) as given, with (1) and (4) serving as the primary goals or purposes of reduction. 22 In this chapter, I will discuss the formal account of theory reduction developed by Ernest Nagel (1961), and the "unity of science" program defended by Oppenheim and Putnam (1958). I will then discuss the relation between intertheoretic and property reduction, paying special attention to some arguments of Nagel (ibid.) and Hempel (1969). Then, I will briefly lay out the main argument against property reductionism based on the "multiple readability" of "functional" properties developed by Putnam (1967) and Fodor (1968), focussing in particular on the influential treatment of it given inFodor(1974). Intertheoretic Reduction and Logical Empiricism Logical empiricists (or positivists, as they are sometimes called) are credited with giving philosophy of science many of its contemporary concerns and even some of its methods, including the insistence on logical rigor in philosophical argumentation. Most empiricist approaches to issues in the philosophy of science have been rejected by post-Kuhnian and Sellarsian philosophers as too narrow, historically naive, or at best unfortunately misguided. A central task of empiricism was to distinguish between sentences which made genuine, meaningful claims, and sentences which did not; Carnap's (1934) "syntacticalism" was one such attempt17, but more often, empiricists opted for some form of verificationism. Currently received opinion has it that empiricist attempts to draw a sharp boundary between the sensical and nonsensical ended in failure 1 . Carnap's syntacticalism was the doctrine that the only meaningful philosophical questions concern the syntactical relations between scientific statements. As Hempel (1969) points out, Carnap later rejected this thesis. 23 (Hempel 1965a). However, a preoccupation with the language(s) of science remained, and we find sophisticated empiricists (e.g. Hempel 1969, Nagel 1961), mamtaining that questions which concerned linguistic matters were more precise than questions which concerned things such as ontology: [The] linguistic turn is characteristic of philosophical misgivings about the ontological construal of the issues in question, and it is an attempt to explicate the latter by restating them in a clearer and philosophically more satisfactory fashion. (Hempel 1969, p. 180) If the basic materialist metaphysical view is correct, then all scientific theories are about microphysical things; to call something a "non-physical" entity is not to comment on its ontological status, but rather to comment on the linguistic status of the terms used to describe it (Hempel 1969, p. 181). Strictly speaking, there are no biological, social, or chemical entities, just biological, social, and chemical descriptions of entities which are also described by physical terms. The privileged status of physical theory stems from the fact that, for every correct description of some object or state of affairs in the language of any science, there will be a correct description of that same object or state of affairs in the language of (a true and complete) microphysics. The same cannot be said for any other scientific theory; meiosis, for example, is not a process which can be correctly described in cosmological or psychological terms. The ideal state of science, on the view developed by Oppenheim and Putnam (1958), is unitary science -- a single theory stated in a single language (i.e. some extension of microphysical theory), complete with explanatory and predictive principles to serve all scientific needs in those areas18. As A popular view among empiricists (e.g. Kemeny and Oppenheim 1956, p. 8) was that the only difference between prediction and explanation is that "prediction" is used to characterize a relation 24 Poland (1994) has remarked, this is a rather strong eliminativist position, since it states that all the purposes currently served by the special sciences could be served by a completed microphysics. On Oppenheim and Putnam's view, unitary science can be achieved by a series of /w/'croreductions; theory T B is a potential microreducer of theory T R provided that T B deals with the parts of objects dealt with by T R . In the language of "levels", microreductions take place between theories at adjacent levels. Microreductions are assumed to be transitive, so that it would not be necessary to reduce, say, cognitive psychology directly to microphysics; one would indirectly reduce such a theory to microphysics via "local stops"19 in cellular and then molecular biology, continuing on "down" through to microphysics. One upshot of a successful microreduction is reduction of ontology; the phenomena with which the reduced theory deals are shown to be explained and predicted by the reducing (lower-level) theory. Thus, only the lower-level entities need be admitted to exist. However, the main purpose of reduction is to achieve a greater "integration in scientific knowledge" (Oppenheim and Putnam 1958, p. 405), because the concepts (terms) employed in the reduced theory are (supposedly) shown to be eliminable in favor of the concepts (terms) of the reducing theory. With this general picture of the reductionist methodological program and its rationale in hand, I will now attempt to characterize what is meant by claims that one theory (micTo)reduces another20. between a theory and an event which has not yet occurred, while "explanation" applies to events which already have occurred. "The term is Fodor's (1974, p. 430) 20Microreduction takes place between theories at different levels; "same-level" reduction often involves historically successive theories (such as special relativity and classical space-time theory), and appears to possess a different "logic" (Nickles 1973, Wimsatt 1976). The main difference between microreducuon 25 A Formal Model of Reduction In the most general sense, reeuction is the explanation of one theory by another. There are, however, different accounts of what is involved in the explanation of one theory by another; different views on explanations and on the nature of scientific theories all but guarantee a multiplicity of distinct models of intertheoretic reduction. Nagel's own formal account is, I believe, based on the "received" views of theories as sets of sentences expressed in a formal language and the "deductive-nomological" account of explanation. On the "syntactic" view, a theory is an axiomatized formal system. The non-logical vocabulary of the language in which a theory is framed is divided into two parts, an "observation" and a "theoretical" vocabulary21. The theory is partially interpreted by "correspondence rules" which link terms in the observation vocabulary to observable states of affairs; the theoretical terms are implicitly defined by the basic axioms of the theory. In discussing intertheoretic reductions, it is necessary to assume that both theories involved have been formalized in the manner described above. This requirement is one of those idealizations philosophers permit themselves, and it is often not met in 22 practice . Let us assume, for the nonce, that every scientific theory can be axiomatized, its correspondence rules explicitly stated, and its observational and theoretical vocabulary clearly distinguished. The law sentences (or statements) of a theory are typically and so-called "successional" reduction is that the former relation is presumed to be transitive while the latter is presumed to be intransitive. 2 1 For brevity, I will sometimes refer to the terms of a theory. 22The importance of this point should not be underestimated; Rosenberg (1994) argues that the lack of axiomatized theories in biology reflects that science's instrumental character. 26 assumed to take the form of universal conditionals (or, in some cases, biconditionals). The sentences entailed by a theory which contain only observation terms comprise the theory's set of observational consequences. Nagel calls law statements which contain only observation terms experimental laws (1961, ch. 5). Supposing that we are given two such theories, T R and T B , we can now state Nagel's formal condition for reduction: Reduction^ T B reduces T R provided T R (or the experimental laws of T ^ can be derived from the conjunction of T B and a set of 'connecting principles'23 The introduction of 'connecting principles' into this definition appears to stem from formal considerations. In most interesting cases of microreduction, the theory to be reduced (T&) will contain terms not occurring in the theory T B which is supposed to reduce it24. In such cases, it is a mere point of logic that (non-trivial) sentences in T R will not be derivable from the sentences of T B (the primary science) alone25. For example, one 'experimental' law of thermodynamics tells us that increasing the pressure on an ideal gas while keeping its temperature constant results in a decrease in the volume of the gas. None of the terms "temperature", "pressure", or "volume" occur in the vocabulary of statistical mechanics. In order for the derivation to go through, then, additional "connecting" assumptions must be produced which link the proprietary terms of T R to terms in T B . It is often assumed that Nagel held that connecting principles must be biconditional in form, although this assumption is manifestly false (see also ^Kemeny and Oppenheim (1956) proposed a weaker formal account on which T B reduces T R provided the observational consequences of T R can be derived from T B. O & P (1958) employed a weakened version of the Kemeny-Oppenheim model. 2 4I will adopt the convention of calling the higher-level theory the "reduced" theory or T R and the lower-level theory the "reducing" theory or T B . 2 5 By "non-trivial", I mean to exclude such examples as disjoining some theorem of T B with some expression which contains terms from T R (cf. Nagel 1961, p. 353n) 27 Richardson 1979); Nagel clearly allows for the existence of 'one way' conncecting principles (Nagel 1961, p. 354, p. 355n.). Nagel's connecting principles are now usually referred to as "bridge laws". The contemporary term is metaphysically loaded, if taken literally: laws relate natural kinds or properties. Talk of "bridge laws" should thus be taken with a grain of salt when it is used to characterize empiricistic accounts of intertheoretic reduction. For his part, Nagel refused to take a stand on what connecting principles signify; this is, I submit, a reflection of a Hempelian attitude towards ontological construals of reduction. Nagel considers three construals of connecting principles: first, that the term linked by such principles are (prior) synonymy claims, second, that they are conventional redefinitions (e.g. "Henceforth, 'temperature' will stand for 'mean molecular kinetic energy'"), and third, that they are "factual hypotheses" which state only that the presence of conditions describable in terms of T B are sufficient (or necessary and sufficient) conditions for the presence of conditions describable in terms of T R (Nagel 1961, p. 354). He rejects the first on the grounds that it is extremely implausible to maintain, e.g. that "temperature" means, and always meant, "mean molecular kinetic energy". However, Nagel argues that both of the remaining hypotheses have something to be said for them, and ends up maintaining that deciding between them depends on the context in which the reduction takes place (Nagel 1961, p. 356). Other empiricists were not so cautious as Nagel; Hempel (1969, p. 189), in particular, argued that if intertheoretic reductions were to result in greater conceptual simplicity, bridge laws must be assumed to be biconditionals. 'One-way' bridge laws, on Hempel's view, simply add to the stock of 28 laws of the reduced theory (Hempel 1969, p. 189). Biconditional bridge laws (BBLs), however, allow for the replacement of the terms of the reduced theory by terms from the reducing theory. Theory and Ontology The bridge between the linguistic construal of reduction and the ontological construal is forged out of the 'sophisticated' empiricist approach to metaphysics. Unlike earlier empiricists, Hempel and Nagel did not argue that ontological claims (such as claims of property identity) were meaningless. Rather, they emphasized the epistemological difficulties such claims faced when they were made without the backing of a theory. Nagel expresses the view in the following passage: ["Nagel's principle"] ... the "natures" of things, and in particular of the "elementary constituents" of things, are not accessible to direct inspection and ... we cannot read off by simple inspection what it is they do or do not imply. Such "natures" must be stated as a theory ... (Nagel 1961, p. 364)26 In other words, questions about the metaphysical relations between objects can be fruitfully discussed only by reference to the linguistic relations which hold between the descriptions of those objects. Nagel's principle asserts a kind of isomorphism between metaphysical relations and linguistic ones, and is closely related to Hempel's claim, above, that "biological" and "chemical" apply to different kinds of descriptions of objects, not to different kinds of objects. Nagel's principle is, I believe, at the heart of 26Hempel's version of Nagel's principle reads: "... the distinction of biological from physical, chemical, and other kinds of items applies to individual things and events and to kinds or classes of things only "under a specific description", i.e., only in so far as they have been characterized by means of a terminological apparatus distinctive of a certain scientific discipline." (Hempel 1969, p. 181) 29 the view that materialism implies reductionism. This is a point to which I will return below. With the resurgence of metaphysics which accompanied the waning of empiricism, the focus of the debate over reductionism shifted to talk of the relations between properties at various levels. It became reasonable to ask what connecting principles say about the world, and one answer gained popularity: "bridge laws" assert property identities. The most famous putative example of such an identity comes from Nagel's discussion of thermodynamics and statisitical mechanics: temperature is mean molecular kinetic energy21. Property identities were also held to be at the heart of the so-called Identity Theory of mind. We get, from Identity Theory, another famous (albeit fictitious) property identity, namely that being in pain is having one's c-fibers fire. In the literature, Identity Theory (and thus property reductionism) are often referred to as "type-type" or simply "type" physicalism (Fodor 1974, Horgan 1982, Wilson 1985). Why did the property identity answer become so popular? First, it was thought that 'mere' BBLs were simply too weak; even if being in pain is everywhere and always accompanied by the firing of c-fibers, it was thought to be possible that being in pain could be a non-physical property, a la classical property dualism. The same reasoning applies to "law-like" BBLs, which state that the coextensivity of (e.g.) pain-havings and c-fiber firings is necessary. Further, the explanatory value of 'mere' BBLs was questioned ~ reductions are supposed to explain higher-level theories, rather than merely allow their derivation. Causey (1977) argues that the introduction of 'mere' 2 7 Although, as I have noted, Nagel himself does not make this identity claim. 30 biconditional bridge laws introduces another mystery at least as objectionable as the one the reduction is supposed to remove. Here, we see another facet of the rejection of the formalistic philosophy of the empiricists. On the "D-N" model of explanation (Hempel 1965b, Nagel 1961, ch. 4), an explanation consists of a set of sentences (collectively known as the explanans) which includes at least one law-sentence and which deductively entail the sentence describing the fact to be explained (the explanandum-sentence). The D-N model is now widely rejected, and one of the reasons is that there are derivations which meet the criteria of the model, yet are not (intuitively) explanatory. Hooker (1981) considers the Wiedmann-Franz law, which says that (for a wide range of metals), the thermal conductivity is proportional to the electrical conductivity of the metal. Given the thermal conductivity of, say, copper, we could derive its electrical conductivity. However, it does not seem as if such a derivation tells us why copper has such and such electrical conductivity. So it appears that derivation is insufficient for explanation28. In contrast, property identities pack the right sort of explanatory punch the reuductionist is looking for: what better explanation could one have for the universal copresence of property A with property B than that A and B are really the same property? And, unlike BBLs, identities do not themselves require explanation. Insofar as one is a realist about properties, identities also yield ontological parsimony29. Incidentally, the same example can be used to show that the difference between prediction and explanation is not merely pragmatic, for we could predict the electrical conductivity of a bit of metal given its thermal conductivity and the Widemann-Franz law. For detailed arguments concerning the importance and role of property identities in intertheoretic reduction, see (Hooker 1981, pp. 201-236). 31 The Functionalist Challenge The demise of both property and theory reductionism coincided with the rise of functionalism in the philosophy of mind. Early functionalists drew an analogy between minds and computers — Putnam (1960, 1967) went so far as to suggest that mental states are analogous to Turing machine states. More recently, functionalists have put a more teleological (in the Darwinian sense) spin on "function" (e.g. Lycan 1987, Papineau 1993). In the most general sense, a view is functionalist if it individuates types of states, properties, or things on the basis of the causal relations those states, properties or things typically bear to other states etc. To make a long story short, functionalists individuate types of things on the basis of what they do, not on what they're made of. Functionally individuated properties are supposed to resist reduction because they can be instantiated by particulars which are (intuitively speaking) quite dissimilar. Functionalism has crept outside of the philosophy of mind, and some are advocating the view that all special-scientific properties are functional properties (e.g. Lycan 1987, Melnyk 1993). If functional properties really do resist reduction to lower-level (aka "structural") properties, then non-(property) reducibility may be a fairly widespread phenomenon. Functional types are multiply realizable: an object which has a specified range of causal powers could be constructed in any number of ways. As we all know, there is more than one way to skin a cat — a cat-skinner could be made out of wood, metal, plastic, ceramic, or for the wealthy taxidermist into automation, out of semiconductors and lasers. There are less macabre examples in the literature; something is an adder if it takes two numbers as input and outputs their sum as output. An adder could be made out 32 of pencil and paper, neurons, beads and wire, semiconductors (they're so versatile!), or, for the really ambitious, out of string and beer cans. As Putnam pointed out early on in his writings on the subject, functionalism pushes the ontological debate between dualists and monisitic materialists to one side: "we could be made of Swiss cheese and it wouldn't matter" (Putnam 1973, p. 291) Putnam's (1967, p. 436) example is pain. Pain-states are caused by bodily damage, give rise to beliefs that one is in pain and desires to be away from the offending stimulus, and, usually, though not always, give rise to 'avoidance behavior' (i.e. attempts to get away from the thing causing the pain). It seems obvious that creatures with different sorts of brains experience pain — mammals (themselves a heterogeneous group), reptiles, and even some molluscs (octopi). Additionally, Putnam points out, it seems plausible that aliens with brains based on some other kind of chemistry or even robots could experience pain. Moreover, the neurological or semiconductor states which "realize" pain in a given individual may change over time (due to brain damage or the effects of aging, perhaps). So, to identify being in pain with (as Smart (1959) seemed to do) having one's c-fibers fire is apparently not on. In one of the most prominent and influential papers on reduction, Fodor (1974) argues against a view he takes to be the standard account of reduction; said view bears a close resemblance to reductionN with the requirement that bridge laws express law-like or nomological coextensions or property identities. To keep the discussion simple, Fodor fixes his attention on the derivation of the law-statements of the reduced theory from the law-statements of the reducing theory. Suppose we have some law-statement in the 33 language of some special science "(x) (S,x -> S2x)". Call this statement L s . Fodor claims that the reductionist requires that there be a single law statement of microphysics of the form "(x) (F^ x -> P2x)" (LP) and two bridge laws of the form "(x) (T x^ <-> Six)", "(x) (P2x <-» S2x)", in order to permit the derivation of L s from L P . Fodor points out that, on the standard conception of laws and law-statements, P t and P 2 must be "kind-predicates" of microphysical language30. Multiple readability supposedly counts against reductionism because it forces the bridge laws to take on a disjunctive character, e.g. (x) (Sx <r+ P tx v P2x v P3x v...v Pnx), where S denotes the multiply realizable special scientific property and the individual P; stand for microphysical configurations which realize S. Worse yet, we have no guarantee that the disjunction on the right hand side (contrary to the way I have represented it here) is not indefinitely long. Fodor claims that whether or not it is indefinitely long, the expression on the right-hand side of the biconditional (which I will denote by P) does not denote a "natural kind", and many philosophers have followed him in this contention. Another way fo saying this is that P and its ilk are supposed to tell us what it is that things which satisfy S have in common, and disjunctive expressions do not do this (at least not in any intuitively satisfying way). The general consensus among philosophers is that Fodor's argument establishes that many, if not most, special-scientific properties cannot be identified with physical 31 properties . For Fodor, the natural kinds of a given theory are expressed by the predicates which appear in the law-statements of that theory. Property identities J U Of course, SI and S2 must be kind-predicates of the special-scientific language in question. 31Fodor uses the terms "property", "kind" and "natural kind" interchangeably. 34 supposedly require that bridge laws link the kind-predicates of physics with the kind-predicates of the special sciences in a one-one fashion, but such one-one matchings are precisely what the multiple realizability arguments preclude. It certainly does not seem as if the law sentences of present or even a 'completed' microphysics will contain wildly disjunctive expressions such as the ones which would have to appear in bridge laws. In cases where the number of disjuncts in P is infinite or even indefinite (as is presumably the case with any bridge law relating 'is a monetary exchange' to microphysical expressions), the special-scientific predicate will not even be definable in the language of microphysics. Various objections to Fodor's argument have been raised. Paul Churchland (1981, 1985) argues that multiple realizability arguments establish that functional kinds are not natural kinds, disputing Fodor's relativization of the notion of natural kinds to particular sciences. Jaegwon Kim (1984, 1989) argues against the claim that disjunctive expressions in the language of a given science do not denote natural kinds of that science. There are also good questions about whether explaining one theory in terms of another requires identifying the kinds of one theory with the kinds of another. Certainly, the term "natural kind" is absent from the empiricists' discussions of intertheoretic reduction32. For these reasons, I find it quite difficult to establish the import of Fodor's argument. One thing which is clear, however, is that the inference from the disjunctive nature of 32 * * * * I think it is particularly embarassing for Fodor's argument, and those who have accepted his claims about "reductionism" that Oppenheim and Putnam, at whom Fodor's critique is implicitly directed, explicitly disavow that bridge laws must be expressed by biconditionals, and (implicitly) that predicates must be matched one-one by bridge laws: "... the notion of reduction we shall employ ... is designed to include reduction via biconditionals as a special case" (Oppenheim and Putnam 1958, p. 405). The issue of "natural kinds" and Fodor's definition thereof, will make a brief reappearance in the fourth chapter. 35 bridge laws to the failure of property reductionism requires a principle of the following sort: [Pred] all microphysical properties are expressed by predicates which are definable in terms of predicates which appear in the law sentences of microphysics. On the face of it, Pred gives a syntactic criterion for membership in what is ostensibly a metaphysical category, which is one good reason to be suspicious of it. As Clifford Hooker puts it, ... it is alleged that infinitely many, worse, indefinitely many different bio-chemo-physical states could correspond to the economic predicate 'has a monetary system of economic exchange; and similarly for the property [sic] 'has just won a game of tennis'. Yet one doesn't want an economic system or a game of tennis to be some ghostly addition to the actual bio-chemo-physical processes involved ... (Hooker 1981, p. 496) I propose, however, to provisionally accept it and see where the conclusion that there are non-physical properties takes us33. If we do accept Pred, it seems that Fodor's argument establishes that the representational power of the language of microphysics may not be up to the task of providing explicit definitions for every term in a true special science (cf. Hellman and Thompson 1975, p. 551). The latter conclusion certainly does count against Oppenheim and Putnam's dream of unitary science and against Hempel's ideal of achieving conceptual simplicity via finding such definitions. As Fodor puts it: One reason in favor of accepting Pred is that we expect the language of physics to be adequate to its subject matter, what motivation could we have for calling a property for which no expression can be found in the language of physics a physical property? Mark Wilson's (1985) challenges the assumption that syntactic criteria delimit the range of physics' properties; Wilson's view will be discussed in the fourth chapter. 36 Physics develops the taxonomy of its subject matter which best suits its purposes ... [b]ut this is not the only taxonomy which may be required if the purposes of science in general are to be served: e.g., if we are to state such true, counterfactual supporting generalizations as there are to state. (Fodor 1974, p. 440) In a nutshell, Fodor is claiming that physics lacks sufficient representational power to perform the predictive and explanatory tasks currently performed by the various special sciences, and thus that the special sciences are indispensible. Fodor's goal is to establish the epistemological claim that the special sciences are conceptually and methodologically autonomous from physics (and thus that the former are ineliminable in principle). The metaphysical claim that there are non-physical properties is regarded as a consequence of the autonomy of the special sciences (and Pred). Traditionally, however, materialism has been regarded as inconsistent with property dualist views (and, by extension, property pluralist views). Non-reductive materialists generally accept property pluralism, but argue that non-physical properties are materialistically acceptable if they supervene on physical properties (i.e. no two objects can differ in any of their non-physical properties unless they differ in their physical properties). In the next chapter, I will discuss some arguments directed at the claim that "supervenience materialism" can deliver on this promise. 37 C h a p t e r T h r e e : S u p e r v e n i e n c e a n d E m e r g e n c e Convinced of the bankruptcy of both property and strong methodological reductionist views, would-be materialists have turned to the notion of supervenience in hopes of specifying an asymmetric depedency relation between microphysical properties or facts and those of the special sciences. The slogan of supervenience materialism (SM, for short34) is "no difference without a microphysical difference". However, as time and much hard work have shown, this slogan can be cashed out in a bewildering number of ways. Recent writings on SM evince a growing distrust of the view: would-be materialists are wondering whether SM really expresses what materialists want to say, or whether claims of supervenience need to be strengthened in some way. A prominent worry along these lines concerns the related notions of causal completeness and emergence; the more popular versions of supervenience seem compatible with the existence of "emergent" properties whose instantiation would violate the causal completeness of physics. In this chapter, I will examine this worry and a few others which have been raised in the recent literature, paying particular attention to the arguments in (Horgan 1993 a). Supervenience Materialism: The Consensus View The different ways of cashing out the sloganistic account of supervenience amount to specifications of the relata of the supervenience relation. The idea the 3 41 will refer to proponents of SM as "SMists" 38 supervenient materialist is trying to capture is that the microphysical facts determine or fix all of the facts. I think it is useful to compare supervenience to what John Earman (1986, p. 13) refers to as "futuristic determinism" (cf. Horgan 1982, p. 34). The inmitive idea behind futuristic determinism is that the way the world is now determines or fixes what will happen in the future. Let £ be the class of all physically possible worlds, i.e. the set of worlds which satisfy the laws of physics which obtain in the actual world Then world W e £ is futuristically deterministic if and only if all Wei, that agree with W on any time-slice agree with W on all future time-slices. Earman's definition is only useful insofar as the notion of two worlds "agreeing on a time-slice" is clear; similarly, supervenience makes sense insofar as we have a relatively clear account of what differences of a certain type are. Supervenience theses are supposed to assert what we might call "vertical determinism": the microphysical facts fix all the higher-level facts. SMists typically reject global property reductionism but accept that some special-scientific properties may indeed reduce to microphysical ones (e.g. the famous temperature I mean kinetic energy case). Thus, SMists want to define a relation which is compatible with, but does not require, the reducibility of higher-level properties to lower-level ones. SM usually (although not invariably) consists in two distinct theses. One is an ontological principle to the effect that "everything is physical"; this might mean something like CP ("everything is completely composed out of microphysical particles") or, more perspicuously, Hellman and Thompson's (1975) "Principle of Physical Exhaustion" (PPE), which states that everything belongs to a set-theoretic hierarchy 39 based upon microphysical individuals35. The second principle is the supervenience thesis itself. This is the "consensus view" I described briefly in the first chapter. One philosopher who lays claim to materialism and disputes the consensus view is John Haygeland (1982, 1984). Haugeland claims that materialism merely requires a version of DP, and that stripped-down "identity theses" such as CP or PPE introduce extraneous elements into the formulation of materialism. One point in favor of Haugeland's view has been touched upon already: how is the materialist to construe talk of non-concrete entities such as national economies, clubs, bank accounts, and penchants for the macabre? Haugeland's attitude towards PPE is summed up in the following passage: Who knows what gawdawful set (or number) is identical (by someone's lights) with "The Sour Grapes" or the Pittsburgh Quilters' Triangle? Who knows how predicates like 'is a temporal individual' might be construed so as to apply to them? And who cares? (Haugeland 1984, p. 12; emphasis original) Whether or not Haugeland knows (or cares) which set in a Hellman-Thompson hierarchy is identical with Milton Berle's favorite joke, Hellman and Thompson care very much that there is such a set. Materialists want to say something like the following: true sentences containing apparent references to "bank accounts" and the like are made true by the way the fundamental microphysical constituents of the world are arranged. The materialist's acceptance of DP and CP are linked: microphysical facts determine all the " Specificially, Hellman and Thompson define "mathematical-physical entity" (my "microphysical individual") as "anything satisfying any predicate in a list of basic positive physical predicates of [a specific language of physics] L" (Hellman and Thompson 1975, p. 553). The hierarchy is based upon all parts of the fusion of such entities (i.e. all parts of the physical universe) so as to include not only 'standard' physical objects such as tables but also "crazy" objects like O (see chapter 1). Properties and even possibilia occur as sets in the upper reaches of the hierarchy. See Hellman and Thompson (1977) for a full presentation. 40 facts because, roughly speaking, there's notiiing outside of the entities countenanced by microphysics to serve as 'truthmakers' for assertions about bank accounts, snide comments, and so forth. , Besides the rather vague worry about truthmakers, however, there is more to be said for the consensus view. Haugeland's main argument for junking PPE and its ilk is that such principles serve only to rule out metaphysical doctrines which are "brilliantly kooky" (1984, p. 10) « Spinozistic monism, parallelism, epiphenomenalism, and so forth. However, my interests in this essay are with defining materialism, which involves setting it apart from other doctrines. The nearest Haugeland comes to defining materialism is: "Any metaphysical position which a) asserts that the microphysical facts determine all the facts and b) is not brilliantly kooky". I take it that such a definition is unsatisfactory, not least because it's not clear what qualifies a doctrine as "brilliantly kooky" — until recently, materialism was thought to fall into this class by the majority of philosophers. If what we're interested in is defending materialism, on the other hand, then maybe all we'd need to show is that special-scientific differences are invariably accompanied by microphysical differences, although how we'd do that is in itself a difficult question. Supervenience Much of the debate over supervenience concerns the domain of "the" supervenience relation. Hellman and Thompson (1975) define determination36 over the Determination is the converse of supervenience; B determines A if and only if A supervenes on B. 41 class of models representing scientific possibility; Kim (1984, 1987) and Horgan (1982, 1993a) define it over sets of properties and possible worlds. John Haugeland (1982) defines it over sets of truths expressible in the languages of the various sciences. By varying the domains for which the relation holds, different theses are generated. Letting A stand for one set of properties and B for another, A-properties supervene on B-properties if and only if two objects cannot differ in their A properties unless they also differ in their B properties. In such cases, A is the supervenient domain and the properties included in it are the supervening properties, while B is the subvenient or base domain and the properties included in B are the subvening or base properties. Haugeland-style definitions range over languages: A and B stand for specified languages, and say that A-statements supervene on B-statements if and only if any worlds or objects which are discernible with A are discernible with B31. Hellman and Thompson define determination over a set of model-theoretic structures which represent scientific possibility (which they denote by a) — given the languages of two sciences, 4> and Y|/, <b-truth determines \\f-truth if and only if any two structures in a which assign the same truth values to the same sentences of <j) assign the same truth-values to the same sentences c 38 of \}/ . As if these variants weren't enough, James Klagge (1988) distinguishes between "ontological" and "ascriptive" supervenience. Ontological supervenience holds between Two things are discernible with language L if and only if there is a sentence in L which is true of one of those things and which is not true of the other (Haugeland 1982, p. 97). I take it that this (more or less) ordinary language formulation captures the essence of Hellman-Thompson determination. For a more rigorous treatment, the reader is reffered to Hellman and Thompson's (1975) 42 domains of properties or facts, while ascriptive supervenience is a normative constraint on the application of certain kinds of terms to objects. For example, Hare (1952) claimed that normative (ethical or aesthetic) judgements supervene on non-normative ones, by which he meant it is a misuse of language to say of two things which are exactly alike in all non-normative respects differ in that one is good or beautiful and the other is not. Ascriptive supervenience is often held in conjunction with non-realism about the supervening properties or facts, although it need not be (Klagge 1988). Most materialists want to be realists of some sort, and hence intend ontological supervenience. I do not think the Hellman-Thompson/Haugeland definition specifies a variety of ascriptive supervenience, since the former is apparently not intended as a normative constraint only 39 as a statement of fact . I will refer to Hellman-Thompson/Haugeland varieties of supervenience as descriptive supervenience. Non-reductive materialists hold that we can accept that the microphysical facts determine all higher-level facts, but that such determination does not require that higher level theories be classically reducible to microphysical theory (i.e. that all the terms of special-scientific theories can be defined in the langauge of microphysics). Since the determinative aspect of non-reductivism is supposed to be fulfilled by supervenience, it appears as if we should prefer ontological supervenience theses over their ascriptive and descriptive counterparts. One of the more popular versions of supervenience, "strong" supervenience, is defined as a relation over sets of properties: i.e. on the Hellman-Thompson/Haugeland view, it need not be a misuse of language to apply a special-scientific predicate to a and not to a's 'microphysical twin' b. Rather, to do so is to properly apply the language of a faulty theory. 43 (SS) A set of properties A strongly supervenes on a set of properties B if and only for any two worlds Wx and W2 and any two objects x andy, if x in Wj has the same 5-propertiesy has in W2, then x in Wx has the same ^ -properties y has in ^ 2 4 0 The merits of strong supervenience (as against other forms of supervenience) have been, and no doubt will continue to be, debated in the philosophical literature (Horgan 1982, 1993a, Kim 1984, 1987, Melnyk 1991). I want to draw out a curious feature of strong supervenience, and of ontological supervenience in general: the formulations of ontological supervenience suggest that there are two metaphysically distinct domains of properties (or facts), and that one of those domains somehow determines the other. For the anti-property reductionist, this is as things should be, since she accepts the claim that there are non-physical properties. Andrew Melnyk derides ontological supervenience as "metaphysical superglue, cementing together the domains of the many sciences" (Melnyk 1993, p. 238). Horgan (1993a, p. 555) notes that the etymology of the term "supervenience" suggests that supervening properties are properties "over and above" the properties which subvene them. But what, exactly, is involved in this determination? , u Kim (1987, p. 81) attributes this formulation to Brian McLaughlin. In other papers, (Kim 1984, Horgan 1993a) strong supervenience is defined with the aid of the assumption that A and B are closed under Boolean operations (conjunction, negation, and disjunction). However, I find this approach unsatisfactory for several reasons: first, it seems to me that conjunction applies primarily to expressions in a given language rather than to properties; second, even if we could make sense of (for example) negated properties, it seems that not being a lepton is not a microphysical property, although the predicate which expresses it can clearly be defined in the language of microphysics (cf. van Fraassen 1980 on 'observation languages'). 44 Emergence Emergentists such as C D . Broad (1925) sought a half-way point between orthodox materialism and outright substantival dualism or pluralism. Broad accepted the basic materialist claim CP, but added that the behavior of chemical biological particulars is not determined solely by the laws which govern the behavior of their microphysical parts. Broad thus denied the orthodox materialist principle of the causal completeness of physics (PCC). However, it also appears that Broad would have accepted the claim that biological facts supervene on microphysical facts, in accordance with, say, (SS). Indeed, emergentists and their opponents apparently used the word "supervenient" as a variant of "emergent" (see e.g. Meehl and Sellars 1956, Kim 1992a, p. 132 n. ). As in the case of supervenience, the 'unit' of emergence has been a subject of debate; as in the case of supervenience, I will focus my discussion on the question of whether there are emergent properties. In the broadest sense, an emergent property is a property of a whole W which is not possessed by the parts of W in isolation from one another. The instantiation of an emergent property by W depends upon the parts of W, and most importantly, upon the relations between those parts; emergent properties are instantiated by all and only complex wholes which have certain microphysical structures. The "levels" view which is so popular among contemporary materialists actually depends upon the existence of emergent properties. When a property is classified as emergent, there is some level L at which the property emerges, i.e. the property is not and cannot be instantiated by objects at levels 45 lower than L. There may be some question about which particular level a given property (or property-type) emerges — quarks and leptons are definitely not coloured, and (structured) collections of (say) 10 1 0 mercuric oxide molecules definitely are. But are individual mercuric oxide molecules coloured? What about smaller (say, 105 molecule) collections? A discussion of these problems would take me too far afield, however, and I will assume that emergent properties emerge at some definite level. The difference between diamond and graphite is not what they're made of — they're both forms of pure carbon — but how their constituents are arranged. The carbon atoms in a diamond lattice bond in a three-dimensional structure (tetrahedrally), while the carbon atoms in a sheet of graphite bond in a planar structure. Individual carbon atoms are neither hard (like diamond) nor opaque (like graphite). So at least some of the macroscopic properties of diamond and graphite emerge from structured collections of carbon atoms. It is actually quite hard to find macroscopic properties which are not emergent in this broad sense. The mass-energy of a macroscopic object is the sum of the mass-energies of its constituent (non-overlapping) parts, and is thus not emergent in this sense (having mass-energy is a property possessed by the parts). Emergent properties can be divided into two classes, which I will refer to, following Smart (1987), as 'weak' and 'strong' (or emergentw and emergents41). 4 1 I will use the term "emergentist" with the appropriate subscript to denote proponents of the view that there are emergent properties of the type in question. 46 Unfortunately, it is hard to characterize these classes by giving examples. Orthodox materialists deny that there are emergents properties, because their instantiation would involve a violation of PCC 4 2 . Like Smart (1987), I think that any materialist should be happy to admit the existence of emergentw properties, which do not violate PCC; the behavior of complex objects which instantiate emergentw properties is determined solely by (and is thus in accordance with) microphysical laws. Although emergentw properties are 'novel' in the relevant sense, there is some sense in which they are 'nothing but' (i.e. are reducible to) the properties of and relations between43 the parts of the objects which instantiate them44: Of course, 'nothing buttery' is often said to be a heinous metaphysical crime, but I see nothing wrong with it: in stating that a complex is nothing but an arrangement of its parts, I do not deny that it can do things that a mere heap or jumble of the parts could not do. (Who would want to deny this? If 'nothing buttery' had such an absurd consequence it would be a view that no one has ever held.) (Smart 1987, p. 248)45 Emergentg properties are another matter (!) entirely. Strong emergentists argue that when microphysical particles arrange themselves into certain kinds of structures, the structure instantiates a novel property, which is 'irreducible' to the properties of the parts. Broad (1925), I believe, defends a version of strong emergentism, although he displays the unfortunate tendency to frame the issue of "irreducibility" in epistemological terms, as evinced in the following claim: 4 2 For brevity, I will say that the properties violate causal closure. 4 3 Again, for brevity I will say "the properties of the parts." 4 4 This is vague, but I think it will be cleared up below in the discussion of Kim's attack on non-reductivism. 4 5 Another orthodox materialist who acknowledges the existence of weakly emergent properties is Patricia Churchland (1986, p. 324), who calls them "network properties". 47 (B) No amount of knowledge about how the constituents of a living body behave in isolation or in other and non-living wholes might suffice to enable us to predict the characteristic behavior of a living organism. (Broad 1925, p. 67) A plausible way to pare away the epistemology from this claim is to assume that Broad is claiming that in biological (living) wholes, the behavior of microphysical particles does not conform to the laws which govern their behavior in non-living wholes. Thus, a theory adequate for the description and explanation of the behavior of microphysical particles outside of biological wholes will apparently be unable to explain the behavior of those particles when they are in biological wholes. The following passage, I believe, bears this reading out: ... we have no right to suppose that the laws which we have discovered by studying non-living complexes can be carried over without modification to the very different case of living complexes. (Broad 1925, p. 69) Broad is careful to emphasize that he believes the difficulty here is an "in principle" one, and not one faced by human beings (Broad 1925, p. 70). It is not that we are not intelligent enough to predict the behavior of wholes or that no computer we could build could crunch enough numbers to yield the right prediction, but simply that there's more going on in living wholes than is dreamt of in the microphysicist's philosophy. In mechanical terms, we would say that wholes which instantiate emergents properties exert forces on the particles which comprise them that those particles do not exert on each other. The instantiation of both emergentw and emergents properties depends on the way the parts of the wholes which instantiate them are related to each other. To continue quoting from Broad, immediately after where passage (B) ends: 48 This possibility is perfectly compatible with the view that the characteristic behavior of a living body is completely determined by the nature and arrangement of the chemical compounds which compose it, in the sense that any whole which is composed of such compounds in such an arrangement will show vital behavior and that nothing else will do so. (Broad 1925, pp. 67-8, emphasis added) One of the distinctive features of emergentist views is that they try to retain some version of the thesis I called DP in the first chapter. Emergent properties are dependent upon the structures from which they emerge, in the sense that the existence of certain kinds of microphysical structures are necessary and sufficient for the instantiation of those properties. And Broad's comments, in particular, seem to commit him to some form of supervenience view. Emergent materialism (EM) begins to look very much like SM (which is not all that surprising, given the history of the terms) once we realize that both the EMist (taking Broad as an exemplar of the view) and the SMist subscribe to both CP and some version of DP. Why Supervenience? The apparent compatibility of the more popular supervenience theses with the existence of strong emergentM properties has sparked a reassessment of the value of supervenience in the formulation of orthodox materialism. In light of the diffculties raised by his consideration of emergentisms, Terence Horgan claims that materialism must be formulated in terms of "superdupervenience", "... ontological supervenience that is robustly explainable in a materialistically acceptable way" (1993a, p. 577)46. In the The problem of emergence is not the only reason Horgan gives for moving to superdupervenience; he also feels that if supervenience is held to be sui generis, a metaphysically fundamental fact, then the relation itself seems to violate the spirit of materialism (1993a, p. 565). 49 terminology I have adopted, Horgan's "materialistically acceptable" means "acceptable to orthodox materialists", since I think emergentisms ought to count as a kind of materialism. Horgan's demand for explanation of the supervenience relation itself amounts to a demand that the materialist show why there can be no difference without a physical difference. The emergentists's account of supervenience is not acceptable in this sense because it entails the denial of PCC. My worries about putting the issue between emergentistss and orthodox SMists as one of the explainability of supervenience relations should be obvious — the issue is not an epistemological one, but rather a metaphysical one: The difference between orthodox materialism and emergentism must be found in the 'mechanism' (for lack of a better word) each gives for the determination of higher-level properties. The emergentists, m effect, is saying that it's just a metaphysically fundamental, "brute" fact that when you put microphysical bits together in certain structures, the behavior of those bits fails to conform to the laws which govern their behavior in isolation. What sort of 'mechanism' can the materialist propose? Horgan is well aware that the problem he raises for SMist is not an easy one to solve; one thing which makes it such a difficult problem is that it's hard to specify exactly what would count as solving the problem: What facts specifically need explaining in order to explain a given inter-level supervenience relation, and why would a materialistic explanation of these facts constitute an explanation of that supervenience relation? (Horgan 1993 a, p. 578) I must admit to sharing Horgan's puzzlement here. The point of bringing in supervenience in the first place was in the hopes that it could be used to formulate materialism in such a way that it did not imply property reductionism. If Horgan is 50 correct, before the materialist can invoke supervenience in a formulation, she must already have formulated theses concerning "materialistic acceptability". How the latter project differs from formulating materialism itself, I do not know. A proposal The problem for Horgan's SMist seems to arise in adopting the view that supervenience must be formulated ontologically, as a relation between distinct domains of properties. Rather than join Horgan's call to elucidate superdupervenience, I would like to advocate a different tack, based on the Hellman-Thompson/Haugeland descriptive supervenience theses. Descriptive supervenience is relatively neutral, metaphysically speaking; while it is compatible with ontological supervenience, unlike the latter, it neither entails nor suggests that there are non-physical properties. Descriptive supervenience also seems compatible with strong emergentism, and so the problems Horgan identifies do not go away completely. We can resurrect the Hempelian idea that the special sciences offer different descriptions of the world described by microphysics, while abandoning Hempel's claim that special-scientific predicates can be defined in terms of physical predicates. • The functionalist 'multiple readability' argument sketched in the previous chapter did not apparently rely on any failure of PCC, but rather, as I put it there, on the claim that the language of physics lacks the representational power necessary to capture the generalizations formulable in the languages of the various special sciences. The (unadvisable) conclusion many anti-reductionists drew from these arguments is that there are non-physical properties, i.e. genuine properties which are irreducible to physical 51 properties, although those properties supervene on physical properties. The problem is that emergentistss make analogous claims. The problem here evidently stems from an ambiguity in the term "irreducible property". Emergentists evidentiy use the term in a metaphysical sense, while functionalists (who rely on the principle I call Pred) have some linguistic notion of 'irreducibility' in mind. The moral of the last two chapters is that the empiricist's linguistic model of reduction does not fully capture the notion of metaphysical reducibility. Fortunately, Mark Wilson (1985) has given an account which does, I think, give a reasonable formulation of metaphysical reductionism. 52 C h a p t e r F o u r : R e d u c t i o n i s m R e v i s i t e d One possibility that does not seem to have occurred to anti-reductionists over the years is that the empiricistic analysis of intertheoretic reduction is incorrect, or at least incomplete. The "syntactic" view of theories Nagel's model relies on has come under fire from various quarters (see e.g. van Fraassen 1980, ch. 3, Giere 1988, ch. 3). So, too, has Hempel's "D-N" model of explanation (for a history of accounts of scientific explanation, see Salmon 1989). Of course, this is not to say that by switching to the "semantic" view of theories van Fraassen and Giere advocate, adopting a non-Hempelian view of explanation, one can automatically give an account of intertheoretic reduction which deals with the general misgivings about reductionism. I have no such original account to offer, at any rate. In this chapter, I would like to deal with the claim that functionalist multiple realizability arguments show that there are non-physical properties, and Fodor's worries about methodological reductionism. Recall that one premise of the "multiple realizability" argument is that it is not always possible to find 'bridge laws' of the form (x) (Sxo<t>), where 'S' is a predicate from the language of some special science, and <J) is a finite expression in the language of physics. Even if the expression is finite, § may not be a 'natural kind predicate', one which figures in a law-statement of physics. In the former case, the bridge laws required to derive the laws of the science we wish to reduce science cannot even be formulated. Either way, claims Fodor, the aims of reductionists are thwarted. In this chapter, I'll 53 focus on the claim that the conclusion counts against intertheoretic reduction. The negative answer, surprisingly, is provided by Fodor himself: [PR] The point of reduction is not primarily to find some natural kind predicate of physics coextensive with each kind predicate of a special science. It is, rather, to explicate the physical mechanisms whereby events conform to the laws of the special sciences. (Fodor 1974, p. 435) I could hardly put it better myself! Not surprisingly, neither Nagel's discussion nor Kemeny and Oppenheim's contains one word about "natural kinds". Nor does the conception of reduction as explanation imply in any straightforward fashion that the 'kind-predicates' of the theories involved be matched up in the one-one fashion Fodor describes. Of course, empiricists such as Hempel who require that reductions yield conceptual simplicity would not agree. In sum, the functionalist challenge depends upon two claims, both of which I think are false. The first is that successful epistemological reduction requires the definition of the terms of the reduced theory in terms of the reducing theory. The second is that successful property reduction requires such definition. These points can be illustrated by taking a brief look at the paradigm case of a successful reduction, that of thermodynamics to statistical mechanics. A Brief Science Lesson When I consulted an elementary text on thermodynamics, I was quite surprised to find the following passage: Thermodynamics deals only with macroscopic variables, such as pressure, temperature, and volume. Its basic laws, expressed in terms of such quantities, say nothing at all about the fact that matter is made up of atoms. (Halliday and Resnick 1986, p. 407, emphasis original) 54 This sounds suspiciously like something a functionalist might say about terms like "pain", "belief and so forth, yet functionalists typically draw on the temperature/mean molecular kinetic energy relation in order to contrast reductionism with their own view. By itself, this observation does little to damage functionalism or antireductionism; however, other facts about the reduction in question do. The derivation of the laws of thermodynamics from the laws of statistical mechanics involves a number of idealizations, limiting assumptions, and the like. For example, we asssume that the molecules which compose the gas are: a) dilute (few molecules to a given unit of volume) b) moving with variable speeds and in all directions c) all collisions betweeen the molecules are completely elastic d) free from forces except those which occur during collisions (this list is adapted from Halliday and Resnick 1986) None of these assumptions are literally true of the molecules of the gas. Nor, for example, does the derivation go through if the gas is at a high temperature, for then the molecules ionize, and there will be nonnegligible intermolecular (electrostatic) forces47, which contradicts (d). The point here is that reductions are not as "neat" as (e.g.) Nagel's presentation makes them seem48; they are also, as Patricia Churchland (1986) puts it, domain-relative. 4 7 As opposed to the negligible amount of gravitation which obtains in the 'normal' case. 48Although the point is tangential, I should note here the influential objection pressed by Feyerabend (1962) that what gets deduced from the reducing theory is often not the reduced theory, but a corrected version thereof. For example, the second law of thermodynamics is time-asymmetric (i.e. it rules out the increase of entropy in one direction of time change), while the laws of statistical mechanics are all time-symmetric. 55 What the reduction in question shows, at best, is that the temperature of an ideal gas at thermal equilibrium is the mean kinetic energy of its constituent molecules. "Temperature" applies not only to ideal gases, but to solids and even to empty regions of space. Thus, "mean molecular kinetic energy" is not even coextensive with "temperature" (Hooker 1981, En? 1983, Wilson 1985). So, it seems as if temperature is multiply-realizable! But if anything is an example of a successful reduction, the thermodynamics-statistical mechanics case is. I think we would rather say that if this case does not conform to the quasi-standard model of epistemological reduction via biconditionals, then so much the worse for that model and its close relatives. Since we have a prima facie reason to think that epistemological reductions do not require the particular sorts of property- or kind-identity functionalist arguments are directed at, we have a prima facie case against the functionalist's rejection of reductionism tout court; it would be better to say that the multiple readability argument counts only against the model of reduction via property identities. Statistical mechanics does give us an explanation of why ideal gases satisfy the laws of thermodynamics (thus, Fodor's comment PR, quoted above, seems essentially correct in this case), but it does not identify temperature tout court with any statistical mechanical property. Property Reduction Again In an excellent but difficult paper, Mark Wilson (1985) presents his brief against 'functionalist' rejections of type-physicalism. Essentially, Wilson's argument is that the mathematical techniques required to do physics are powerful enough to 'construct' 56 properties which correspond to functionally defined predicates. Physics requires these techniques in order to deal with macroscopic phenomena, like bridges, ideal gases, and panes of glass. His main contention is that the functionalist argument against property reductionism rests on a claim which is at odds with physical practice ~ the ^definitely many structures which 'realize' a given functional property or state do in fact constitute a 'physical' kind or property. One good reason to reject the 'definability' view of physical properties is that no language possesses the resources to construct all the definitions required to do physics. Languages can contain only countably (denumerably) many expressions. However, physical theory must be able to deal with such properties as the shape of objects, and there are non-denumerably many possible shapes. Ergo, there are some physical properties which are not definable in the sense in question. Wilson argues that physical properties are those which correspond to mathematical relations which satisfy the most general equations of physics (such as Lagrange's equation). He calls this principle (3) : ... physics does not construct its traits on the basis of definitional form at all, just as mathematics does not build its full universe of functions in an analogous manner either. One's average real-valued function, for example, owns no definition simply because there are too many of them. The realm of physical traits, as carved out by (3), simply inherits this "nondefinability" from the mathematics upon which it depends. (Wilson 1985, p. 233). One example of a property picked out by (3) which seems "multiply realizable" is that of being a simple harmonic oscillator. All and only simple harmonic oscillators satisfy a (certain) second-order partial differential equation. Of course, "is a harmonic oscillator", so defined, looks like a functionally defined predicate. But then, so are many predicates which express properties physicists call their own: "is ergodic", "slides without friction", 57 and, of course, "temperature". Given the general applicability of, say, Lagrange's equations to physical systems49 and the entire realm of functions to choose from, it is trivially true that there will be some relation corresponding to any property which can had by any set of physical objects. Interestingly, Wilson's 'type-physicalist' view, by itself, seems to have only minimal consequences for the confirmation and development of the special sciences. The existence of the requisite mathematical functions implies virtually nothing about our actual capability to compute and use Lagrange's equation for predicting and explaining, say, biological phenomena. In this respect and others, Wilson's position has much in common with Hellman-Thompson (1975, 1977) physicalism. The main differences between these views is Wilson thinks that Hellman and Thompson's use of 'determination principles' (a variant of 'supervenience') is unnecessary. The fact that physics + mathematics does possess the resources necessary to capture the generalizations of the special-sciences in no way counts against claims of descriptive supervenience. Wilson's type-phyiscalism in fact provides an excellent answer to the question "why do special-scientific truths supervene on physical truths?" ~ special-scientific predicates pick out functions which satisfy the most general laws of physics — and thus seems to count as a materialistically acceptable explanation of supervenience. Very roughly, this is a restatement of the principle of causal closure. Lagrange's equation is a re-formulation of Newtonian mechanics in the form of a single equation. 58 Methodological Reductionism In one sense, it is crashingly obvious that materialism does not imply reductionism of the epistemological or methodological sort, for the same reasons that statistical mechanics does not imply thermodynamics ~ the word "explain" does not appear in the formulation of a metaphysical doctrine such as materialism. However, the empiricist blurring of the distinction between what is metaphysically the case and what we could come to know "in principle" seems to support the inference from materialism to epistemological reductionism. The blurring of this important distinction is what needs to be rejected. Fortunately, there is a way out for the would be "realist". Fodor's first concern in "Special Sciences" is that epistemological reduction, or at least reducibility, not be taken to be a necessary condition for the acceptability of special scientific theories. However, a reductionist need not make such stringent demands; certainly Nagel does not: ... the possibility should not be ignored that little if any new knowledge or increased power for significant research may actually be gained from reducing one science to another at cerain periods of their development... (Nagel 1961, p. 362) Whether or not theories ought to reduce to other, lower level theories depends on the maturity of the theories involved. Maturity consists, at least partially, in the availability of relatively clear formulations of the theory. As Nagel points out, questions about reducibility cannot even be asked in the absence of such formalizations. Further, the notion that higher level theories must reduce suggests that lower-level theories are, by their very nature, better confirmed. In cases where a relatively new and speculative theory seems to be a likely reducer of a more well entrenched, well confirmed theory, 59 one would expect that it is a constraint on the lower-level theory that it be able to reduce the higher-level theory! This last point brings me to another: recent reductionists have emphasized the dynamic character of most reductions and have proposed "dialectical" models of reductions in which a higher-level theory and a lower level theory co-evolve. This approach was first developed in detail by Hooker (1981), although the groundwork was laid by Wimsatt (1976); more protean versions of the idea are present in Dobzhansky (1969) and hints of it can even be found in Nagel (1961) (!). Most recently, the dialectical approach has been championed by Patricia Churchland (1986). On the dialectical account, reductions do not happen "all at once". It is not as if a theoretician sits down at a table with two theories, does a little work, tests some predictions, and cries out "Eureka! I have reduced it!". Rather, reductions take place over periods of time; the connections between the two theories are noted, work begins on making those connections a little clearer, which involves in minor revisions being made to the theories. This process is iterated, resulting in further revisions to the theories until one day, it is possible to construct an analogue of the higher level theory using the lower level theory which warrants calling the relation between the theories a reductive one (see Hooker 1981, pp. 50-52, for a more detailed account) The second worry expressed by Fodor is that a successful reduction results or ought to result in the elimination of the reduced theory. He notes in several places that such eliminations are relatively rare while the "the development of science has witnessed the proliferation of specialized disciplines at least as often at it has witnessed their 60 elimination" (Fodor 1974, p. 429). He takes this as constituting a point against classical ("concept elimination") epistemological reductionism. As I have noted, Oppenheim and Putnam's account of the Unity of Science is eliminativistic in principle, but they also claim that there is in fact a trend in science toward unitary science; so, Fodor's charge that reductionism, in their sense, "flies in the face of the facts", seems to be warranted. However, Fodor seems to conflate the ought of methodology with the is of history. While it may be true that many sciences are in fact autonomous, that new sciences are constantly springing into being, and relatively few are actually eliminated, the consequences for 'in principle' epistemological reductionists are virtually nil. "In principle" reductionistsE, such as Oppenheim and Putnam, can always fall back on the argument that even reduced theories continue to be used for pragmatic reasons. The increase in mathematical and computational complexity "enjoyed" by those who use more general, lower level theories may speak against the use of such theories in contexts where the extra degree of accuracy would not compensate. Geometrical optics and thermodynamics are still with us, even if they are in principle reducible to electromagnetic theory and statistical mechanics, respectively. And nobody, I take it, would advocate the use of quantum mechanics in the place of transmission genetics or cognitive psychology! This "pragmatic" line, however, opens up some room between 'in principle' epistemological (TPE) and methodological reductionism. Even if materialism is true and materialism implies IPE reductionism, it does not follow that reductionist methodology gives the best strategy for scientific research. There are obvious limits to human 61 computational powers, even when they are augmented by modern digital computers. This point has been forcefully pressed by Wimsatt (1976) and, more recently, by Rosenberg (1994). Note that Wilson's defense of 'type' physicalism yields this conclusion too; although the physicist can appeal to the realm of mathematical functions in picking out physical properties, she cannot guarantee that the functions which correspond to properties like pain and being a monetary exchange could be computed by humans. Secondly, the amount of resources humans are able to devote to scientific research are finite and limited. Social and scientific goals might be better served by devoting resources to the development of relatively autonomous sciences rather than to reducing extant theories. Conclusion This essay has covered much ground, and in doing so, has certainly ridden roughshod over some of the finer points involved with materialism. Horgan is certainly correct when he writes "[i]t is not easy formulating a metaphysical position that meets the demands of a material world; there is still a lot of philosophical work to do" (Horgan 1993a, p. 582). However, I do hope that I have shown the plausibility of the following claims: 1) 'In principle' epistemological reductionism of the sort advocated by empiricist-minded philosophers does not provide unqualified support for methodological programs. 2) Epistemological reductionism of one theory to another does not require identifying the properties associated with the two theories. In the most discussed case of reduction, 62 what was shown was not that temperature=mean molecular kinetic energy, but that the temperature of ideal gases=mean kinetic molecular energy. This reduction seems to be a case of explaining why certain kinds of systems obey certain kinds of laws, without broad-scope property identities. 3) Property reductionism does not imply methodological reductionism, if we follow Wilson's method for determining the range of physical properties. 63 B i b l i o g r a p h y Beckermann, A., Flohr, H., and Kim, J.: 1992: Emergence or Reduction?, de Gruyter, Berlin Boyd, R., Gasper, P., and Trout, J.D. (eds.): 1991: The Philosophy of Science, MIT Press, Cambridge, MA Broad, CD.: 1925: The Mind and Its Place in Nature, Routledge and Kegan Paul, London, UK Callebaut, W: 1993: Taking the Naturalistic Turn, University of Chicago Press, Chicago, IL Carnap, R.: 1933: Unity of Science, Routledge and Kegan Paul, London, UK Causey, R.: 1977: Unity of Science, D. Reidel, Boston, MA Charles, D. and Lennon, K. (eds.): 1992: Reduction, Explanation, and Realism, Clarendon Press, Oxford, UK Churchland, P. S.: 1986: Neurophilosophy, MIT Press, Cambridge, MA Churchland, P. M.: 1981: "Eliminative Materialism and the Propositional Attitudes", reprinted in (P.M. Churchland 1988), pp. 1-22 1985: "Conceptual Progress and Word-World Relations: In Search of The Essence of Natural Kinds", in (P.M. Churchland 1988), pp. 281-296 1988: A Neurocomputational Perspective, MIT Press, Cambridge MA Collier,!: 1988: "Supervenience and Reduction in Biological Hierarchies", Canadian Journal of Philosophy Supplement 14, pp. 209-234 Crane, T.: 1994: "Reply to Pettit", Analysis 53, pp. 224-227 64 Darden, L., and Maull, N.: 1977: "Interfield Theories", Philosophy of Science 44, pp. 43-64 Dobzhansky, T.: 1969: "On Cartesian and Darwinian Aspects of Biology", in (Morgenbesser et al. 1969), pp. 165-178 Dupre, J.: 1993: The Disorder of Things, Harvard University Press, Cambridge MA . Earman, J.: 1986:^ 4 Primer on Determinism, D. Reidel, Boston MA Feigl, H. and Maxwell, G. (eds.): 1962: Minnesota Studies in the Philosophy of Science 3: Scientific Explanation, Space, and Time, University of Minnesota Press, Minneapolis, MN Feyerabend, P.K.: 1962: "Explanation, Reduction, and Empiricism", in (Feigl and Maxwell 1962), pp. 28-97 Field, H.: 1991: "Physicalism", in Earman, J (ed.): Inference, Explanation, and Other Frustrations, Berkeley Press, pp. 271-291 Fodor, J.A.: 1974: "Special Sciences", reprinted in (Boyd et al 1991), pp. 429-441 1989: "Making Mind Matter More", in Fodor, J.A., A Theory of Content, MIT Press, Cambridge, MA, pp. 137-159 Globus, G , Maxwell, G., and Savodnik, E. (eds.): 1976: Consciousness and the Brain, Plenum Press, New York, NY Goodman, N.: 1954/1983: Fact, Fiction, and Forecast (fourth edition), Harvard University Press, Cambridge, MA Grimes, T.: 1988: "The Myth of Supervenience", Pacific Philosophical Quarterly 69, pp. 152-160 Hare, R.M.: 1952: The Language of Morals, Clarendon Press, Oxford, UK 65 Haugeland, J.: 1982: "Weak Supervenience", American Philosophical Quarterly 19, pp. 93-103 1984: "Ontological Supervenience", Southern Journal of Philosophy Supplement 22, pp. 1-12 Hellman, G.: 1985: "Determination and Logical Truth", Journal of Philosophy 82 Hellman, G. and Thompson, F.: 1975: "Physicalism: Ontology, Determination, Reduction", Journal of Philosophy 72, pp. 551-567 1977: "Physicalist Materialism", Nous 11, pp. 309-345 Hempel, C.G.: 1965a: "Empiricist Criteria of Cognitive Significance: Problems and Changes", reprinted in (Boyd et al 1991), pp. 71-84 1965b: Philosophy of Natural Science, Prentice-Hall, Englewood Cliffs, NJ 1969: "Reduction: Linguistic and Ontological Facets", in (Morgenbesser et. al 1969), pp. 179-199 Hooker, C.A.: 1981: "Towards a General Theory of Reduction I-III", Dialogue 20, pp. 38-59, pp. 201-236, pp. 496-529 Horgan, T.: 1982: "Supervenience and Microphysics", Pacific Philosophical Quarterly 63, pp. 29-43 1993 a: "From Supervenience to Superdupervenience: Meeting the Demands of a Material World", Mind 102, pp. 555-586 1993b: "Non-Reductive Materialism and the Explanatory Autonomy of Psychology", in Wagner, S.J. ed.) Kemeny, J.G and Oppenheim, P.: 1956: "On Reduction", Philosophical Studies 7, pp. 6-19 Kim, J.: 1984: "Concepts of Supervenience", in Kim (1993a), pp. 53-78 1987: "'Strong' and 'Global Supervenience Revisited", in Kim (1993a), pp. 79-91 66 1989: "The Myth of Nonreductive Materialism", in (Kim 1993a), pp. 265-284 1992a: "Multiple Realization and the Metaphysics of Reduction", in (Kim 1993a), pp. 309-335 1992b: "Emergence and 'Downward Causation' in Nonreductive Materialism", in (Beckermann et al 1992), pp. 118-138 1993 a: Supervenience and Mind, Cambridge University Press, New York, NY 1993b: "The Nonreductivist's Troubles With Mental Causation", in Kim (1993a), pp. 336-357 (forthcoming): "Emergence, Supervenience, and Realization in the Philosophy of Mind", ms. Kitcher, P.: 1983: "Explanatory Unification", reprinted in (Boyd et al 1991) Kitcher, P. and Salmon, W. (eds.): 1989: Scientific Explanation, University of Minnesota Press, Minneapolis, MN Klee, J.: 1984: "Micro-Determinism and Concepts of Emergence", Philosophy of Science 51, pp. 44-63 Leggett, A.J.: 1987: "Reflections on the Quantum Measurement Paradox", in Hiley, B.J. and Peat, F.D. (eds.) Quantum Implications: Essays in Honor of David Bohm, Routledge and Kegan Paul, New York, NY Lewis, D.: 1983: "New Work For a Theory of Universals", Australasian Journal of Philosophy 61, pp. 343-377 Lycan, W.G.: 1987: Consciousness, MIT Press, Cambridge, MA Marras, A.: 1994: "Non-Reductive Materialism and Mental Causation", Canadian Journal of Philosophy 24, pp. 465-494 Matthen, M. and Ware, R.X. (eds.): 1995: Biology and Society, University of Calgary Press, Calgary, AB 67 McLaughlin, B.: 1992: "The Rise and Fall of British Emergentism", in (Beckermann et al 1992), pp. 49-93 Meehl, P., and Sellars, W.: 1956: "The Concept of Emergence", in Feigl, H., and Scriven, M. (eds.), Minnesota Studies in the Philosophy of Science 1, University of Minnesota Press, Minneapolis, MN, pp. 239-252 Melnyk, A.: 1991: "Physicalism: From Supervenience to Elimination", Philosophy and Phenomenological Research 61, pp. 573-587 1993: "Being a Physicalist: Why and (More Importantly) How", Philosophical Studies 7 4 , pp. 221-241 Morgenbesser, S., Suppes, P., and White, M. (eds.): 1969: Philosophy, Science, and Method: Essays in Honor of Ernest Nagel, St. Martin's Press, New York, NY Nagel, E.: 1961: The Structure of Science, Hackett, Indianapolis, IN Papineau, D.: 1993: Philosophical Naturalism, Blackwell, Cambridge, MA Passmore, J.: 1970: Philosophical Reasoning (2nd ed.), Duckworth, London, UK Pettit, P: 1994: "A Definition of Physicalism", Analysis 53, pp. 213-223 Post, J.: 1987:777e Faces of Existence, Cornell University Press, Ithaca, NY Putnam, H.: 1960: "Minds and Machines", reprinted in (Putnam 1975), pp. 362-385 1967: "The Nature of Mental States", reprinted in (Putnam 1975), pp. 429-440 1973: "Philosophy and Our Mental Life", reprinted in (Putnam 1975), pp. 291-303 1975: Mind, Language, and Reality, Cambridge University Press, New York, NY 68 Richardson, R.C.: 1979: "Functionalism and Reductionism", Philosophy of Science 46, pp. 533-558 (unpublished): "Dualism, Interaction, and Emergence" Rosenberg, A.: 1994: Instrumental Biology, University of Chicago Press, Chicago, IL Sellars, W.: 1963: Science, Perception, and Reality, Ridgeview Press, Atascadero, CA Simon, H.: 1969: The Sciences of the Artificial, MIT Press, Cambridge, MA Smith, P.: 1992: "Modest Reductions and the Unity of Science", in (Charles and Lennon 1992), pp. 19-44 Sperry, R.W.: 1976: "Mental Phenomena as Causal Determinants in Brain Function", in (Globus et al 1976), pp. 163-177 1986: "Discussion: Macro- Versus Micro-Determinism", Philosophy of Science 53, pp. 265-270 Smart, J.J.C.: 1959: "Sensations and Brain Processes", Philosophical Review 68, pp. 141-156 1987: "Physicalism and Emergence", in Smart, J.J.C., Essays Metaphysical and Moral, Blackwell, New York, NY, pp. 246-255 Teller, P.. 1984: "A Poor Man's Guide to Supervenience and Determination", Southern Journal of Philosophy Supplement 22, pp. 137-162 1992: "A Contemporary Look at Emergence", in (Beckermann et al. 1992), pp. 139-153 van Fraassen, B.C.: 1980: The Scientific Image, Clarendon Press, Oxford, UK Wimsatt, W.: 1976: "Reductionism, Levels of Organization, and the Mind-Body Problem", in (Globus etall976) 1995: "The Ontology of Complex Systems", in (Matthen and Ware 1995), pp. 207-274 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items