UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Affective qualities : what makes objects pleasant Sandlin, Richard 2020

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata


24-ubc_2020_may_sandlin_richard.pdf [ 1.07MB ]
JSON: 24-1.0389587.json
JSON-LD: 24-1.0389587-ld.json
RDF/XML (Pretty): 24-1.0389587-rdf.xml
RDF/JSON: 24-1.0389587-rdf.json
Turtle: 24-1.0389587-turtle.txt
N-Triples: 24-1.0389587-rdf-ntriples.txt
Original Record: 24-1.0389587-source.json
Full Text

Full Text

   Affective qualities: what makes objects pleasant  by  Richard Sandlin  BA, University of California, Berkeley, 2005 MA, Brandeis University, 2011    A DISSERTATION SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY  in   THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES  (Philosophy)   THE UNIVERSITY OF BRITISH COLUMBIA  (Vancouver)    March 2020      © Richard Sandlin, 2020    ii  The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled: Affective Qualities: what makes objects pleasant Examining Committee:                                                                   submitted by Richard Sandlin  in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Philosophy Murat Aydede, Professor, Department of Philosophy, UBC Supervisor Dominic Lopes, Professor, Department of Philosophy, UBC Supervisory Committee Member Christopher Mole, Professor, Department of Philosophy, UBC Supervisory Committee Member Roberta Ballarin, Professor, Department of Philosophy, UBC University Examiner Barbara Weber, Professor, Department of Interdisciplinary Studies, UBC University Examiner  iii  Abstract        The smell of freshly baked bread, the flavor of chocolate cake, the feeling of a cool breeze on a hot day: these are paradigmatic pleasant sensations. My question is this: what makes these objects pleasant? In other words, what kind of property is sensory pleasantness? I focus my discussion on pleasant smells and pleasantness attributed to objects (as opposed to experiences). I canvass four views. Two views are objectivist: physicalism and primitivism. On these views, pleasantness is an experience-independent property. The other two views are subjectivist: projectivism and the relational view. On these views, pleasantness is an experience-dependent property. I argue that physicalism is circular and cannot explain a core aspect of pleasantness. I argue that primitivism leads to unacceptable pleasantness property proliferation. I conclude that pleasentess must be a subjective property. However, I argue that projectivism won’t work because the view cannot explain why we would have evolved systems to sense the pleasantness of objects. I conclude that pleasantness must be a relational property. On this view, we can explain core aspects of pleasantness in a non-circular way without undesired property proliferation, while also explaining why we evolved systems to sense pleasantness. In particular, I argue that pleasantness is the property of objects that dispose us to classify certain information in particular ways.           iv  Lay Summary       We experience certain objects as pleasant: the smell of freshly baked bread, the flavor of chocolate cake, a cool breeze on a hot day. I explore what makes experiences like this pleasant. In particular, I explore whether pleasantness is something that is dependent or independent of our experience. I argue that pleasantness cannot be independent of our experience. The reason: either key aspects of pleasantness cannot be explained or bizarre consequences result. If pleasantness depends on our experience, we must be careful to explain why we would have evolved to think that objects themselves are pleasant. To that end, I argue that pleasantness is a relationship between our experience and the world. Objects provide certain types of information and our experience of this information gets classified in particular ways.                                                                                                                                                      v  Preface  This dissertation is original, unpublished, independent work by the author, Richard Sandlin.                                                                             vi  Table of Contents   Abstract ..........................................................................................................................................  iii Lay Summary ................................................................................................................................... iv Preface ............................................................................................................................................. v Table of Contents ............................................................................................................................ vi Acknowledgements ........................................................................................................................ vii Dedication ....................................................................................................................................... ix Introduction .................................................................................................................................... 1  CHAPTER 1: Smelling Pleasure 1.1 Smelling Pleasure .......................................................................................................... 8 1.2 Study 1: Pleasantness reflects the physical world ...................................................... 12 1.3 Study 2: Pleasantness and molecular size .................................................................. 16 1.4 Study 3 Pleasantness and the electronic nose ........................................................... 18 1.5 Study 4: Pleasantness and olfactory receptor surfaces .............................................. 21 1.6 Study 5: Categorical odors .......................................................................................... 23 1.7 Interpretation.............................................................................................................. 27  CHAPTER 2: Affective Physicalism  2.1 Default parameters ..................................................................................................... 37 2.2 Methodological problems ........................................................................................... 42 2.3 Affective neuroscience................................................................................................ 46 2.4 Biological significance ................................................................................................. 50 2.5 Function ...................................................................................................................... 59  CHAPTER 3: Primitive Pleasantness 3.1 Attribution ................................................................................................................... 68 3.2 Affective primitivism ................................................................................................... 74 3.3 Johnston ...................................................................................................................... 78 3.4 Strengths ..................................................................................................................... 82 3.5 Weaknesses ................................................................................................................ 84 3.6 Irrealism ...................................................................................................................... 94  CHAPTER 4: Affective Relationalism(s) 4.1 Affective Relationalism ............................................................................................... 97 4.2 Particular Relational approaches .............................................................................. 102 4.3 Problems for affective relationalism ........................................................................ 111 4.4 What makes objects pleasant ................................................................................... 121 Conclusion ................................................................................................................................... 133 Bibliography ................................................................................................................................ 136     vii  Acknowledgements        I am the sole author of this dissertation. But I am far from its only contributor.      My greatest debt is to Murat Aydede. Not only is he a great advisor and philosopher but his support, encouragement, patience and care were incalculable in my time at UBC. It’s hard to imagine what this work would be without Murat’s guidance.       I thank my other committee members: Dominic Lopes and Chris Mole. Dom helped me frame and think through important sections of this dissertation. A discussion with Chris lead to a crucial tweak and strengthening of my positive argument.      I also thank my external examiners for their insightful comments: Christine Tappolet, Roberta Ballarin and Barbara Weber.       Several of my graduate student colleagues deserve mention. I left every meeting with Aleksey Balotskiy with ideas on how to improve my dissertation. He is also partly responsible for the title. Kousaku Yui, Servaas van der Berg and Madeleine Ransom provided penetrating thoughts on how to improve this work and words of encouragement when I most needed them.       Chris Stephens provided structure and guidance when I most needed it. Other UBC colleagues who deserve thanks: Emma Esmaili, Sina Fazelpour, David Friedell, Jasper Heaton, Rebecca Livernois, Eric Margolis, Phyllis Pearson, Matthew Smithdeal and Jerry Viera. I thank Mark Vessey and the 2012-2014 residents of Green College, UBC for stimulating discussion and much memory making.   viii       I first started researching the chemical senses when I was an MA student at Brandeis University. I was helped a great deal by my committee members Jerry Samet and Berislav Marusic and by my fellow MA students Kevin Lande and Derek Leonard. I thank all four for making my time in Boston so positive.           I first fell in love with philosophy in Joseph Monast’s courses. That love was bolstered in classes with Hubert Dreyfus, Barry Stroud, Mike Martin and John Campbell. I thank all of them for the excellent instruction.       I am blessed to have a group of philosophically inclined friends from undergrad who have been steadfast in their love and support all these years. I thank Michael Karimi, Mahdi Gad, Nicholas Marcotte, Richard Parkin and Daniel Wilson for helping me grow both philosophically and personally. I want to thank my Vancouver friends Jennifer Yip, Jennifer Na and Roger Revell; friends who not only challenge me to be a better writer and thinker but a better person.            ix                                                              Dedication       I dedicate this dissertation to three people. My parents, Andrew and Sharon, have provided lifelong love and support. I never once had to justify to them why I studied philosophy instead of a field with greater social status. To plagiarize Barry Stroud, my parents made everything possible so all I had to do was make things actual. Samantha Matheson has been by my side every step of this dissertation. I can’t think of anyone else I’d want on the journey. She came into my life during a dark time and I would not have finished this work without her love, support and encouragement. A typed page is the wrong medium to express how much love and gratitude I have for her.               1                                                           Introduction       What makes objects pleasant? We attribute pleasantness to many things: the feeling of a cool breeze on a warm day, the touch of a soft feather on the skin, the smell of a lavender bed, the taste of a cappuccino. How are these objects pleasant? In other words, what is the nature of affective qualities? Yet another way to ask the question: if pleasantness is a property, what kind of property is it?        We know of the paradigmatic perceptual qualities: shape, size, extension, motion, color, sound, feels, smells, tastes. Could affective qualities be given a similar philosophical treatment to at least some of these? I will argue in the affirmative, particularly in relation to the paradigmatic secondary qualities (color, sound, feels, smell and taste).       There is a viable view immediately available. It follows from a crucial fact about pleasantness. We don’t just attribute pleasantness to objects. We also attribute pleasantness to experiences; we not only say that the lavender smells pleasant but that our experience of the lavender is pleasant as well. Though I discuss this distinction in more detail later, I call the former O-P (object pleasantness) and the latter E-P (experiential pleasantness). This opens the door for a subjectivist view of pleasantness. The idea is this: pleasantness is a property of experience that we attribute to objects. The objects are not actually pleasant; only our experiences are. This is Projectivism. Sydney Shoemaker (1997) claims that Projectivism comes in two varieties: literal and figurative. I just gave a definition of literal Projectivism: pleasantness is a property of experience that we project onto objects (for the case of color, see Boghossian  2  and Velleman, 1997; Avrill, 2005). Figurative projectivism is the following: pleasantness is not a property of objects or experiences; it is a property of our intentional contents (for the case of color, see Pautz, ms.; chapter 1 of Mackie on Locke, 1976; p. 86-87 of Stroud on Hume, 1977).       I discuss a version of figurative Projectivism later in this work. But there is a reason I bring up literal Projectivism here. There is an undeniable attraction to the view. The Galilean picture tells us that colors, sounds, smells and tastes have no place in the scientific description of the physical world. If these qualities have no place, it is an easy move to say the same about pleasantness. Literal Projectivism gets off the ground by pairing the Galilean intuition with the fact that we attribute pleasantness to both objects and experiences. I return to this view later.       Throughout this work, I discuss views that push against Projectivism. I start with two objectivist views: Physicalism (chapters 1 and 2) and Primitivism (chapter 3). I then discuss a sophisticated subjectivist alternative to Projectivism: the relational approach (chapter 4). Ultimately, I argue for a particular relational approach that is close to, but importantly, not identical with Projectivism. Before I discuss these theories, an important note about terms and method.  Terms and Method             Before I continue, several important terms need to be clarified. I focus on the pleasure and displeasure that is indexed to sensory percepts (as opposed to pleasure and displeasure more generally). By ‘index’ I mean: a property that accompanies, or attaches to, a token sensory  3  percept but is not itself a sensory property (see Robinson, 2006). I will mostly focus on olfactory cases. Different terms capture these phenomena. These include: sensory pleasantness/unpleasantness, pleasantness, sensory affect, liking, disliking, etc. I will use the terms ‘pleasant’ and ‘unpleasant’ (or sometimes ‘displeasure’, depending on the context). When I use these terms, I mean ‘pleasure or displeasure that is token indexed to a particular sensory percept’.       Sometimes I will use ‘pleasantness’ to stand for both pleasantness and unpleasantness. This will usually occur when I am speaking more generally of these affective property types. ‘Pleasant’ will only pick out pleasantness in more specific examples. For example, the pleasant smell of lavender versus the variance found in pleasantness more broadly. In the former case, ‘pleasant’ picks out the specific quality of pleasantness. In the latter case, ‘pleasantness’ picks out both pleasantness and unpleasantness. After all, unpleasantness is prone to variance just as much as pleasantness is.                When discussing a particular sensory type ST, I will use ‘ST pleasantness’ or ‘ST displeasure.’ So, for example, you will see the terms ‘olfactory pleasantness’ and ‘olfactory displeasure’ throughout. But you may also see ‘gustatory pleasantness’ and so on.       I will use the phrase ‘affective X’ when discussing particular theories. I do this for reasons of readability. So, for example, you’ll see the phrase ‘affective relationalism.’ This picks out a particular theory of sensory pleasantness, the relational view. The same holds for other theories.     4       There is another set of important terms. These pick out the nature of properties in the debate over sensory qualities/sensible properties. But before I do this, I’d like to clarify the terms I just gave. ‘Sensory qualities’ refer to properties of experience. ‘Sensible properties’ refer to properties of the object. When you see red, your subjective experience of seeing that color is a sensory quality and it corresponds with the sensible property of the object that causes your particular subjective experience.       I give a more sustained discussion of these properties later but want to clarify a few more terms beforehand. A subjective view of the sensible properties or (sensory) pleasantness will, no matter how the particular view is spelled out, rely on the idea that the sensible property type is (sensory) experience-dependent. Subjects are the owners of experience; hence, the subjective view. I define dependence in the following way. In every possible world where we find a particular sensible property type SP (like the color red), the existence of SP will be constituted by some (or other) experience type (corresponding to a particular sensory quality). Again, by experience I mean sensory experience. More formally, a subjective view holds that necessarily SP is subjective if and only if SP is constituted by some (or other) experience property type.           An objective view of the sensible properties or (sensory) pleasantness will, no matter how the particular view is spelled out, rely on the idea of the sensible property type as experience-independent. It won’t quite do to use the term ‘object-dependent’ as certain sensible property types, like smell (Batty, 2010), might not be properties of objects or experiences. Another way to spell this out is to claim that the above definition of experience-dependence does not hold  5  for the sensible properties in question. On the objective view, in all possible worlds where we find SP, the existence of SP will not be constituted by some (or other) experience type or any other property of subjects. The objective view can be defined more positively: in every possible world where we find SP, the existence of SP will be constituted by some (or other) experience-independent property type. Objectivist views will differ on which property plays this role. More formally, an objective view holds that necessarily SP is objective if and only if SP is constituted by a non-experiential property of the subject-independent world.           Certain views, like the relational views I mention later in this work, are complex examples of the above views. These views come in many shapes and sizes but the underlying idea is that the sensible property type is found in the relationship between certain subjective quality types and objective conditions. The views differ in how they spell out the nature of the relation. Irrealist views, like figurative Projectivism, deny that pleasantness is instantiated at all: it is not physical or relational or primitive.  Some of these views will be more relevant than others in the subsequent discussion. However, it is important to clarify the terms and conceptual terrain that I’ll be working in.       Consider this first pass at, an admittedly general, theory of pleasantness. On objective views, pleasantness will be constituted by a particular non-experiential, non-subjective property type. These could be physical properties or primitive properties. Both of these property types are objective in that they posit the existence of pleasantness properties that bear no constitutive relationship to subjects. On subjective views, pleasantness will be constituted by a particular experience quality type. These could be properties of the experience itself, mental particulars  6  (like sense-data) and so on. These quality types are subjective because they posit that the existence of pleasantness properties are dependent on subjective sensory experiences.        On most relational views, pleasantness will be constituted by both non-experiential, non-subjective property types—usually physical properties—and particular experience property types. Relational views posit that certain sensible properties are constituted in-between the objective and subjective; that pleasantness is an objective property that bears certain relationships to perceivers. By introducing relations to perceivers, relational views introduce the idea that subjective experiential quality types play a constituting role for pleasantness. There is sense in which once a subject and/or experience property type is introduced into the analysis at all, your view ceases to be objectivist. We can see relational views as, in a way, sophisticated forms of subjectivism. Instead of positing that pleasantness is a type of sense-data, say, the relationalist will claim that pleasantness exists when subjective qualities encounter certain objective properties.       I need to say something about method. I am interested in the nature of pleasantness qualities. To understand them, I look at two perceptual qualities: smell and color. I use the color debate to structure the general discussion. The color ontology debate is well-developed and it provides the theoretical structure needed for me to make claims about particular theories of affective qualities. Given this, I use the example of color in a general way; I don’t say anything about the nature of color or color science. Rather, I use examples of theories of color to probe theories of pleasantness.   7       But I also talk about smell, especially in the first two chapters. I do this to motivate affective physicalism. In other words, I use the example of smell to motivate one particular theory of affective qualities. I use smell for several reasons. First, there is a scientific literature on the topic of smell and pleasantness in which one can motivate a type of affective physicalism. Second, as will become clear, affective physicalism is a tough theory to motivate; it’s not easy to find adherents. The smell literature I engage with is the best and perhaps only way, in my view, to motivate the theory. Third, smell is both tightly connected to pleasantness and a paradigmatic perceptual quality that, unlike taste, is, in some sense, distal. Therefore, it is just the right sensory quality type in which to understand how you could be a physicalist about pleasantness.         So I discuss smell to motivate a particular theory of affective qualities and then use theories of color as a template for when I discuss the nature of affective qualities more generally. In the first two chapters, I motivate a physicalist approach to pleasantness. I focus on the case of olfaction for reasons just mentioned. I motivate via studies in olfactory psychophysics.       Ultimately, affective physicalism will not work. I discuss why in chapter 2 (preview: same stimulus, different perceptual effects). However, I tell the story of how it could work. This chapter is meant to motivate the affective physicalist framework. By doing so, we can explore the logical space of affective qualities. I believe this is an important task. Each major view in the logical space should be explored and motivated. This will, I believe, make the case for my positive view stronger.     8                                                      Chapter 1: Smelling Pleasure     1.1 Smelling Pleasure            Good and bad smells permeate our lives. The smell of lavender is pleasurable and we seek it out. The smell of ammonia at a high intensity is unpleasant and we seek to avoid it. Smell, like its close cousin taste (see Rozin, 1982; Smith, 2015), seems drenched with pleasantness. My task is to understand the philosophical implications of this fact for the purpose of a more fundamental question: what type of property is pleasantness?         Smell and taste (and flavor) are the chemical senses. These senses have been given short shrift in the history of philosophy (though see Reid, 1764). There could be any number of reasons for this (see chapter 1 of Korsmeyer, 1999). Fortunately, the non-visual senses have been getting philosophical attention in recent years (see O’ Callaghan, 2007 on sound and Fulkerson, 2014 on touch), including the chemical senses (see Batty, 2011 and Young 2016 on smell; and Smith 2015 on taste and flavor).       Recent years have also seen a growing interest in theories of sensory affect. There are other forms of affect—emotional and homeostatic (see Panskeep and Briven, 2012). But I will focus on the affect that accompanies sensations, not those that accompany emotional states or states of the body seeking internal balance. Sensory affect describes the pleasure you ‘feel’  9  when eating chocolate peanut butter ice cream or the displeasure you experience when smelling popcorn. Various theories of sensory affect have been proposed (see Aydede and Fulkerson, 2014; Carruthers, 2018; Heathwood, 2007, Schroeder, 2004; Robinson, 2006; Smuts, 2011; Bramble, 2013; Bain, 2013; Helm, 2002; Klein, 2015, Cutter and Tye, 2011).       My focus is on the chemicals senses. The relationship to pleasantness seems to be a special one. Smells and tastes seem more directly connected with pleasure than the other senses are.      My starting point is captured nicely by Austen Clark. He makes the following important point about the function of the chemical senses (and pain):   In motivational and emotional terms, the limbic system is a potent place. Pain wraps its tendrils around all parts of it. Unlike most other sensory systems, pain has direct access into the innards of our preference functions. Only a few systems do this: pain, smell and taste…These are arguably the most primitive of all our mental states…All three have direct and strong connections to motivational states…In these primitive feeling states the sensory and motivational components really do work in tandem: they do not arise entirely independently of one another, nor is one entirely antecedent to the other…The primitive worm at our functional core did not have the need to distinguish between its sensations and desires (2005, p.191-2).              I think Clark is onto something. Essentially, what he says is this: smell and taste function more like pain than they do seeing or hearing or touching. Smelling the rotting fruit and tasting the rotten cream are more like feeling the searing backache than they are like seeing a sunset or hearing a tree fall in a secluded forest; smell and taste are a sibling to pain but cousins to seeing, hearing and touching.   10       Notice that Clark commits to the idea that these ‘primitive feeling states’ (pain, smell, taste, etc.) have ‘direct and strong connections to motivational states’. Of course, pain perception is both sensory and affective. We sense something bad. But can the same be said for smell (and taste)? In Clark’s tandem model, we find a nascent framework; sensory and affective processes work in tandem to create an experience. But that is not yet good enough. The important question is: what is the exact nature of the direct and strong connections to motivational states evidenced by the chemical senses? The answer will determine much; whether, for example, there is a simple continuum between the traditional senses—seeing has weaker connections to sensory affect, smell stronger connections, pain the strongest connection—or a fundamental difference—the chemical senses are partly affective, the other traditional senses are not. If the difference is fundamental, then we can genuinely christen a group of evaluative senses.      The relationship between perception and cognition receives much attention and righty so (MacPherson, 2011; Siegel, 2010). However, Clark is probing a different relationship, that between perception and affect. Much less attention has been garnered by this question. The issues are similar. In the former case, we seek a clear understanding of how cognitive information can penetrate or influence perceptual information. Can my beliefs influence what and how I see, for example? In the latter case, we seek a clear understanding of how affective information can index to perceptual information. When in pain, do we perceive painfulness; when eating chocolate, do we perceive pleasantness?        Andreas Keller notes (2016), ‘Whether something that is directly wired to perception is part of perception or not is part of the difficult problem of drawing the line between perception and  11  cognition…’ (p.33). It is telling that Keller is responding to the above quote by Clark. The same strategy used to understand the distinction between perception and cognition can be used to understand the relationship between perception and sensory affect.       In this chapter, I explore the connections between a particular sensory system and its connections to our affective states. The goal is to understand the question: what is the nature of pleasantness; what kind of property is it?             Sensory systems—touching, hearing, seeing, tasting, smelling—bear certain relationships to affective systems—emotions, mood, feelings, longings, pleasure, displeasure (see the intro for a more detailed discussion of affect). In this chapter, I probe several important empirical studies purporting to show a tight relationship between olfactory perception and pleasantness. To be clear, these studies focus on monomolecular properties of orthonasal olfaction (olfaction of environmental stimuli). We smell food objects in the mouth (see Rozin, 1982 for a discussion of retronasal and orthonasal olfaction) and most smell objects are teeming with molecules. However, my strategy is one of a builder laying foundations. We must start somewhere.       In this chapter, I focus on five studies in olfactory psychophysics. The studies are discussed in several reviews (Yeshurun and Sobel, 2010; Castro and Seeley, 2014) and there are several other important studies on olfaction and pleasantness (Zarzo, 2008 and Haddad et al., 2010 are representative).1 My focus is not merely expository but critical. Interesting results abound but the conclusions remain inconclusive.                                                           1 For a good introduction to general issues in olfactory science see Gottfried (2010) and Stevenson and Wilson (2007). For a good introduction to general issues in the philosophy of olfaction see Batty (2011) , Young (2016) and Keller (2016).     12       This discussion contains two substantive parts. In chapters 1 and 2, I discuss said studies and provide an interpretation of what the researchers purport to show. I then provide a critique of the relevant conclusion. The goal is to motivate a physicalist approach to pleasantness. In chapters 3 and 4, I situate the previous discussion in the context of rival theories of pleasantness. In other words, I ask the questions: is O-P (object pleasantness) like a primary or secondary quality; is O-P ‘located’ in the perceiver or in the world?              1.2 Study 1: Pleasantness reflects the physical world       The first study of focus is Khan et al. (2007). In this study, the researchers seek to map the relationship between molecular properties and the olfactory percept. The perceptual space is marked by verbal descriptors obtained in previous olfactory studies (Dravnieks, 1985). The physiochemical space is marked by physiochemical structures obtained from chemical databases. Researchers tested for a systematic relationship between these two spaces. In simpler terms, the researchers are looking to study the relationship between, say, the verbal descriptor ‘rose’ and the particular chemical structure PEA (the dominant monomolecule found in roses) and so on for various verbal descriptors and chemical structures.        One hundred and eighty five subjects sniffed odorants in jars. Subjects subsequently rated stimuli by using a computer. Evaluative dimensions included similarity, intensity, pleasantness,  13  edibility and various qualities (‘sweet’, ‘flowery’ etc.). As a subject in this study, you would have encountered the following. You sniff a jar containing a particular odorant. You are then asked to rate the odorant for pleasantness (on a scale from ‘very pleasant’ to ‘very unpleasant’), edibility (‘edible’ to ‘toxic’), intensity (‘very low intensity’ to ‘very high intensity’) and various qualities (‘not at all flowery’ to ‘extremely flowery’). Researchers also conducted a reaction time study where subjects had to quickly rate whether two particular stimuli were the same or different.       A brief methodological interlude before I continue. Principal component analysis (PCA) is the method used to analyze the data from this study. This method is important for the current study and the subsequent two I discuss in this chapter. As such, it is important for me to spell out how PCA is used. PCA is a dimension reduction technique (for a good introduction see Bro and Smilde, 2014). Researchers obtain a large amount of data. Dimensions can enlarge quickly. However, using PCA transforms all this data into a new and small dimension space. The new space has two important properties: 1) the dimensions are orthogonal to each other and 2) the new dimensions are grouped as principal components (hence the name of the method). This latter property is crucial. Principal components are ordered according to how much original data variability they explain. PC1 explains the most variability, PC2 the next most and so on. Given orthogonality, the entire set of PC’s explains most of the original data set variance.       In the study under review, the original data set consisted of 160 odorants in a 146 odor descriptor space. Applying PCA, the researchers transformed the 144 monomolecular odors (two odors were dropped) into a subspace made up of the first four PC’s. So each odorant is now described by four numbers as opposed to the original 146. That is quite the reduction.  14  Each perceptual PC was modeled independently of each physiochemical PC in order to test the relationship between perceptual PC’s and physiochemical PC’s.       The perceptual odor space was broken down into the first ten PC’s. The first two PC’s—PC1 and PC2—accounted for 40.1% of the total variance and the first four PC’s accounted for 54%. The last six PC’s accounted for the vast majority of the rest of the data. So the first two PC’s—whatever they turn out to be—explain the majority of variation in odor perception space. Subsequent experiments revealed that the researcher constructed perceptual space corresponded to subjects’ judgements of similarity between pairs of olfactory stimuli (p. 10017). An important question remains: what is PC1?       Researchers in this study identify pleasantness as the first principal component of olfactory perception space. Various experiments were used to test this claim and all show a strong correlation between pleasantness and PC1. To highlight one study, twenty-two subjects rated the pleasantness, floweriness and sweetness of 22 randomly selected odors. These latter two odor qualities were chosen because they had, along with pleasantness descriptors, the highest weights (or loadings) in the perceptual PC1 (that is, they were the most relevant variables in the construction of PC1). Of course, all three showed a significant correlation with PC1 values. However, the correlation between PC1 and pleasantness was much higher than either of flowery and sweetness was to PC1. In other words, pleasantness captures the greatest amount of PC1 variance, more than its top two competitors. The important conclusion is this: pleasantness explains the largest amount of variability in odor perception space.   15       The researchers were only half done. A second space was built, this one around physiochemical properties of molecules. Researchers started with 1514 physiochemical descriptors for 1565 odorants. After applying PCA, researchers found that the first ten PC’s accounted for 70% of the variance, with PC1 accounting for 32%. However, due to the very large set and descriptor size, constructing a physiochemical space is much harder than constructing a perceptual space. PC1 seems to correlate with proxies for molecular size and weight, the amount and distribution of mass within a molecule. Still, given the above difficulties, we do not yet fully understand the true nature of the PC1 of physiochemical space.       However, we can still gain some understanding of the PC1 of psyiochemical space. We can compare it with PC1 of odor perceptual space. Essentially, we can compare two PC1’s. The researchers took the first four PC’s of odor perception space and looked for a correlation with the first six or seven PC’s of psychochemical space. In the most important result for my purposes, the first perceptual PC and the first physiochemical PC—the two PC1’s—were significantly correlated with a correlation coefficient of 0.49. This result provides some evidence that there is a relationship between olfactory pleasantness (PC1 of olfactory percept space) and molecular weight (PC1 of olfactory physiochemical space).       Crucially, none of the other PC relationships showed a significant correlation. The next strongest correlation coefficient was between physiochemical PC1 and perceptual PC4 and that a mere 0.20. None of the other PC relationships had a correlation coefficient above 0.12. The results are clear: a strong correlation exists between the PC1’s and all other correlations between the two spaces are rather weak. Pleasantness has been established as the PC1 of odor  16  perceptual space. As such, pleasantness correlates strongly with PC1 of physiochemical space (which may be molecular weight).   The idea that olfactory pleasantness is located in an objective property gets off the ground with a result like this, as oderent structure is an objective property.        To test these results, researchers built a model that predicted novel odor perception from odorant structure. They used odors commonly used in experiments but not any of the 144 they used in their previous experiments. The researchers then predicted the values for the PC1 of the perceptual space. Subjects were given a novel odor and then asked for a pleasantness rating (intensity was held constant). The predictive model did well, garnering a rank correlation of 0.72. The moral: the perceived pleasantness of odorants can be highly predicted by looking at the physiochemical properties of molecules. The researchers validated this relationship in cross-cultural studies (p. 10019-20).       The researchers conclude:  …those physiochemical measures that were best at discriminating a set of molecules were found to be those that were most correlated with the perception of olfactory pleasantness. In other words, when one orders a set of odorants based on the variance in their physiochemical properties alone, they end up roughly ordered by perceptual pleasantness as well (p. 10021).        1.3 Study 2: Pleasantness and molecular size    17       The second focus of study is Zarzo (2011). The researchers in study 1 showed a significant correlation between odor pleasantness and odorant structure. The following study takes off from this point. Zarzo assumes the results of Khan et al. (2007) and works with the idea, mentioned in section 1, that non-hydrogen atoms were one of the highest loadings in odor physiochemical PC1. In this study, Zarzo tests the correlation between non-hydrogen atoms and pleasantness judgments deduced from the odor character map (Dravniek, 1985). In other words, is there a strong correlation between molecular size and odor pleasantness?      Zarzo obtained pleasantness scores from five sets of chemical stimuli. The following variables were obtained for all the chemicals in the set: molecular weight, total number of atoms, total number of atoms sans hydrogen, number of carbon atoms, number of nitrogen atoms, etc. Importantly, oxygen atoms are much heavier than hydrogen atoms, about 16 times heavier. Therefore, a molecular structure composed of oxygen plus any number of non-hydrogen atoms will of necessity be heavier than atoms composed strictly of hydrogen atoms. The above molecular descriptors were compared with pleasantness scores and a correlation coefficient of 0.43 was found. However, the correlation between pleasantness scores and non-hydrogen atoms was stronger and more significant, sitting at a robust 0.59. This is very similar to the results of study 1.       The relationship between pleasantness tones, odor character description and functional grouping is close. According to Zaro, certain functional groupings determine specific odor character and we can find pleasantness scores for odor characters. Esters, ketones and lactones usually smell pleasant. This indicates the importance of an oxygen link in pleasant smelling  18  odors. In other words, the presence of oxygen (except in acids) increases perceived olfactory pleasantness. Many sulfur and acidic compounds smell unpleasant, as various equations in this study seem to indicate.       All in all, Zarzo found that molecular size accounts for between 0.46 and 0.71 of the pleasantness judgments made by subjects of the five analyzed studies. The moral: molecular weight best explains the discrimination of pleasant versus unpleasant odors. Furthermore, if a molecule has oxygen plus at least six additional non-hydrogen atoms, it will most likely be perceived as pleasant. If a molecule has sulfur or carboxylic compounds, it will most likely be perceived as unpleasant. Heavy, oxygen drenched molecules will be perceived as pleasant; sulfurs and acids generally won’t. Zarzo has shown how molecular size strongly correlates with perceived odor pleasantness. Perhaps the PC1’s of olfactory perception and olfactory physiochemical space are more strongly correlated than Khan et al. (2007) originally thought.  1.4 Study 3: Pleasantness and the electronic nose        The third study of focus is Haddad et al. (2010). In this study, researchers test the link between novel odors and pleasantness. An electronic nose device (eNose, hereafter) is used to test this link. ENoses are set up to closely mirror biological olfaction, even encoding the chemical mixture that makes up an odor. Like in biological olfaction, an e-sensor type can respond to more than one odorant and one odorant type can activate more than one e-sensor. The authors mention that almost all eNose studies focus on odor detection and discrimination.  19  The current study shifts emphasis to odor pleasantness. Specifically, the researchers test for the link between eNose output and pleasantness and then test for predictions of the pleasantness of novel odors.      Researchers needed to train the eNose. They measured 76 odorants with an eNose. These odorants were measured six times at the same concertation. After accounting for signal failures, 378 odorant samples were procured. 16 sensors were used and 120 features were extracted. A 378 X 120 matrix resulted from this experimental setup. 14-20 subjects were used to rate the pleasantness of each odorant. In this training run, the correlation between the human rating and eNose predication was 0.46; a significant correlation. More importantly, a p-value of less than 0.05 was obtained in 100% of the trial runs. The researchers conclude that an eNose can be tuned to olfactory pleasantness.      This, however, is not the big result of the study. The researchers were especially interested in the eNose ability to predict the pleasantness of novel odors. One could see why: if an eNose can successfully complete this task, then an argument can be made that pleasantness is somehow ‘written into’ the molecular structure of odorants themselves. To test this, the eNose was given odorants that it did not encounter during the training period; 22 novel odors. The same parameters were used in the training and novel odorant experiments. 14 human subjects were asked to rate the pleasantness of said novel odors. The correlation between the machine predictions and human ratings was 0.64. The researchers then calculated the correlation between each human subject rating and the overall median human rating; a correlation of 0.72  20  resulted. Importantly, the machine-human correlation was 88% of the human-human correlation. The presumed result: the eNose generates human like odor pleasantness ratings.       The researchers weren’t done. In the previous experiments, a continuous pleasantness scale (30 points) was used. In the following experiment, categorically pleasant and unpleasant odors were used. Researchers cast aside the odorants with pleasantness ratings in the 10-20 point range and so tested odorants that had previously rated 0-9 (unpleasant) or 21-30 (pleasant). The eNose discriminated between the two groups with 99% accuracy. Importantly, given the experimental set up, the eNose pleasantness discriminations did not depend on odor intensity; so no argument can be made that eNose abilities to discriminate odor pleasantness depend on perceived odor intensity.       Given the various contextual factors (culture, learning, etc.) that influence odor perception, the researchers wanted cross-cultural validation. Two cultures were studied: Israeli and Ethiopian. The human studies used in the previous experiments were used in the cross-cultural experiment. The correlation between the two cultural groups was 0.75. The correlation between the machine predictions and the Ethiopian ratings was 0.52. The correlation between the machine predictions and the Israeli ratings was 0.49. The researchers conclude that the eNose is tuned to culture-independent properties of olfactory molecules.       The upshot of the study is this: an eNose can be trained to successfully predict the pleasantness character of particular odorants and can do so regardless of cultural influence. As such, eNoses can be trained to ‘pick up’ the pleasantness character of particular odorant molecules. Given the significant correlation between the machine predictions and human  21  ratings, we humans must also have this ability. As such, an argument can be made that pleasantness is ‘written into’ the very structure of odorant molecules. In my discussion below, I discuss these results and the resulting picture they engender.                                                1.5 Study 4: Pleasantness and olfactory receptor surfaces        The fourth study of focus is Lapid et al. (2011). In this study, researchers test the hypothesis that the organization of the olfactory epithelium reflects olfactory perception. In other words, does the organization of the olfactory receptor surface (the olfactory epithelium, tissue at the nasal cavity roof that is small in measurement and transduces incoming information from odorant molecules) reflect the primary axes of olfactory perception?      The experimental setup included placing an EOG (Electrooculography) electrode into the left nostril of subjects (57 of the 81 subjects elicited measurable recordings). Three batches of six odorants were used for each experiment. The researchers of this study tested the following assumption: ‘…in an epithelium consisting of uniform and randomly distributed receptor subtypes, one would expect a similar EOG tuning profile no matter where the electrode was placed (p.1456).’ This means that the same odorant would generate the greatest response regardless of recording location. However, researchers found this to be incorrect. The maximal response was generated by a distribution of odorants: ‘…for each location, there was a subset of receptors that best responded to one of the six odorants (p.1456).’ The differences observed  22  were differences in recording locations and not differences across subjects. In other words, recording from different locations in the same subject does not generate the same response.       Another experiment was run to bolster this result. In the first experiment, each subject was given six odorants at one maximally responsive location. Of course, each location between subjects differed; each particular subject’s epithelium being maximally responsive at different locations. The result of the first experiment is vulnerable to the following criticism: only differences between subjects were tracked and not any differences across recording locations. In the second experiment, each subject was tested at three different locations. So for a subject S, location L and odorant O, O1 was placed at L1 in S1, O2 was placed at L2 in S1, O3 was placed at L3 in S1 and so on for each subject. As in experiment 1, researchers observed ‘an independent location that was maximally tuned to each of the three odorants that [were] used to identify a responsive location (p.1456).’               The conclusion reached: olfactory receptor subtypes are concentrated in olfactory epithelium patches and are not uniformly and randomly distributed. There is an important functional response pattern in a given patch of epithelium. Importantly, this is true regardless of subject. If only one odorant was used to identity a recording site, the same response profile would occur across subjects. Researchers found that a particular odorant, when used to identify a responsive location, induced a maximal response in almost all subjects tested.       Researchers then tested what this response organization was tuned to. They calculated EOG response distances with distance in odorant perceptual and physiochemical space. Crucially for my purposes, pleasantness was significantly correlated with odorant EOG distance, indicating a  23  correlation coefficient of 0.45. By contrast, intensity and trigeminality (chemical irritation) were significantly less correlated with odorant EOG distance, indicating correlation coefficients of 0.09 and 0.23 respectively. In simpler terms, a location that responded maximally to a pleasant odorant was more likely to respond maximally to other pleasant odorants. The same structure holds for unpleasant odors. It seems, in other words, that we have a location in our nose that is tuned to odor pleasantness.       The correlation between EOG response and physiochemical structure was significantly weaker, though still significant, than the correlation between EOG response and odorant perception. The PC1 of physiochemical space indicated a correlation coefficient of 0.22 with EOG response space. In a previous section, I discussed the finding (Khan et al., 2007) that the PC1’s of odorant space (pleasantness) and odorant physiochemical space (most likely molecular weight) were strongly correlated. However, notice that the PC1 of physiochemical space is only weakly correlated with EOG response space. EOG responses are more tuned to olfactory pleasantness than odorant structure/molecular weight. However, odorant structure is strongly correlated with olfactory perception. Interesting connections abound. I discuss these interconnections in greater depth after I discuss the final study. But the moral of the second study is this: the organization of the olfactory receptor surface reflects the key olfactory perceptual axis, pleasantness.   1.6 Study 5: Categorical Odors   24       The fifth and final study of focus is Castro et al. (2013). Up till now, I’ve focused on studies linking olfaction and pleasantness. This study stands out from the others, not only in focus but also method. In this study, the researchers are seeking to understand the structure of odor profiles and whether odor qualities can be predicted from molecular structure, tasks not too different from the previously analyzed studies. However, the researchers in this study do not  test for olfactory pleasantness directly; hence the difference of focus. The previous studies relied heavily on the PCA method when analyzing the relationship between odor qualities and odor molecules and between each of these and olfactory pleasantness. In this study, researchers use non-negative matrix factorization (NMF) instead of PCA; hence the difference of method.       And the difference is important. Using NMF allows the researchers to test whether odor space can be described with sparse and discrete perceptual categories. Both PCA and NMF are dimension reduction techniques. They both take a large data set and reduce the number of dimensions, allowing for a greater understanding of said data. However, each method reduces in a different way. As previously discussed, the vectors of PCA are used to maximize variance. The vectors of NMF are used to obtain non-negative values (zeros and ones allowed, negative values disallowed); hence the method’s name. The goal is to derive a matrix from multiplying two other matrices: one that represents feature vectors and one their weightings. This can be clarified by jumping right into the study.       The Dravnicks (1985) olfaction database—discussed in the first two studies—is constituted by non-negative data—a given semantic label applies or it doesn’t—making it apt for a NMF  25  analysis. The matrix sought was composed of 146 descriptors (like ‘sweet’, ‘floral’ or ‘earthy’) X 144 odors. Importantly, this matrix M is the product of two other matrices, call them sub-matrix1 (SM1) and sub-matrix2 (SM2). SM1 is composed of semantic descriptors and represents the feature vectors. SM2 is composed of odorants and represents the weightings of the feature vectors. Multiply SM1 and SM2 and you get M.       Using NMF, the researchers obtained a 10-dimensional representation of odor dimension space. The vectors of said space were sparse, meaning the key vectors contained several large values with the remaining values near or equal zero. Four particular axes stood out as defined by the key vectors: ‘fragrant’, woody’, ‘fruity’ and ‘sickening.’ These represent the first four dimensions of olfactory space as determined by the NMF method. The remaining six in order: ‘chemical’, ‘minty/peppermint’, ‘sweet’, ‘popcorn/burnt/smoky’, ‘sickening/garlic/onion’ and ‘lemon/citrus’. Crucially, the authors point out that, despite there being no orthogonality constraint on NMF as there is for PCA, the olfactory data set of this study turned out nearly  orthogonal (see the study for more details).       Once the dimensions were established, the researchers looked at how the odorants were distributed within the odor descriptor dimension space. Each piece of odorant data found in SM2 was scanned until the odorant with the largest coordinate in dimension 1 (‘floral’) was found. And so on for each of the 10 subsequent dimensions. An important finding resulted. Each odor was strongly characterized by one dimension and each of the 10 dimensions was occupied. The first indication of the categorical nature of odorants was thus revealed.   26       Indeed, this is the key finding of the study: odor dimensions apply categorically, they are discrete and clustered. Interestingly, the data shows strong clustering of both descriptor space (SM1) and odorant space (SM2). For example, NMF revealed 8 discrete, non-overlapping clusters in the descriptor space. ‘Fruity/Citrus’ occupied the first cluster and is not anywhere close to overlap with any of the other clusters. NMF revealed 10 discrete, non-overlapping clusters in the odorant space. Cluster 4, occupied by Amyl Cinnamic and other chemicals, is likewise not close to overlap with any of the other clusters.       Comparing this study with the four previous allows an important distinction to be made. Relying as they do on PCA, the researchers in the first four studies can be said to give a distributional analysis of odor dimensions. Odors span multiple dimensions, they do not apply categorically. In contrast, the researchers of study five give a categorical analysis of odor dimensions. Odors apply categorically and do not span multiple dimensions. That is, a particular odor belongs in one dimension at the exclusion of other dimensions (‘floral’ and ‘woody’ being two examples). The key point for my purposes: the same odorant data can provide variable conclusions depending on which dimension reduction technique is used (PCA or NNMF). This will prove an important point in my subsequent discussion.              To sum up so far: PC1 of olfactory perceptual space is pleasantness. PC1 of olfactory physicochemical space is most likely molecular weight. These two PC1’s are strongly correlated (study 1). Chemicals with higher molecular weights—oxygen plus at least six additional non-hydrogen atoms—are strongly correlated with perceived olfactory pleasantness (study 2). A trained eNose can predict olfactory pleasantness from odorant structure (study 3). Neural  27  activity at the olfactory epithelium is ‘tuned’ to reflect the PC1 of olfactory perceptual space—pleasantness (study 4). Using NMF, we see that odor dimensions are categorical; that is, odors group into discrete and exclusive clusters (study 5).  1.7 Interpretation            A tempting philosophical picture of olfaction emerges, supported by many of the researchers whose studies I analyzed in the previous sections. Four of the following quotes appear in the studies just analyzed; two appear in reviews that discuss said studies:   Pleasantness is the primary perceptual aspect humans use to discriminate odorants…it is clearly the pleasantness meaning of odor that dominates odor perception (Khan et al., 2007, p. 10020).   This revealed a…correlation between the response pattern and the perception of odorant pleasantness. In other words, as in vision and audition, organization of the receptor surface in olfaction reflects a primary perceptual axis [pleasantness] (Lapid et al., 2011, p. p.1460).   …a fundamental question still unsolved in the field of olfaction is what makes an odorant smell pleasant or unpleasant. One theory is that acquired semantic knowledge is one of the important factors…An alternative view [and one the author favors]…is that…the pleasantness of  28  odors may be partly explained by the physicochemical properties of the odorant molecules themselves (Zarzo, 2011, p. 3679).  …unlike in vision and audition, in olfaction there is a systematic predictable link between stimulus structure and stimulus pleasantness. This goes in contrast to the popular notion that odorant pleasantness is completely subjective…These findings imply that unlike in vision and audition, in olfaction pleasantness is written into the molecular properties of the stimulus…This finding of hard-wired odorant pleasantness is in contrast to the popular notion that odorant pleasantness is both subjective and learned…olfactory pleasantness is hard-wired to molecular structure (Haddad, 2010).   The resultant axis, the first principal component of molecular structure, i.e., the axis that best explains the variance in odor structure, was significantly correlated to the perception of odorant pleasantness. This implies that pleasantness is indeed “written into” the molecular structure of odors (Yeshurun and Sobel, 2010, p.229).   Building off this work, we speculate that the affective (pleasantness) dimension of olfactory experience is fundamental, and that olfactory categories, as a result, carry information about the behavioral salience of their sources as opposed to the specific identities of either the odorant perceived or its source (Castro and Seeley, 2014, p. 299).   29      It is not immediately clear how to interpret these statements. However, two important conceptual categories emerge as important. Firstly, there is the objective-subjective category. Secondly, there is the innate-learned category. Most of the requisite studies mention these terms as capturing what the data show. As such, four views emerge as possible interpretations: objective/innate, objective/learned, subjective/innate, subjective/learned. The truth is not so simple, as I will later argue. But these possible views will help get at the best interpretation. Before I discuss these four views, I need to clarify what is meant by the relevant concepts.      I will define the concepts of ‘objectivity’ and ‘subjectivity’ operationally as outlined in the ‘terms’ section above. Put simply, objective properties are experience-independent, subjective properties are experience-dependent. As I said above, it is not just the objective-subjective distinction which looms large in the analyzed olfactory psychophysics literature. The innate/learned distinction looms large as well. There are two ways to define ‘innate.’ The first comes from psychology, the second from computer science. On the psychological view, ‘innate’ is defined as genetically endowed (this is general; I discuss a more specific rendering below). On the computer science view, ‘innate’ is defined as involving two or more permanently connected circuits, i.e. circuit connections that are unmalleable. In the olfactory psychophysics literature analyzed, ‘innate’ is typically contrasted with ‘learning’ which is consistent with both definitions. However, the researchers typically mention how olfaction is modulated by learning (I’ll discuss this more below). This suggests that the olfactory psychophysicists believe some aspects of olfaction are malleable. Therefore, I interpret them as using ‘innate’ in a way consistent with both the psychological and computer science definitions. Genetically endowed  30  properties and fixed circuits can both be modulated vie learning. The definition of ‘learned’ can be defined in contradistinction: endowed through experience, malleable, etc. I can now unpack the four possible interpretations mentioned above.       Olfactory pleasantness is the target phenomenon. The psychophysics studies I analyzed engender tentative conclusions about the nature of olfactory pleasantness. The questions: is olfactory pleasantness objective or subjective and are the olfactory mechanisms that encode these properties innate or learned? To be clear, objectivity/subjectivity applies to pleasantness properties and innate/learned applies to the connection between olfactory mechanisms and olfactory pleasantness. To be clear: for the purposes of interpretation, I am exploring the logical space of the aforementioned literature on olfactory psychophysics.         As mentioned previously, four potential views emerge directly from comparing these axes. To be clear, I think the interpretative truth will lie somewhere between these views. But I think this exercise will be helpful in understanding what the olfactory psychophysicists take their studies to show.      Olfactory pleasantness could be objective and innate. On this view, olfactory pleasantness is located in the molecular structure of odorants and we have innate mechanisms that are tuned to encode these properties. That is, pleasure is a perceiver independent property that we don’t learn to perceive but are built to perceive (I use ‘built’ synonymously with ‘innate’). This is a strong view. It treats olfactory pleasantness like a sensible property, similar in certain respects to colors, sounds, touch, taste and felt temperature. Smelling pleasure is a matter of being confronted with a heavy odorant molecule which our olfactory system is innately built to  31  encode. Clearly, this view precludes any role for learning in the explanation of our olfactory pleasantness experience. An analogous view could be that colors are surface spectral reflectance properties of objects (and so objective) and we have innate mechanisms for encoding these light properties.        Olfactory pleasantness could be objective and learned. On this view, olfactory pleasantness is located in the molecular structure of odorants and we learn to encode these properties through experience. That is, pleasure is a perceiver independent property that we learn to perceive. As in the previous view, olfactory pleasantness would be akin to the sensible properties; pleasantness objectivism entails this. Smelling pleasure is a matter of being confronted with a heavy odorant molecule which our olfactory system learns to encode. This view precludes any role for innate mechanisms in the explanation of our olfactory pleasantness experience. A similar view could be that tastes and flavors are chemical properties of objects (and so objective) but we learn through experience to pick up on different flavor properties of the object (Barry Smith holds this view but would not extend it to pleasantness properties; 2007, 2013).      Olfactory pleasantness could be subjective and innate. On this view, olfactory pleasantness is located in subjective experience and we have innate mechanisms that are tuned to “encode” these properties. That is, pleasure is a perceiver dependent property that we are built to perceive but do not learn to perceive. Unlike the previous two views, olfactory pleasantness would not just be a sensible property. Smelling pleasure is a matter of subjective processes of the perceiver which our olfactory system is innately built to encode. This view seems promising.  32  On it, we have innate mechanisms that are tuned to encode molecular properties in a way that allows us to attribute pleasantness to our experience of said properties. I’ll say more about this approach in my subsequent discussion.         Finally, olfactory pleasantness could be subjective and learned. On this view, olfactory pleasantness is located within the perceptual system of the animal or in subjective experience and we learn to “connect” these properties with olfactory perceptual mechanisms through experience. That is, pleasure is a perceiver dependent property that we learn to ‘perceive’ but are not built to perceive. As in the previous view, olfactory pleasantness would not just be a sensory quality; pleasantness subjectivism entails this. Smelling pleasure is a matter of subjective processes of the perceiver which our olfactory system learns to ‘encode.’ This view precludes any role for innate mechanisms in the explanation of our olfactory pleasantness experience. On this view, our olfactory system learns through experience which sensory qualities are appropriate to index as pleasant and which to index as unpleasant. So it is not so much that we learn to ‘encode’ subjective processes per se. Rather, our olfactory system learns to connect the right kind of pleasantness experiences with the right kind of (olfactory) sensory experiences. So it’s not quite correct to speak of ‘encoding’ these subjective pleasantness qualities per se.         There is no need to look for similar or analogous views. This is the default view of olfactory pleasantness and sensory affect more generally. The subjective/learned view is especially prominent in affective neuroscience (more on this in chapter 2). The view is also culturally  33  prominent, though we also attribute pleasantness properties to mind-independent objects, like food and drink items (see Aydede and Fulkerson, 2014).      The question is: where does the previously discussed literature fit in? How should we interpret the results? I think there is a negative and positive element. Clearly, the aformentioned olfactory psychophysicists see their results as undermining the default subjective and learned view of olfactory pleasantness. The Haddad quote at the beginning of this section is very clear on this point. This is the negative element of the interpretation: olfactory pleasantness is not subjective or learned. The positive element of the interpretation is more nuanced and deserves more discussion.       The subjective/learned view is out. The subjective/innate view might fit what these researchers are going for but it is not clear how. I suspect one issue is that these theorists conflate O-P (object-pleasantness) and E-P (experiential-pleasantness) but it is not totally clear. Both objectivist views get closer to the correct interpretation but are too coarse-grained. The correct interpretation is more nuanced. In the literature under discussion, the results provide evidence for the following conjoined claim: 1) there is an innate connection between our olfactory perceptual mechanisms, heavier odorant molecules and subjectively experienced pleasure, and 2) sometimes olfactory pleasure can be modulated by learning and experience.   …it is notable that although our predictive power for pleasantness was significant, we explained only a portion of the variance. This leaves open the possibility that individual differences and plasticity in olfactory pleasantness make important contributions to olfactory experience (Khan et al., 2077, p. 10022).   34   …experimental evidence supports the view that the pleasantness perception of odorants is a complex process which involves both innate and learned components (Zarzo, 2011, p. 3680).2      …we argue that olfactory pleasantness is hard-wired to molecular structure. That this link is modified through culture, context, and learning does not preclude the initial hard-wired aspects of this link, and it is this link that we have captured (Haddad et al., 2010).   Although our results imply a hard-wired aspect of odorant pleasantness, they do not imply that odorant pleasantness is unmalleable, as a portion of perception that is hardwired at the receptive surface does not rule out later remapping…this mapping is by context and experience (Lapid, 2011, p.1460).   …pleasantness does not explain all the variance in olfactory perception, nor does physiochemical structure explain all the variance in pleasantness. For example, odor pleasantness is dependent on odor intensity and familiarity, and it varies across individuals and cultures as well as within individuals over time (Yeshurun and Sobel, 2010, p. 229).            So olfactory pleasantness can be modulated by learning and experience but there is an innate connection between olfactory perception and the pleasantness found in heavier odorant molecules. Perhaps the researchers think there is some subjective aspect when learning modulates the olfactory pleasantness experience. It is not clear. However, what is clear is that                                                           2  Note the language of pleasantness ‘perception’ which might already assume a type of pleasantness objectivism.   35  said researchers think they have found an innate connection between olfactory perception and olfactory pleasantness, where the latter are, at least partly, objective. Of course, it does not follow that this explains the entire relationship. Learning and experience play some role. But the positive interpretation, if true as a theory, is certainly enough to undermine the subjective and learned view. This latter view precludes the idea that there is an innate connection between olfactory perception and olfactory pleasantness.       The true interpretation can be compared with the four possible views. There is a sense in which the correct interpretation is compatible with all the views. The analyzed results have, at most, shown that there is some kind of innate link between olfactory percepts and olfactory pleasantness; the latter being, at least in part, objective. However, learning and experience can and do modulate this innate link. The following view is then in play: the relationship between olfactory percepts and olfactory pleasantness is innate and objective till it is modulated by learning and experience whereupon said link adds elements of learning and subjectivity.       Of course, the four interpretive views under discussion are crude and too coarse-grained. It is unlikely that the requisite olfactory psychophysics literature would fit in to such interpretive schemes. However, the correct interpretation shows something interesting. The positive view from olfactory psychophysics is permissive. Olfactory pleasantness is both subjective and objective; the link between olfactory percepts and olfactory pleasantness both innate and learned. What to make of this? I think the positive view collapses into the negative view. The researchers have shown that there is some kind of innate link between olfactory percepts and olfactory pleasantness. The goal is not to push a particularly tight positive view per se. Rather,  36  the goal is to undermine the dominant view; the view that olfactory pleasantness is both subjective and learned and nothing else. In a way then, the literature under discussion can be seen as a nuanced counterweight to the seemingly less nuanced dominant view. So what seems to be a strong and unlikely view—that olfactory pleasantness are objective—is, in actuality, the more moderate and seemingly likely view.       Unfortunately, things are not so simple. In chapter 2, I provide what I think is the most charitable interpretation of the results from studies 1-4. I then argue that there some flaws in pleasantness objectivism.3 After expressing concerns over methodology, I discuss how results from affective neuroscience present a potentially fatal problem for pleasantness objectivism. I then provide alternatives to the affective objectivist view I discuss.                                                                                                                                                                        3 In chapter 3, I present another type of affective objectivist view. However, in chapter 2 I assume that affective objectivism is physicalist.   37                                                    Chapter 2: Affective Physicalism    2.1 Default Parameters        I created a logical space in my interpretation of the psychophysics studies. Despite this, the most charitable interpretation of the results has not been discussed yet. In this chapter, I discuss what the above results show and how they motivate a type of pleasantness objectivism. I then push several arguments against this pleasantness objectivism. An important point: I am not arguing that the above researchers hold this type of pleasantness objectivism. I am using their results to motivate a type of objectivism; affective physicalism.        Consider the following analogy. The HBO series Westworld features a player piano. The piano functions in a unique way. Patterned marks are made on a punch card which is subsequently fed through an encoding mechanism on the piano. The piano then plays a tune without any human hands playing the keys. In other words, the tune is programmed to play on the piano and the piano is ‘tuned’ to encode the patterned marks and play accordingly.       We can think of olfactory pleasantness in a similar way. Heavier odorant molecules are like the patterned marks on the punch card. Said molecules exhibit a pattern found in nature. Our olfactory mechanisms are like the encoding mechanism on the piano. Heavier odorant molecule information is fed through the olfactory epithelium (the smell receptor) and then encoded. The music played on the piano is a response to the encoded patterned marks fed through the  38  encoding mechanism. Likewise, olfactory pleasantness is a response to the encoding of odorant molecular structure. Essentially, you take molecule X, push it through the perceptual system and it comes out pleasant.       The analogy clarifies what default parameters mean for olfactory pleasantness. It does not (necessarily) mean that we are genetically endowed to encode heavier odorant molecules as pleasant. It certainly does not mean that our olfactory mechanisms must always encode heavier odorant molecules as pleasant, learning be damned. Parameters can always be reset. Rather, a key lesson of the analyzed studies is this: our olfactory pleasure tracking mechanisms have built in defaults that are tuned to heavier odorant molecules. Like the player piano, if the mechanisms encode the ‘right’ kind of pattern, a particular response will follow. However, unlike the player piano, our olfactory mechanisms can learn to develop a different response to the same pattern of odorant structure that previously generated a pleasant response.       We can now get a better grip on the idea that pleasantness is ‘written into’ the molecular structure of odorants. ‘Written into’ can be thought of like the player piano punch card; a symbolic, response generating structure. Patterns are found all over nature. In this case, the patterns are found in the molecular structure of odorants; the heavier the odorant, the more pleasant it will be. The pattern hits an encoder and generates a response.      This conclusion engenders a kind of pleasantness objectivism, affective physicalism. On the face of it, an affective physicalist thinks that pleasantness is a property of the world and not a property of the subject. That is, pleasantness is not a property of experience but of a physical property type, in this case, molecular weight. This is affective physicalism. The idea of objective  39  patterns motivates the view. The above studies seemingly show that our olfactory mechanisms are partly pleasure tracking mechanisms. Pleasure is found in heavy odorant molecules. There is no reference to the subject here. The subject only enters when we discuss how our olfactory mechanisms function as pleasure tracking mechanisms (among other olfactory functions).       The subjective view under attack can now be clarified. On this view, according to the above researchers, olfactory mechanisms have no built in pleasantness encoding default parameters. We only and always learn to set the pleasantness encoding parameters; the parameters are not default. In this case, the subject must learn to set their own parameters and so subjectivism gets off the ground. This is the view the affective physicalists are challenging. The key question is: how can affective physicalism emerge from the idea that we have default parameters governing the relationship between olfactory perception and olfactory pleasantness? The pleasantness is not somehow ‘in’ the molecules themselves. Rather, our responses are tuned to objective patterns in nature. Our pleasure tracking mechanisms are tuned to objective patterns in the world.       A relational theory of pleasantness would hold that olfactory pleasantness is a property of the relationship between heavier odorant molecules and our olfactory mechanisms. This view is a type of weak subjectivism.4 The subject enters the analysis but it’s the relationship to odorant structure that matters and not so much the subject’s own affective mechanisms modulating incoming sensory information.                                                            4 I’ll differentiate between subjectivism and the relational view even though the latter is a type of weak (or complex) subjectivism. I do this for reasons of readability but the views are also importantly different enough to warrant a differentiation of theory name.     40       So the debate over sensory affective qualities mirrors the debate over sensory qualities. Are sensory/affective qualities objective, subjective or relational? One could hold that the sensory qualities are objective but pleasantness qualities are subjective or relational. One could hold that both sensory and pleasantness qualities are relational. So, for example, smells could be a relation between subjects’ sensory mechanisms and odor plumes (Batty, 2011) or molecular structure (Young, 2016) just as the pleasantness of smells could be a relation between subjects’ affective mechanisms and molecular structure.         Echoes of pleasantness objectivism appear in the history of philosophy. ‘Odor perception is largely the perception of odor valance. Plato suggested that “pleasant” and “painful” are the only odor categories’ (Keller, 2016, p. 124). Likewise, Lisa Shapiro (2014, p.405) shows that Bishop Berkeley held ‘…that the character of immediate sensory experience is constituted by pleasure and pain. All sensible things are perceived immediately, and what we perceive immediately is just pleasure and pain. So all sensations just are pleasures and pains’ (see Rickles, 2013 on mediate/immediate perception in Berkeley). Plato and Berkeley are not pushing pleasantness objectivism per se. Rather, Plato claims that sensory affect determines odor categorization and Berkeley claims that all instances of immediate perception are constituted by pleasantness. However, Plato’s point fits nicely with the above psychophysics literature: pleasantness is the first component of odor perception.       Berkeley makes a constitution claim: percepts, including olfactory percepts, are constituted by pleasantness. However, Berkeley is an absolute idealist; to be is to be perceived. So if olfactory percepts are governed by affect, olfactory objects are as well. That is, smells just are  41  affective states. Though Berkeley ends at idealism, it is a short step from his claims about sensory affect to the claim that pleasantness is an objective property. All one needs to do is argue that Berkeley got perception right but his ontology wrong. We do directly perceive objects, including their pleasantness qualities, but these are material and not ideal. Indeed, this seems just what Thomas Reid does (see Copenhaver, 2013 for a good discussion of Berkeley’s influence on Reid).       Plato and Berkeley are both making claims about olfactory perception. Echoes of pleasantness objectivism can be found in their views but to get a full on pleasantness objectivism we need something more; that something is the idea of pleasantness as sensory quality. This idea brings affective physicalism to the fore. The aforementioned psychophysicists seem to show that our olfactory mechanisms are partly pleasure tracking mechanisms tuned to heavier odorant molecules. Now consider the idea that sensory affect is a kind of sensory quality. If true, olfactory pleasantness is a sensory feature of our experience.  And if the quality in question is found in heavier odorant molecules, then we have found the sensible property (molecular structure) that corresponds to our sensory quality (experienced pleasantness).      There are two ways to imperil this argument. I could show that the empirical results have not established it. I could also show that, even if the empirical results are sound, the claim falters for independent reasons. That is, I could show that, despite the empirical evidence, it cannot be true that olfactory pleasantness is a sensory feature. In the rest of this chapter, I aim to do both. Along the way, I’ll re-engage periodically with the question of pleasantness properties as sensible properties.      42  2.2 Methodological Troubles        Two methodological sticking points emerge in the olfactory psychophysics literature I canvassed. These problems are not enough to imperil the objectivist approach per se. However, they do serve to initiate the destabilization of the view. One problem emerges from comparing the results of studies using PCA and one using NNMF. Another problem emerges from comparing the results of study 4 (Lapid et al., 2011) with the results of study 1 (Khan et al., 2007), two studies that rely heavily on the PCA approach. So there are problems internal and external to the PCA approach. I discuss each in turn.       I start with the internal problem. In study 1, the researchers took the first four PC’s of odor perception space and looked for a correlation with the first six or seven PC’s of psychochemical space. The first perceptual PC and the first physiochemical PC—the two PC1’s—were significantly correlated with a correlation coefficient of 0.49. Crucially, none of the other PC relationships showed a significant correlation. The results are clear: a strong correlation exists between the PC1’s and all other correlations between the two spaces are rather weak. Pleasantness had been established as the PC1 of odor perceptual space. As such, pleasantness correlates strongly with PC1 of physiochemical space (which may be molecular weight).        However, note one of the most important results of study 4. The researchers calculated EOG response distances with distance in odorant perceptual and physiochemical space. Pleasantness was significantly correlated with odorant EOG distance, indicating a correlation coefficient of 0.45. The correlation between EOG response and physiochemical structure was, however,  43  significantly weaker, though still significant, than the correlation between EOG response and odorant perception. The PC1 of physiochemical space indicated a correlation coefficient of 0.22 with EOG response space. In the previous paragraph, I discussed the finding (Khan et al., 2007) that the PC1’s of odorant space (pleasantness) and odorant physiochemical space (most likely molecular weight) were strongly correlated. However, notice that the PC1 of physiochemical space is only weakly correlated with the EOG response space.       Conjoining studies 1 and 4 we get: EOG responses are more tuned to olfactory perception than odorant structure but odorant structure is strongly correlated with olfactory perception. So in one study odorant structure correlates strongly with olfactory perception and in the other study it only correlates weakly. Both studies rely on the PCA approach. As such, this is a problem internal to the PCA approach. Both studies compared olfactory perception with odorant structure and each finished with differing correlation strengths. The takeaway: using the PCA approach is not enough, by itself, to establish any conclusion about the correlative, let alone the causal, relationship between odorant structure and odorant perception. The claim that olfactory pleasantness is written into molecular structure needs far more support.       As if that weren’t enough, there is an external problem with PCA. Both PCA and NNMF are dimension reduction techniques. They take a large data set and condense it for analysis. Studies 1-4 (chapter 1) rely on PCA. Study 5 (Castro et al., 2013) relies on NNMF. But notice the very different conclusions found. In study 5, we learn that each odor is strongly characterized by one dimension and each of the 10 dimensions studied was occupied. The key finding of the study:  44  odor dimensions apply categorically, they are discrete and clustered. Interestingly, the data shows strong clustering of both descriptor space and odorant space.       Comparing this study with the four previous allows an important distinction to be made. Relying as they do on PCA, the researchers in the first four studies can be said to give a dimensional analysis of odors. Odors span multiple dimensions; they do not apply categorically. In contrast, the researchers of study five give a categorical analysis of odor dimensions. Odors apply categorically and do not span multiple dimensions. That is, a particular odor belongs in one dimension at the exclusion of other dimensions.       Here we have two dimension reduction techniques: PCA and NNMF. Using PCA, researchers find data that seemingly shows a strong correlation between olfactory pleasantness, olfactory perception and even odorant structure (though remember the internal problem of PCA); that is, researchers found that unique dimensions, like pleasantness, apply to odors. Using NNMF, researchers found that odors apply categorically, in discrete exclusive clusters. Olfactory pleasantness is not one of those categories. Using NNMF, we find little support for the idea that olfactory pleasantness is ‘written into’ the molecular structure of odorants themselves.        There are two dimension reduction techniques analyzed and different conclusions reached depending on which technique is used. Comparing PCA with another dimension reduction technique leaves us at an impasse. Different conclusions will be found depending on which technique is used to analyze the data. Given that both are dimension reduction techniques working on roughly the same data, there is no principled way to claim that one approach should be more trustedl than the other. Both are valid techniques and so both results are equally valid.  45  We are left with the following: on the dimensional approach pleasantness is written into odorant structure (leaving aside the internal problem) and on the categorical approach pleasantness does not appear as an odor category. There is no principled way to choose one dimension reduction technique over the other. Therefore, objectivism is not supported by studies 1-4.       The crucial issue is this: the resultant analysis might be determined by the structure of the method. That is, a categorical analysis results in study 5 because NNMF does not work with negative values. A dimensional analysis results in studies 1-4 because PCA works with positive and negative values. In other words, your results will depend on which dimension reduction technique is used. If using only positive values, you get categorical values. If using both positive and negative values, you get values in a dimension.       I don’t know if this is true, hence my qualification. I could not find anything in the literature. However, I contend that there is a methodological discrepancy and this is a problem for pleasantness objectivism.        As mentioned in the previous paragraph, study 5 uses the same data as the studies 1-4. The results are arbitrary; they rely on the structure of the type of dimension reduction technique used. If you use PCA, you get pleasantness as the PC1 of olfactory perception space. If you use NNMF, you don’t. Given this, the idea that we have pleasure tracking olfactory mechanisms has not been established, for the very data used to establish said idea is relative to the type of statistical method used. And if this is true, then P1 is destabilized.    46       This result may not imperil pleasantness objectivism but the empirical results are not well supported. I have not disproven P1. However, the premise is seriously under question. This is not to cast aspersions on the studies themselves. It is, however, to cast aspersions on the idea that we can use the results of studies 1-4 to establish P1. There is a further worry that does, in my view, undermine pleasantness objectivism. The worry follows from results in affective neuroscience. There is tension between olfactory psychophysics and affective neuroscience. I turn to that worry now.       2.3 Affective Neuroscience                        Pleasantness is a core component of reward systems. According to Berridge and colleagues (2003, 2004), the three components of our reward system are: motivation, learning and affect. These three components each have conscious and unconscious aspects to them. Affect itself can be broken down into three types: homeostatic (e.g. hunger, thirst), emotional (e.g. anger, joy) and sensory (pleasantness). Sensory affect is tied to other components of the reward system. Affective neuroscientists study the brain mechanisms involved in these systems. Two relevant points permeate this work: affect is biologically significant and brain based.       There are several pleasantness hotspots in the brain (Berridge, 2003, 2004). All three reside in subcortical regions. The primary hotspots are: the nucleus accumbens shell, the ventral pallidum and the brainstem parabrachial nucleus. Positive affective reactions—pleasantness—occur via an increase in opioid neurotransmitter receptors. This increase is found to occur most  47  in the nucleus accumbens shell. Positive affect also occurs in the parabrachial nucleus of the brainstem. This region contains a circuit for benzodiazepine, a substance that increases pleasantness. Finally, the ventral pallidum: all positive affect is lost when neurons to this region are lost. The ventral pallidum is the primary target of neurons from the nucleus accumbens shell.  These three areas interconnect with each other and with higher cortical pathways. In that way, signals are sent for cognitive representation in the orbitofrontal cortex (taste signals go through the thalamus before reaching the OFC but smell signals do not go through the thalamus at all; see Rolls et. al, 2003).       These are the brain mechanisms of sensory affect. But an important question still remains: what is the function of sensory affect? To answer that question, I focus on two important theories of sensory affect. Though the views ultimately disagree on the metaphysics of sensory affect, both agree on the function.       Edmund Rolls (2013) stakes a behaviorist line. To him, positive affective states are those elicited by rewards and punishments. Animals work for rewards and seek to avoid punishers. Cleary, pleasantness is a reward and displeasure a punishment. On this view, sensory affect is a kind of behavioral response reinforcement; the behavioral event causes the affective psychological reaction. If you smell the sweet vanilla, you will work to experience more of it. Consequently, your behavior will reinforce subsequent seeking behavior by causing pleasantness. Likewise, if you smell ammonia at a moderate to high intensity, you will seek to avoid experiencing more of it. Consequently, your behavior will reinforce subsequent avoidance  48  behavior by causing displeasure. On this view, sensory affect is a brain based response to reinforced behavior.        Jaak Pankseep (1998) stakes a subjectivist line. To him, humans and animals share similar affective profiles. Affect is found in deep subcortical regions, brain regions that are capable of eliciting subjective feeling states with a valance profile. So the sweet smell of vanilla is pleasant if it feels good to the subject. The chemical smell of ammonia is unpleasant if it feels bad to the subject. Feeling good or bad is determined by which brain regions are modulating the incoming sensory information.       Notice that both views are brain based. Even Rolls’ view provides evidence for affective subjectivism. If pleasantness is found in brain states than, by definition, it is not found in, say, molecular structure. Pleasantness is not an objective property for Rolls and so must be subjective or relational (a complex subjective property, remember). These are not fringe views. Rolls and Pankseep are pioneers of affective neuroscience and very much represent the mainstream. The outlook is very opposed to the idea that pleasantness can be found in, say, the molecular structure of odorants or any other perceiver-independent property. Of course, neuroscientists and psychophysicists do not share fundamental starting points; the former start with brains, the latter with the transformation of perceiver independent properties (like light or molecular odorant structure) into sensory perceptual processes. That is, psychophysicists start with the world, neuroscientists with the organism.      Consider the following quote from two prominent affective neuroscientists: ‘From this evidence, it is clear that actual reward pleasure lies in active processes of the brain and mind as  49  a reaction to a stimulus, rather than any physical stimulus itself. Pleasure is thus never merely a sensation, but rather something that the brain adds to sensations and experiences’ (Kringelbach and Berridge, 2015, p. 133). The subjectivism is evident. However, the learning component of sensory affect is also implied.       On this view, sensory affect (pleasantness) is a gloss on sensations. That is, pleasure is not strictly speaking sensed but is a psychological process that modifies incoming sensory information (see Aydede; 2014, 2018).We are back at the two subjectivist views mentioned last chapter. Our affective systems learn when and how to apply the right pleasantness gloss to the right kind of sensory information. In fact, there are important biological reasons why sensory affect would need the kinds of flexibility provided through learning. I discuss these issues more below. The important point is this: affective neuroscience provides a clear subjective and learned view of sensory affect, the very view that the olfactory psychophysicists canvassed in chapter 1 purported to disprove.        As of yet, there is still no principled reason to prefer the subjectivist view of sensory affect over the objectivist view. However, we have seen that the objectivist view motivated by olfactory psychophysics is already weakened by methodological problems. And we have also seen that affective neuroscientists, researchers who—unlike those in psychophysics—study the nature of affect in depth, provide a subjectivist counterpoint to objectivism that must be taken seriously. There is, however, one question that can seriously undermine objectivism: what is the function of sensory affect? Affective neuroscientists have much to say about this question. Psychophysicists do not study affect directly. However, one can extrapolate what an objectivist  50  would say on this question. Understanding affective function will, I argue, open the door to the idea that pleasantness is a subjective property. The olfaction studies discussed earlier will then be seen in a new light.  2.4 Biological Significance        In a sense, affective states function to keep us alive. Pleasant smells (and tastes) typically signal nutrient rich edible items that help us preserve bodily integrity. Unpleasant smells (and tastes) typically signal toxic items that can harm our bodily integrity. Ingesting vanilla typically preserves or enhances bodily integrity; ammonia has the opposite effect. The important question is: how does affect do this? Here, again, are Rolls and Panskeep, respectively:   …we are built to have [affective states] because we (and many other animals) use rewards and punishers to guide or determine our behavior, and that this is a good design for a system that is built by genes where some of the genes are increasing their survival by specifying the goals for behavior…this is a very adaptive way for evolution to design complex animals without having to specify the details of the behavioral responses, the actions, as it is much more flexible in an uncertain environment for responses and actions to be learned (Rolls, 2013, p.45).   …affective capacities were retained as various species emerged through natural selection because these brain functions provide efficient ways to live and reproduce. These brain functions provide selective advantages in that they effectively anticipate universal, future survival needs…Affects, from this perspective, are inbuilt anticipatory  51  neuropsychological mechanisms of the brain. Just imagine how useful pain is for your survival. Affects provide a flexible guide for living. Prior to the evolution of [affect], animals must have behaved in more stereotypical ways…The full-blown affective capacities of mammals, however, allow animals to respond to the here-and-now challenges of life in highly, flexible ways (Pankseep and Biven, 2012, p. 42-43).         Both Rolls and Pankseep express an important point. Affect is built to guide and promote flexible, lifesaving, behavior. Affective capacities do this through learning. Suppose our olfactory pleasantness mechanisms were purely innate. An unmalleable connection would hold between olfactory percepts and olfactory pleasantness. However, notice how inflexible, almost by definition, our behavior would be. We would, say, always experience heavier odorant molecules as pleasurable and would be motivated to gain more of the stimulus. However, imagine an insidious olfactory scientist who introduces an artificial fragrance SX that smells pleasurable, is composed of fatal toxins but also has a heavy odorant structure. If introduced ubiquitously, we would not be long for this world.       This is the peril of a purely innate link between our percepts and affective mechanisms. However, this is just the puzzle our affective mechanisms solve. Through flexibility and learning, our affective mechanisms might quickly adjust. We might be motivated to avoid SX. We might learn to associate unpleasant sensations with SX even though it fits the profile of a pleasant sensation on the innateness view. My discussion is meant to draw out a crucial point: if it is anything, pleasantness (and affect more generally) is biologically significant. Pleasantness  52  motivates.5 It furthers the biological interests of organisms. I think this is what Rolls and Panskeep and Berridge are getting at. Pleasantness is learned in order to promote the biological interests of the organisms experiencing said pleasantness.       I’ve spent much of this chapter discussing the idea that olfactory pleasantness could be objective and innate, promoting a kind of pleasantness objectivism. At the very least, the subjective and learned view was not supposed to be the entire story between olfactory percepts and olfactory pleasantness. However, I’ve just shown that the subjective and learned view explains the biological significance of olfactory pleasantness (and sensory affect more generally). Can the affective physicalist do the same? That is, can pleasantness objectivism explain the biological significance of olfactory pleasantness?      The affective physicalist argues that we have some innate capacity to perceive the pleasantness properties of odorants. We even have a candidate physical property for olfactory pleasantness: molecular weight. This is a property our olfactory system is attuned to, at least according to the affective physicalist. Again, they do not suppose the entire connection to be innate and objective. There are some learned (and possibly subjective) aspects to the connection. Either way, the affective physicalist must explain the biological significance of olfactory affect or explain it away. As will become clear, both present major problems for the view.       What would the biological significance of olfactory pleasantness look like on the affective physicalist view? To answer, we must look at what is supposed to be objective about olfactory                                                           5 I am assuming a Humean theory of motivation. This follows from the idea that pleasantness both motivates and is not an experience-independent property.   53  pleasantness. The answer is clear from sections 1-6. Heavier odorant molecules are, in some sense, the objective (olfactory) pleasantness properties. These are the properties that show a significant correlation with the PC1 of olfactory perceptual space—pleasantness. These are the properties that both exhibit pleasantness and perceiver independence. Given this, the objectivist view under discussion is a type of affective physicalism; an olfactory affective physicalism to be more precise.  However, there is a big question lurking in the discussion: what explains the fact that heavier odorant molecules are experienced as pleasant? How can the objectivist explain the presumed motivational characteristics of heavier odorant molecules?          The objectivist has three options. Explain how heavier odorant molecules are pleasurable and motivate, explain this fact away or leave it unexplained. The final option—leaving the fact unexplained—is theoretically undesirable. On this option, the physicalist would claim that it is simply a brute fact that sensory pleasure is found in heavy odorant molecules. Our motivation to get more pleasant smells and avoid unpleasant ones would be mysterious on this view. There would be no motivated reason why I should seek out the smell of freshly baked bread other than to simply state that I am ‘tuned to’ heavier odorant molecules in the bread. That is, this option leaves unexplained the biological significance of the pleasant smell of freshly baked bread. The subjective view has a clear answer on this question. Pleasantness is built to guide and promote flexible, life giving, and saving, behavior. Sensory affective capacities do this through learning by brain based organisms. The subjectivist can explain the biological significance of olfactory sensory affect but, on this option, the physicalist cannot. Another option is more preferable.  54       The physicalist could explain the facts away. On this option, the objectivist would need to eliminate the role of biological significance itself. We experience heavier odorant molecules as pleasurable but there is no biological significance that accompanies this fact. This option allows the objectivist to block any argument that relies on the idea that subjectivism best explains the biological significance of sensory affect. However, this option is, like the last, undesirable. It amounts to claiming that pleasantness does not motivate and promote life enhancing behaviors. The question naturally follows: what would the function of sensory pleasure be?6 Perhaps pleasantness allows us better olfactory object recognition or enhances olfactory object intensity. Firstly, it is simply fact that we seek out pleasant smells and avoid unpleasant ones. Secondly, it is not clear that there are olfactory objects as such (see Barwich, 2014, 2016, and Keller, 2016). Thirdly, even presumed objectivists assume the biological significance of sensory affect. Zarzo (2008)—he of study 2—argues that edibility is possibly one of the key dimensions of olfaction. Edibility is a paradigm case of biological significance. Objectivists should reject the option of explaining away.       There is one option left. Explain how objective (olfactory) pleasantness properties can be biologically significant. In my view, the objectivist must seek out this task. However, this option comes with its own perils. In this case, the physicalist must explain what makes heavier odorant molecules the ‘vehicles’ of sensory pleasure. I argue that the physicalist can only offer a circular explanation. Pleasantness and biological significance is the relationship to be explained. The                                                           6 Lisa Shapiro argues that certain early modern figures saw sensory affect as a way of gaining knowledge about the world (see her 2014). Of course, this is compatible with the idea that pleasantness motivates. The claim we must reject is the exclusivity claim: that pleasantness only functions to give us knowledge of the world (through e.g. recognition).   55  objectivist explains the biological significance of pleasantness by appeal to the pleasantness found in heavier odorant molecules. We want to experience more of the fruity smell because it is pleasurable. However, how best to explain pleasantness? The objectivist can only explain the pleasantness found in heavier odorant molecules by the fact that said molecules are biologically significant. We know that heavier odorant molecules are pleasurable because subjects want to experience more of them. This is circular. Pleasantness is explaining the biological significance. Biological significance is explaining the pleasantness. Essentially, the physicalist can only give a circular explanation of the biological significance of (olfactory) sensory affect.        This suggests a defect in the theory. Motivation is a key characteristic of all affect. The affective physicalist can only give a circular explanation of this fact. Therefore, they have not explained how heavier odorant molecules can be the vehicles of (olfactory) sensory pleasure. Essentially, the affective physicalist cannot explain how we are motivated to seek more pleasant odors. Conjoin this with the methodological troubles mentioned last section and P1 is not yet established. Our olfactory mechanisms do many things but one thing they almost certainly don’t do: track pleasantness. Given this, olfactory pleasantness is not, contra P1, a sensory feature of olfactory experience.       A sensible property must be what is tracked by a sensory system. If said sensory system does not track the candidate property, then said property is not a sensible property. Perhaps there is an interesting connection between molecular weight and olfactory pleasantness. But if there is no pleasantness property tracked by our olfactory system, then clearly molecular weight cannot  56  be the candidate sensible property. Molecular weight will always be mind-independent; it will never be the vehicle of olfactory pleasantness.       The situation is worse when you compare affective physicalism with pleasantness subjectivism, even a generic pleasantness subjectivism. On the latter view, pleasantness properties are properties of the subject. This could be a property of experience, a brain state, a relational property (subjective as I’ve defined it in Chapter 1) between one of these subjective states and molecular weight and so on (the metaphysics of pleasantness properties is the theme of chapters 3-4 ). It is not very difficult to explain the motivational nature of pleasantness tracking biological significance on these views. You want more of the freshly baked bread because your experience of the smell is pleasurable. The smell is pleasurable because your subjective experience or a particular brain state or some relation between these and the molecular weight cause you to feel pleasure.       Notice there is no circularity afoot. Pleasantness is still explaining the motivation; you want more of the smell because it feels good. But notice that, unlike the physicalist view, motivation is not explaining what is pleasurable. If pleasantness is found in a subjective state or process, then there is a non-motivational property that explains what is pleasant: a subjective experience, a brain state, a relation, etc. One of these is the vehicle of pleasure. And unlike molecular weight, these properties are not themselves intrinsically motivational. It is not the motivational characteristics that explain why they are pleasurable.       Consider the theory of sensory affect proposed by Aydede (2018). On this view, sensory affect is located in the sub-personal modification of sensations (though Aydede also gives a  57  story for how affect is also attributed to external objects). The details of said modifications are not crucial for my point (see Aydede, 2014, 2018). Aydede locates sensory affect in properties of the subject and not in any mind-independent objects. His view is subjectivist in the sense I’ve been using. In the olfaction case, we find the freshly baked bread pleasant to smell because, according to this view, certain modifications are made to the freshly baked bread olfactory sensation. We are motivated to seek out this smell because it is pleasurable. And, on this view, the modifications to the sensation are what make said sensation pleasurable. The motivational component of olfactory pleasantness is explained in a non-circular way. This is just another way of saying that subjective views, like the one postulated by Aydede, can, unlike affective physicalism, explain motivation.     A metamer-like objection            It gets worse for the affective physicalist. The charge of circularity is ultimately an empirical question. Theoretically, the physicalist could find a property that explains how heavier odorant molecules are pleasant and not just that they are. As long as this remains unexplained, however, the physicalist is in trouble. But there is a more general theoretical worry that threatens affective physicalism. The affective physicalist has a similar problem to the color physicalist. The latter have to explain metamers. The former do not, of course, but there is a ‘metamer-like’ objection that threatens affective physicalism. I explain each in turn.7                                                            7 I thank Aleksey Balotskey and Kousaku Yui for pushing me on this point.   58       Color physicalists (see Hilbert, 1987; Byrne and Hilbert (ed.), 1997) identify colors with light-reflecting properties of objects; clearly, a physical property type. However, the existence of metamers throws doubt on the view. Metameric pairs are surfaces that look the same in color despite having different, sometimes wildly different, reflectance profiles (see Fish 2010, pgs. 140-45 for a good introduction to metamers and color realism). On the face of it, this should not happen if color physicalism is true, for if color is identified with light-reflectance properties of objects this should guarantee that a difference in reflectance profile will ensure a perceptual difference. But metamers prove that this is not the case. The color physicalist has potentially many responses (Hilbert, 1987). We do not need to dwell on them. The crucial issue for us is: does a similar problem arise for the affective physicalist?      Back to smell. The affective physicalist claims that (olfactory) pleasantness is identified with heavier odorant molecules. This immediately opens the door to metamer-like worries. If the affective physicalist is right, then the same molecular weight profile will ensure sameness of pleasantness. But I have already discussed the literature on the role of variation, learning and context in the relationship between olfaction and pleasantness. It is very likely that the same molecular weight profile will not ensure sameness of pleasantness. The particular pleasantness response can be easily modulated through learning, context and experience. If true, and the evidence shows it likely is, this fact undermines the idea that sameness of molecular weight ensures sameness of pleasant smells. In other words, olfactory pleasantness cannot be identified with heavier odorant molecules. Affective physicalism is false.                   59       To sum up: the affective physicalist has three options. They can leave the biological significance of pleasantness unexplained, they can explain it away, or give an explanation. Leaving biological significance unexplained is theoretically unsatisfying, explaining away unsuccessful and explaining circular. In section 2, I mention that methodological troubles leave affective physicalism shaken but still standing. Furthermore, there is a metamer-like objection that further threatens the view. However, the analysis of this section makes clear why I think affective physicalism is an unsatisfactory theory. The subjectivist view of pleasantness naturally explains the biological significance of pleasantness. I conclude that said view provides the best explanation of olfactory pleasantness. I’ll say more about subjective views of pleasantness in chapter 4.      As for studies 1-4, the most we can conclude from their results is this: there is a significant and interesting statistical correlation between olfactory perception and olfactory pleasantness; nothing more. We cannot conclude that there is an innate link between the two that the subjectivist has somehow missed, nor can we conclude that olfactory pleasantness is ‘written into’ the molecular structure of odorants. Affective objectivism/physicalism cannot get off the ground, at least not with the results of the olfactory psychophysics studies discussed at the beginning of this chapter. The affective objectivist/physicalist cannot use said results to motivate their theory of pleasantness. Olfactory pleasantness might be correlated with heavier molecular weight. But these cannot be identified with each other.   2.5 Function  60             There is another motivation for affective objectivism, one that does not lie in the olfactory psychophysics literature. Affective objectivism can also seem a natural consequence of the function of perception. Typically, philosophers of perception assume the function of perception is to provide accurate information about the physical world. This idea can motivate a particular metaphysical project; the traditional problem of perception for example (see Smith, 2002; Crane, 2006). It can also motivate epistemically driven projects; asking, for example, how empirical beliefs are possible (McDowell, 1994) or where our very idea of a mind-independent world comes from (Campbell, 2002).       However, there is a different approach to the question of perception’s function. On this view, perception functions primarily to guide behavior; to guard the body and not to provide beliefs about the world. Various philosophers have taken this stance (see Akins, 1996; Matthen, 2005; Simmons, 2015) but I focus on Andreas Keller’s recent work (2016, chapter 4) on olfaction. He argues the following show that the function of perception is to guide behavior: perceptual variability between species (humans and honeybees see the same flower differently), perceptual variability within the same species (pregnant women perceive food items differently than when not pregnant) and the mismatch between same stimuli, same percept (many physically different chemicals are perceived as bitter; same percept, differing stimuli).              The chemical senses naturally fit this model. It is natural to think that smelling and tasting function to keep us away from toxins and to seek out nutritionally dense food items. It is  61  natural to think that these senses do not function primarily to provide us with correct information about the physical world. Spatial resolution might provide the key to understanding why. Castro and Seeley (2014, p.299) tell us that ‘Whereas the receptor epithelia for vision, audition and somatosensation are topographically organized to encode information about spatial proximity and/or basic physical variables pertaining to stimuli, the olfactory epithelium appears to relative pleasantness topographically [they cite study 4 above].’ Olfaction (and gustation) is not in the business of locating objects in space. Hence, there is little need for our olfactory receptors to be sensitive to spatial properties. Rather, olfaction (seemingly) encodes for pleasantness. This would explain the low spatial resolution of olfaction.     Consider another view. Manolo Martinez (2015) argues that olfactory pleasantness carries imperative force. Imperative views (see Klein, 2015) stress the commanding nature of certain content. Certain contents (pain, hunger, thirst, itches, etc.) command you to act or not act in certain ways. For example, pain might carry the following content: stop that and pull your hand away from the fire! Instead of describing the world, or indicating its various properties, imperative contents command. Martinez claims that olfaction is similar. He explicitly attacks the idea that olfaction has duel components: one indicative and one affective (see Clark, 2005). Rather, Martinez claims that ‘What one takes in through the nostrils is (not information about how the world is, but) a command to avoid a certain source of pathogens’ (2015, p.198). Like others before him, Martinez assumes that olfaction is largely pleasantness; that when you’ve given a story about pleasant smells, you’ve largely given the story of smell.        62       What does this have to do with affective physicalism? If the above claim is right and olfactory perception functions to guide behavior, then it is natural to think that coding for pleasantness is the best way to do that. That is, pleasantness is a natural behavior guidance property; it is fundamentally motivational. We want more of things we find pleasurable and want to avoid things we find unpleasurable. Consider the idea that olfaction is partly a pleasure tracking sense. One benefit of this idea is that it would naturally explain the function of olfaction. Olfaction functions to guide behavior towards the heavier odorant molecules in the world that typically correlate with nutrient rich edible items and away from the lighter odorant molecules in the world that typically correlate with toxic items.       Affective physicalism has a further benefit. It can explain the secondary function of olfaction as well. The primary function is to guide behavior. The secondary function is to provide correct information about the physical structure of the world. On the affective physicalist view, olfactory pleasantness is a physical structure of the world. So on this view, we can both explain how pleasantness guides our behavior and explain how we perceive the pleasantness properties of the physical world. The function of olfactory perception seemingly motivates affective physicalism. However, as we’ll see, pleasantness subjectivism has its own take on perceptual function and sensory affect.       There are different ways to undermine the premise that olfactory percepts are partly pleasure tracking. One is the following: show that we do not have a unique olfactory pleasure tracking mechanism. Olfactory affective physicalism is grounded in the idea that patches of the olfactory epithelium are default set to encode heavier odorant molecules as pleasant and  63  lighter odorant molecules as unpleasant. That is, on this view, we have a dedicated pleasantness tracking mechanism.       Matthew Fulkerson (2016; chapter 7 of 2014) is one of the few philosophers of perception to tackle this issue. Fulkerson focuses on pleasant touch. However, his conclusions serve as a moral for olfaction. My task is to reconstruct Fulkerson’s argument and then apply his conclusion to the case of olfaction. Doing so will undermine the idea that we have a dedicated pleasure tracking mechanism. Affective physicalism will be undermined.       A group of researchers conclude that CT fibers are a kind of pleasant touch encoding mechanism. The CT system activation patterns correlate well with pleasantness ratings. The details are less important for my purposes than the conclusion (see Fulkerson, 2016 for the details). Needless to say, the CT fibers are thought to function like the patches of olfactory epithelium. Both carry pleasantness information about sensible properties. Fulkerson challenges this idea: “There isn’t any single channel that simply detects and elicits experiences of tactile pleasures” (2016, p. 332). One reason to think this: pleasant touch can occur without the CT system at all (in the hands, lips, genitals). Another reason to think this: pleasant touch can only occur when accompanied by sensory discrimination. As Fulkerson points out, this dual system functioning ‘…depends on the present and prior state of the overall tactual system, as well as many other systems involved in the evaluation of tactual inputs’ (2016, p. 332).       So not only does context matter (the present and prior state) but so do other systems (such as mood or arousal). Summing up, Fulkerson states: ‘There is reason to think that pleasant and unpleasant touch experiences are not simples, activated automatically by a single type of  64  sensory fiber, but complexes built out of several interacting components’ (2016, p.333). Affective physicalism relies on the idea that pleasantness is activated by a single type of sensory channel that encodes the requisite type of physical properties. Affective subjectivism gets a grip once we acknowledge that pleasantness is a complex experience built out of several interacting components. The reason: once you introduce experience-dependent properties, you have already arrived at subjectivism. Conjoin this with the reasonable claim that if you introduce several interacting components into the ontology of pleasantness, then at least some of these will be experience-dependent components and you can see how affective physicalism is undermined.              Olfaction is similar. The physiochemical structure of odorants—molecular weight—is not the only determinate of olfactory pleasantness. I have argued that it is no determinate at all. But even if you are an affective physicalist, you must acknowledge that olfactory pleasantness is also explained by non-physiochemical properties. The following properties explain aspects of olfactory pleasantness: odor intensity (Doty, 1975), odor familiarity (Delplanque et al., 2008), individual and cultural context (Ayabe-Kanamura et al., 1998) and interpersonal variation (Hudson, 1999).       We know this. Think of the difference between a subtle amount of perfume and dousing oneself in cologne (intensity), smelling skunk odor as a Canadian versus as a European (the latter have been found to find the smell inoffensive or even pleasant; familiarity), liking the smell of popcorn versus finding it offensive (if, say, you worked for years at a movie theatre; individual context) and the smell of coffee when you were a child and now as an adult  65  (interpersonal variation). In each of these examples, you can take two subjects, give them the same physiochemical odorant stimulus and they would differ in their pleasantness rating of said stimulus. Take the skunk example. You and a European friend might encounter an odor with the exact same molecular weight yet, because skunks are not native to Europe, your friend finds the smell inoffensive and you find it repulsive (see Herz, 2008). I even find the smell of skunk inoffensive at low intensity levels but offensive at higher levels; an example of intensity and individual context (skunks are cute and furry) interacting to determine my pleasantness reaction.       Of course, the affective physicalist would not deny these facts. She will say that learning and context play an important role in determining our olfactory pleasantness reactions. All I am saying, proclaims the objectivist, is that there is a default connection between our olfactory mechanisms, olfactory pleasantness and molecular weight. These connections are regularly modified by context and the systems mentioned above but that does not negate the initial objective connection.       The important comparison with touch lies at the beginning, in considering how the transducer works. The relevant comparison here is between CT fibers and the olfactory epithelium. Fulkerson points out two important facts: pleasant touch can be had without the activation of CT fibers and pleasant touch can only occur with activation of tactile discrimination. This is where the affective physicalist can, and the only place they really can, push against the analogy of smell with touch. The affective physicalist will ask: can pleasant smell be had without the activation of the olfactory epithelium and can pleasant smell only  66  occur with the activation of indicative olfactory sensory discrimination? Notice that both need to be true if physicalism is to be bolstered. If we can have pleasant smells without the activation of the epithelium, then there is no single channel that encodes for olfactory pleasantness. If olfactory sensory discrimination occurs with olfactory pleasantness encoding, then we do not have a unique sensory channel that encodes for olfactory pleasantness. And if anything, the latter fact would show that the sensory system is doing its job: it’s encoding indicative sensory information about odors. Either way, the affective physicalist is in trouble.      There is enough of a disanalogy with touch to help and hurt the objectivist. The disanalogy is this: all olfactory information is transduced through the epithelium. That helps the objectivist in this way: all pleasant smells are encoded by the epithelium. But this is just what hurts the objectivist. All pleasant smells are so encoded because the epithelium does all the olfactory encoding, which means it encodes discriminatory information as well. There is no unique sensory channel dedicated to encoding pleasantness.       This is not enough to undermine the physicalist picture totally. The following would need to hold: pleasant smells are encoded only when accompanied by indicative olfactory information. That is, we smell the indicative information ‘fruity’ and experience an accompanying pleasant feel. The epithelium encodes both simultaneously. This would be like the pleasant touch case. However, even if we can have pleasant smells encoded without accompanying indicative information, the physicalist cannot say that we have a unique sensory channel that encodes only pleasantness information.   67       The analogy with pleasant touch leaves olfactory affective physicalism in a perilous place. Like pleasant touch, pleasant smells are influenced, usually strongly, by context, mood, learning and other non-objective states. Though the nature of the tactile and olfactory receptors is not analogous, enough similarity exists between them to undermine the olfactory affective physicalist idea that we have a unique sensory channel that encodes only pleasantness information. And the latter claim is important in claiming that our olfactory system is partly pleasure tracking. If we have a sensory system that tracks pleasantness in the way the physicalist thinks, we would have a unique channel that would only encode pleasantness information. This would ensure that no subjective properties are needed in the making of pleasantness. However, we have no such unique channel. Therefore, it is highly unlikely that our olfactory system is pleasure tracking, even in part. We have no support for the idea that olfactory pleasantness is a sensory feature of olfactory experience.        I conclude that pleasantness cannot be a physical property. This is true in the olfaction case and almost certainly true when applied to any sensory modality. I used studies in olfactory psychophysics to motivate affective physicalism. The remaining two chapters tackle the question: if not affective physicalism, then what; what type of property is pleasantness if it is not a physical property?                                                                                                                68                                               Chapter 3: Primitive Pleasantness            I discuss subjectivist views in chapter 4. However, there is an important issue left standing: the pleasantness objectivist is not out of options just yet. Affective physicalism does not exhaust her options. There is a view according to which O-P properties are not physical properties but are still objective. On this view, physical properties do not exhaust the types of properties that can be objective. Remember that objective properties are those that are not constituted in any way by the subject or subjective experience. Physical properties clearly fit this bill. But if O-P is a non-physical, objective property, what kind of property is it? However, I must say something about attribution before I answer this question.     3.1 Attribution        Mohan Matthen (2017) claims that there are two primary questions in the color debate: 1) what is the constituent structure of color and 2) how do we attribute colors? In the color debate, the two questions come apart. All parties to the debate agree on the answer to the attribution question: colors are attributed to the surfaces of objects. Consider the following three quotes. The first is given by a physicalist, the second by irrealists and the last by a primitivist.    69  I will assume here that colors are where the contents of our visual experiences and our ordinary ways of talking place them—on the surfaces of physical objects, or in expanses of sky or water. Grass is green, the sky is blue, ripe tomatoes are red (Shoemaker, 1997, p.230).   The projection posited by this account has the result that the intentional content of visual experience represents external objects as possessing colour qualities that belong, in fact, only to regions of the visual field (Boghossian and Velleman, 1997, p. 95, emphasis added).     When we see an object as red we see it as having a simple, monadic, local property of the object’s surface. The color is perceived as intrinsic to the object, in much the way that shape and size are perceived as intrinsic (McGinn, 1996, p. 541-542).           The color ontology debate is focused on the first question: what is the constituent structure of color? Take two very different views, physicalism and the sense-data theory. Proponents of both views think that we attribute colors to the surfaces of objects. However, on the former view colors are light-reflecting properties of objects and on the latter view colors are some kind of private, mental particular. Both views need to explain our color attribution practices given their respective ontological commitments. The physicalist must explain how we attribute colors to the surfaces of objects when colors are light-reflecting properties of objects (a seemingly  70  easy task), while the sense-data theorist must explain how we attribute colors to object surfaces’ when colors are properties of mental particulars (a seemingly difficult task). And as I’ll soon show, the same holds for color relationalism and primitivism.       Importantly, the pleasantness debate is not like this. There is a much smaller gap between our attributional practices and the nature of the O-P property. The reason is simple but important: pleasantness is, contra color, attributed twice. We attribute pleasantness both to objects and to their experiences. The smell and flavor cases are most helpful. The lavender experience is pleasant but so is the flower itself, the experience of chocolate ice cream is pleasant but so is the ice cream itself. As in the color debate, all parties will agree on this dual attributional practice. The gap between attribution and constituent structure has closed because pleasantness is attributed to both objective and subjective properties. There is a sense in which all will agree that the nature of pleasantness is both objective and subjective. Aydede and Fulkerson (2014) point this out:   …what do these [pleasantness] qualities qualify in the first instance? It seems obvious that, depending on context, we attribute these qualities both to our perceptual experiences and to their objects. Although it seems to apply in all sensory modalities, the ambiguity is especially robust in the case of smells, tastes and bodily sensations. Indeed, when we talk about how pleasant the taste of strawberry is, we may be attributing the pleasantness to our subjective experience, or we may be attributing it to  71  whatever objective sensible qualities of the strawberry are responsible for our experience (p. 177).                    Notice that this fact immediately presents a puzzle. Objective views must explain the subjective aspect of pleasantness and subjective views the objective aspect. Indeed, this fact undergirds our guiding question: what is the metaphysical nature of the O-P property? This question will be answered by whichever theory best explains the nature of the relationship between O-P and E-P.       There are different approaches to this question. You could say that a good theory of pleasantness will reconcile O-P and E-P. More precisely, the two types of pleasantness attributions are non-accidently related. That can happen if various kinds of explanatory relationships hold between the two. I return to this below.          You could reject the above approach. There are three possible ways to reject it. Firstly, the attributions are accidently related. On this view, O-P and E-P can be any property type you want but they bear no essential connection to each other. Secondly, you could say that one of the attribution types is confused and that O-P is the culprit. On this approach, there is no need to reconcile the two attribution types or even claim that they are accidentally related. The reason is simple: there is no O-P! We think there is but that is a mistake (cue literal projectivism). There is only E-P. Thirdly, you could say that one of the attribution types is confused and that E-P is the culprit. There is only O-P (see below). Notice that these last two options imply that pleasantness is more like color than I initially claimed. We only attribute pleasantness  72  successfully to one property type, like in the color case, and are confused in thinking that we attribute pleasantness to more than one property type.      These are the options. I think we should jettison all three. Start with the first option. The idea that the two attribution types are accidentally related strikes me as a count against a theory of pleasantness. Think of the biological significance of pleasantness (see Berridge, 2003, 2004). The O-P properties will be those that function to nourish and protect our body. One of the best ways they do this is through a connection with E-P. The idea that the connection is accidental is strange; it opens the door for an unpleasant O-P to cause a pleasant E-P. E-P functions to register information about the O-P properties in the world. The world would suddenly become more perilous if the link between the two types of pleasantness were severed. We should reject any theory of pleasantness that leads to the conclusion that the two types of pleasantness are accidentally related.       We should also reject the idea that one of the pleasantness attribution types is confused. These two views violate our attributional practices and introduce an unneeded complexity into the discussion. On these approaches, one of the attribution types is given a positive analysis but the other is given a purely negative thesis. To argue that one of the attribution types is confused requires giving up too much. This is especially true if we give up E-P. To say we don’t actually attribute pleasantness to our lavender smell experiences is simply not true. Experiences feel good or bad. I take this is an empirical truism. It is a bit less problematic to claim that O-P is confused. Perhaps we are always wrong to claim that lavenders themselves are pleasant. However, this is radical irrealism. On this view, there are no O-pleasantness  73  properties. But this leads to an odd view. There is no O-P but E-P is, say, relational. I can’t see a radical irrealist holding this half-way view. The same can be said if there are no E-P properties but we give a positive account of O-P. Later in this chapter, I discuss a more moderate form of irrealism. Needless to say, we should reject the idea that one of the pleasantness types is confused.       I think a theory of pleasantness should reconcile these two pleasantness types into a unified theory. My question for the remainder of this work is: which theory does this best? The best approach will conclude that E-P and O-P are non-accidently related.       Importantly for the two valid approaches, you can privilege one attribution type over the other. For example, you might think that attributing pleasantness to objects is dependent on attributing pleasantness to experience. On this approach, which Aydede and Fulkerson (2014) dub the experience view (EV), affective qualities primarily qualify perceptual experiences and not their objects. Attributing pleasantness to the lavender experience is to attribute it to the olfactory experience of the flower (E-pleasantness). On the other approach, which Aydede and Fulkerson dub the object view (OV), affective qualities primarily qualify the objects of perceptual experience and not the experiences themselves. Attributing pleasantness to the lavender, on this approach, is to attribute it to the flower itself and not to your olfactory experience of the flower (O-pleasantness).      My focus on object attribution does not assume OV. The primacy of attribution does not negate the fact that we attribute pleasantness to objects at all. The important point is, given that we do attribute pleasantness to objects, what theory best explains our attributional  74  practices and what does that tell us about the property of O-P. And importantly, there is a difference here. Mohan Matthen (2017) points out that there are two different questions in the color debate: what are the color properties and how do we go about attributing said properties? This distinction holds for pleasantness: what are the pleasantness properties and how do we go about attributing them? I focus on the former question but also go on to give some answers to the latter question as well. The motivating question is this: which theory can best reconcile O-P and E-P?              3.2 Affective Primitivism        Affective primitivism claims that pleasantness is a primitive property, no surprise there. But what exactly does this mean? Primitive properties are properties of objects that are not physical or relational but sui generis.  In other words, the primitivist claims the world has a richer store of properties than is typically granted by physicalists and relationalists. A brief foray into color primitivism will be helpful here. Color primitivists claim that colors are, what else, primitive properties (prominent primitivists include Campbell, 1993; McGinn, 1996; Watkins, 2005, 2008 and, more recently, Cutter, 2018; prominent critics include Bryne and Hilbert, 2007; Chalmers, 2006; and Pautz, ms.).     75       To say that color is primitive is to say that color is chromatic through and through; colors are not physical properties (subject-independent properties observed and measured by the sciences), they are not relational properties but they are objective properties. Consider the following statements in support of the view:   …[colors] are to be thought of as the grounds of dispositions of objects to produce experiences of colour. This is not a kind of physicalism about colours. To suppose that it must be is to assume an identification of the physical and the objective which the thesis may question. It may instead be that the characters of the colours are simply transparent (Campbell, 1997, p.178).  …[Primitivism] holds that the nature of color is revealed in how colored objects look. [Primitivism] takes colors to be just as they appear to be…To have the property of being red is precisely to have the very property that fixes the look of objects when they appear red. Color properties are what populate our visual impressions, while being (as we might say) expressions of underlying sensory dispositions, of which impressions are impossible (McGinn, 1996, p. 547).  Visual experience represents objects as colored and it represents colors as monadic non-dispositional properties of objects. Moreover, visual experience is veridical, much of the time…But colors are not identical to the properties identified by science as  76  causally responsible for color experience. Therefore, even though colors surely supervene on the properties of interest to color science, and they might even be realized by these properties, they are identical neither to those properties, nor to some property constructed out of those properties. Or so claims Primitivism (Watkins, 2010, p.123).        Color primitivists claim that, in having a color experience, the color itself is transparent to us. We know which property color is when we undergo a color experience.8 Primitive colors may supervene on physical properties or relational properties (I’ll discuss this more when I get to pleasantness) but they are not to be identified with said properties. This is the core claim of color primitivism.       Notice that primitivism is a kind of objectivism. This is an important point, especially for John Campbell who claims: ‘This view of colours would be available even to someone who rejected the atomic theory of matter; someone who held that matter is continuous and that there are no microphysical properties. The view of colors as mind-independent does not depend upon the atomic theory’ (1997, p.178).       Primitivism is an objectivist alternative to physicalism. Both views claim that colors are independent of subjects. The difference lies in which type of objective properties constitute the colors. The primitivist view can be seen as a non-reductive alternative to physicalism and even                                                           8 Campbell (2005) claims that on his view colors are transparent but not fully revealed to us in color experience. The distinction is important for avoiding the commitment to revelation, an idea originated by Mark Johnston (1992), which commits one to claim that the essence of the color property is fully revealed in color experience. Campbell defines transparency as: ‘experience of color provides knowledge of the categorical color property intervention on which changes the experiences of observers’ (2005, p.111).       77  relationalism. Colors may supervene on physical properties but we don’t see light reflectance, we see brown and red and yellow, particulars identified as chromatic. Primitive properties are non-reductive properties that are transparent to the subject. Physical (pleasantness) properties would be reductive properties that are perceptually opaque to the subject.       The view is, however, a bit more complicated than that. There are two varieties of primitivism (see Pautz, ms.). On response-independent primitivism, colors are primitive properties that supervene on physical properties. This is the view of John Campbell. On response-dependent primitivism, colors are primitive properties that supervene on relational properties. This is the view of Colin McGinn. He claims ‘…color properties themselves are categorical, simple, monadic, intrinsic features of objects, despite the fact that the properties they supervene on are dispositional, complex, relational, extrinsic features of objects’ (1996, p. 545). These are the two views I will focus on and transpose to the pleasantness case. Now, response-independent primitivism is a genuine objectivist alternative to physicalism. However, response-dependent primitivism is the non-reductive version of relationalism. The former view has been more popular with realist primitivists but the latter view provides an interesting variant and one that will prove especially interesting in the pleasantness case.        Primitive sensible properties are those whose nature is transparent to us when undergoing the requisite sensory experience. Sticking with the color case, physical properties—like light-reflecting properties—and relational properties—like being disposed to appear red to S1 in C1—are not transparent in this way. In my view, this is one of the primary motivators for  78  primitivism. The view allows color phenomenology, or the phenomenology of any sensory quality, to play a role in the ontological explanation of color (or smell or taste or…).       There are two other motivations I wish to discuss before I move to pleasantness (see the discussion in Cohen, 2009). One motivation for primitivism is the idea that the view gives us epistemic access to the colors (or any other primitive property). This is especially true if you give pride of place to knowledge by acquaintance instead of knowledge by description. On the primitivist view, we are directly acquainted with colors when we undergo color experience. On Campbell’s’ view, we are acquainted with an objective property, while traditionally, theorists claimed we are acquainted with subjective sense-data.         Another prime motivation for primitivism is the idea that it is the view left over after we have worked through all the difficulties with physicalism and relationalism (see Watkins, 2010 and Pautz, ms.). If these standard two views fail, then perhaps we can find what we are looking for in another kind of property. In a sense, this motivation and the one from phenomenology loom large. I don’t think affective primitivism is especially motivated by the idea of epistemic access. Rather, the view is motivated by the idea that phenomenology plays a role in the ontology of O-P conjoined with the idea that affective physicalism and relationalism cannot be true. The question is: is there an affective primitivist view on offer? Fortunately for us, there is.  3.3 Johnston         79      Mark Johnston (2001), in his rich essay, argues for affective primitivism. Affective relationalism and irrealism are his two primary targets. In an earlier work (1998), Johnston claims that relationalism reduces to a kind of irrealism. Though this latter essay does not equate the two views explicitly, the relationship is implicit in the discussion. Johnston thinks relationalism (and so by extension irrealism) cannot explain the authority of affect. By affect, Johnston means something fundamental: sensed attraction and repulsion. These are fundamental in that our aesthetic and moral judgments follow from the fact that we sense things like attraction and repulsion; Johnston calls these, and their ilk, ‘inherently sensuous values’. It is no great stretch to understand that pleasantness (attraction) and displeasure (repulsion) are also captured in Johnston’s analysis.       The following two quotes capture the heart of Johnston’s positive account. The first captures his idea that primitivism is preferable to relationalism, while the second captures his idea that primitivism and not physicalism is the preferred objective account of pleasantness. Both quotes are taken from his (2001):  Against the [Relationalist], I maintain the following: it is because affect can be the disclosure of the appeal of other things and other people that it can have authority in the matter of what we should desire and do…the presence of the affect can make the desire or action especially intelligible to the agent himself. It can make the desire or act seem apt or fitting in a way that silences any demand for justification (p.189).    80  …we desire other things and other people, we are struck by their appeal, we are taken with them. This is part of how things are manifest to us: part of their appearing or presenting is their presenting to us in determinate ways and to various degrees appealing and repulsive. On the face of it, appeal is as much a manifest quality as shape, size, color and motion (p.188) [emphasis added].         This is a striking picture. Notice the use of the term ‘manifest’ in the latter quote. This provides an important marker that Johnston is an affective primitivist. The claim is that pleasantness is manifest to us in our sensory experience. Physical properties are not manifest in this way. Heavier molecular weight is not manifest to you when you smell the lavender. You are presented with pleasantness itself, among other manifest qualitative properties, when you undergo the lavender smell experience. Johnston is arguing for an objectivist theory of pleasantness, which in itself makes the view striking, but does so without appeal to physical properties. We know from other works (2006, ms.) that Johnston is a direct realist who thinks that manifest qualities are directly presented, as opposed to represented, in sensory experience (see also, Campbell and Cassam, 2014). Johnston is a primitivist about all the sensible properties and O-P is no exception.           Notice the term ‘disclosure’ in the first quote. Johnston claims that, on affective primitivism, among other affective qualities, O-P is given in experience. This is what makes intelligible to us our desires and action, why we desire and act as we do. If you are drawn to the bed of lavender and act in ways to get more of the smell, we can explain this desire and action by the idea that  81  the pleasantness of lavender was directly disclosed to you. In other words, Johnston is claiming that O-P directly motivates, the right sensory experience is all you need. The most striking aspect of this view is the claim that pleasantness is not a physical property or a relational property but is an objective quality that directly motivates. The world itself is rich with meaning and significance that is waiting to be detected by perceivers.       Johnston is almost certainly a proponent of response-independent primitivism. On his view, pleasantness is a primitive property that supervenes on physical properties. So Johnston could allow that olfactory pleasantness is supervenient on molecular weight; he just can’t allow for the identifying of the two. If O-P is objective, it is a primitive and not physical property. Johnston explicitly attacks the relational theory enough for me to feel safe in making this assessment.        Christine Tappolet (2016) postulates a view that could be construed as primitivist. On her view, evaluative properties, such as fearfulness or admirableness, are properties of objects. Say I run into a bear in the wild. The bear itself, on this view, instantiates the property of fearfulness. Emotions explain the motivational characteristics of objective evaluative properties. I am motivated to get away from the bear because I feel fear. It is not clear exactly how Tappolet’s theory fits in to affective primitivism more generally but it is at least a close cousin.9           In chapters 1 and 2, I argued that affective physicalism has trouble explaining how a physical property like molecular weight could explain the inherent motivational characteristics of                                                           9 Tappolet focuses on emotion. I am not clear how her picture, even if true of emotion, can explain sensory pleasantness (see her 2016).   82  pleasantness. Affective primitivism does not suffer from this same problem. This counts as a major plus for the view. In fact, affective primitivism should be the default objectivist theory of O-P. If you want to run an objectivist view of O-P, it’s best to start here. This creates a situation where relationalism (see chapter 4) and primitivism are the major contenders in the search for a realist theory of pleasantness.    3.4 Strengths       Affective primitivism has its strengths and weaknesses. I discuss the strengths first. When discussing these issues, I will assume response-independent affective primitivism.       In my view, the biggest strength of affective primitivism is the seeming faithfulness to our phenomenology. By ‘phenomenology’ I mean the ‘what it is like’ aspect of sensory experience. When you see the red apple, there is something it is like, a subjective ‘feel’, to your experience of the redness. There it is something it is like for me to experience the same token apple. I don’t share the same subjective ‘feel’ that you do. I make a distinction between naïve phenomenology and complex phenomenology. The former picks out our pre-reflective, purely descriptive aspect of phenomenology, the latter picks out our post-reflective, introspective aspect of phenomenology.10                                                               10 The Phenomenological tradition uses naïve phenomenology as a method in making various philosophical claims. For a classic account see Merleau-Ponty (1945) and Zahavi (2005) for a more recent account. I thank Barbara Weber for pushing me to clarify my account of phenomenology.     83       O-P phenomenology seems to reveal intrinsic, simple, monadic properties. Relational properties seem not to fit the bill; primitive properties, however, seem to fit them to a tee. Primitive properties are intrinsic, simple, monadic properties. This is the primary motivation for affective primitivism.       The issue here is complex. There are important questions about the relationship between introspection and sensory phenomenology. I discuss this relationship more in the next chapter. For now, it is enough to consider the following: does affective primitivism best explain our naive phenomenology? By naive phenomenology I mean: the world as it is immediately presented to us. This implies that there is an aspect of our phenomenology that is temporally extended and available to introspection. We could call this our complex phenomenology. I claim that affective primitivism has an advantage in explaining our naive pleasantness phenomenology but it doesn’t follow that it necessarily best explains our pleasantness phenomenology overall.       Of course, Johnston claims an additional motivation for affective primitivism: he says that relationalism cannot explain the authority of affect, the ability of sensory affective states to make intelligible our desires and actions. But in many ways, the argument from authority turns on an argument from phenomenology. Johnston thinks affective states like pleasantness have this kind of authority because pleasantness is directly disclosed to us when we undergo a pleasant experience. It seems very unlikely that Johnston would make similar claims if relational, complex, dyadic pleasant properties were disclosed instead. The argument from phenomenology is the prime mover for affective primitivism.   84       I earlier mentioned that primitivism more generally has two motivations: the view grants us direct epistemic access to the mind-independent world and primitivism is the view left over after the failure of the other candidate views. I am not especially concerned with epistemic access in this work, so I’ll move past that potential motivation. The other—that primitivism is the ‘left-over’ view—has some merit. I’ve discussed affective physicalism and found it wanting. I discuss relational views in chapter 4. The reader should hold in mind the idea that if affective relationalism fails, then affective primitivism can gain a foothold as the best candidate theory of O-P.       This is a risky motivation for it opens the door to a kind of irrealism. That is, if primitivism is all that is left over and despite my discussion of the realist motivations for the view, it doesn’t follow that a realist version of primitivism is the result. There is an irrealist version of primitivism. On this view, O-P is a primitive property. However, said property is not instantiated. This opens the door for a kind of figurative projectivism. I discuss this view, and other irrealist views, in more detail below. I bring the view up here to highlight that realist affective primitivism has more work to do even if it is determined that primitivism is the ‘left-over’ view of O-P.       All that said, the two best arguments for affective primitivism are that the view best describes the phenomenology of O-P and that it is the best view left after having gone through the other viable options.   3.5 Weaknesses  85            Affective Primitivism has three glaring weaknesses. The problem arises when you consider that the view is the near converse of literal projectivism.11 The latter view claims there is only E-P, the former that there is only O-P. Both views privilege one of the pleasantness property types at the expense of the other pleasantness property type. Essentially, the affective primitivist denies E-P. In this section, I will argue why that is problematic. I look to affective neuroscience and our folk notions in the undermining of primitivism.       Primitivists must reject E-P because their view relies on transparency. Primitive properties are those mind-independent properties that are made manifest to us in experience. Micro-physical properties, like molecular weight, are not made manifest in experience and so primitivism is an objectivist view that is also non-physicalist. Dispositional properties, such as those postulated on relational views, are also out for they too are not manifest to us in experience; only their realizers are (despite arguments to the contrary from Johnston, 1997). And of course, any subjective experiential properties, like those postulated by literal projectivists, violate transparency. Transparency is a key motivator for the affective primitivist.       You can understand why, then, the primitivist seeks to eliminate E-P. Experiential pleasantness is, by nature, a subjective property. It does not make transparent to us any mind-independent properties. In general, primitivists do not agree with the idea that there could be qualia or any other properties of experience. The reason: primitivists are motivated by strong                                                           11 The views are not the exact converse of each other. Primitivists will not claim that E-P is a projection of O-P, for example.   86  connections to direct realism. Many seek to block skeptical worries by arguing that the mind-independent properties of the world are made manifest to us in sensory experience. In a sense, the world itself is directly presented to you in experience. Claiming that there are properties of experience counters direct realism. The reason: It seemingly adds a mediational element to perception. The direct realist wants direct, non-mediated epistemic access to the mind-independent world. They think this blocks the skeptical worry that we can’t know whether the external world exists. By introducing experiential properties, we make room for the idea that perceivers could be mistaking experiential properties for mind-independent ones. Transparency is crucial to the entire primitivist project.      The affective primitivist is no different than the sensory quality primitivist. There is no room for E-P in the primitivist story. That’s a problem. In chapter 2, I gave a general argument for why we should reject any theory of pleasantness that denies E-P (or O-P for that matter). I want to give a specific argument against primitivism here.      Denying E-P runs afoul of what we know from affective neuroscience (see chapter 2). If it is anything, pleasantness is an experiential gloss on incoming sensory information. The picture is this. We undergo a sensory experience with particular sensory information. This information is then classified in a particular way and a pleasantness property is then indexed to the sensory information (I develop a similar relational theory of O-P in chapter 4). Said pleasantness property is subjective and experiential, in other words, E-P.        On the primitivist picture, pleasantness is directly presented to you. There is no active transaction between stimulus and subject in the constitution of pleasantness. We simply detect  87  the pleasantness of mind-independent properties. Affective neuroscience shows this to be incorrect. We detect mind-independent properties all right. However, pleasantness is added on after this detection has occurred. We don’t detect pleasantness; rather, we detect sensible properties and then experience pleasantness. In other words, you can’t eliminate E-P without violating core discoveries from affective neuroscience.        The primitivist could claim that the neuroscientist overlooks our pleasantness phenomenology. It seems to us as if the pleasantness is ‘out there’ in the world and not an experiential gloss on our sensory phenomenology. However, this does not prove what the primitivist needs, even if true. At best, this claim shows that O-P exists (at worst, not even this, as the coherence of literal projectivism makes clear), it does not show that we can eliminate E-P.       Eliminating E-P also runs afoul of our folk intuitions. You don’t have to think that said intuitions are sacrosanct to think that a theory of pleasantness should not strongly violate them. But primitivism does violate our folk intuitions. I smell the lavender and experience pleasantness. The primitivist must give an error theory for my statement. Perhaps what I really mean, according to the primitivist, is just that the lavender itself smells pleasant. But we commonly attribute pleasantness to our experience. The primitivist runs a direct realist line for O-P but then an error theory for E-P. Perhaps this is the way to go. But I’ll reiterate what I said in chapter 2: a theory that can explain both O-P and E-P is more theoretically virtuous than one that explains one and eliminates the other. We should avoid having to give an error theory of pleasantness. This means primitivism and projectivism are not desirable theories of O-P.   88       The folk intuition point is not meant as a knock-down argument against primitivism. But coupled with the results of affective neuroscience and the fact there are theories that can explain both O-P and E-P, viz. physicalism and relationalism, we have strong reasons to reject affective primitivism. We should look for a theory which explains the affective neuroscience and our folk intuitions of pleasantness. Such a theory is more attractive than one that eliminates one of the pleasantness properties. I develop a good candidate theory in chapter 4. Though I focus on O-P, the theory naturally explains E-P. So the view is compatible with both affective neuroscience and our folk intuitions.      Affective primitivism suffers from several other weaknesses. I’ll discuss two: an inability to explain E-P that undermines the motivation for the view and the inability to explain variation. I discuss each in turn.       Affective primitivism is an appealing theory of O-P. It explains the (naive) phenomenological facts, which are important for understanding the ontology of pleasantness. However, affective primitivism is handicapped when explaining E-P. There are a few reasons for this. Firstly, it is difficult to find an experiential primitive property that corresponds with an objective primitive property. They are properties made manifest in our experiences of pleasantness; they are transparent to us. Primitive properties are simple, monadic and intrinsic properties of objects. Importantly, we do know of an experiential property that has these characteristics: qualia. These are postulated to be simple, monadic and intrinsic properties of experiences. If we apply primitivism to E-P, then E-P is a qualia type. There is, of course, nothing at all wrong with postulating or denying qualia. However, qualia present a problem for the affective primitivist.    89       As my discussion of Johnston makes clear, affective primitivists hold that objective properties are directly presented to us in experience in a pre-content, pre-judgmental way. They are no fans of qualia. Qualitative properties of experience introduce a ‘sensory subjectivism’ that proponents of affective primitivism reject because it makes sensory perception less than fully transparent.  But the affective primitivist is stuck; either she admits qualitative properties of experience for E-P and violates transparency or she claims that E-P is confused—a highly unlikely result.      In other words, on this line of reasoning,  qualia just are the primitive E-P properties. The affective primitivist can either accept this fact and give up full transparency or give an error theory of E-P.      The deep issue is this. Affective primitivism is appealing because it explains the (naive) phenomenology of O-P well. One of the key facts of this explanation is the claim that O-P is transparent to us in our experience; we know which property O-P is when we undergo an experience of it. The problem is the same move cannot be made for E-P. Presumably, qualia are the simple, monadic, intrinsic properties made manifest to us in our experience of E-P. Of course, these are properties of experience. The following result occurs: when I experience pleasantness when smelling the lavender, I attribute a primitive O-P property to the flower itself and a pleasant quale to my experience of the flower. But it then follows that my overall experience of the lavender is not transparent. The affective primitivist must, therefore, drop the commitment to transparency or give an error theory of E-P. Both are undesirable.  90       Transparency is one of the key motivations for primitivism. Primitive properties are often postulated to account for the supposed fact of transparency (Campbell, 1993, 2005). An affective primitivism devoid of transparency is no primitivism at all. The only real option here is to claim that O-P is primitive but give an error theory of E-P. Notice that this is the inverse of projectivism, which gives an error theory of O-P. The primitivist should avoid giving a positive theory of E-P if they want to preserve the consistency of their view.       How would an error theory of E-P work? On this view, there are no pleasant experiential properties. Pleasantness is not instantiated in experience. However, we do still attribute pleasantness to experience but, on the present approach, we do so in error. If the primitivist were to take this line, then the following picture emerges. When I attribute pleasantness to the lavender and my experience of the lavender, I am attributing a primitive property to the former and am in error when I make an attribution to the latter.       I have touched on this general approach above. When discussing the different ways to accommodate O and E-pleasantness, I mentioned this approach as the one where you can affirm the reality of one pleasantness type while claiming the other is confused in some way. I have claimed this is the only path forward for the affective primitivist; but is a fraught path. First, the primitivist will have to give a semantics for E-P that explains what we mean when we say and think that the lavender experience is pleasant. It is easier to see how an error theory semantics would work for O-P than E-P; the projectivist is in better shape than the primitivist here.   91       The reason is this: we are more likely to correctly attribute properties to our experience than to correctly attribute properties to external objects. When I see the green grass and have a corresponding subjective experience, I am more likely to be in error attributing ‘green’ to the grass than I am attributing ‘green’ to a quale. In other words, an error theory is more likely when we are constrained by the appearance-reality distinction than when we are not. Given this, the projectivist is on stronger footing than the primitivist.        Second, the affective primitivist has difficulty explaining why we would sense O-P properties but not their corresponding E-P properties. This is important, for why would have evolved systems of pleasantness ‘detection’ that only picked up on O-P? This opens the door for an odd consequence. Consider that you ‘detect’ O-unpleasant ammonia. The chemical is at a high enough intensity to be toxic to you but you don’t really ‘detect’ any corresponding E-P property. This would be fatal. The problem comes back to this: how could an objective primitive property motivate? Motivation is one of the key characteristics of pleasantness. Unless the primitivist can answer this question, they will not provide a satisfactory account of O-P.         The idea that we have no real E-P properties that correspond in some way to O-P properties makes little sense of why our affective systems function the way they do (see chapter 2).  ‘Detecting’ only O-P properties would not explain why we are motivated to get more fitness enhancing stimuli and avoid more toxic stimuli. E-P is needed to explain the motivation. The discussion of affective neuroscience tells us why. Without E-P, we would not have gained the flexible behaviors needed to navigate the world of nutritious and toxic food items.   92       In other words, this version of affective primitivism—one that gives an error theory of E-P—runs afoul of the same worries over affective physicalism. Neither view can explain how pleasantness motivates us. The goal is to find a theory that explains the nature of O-P properties in a way that accommodates the fact that its corresponding E-P properties are important motivators. I give such a theory in chapter 4. Suffice it to say, the affective primitivist attempt does not succeed on this front.      Both forms of affective primitivism are troubled by the problem of incorporating E-P. Response-dependent affective primitivism says that primitive pleasantness properties supervene on relational properties, like dispositions. It looks like, just maybe, this view could accommodate the idea that O-P is primitive but E-P is relational. This will not do, as the RD primitivist variant does not identify pleasantness with relational properties. Like reponse-independent primitivists, they identify pleasantness with primitive properties. It is therefore likely that both variants of affective primitivism will have to either give up transparency or run an error theory of E-P.        In my view, this is the biggest weakness of affective primitivism. The quotes above by Johnston show someone who thinks pleasantness is an objective property that has a certain authority in our lives. However, he rarely mentions E-P. This is puzzling. Affective primitivists provide an attractive picture of O-P but have little resources to say much about E-pleasantness. More precisely, primitivism plus direct realism entail en error theory of E-P but a positive theory of O-P. The fatal flaw for the primitivist: they cannot give a positive theory of E-P without violating the fundamental motivations for their own theory.   93       Finally, affective primitivism struggles to explain variation. In the next chapter, I will argue that affective relationalism does a good job explaining the variation of pleasantness. This would give us good reason to think affective relationalism is the better theory. David Chalmers (2006) explains the problem of variation for color primitivism. I’ll transpose the case from color to pleasantness.       Two communities are presented with an ammonia olfactory stimulus. In one community, the smell is very unpleasant; in the other the smell is very pleasant. If both experiences are veridical, then the first experience attributes primitive pleasantness and the second primitive unpleasantness to the same olfactory stimulus type. The affective primitivist cannot claim that one of the communities experiences this stimulus in the right way while the other does not. Like in the case of the affective physicalist, this is an ad hoc and unprincipled move.       The problem for affective primitivism comes when we realize the above case can generalize to any pleasantness property types; the problem is not specific to ammonia. Once we grant this and the point about the symmetry of the two communities, we get the following result: every object instantiates every primitive O-P property and every experience instantiates every E-P property. This is a deeply counterintuitive result. It follows from the bare facts of variation and the content of affective primitivism.       Mark Johnston (2001, p.185) has a response to the worry from variation. He claims that the environment is rich with O-P properties. One object could be O-P pleasant to me but O-P unpleasant to you because I detect something in the object that you cannot. On this view, there is no neutral base that disposes me to find pleasant what you do not. However, think of  94  relativizing to circumstances; there are subjects (including non-human animal perceivers), objects and circumstances in our analysis of O-P. If Johnston is right, there would be an explosion of primitive O-P properties. I think we should shy away from theories that postulate massive property proliferation. There can and are massive amounts of relational properties, of course. But remember that, for primitivists like Johnston, O-P is most decidedly not a relational property. Variation continues to be a problem for the primitivist.        Though note that it is response independent primitivism, a la Campbell, that is in trouble here. McGinn’s response-dependent primitivism can handle the facts of variation. On the latter view, O-P supervenes on relational properties so the facts of variation can be explained by the change in the underlying base properties. Be that as it may, McGinn’s view still suffers from the other two weaknesses of affective primitivism: violating affective neuroscience and our folk intuitions and a forced choice between dropping the transparency requirement or giving an error theory of E-P.   3.6 Irrealism            The above three arguments imperil affective primitivism. The costs are high. But before we back our way into the arms of affective relationalism, there is one final option. I’ve discussed two major objectivist theories of pleasantness:  physicalism and primitivism. A third view takes  95  up room in logical space: irrealism (see Pautz, ms; Chalmers, 2006). One way to get to irrealism is through primitivism (see Cohen, 2009; Pautz, ms.). I’ll briefly discuss this option.12      There are two ways you can argue for irrealism about pleasantness. The first is by claiming that there are no pleasantness properties at all; no O-P, no E-P, none. The second is by claiming that pleasantness—both O and E—is a primitive property but is not instantiated in any object or experience. On this latter view, pleasantness exists it’s just not a property of objects or experiences. The first approach is so deeply revisionary and contrary to our everyday experience that I will not discuss it much further. The view should be one of last resort and as Cohen (2009) emphasizes, it’s difficult to ever conclusively prove philosophical viewpoints wrong. It’s very unlikely that there are no pleasantness properties.      However, the second option has been ably defended by Adam Pautz (ms.) in the color case and it transposes nicely to the pleasantness case. In fact, affective irrealism just is a kind of figurative projectivism (assuming they agree with the claim that we attribute O-P to objects). The view is this. O-P is a primitive property that we attribute to objects. However, objects do not in fact have any O-P properties. We are in error when we make O-P attributions. This causes us to mistakenly ‘project’ the property of pleasantness onto objects. A similar story could be given of E-P.  On Pautz’s view, the intentional, experiential content, not any property of the experience itself, of O-P is the closest we get to instantiated pleasantness properties.                                                             12 There are some theorists who claim that affective qualities are not phenomenal qualities. They are phenomenal irrealists about affect but realists about affective qualities. I will not directly deal with these views. See Armstrong (1968), Pitcher (1970), Tye (1997), Heathwood (2007), Robinson (2006, unpublished ms.).    96       How does Pautz arrive at his conclusion? Affective physicalism and relationalism do not work for the reasons I gave above. We should, therefore, hold the view that O-P is a primitive property; it is the only option left. But there is another caveat. As I showed above, a realist version of affective primitivism suffers from major problems. Given all these facts, there is one natural view remaining: O-P is an un-instantiated primitive property. E-P would be as well. We have arrived at O-P irrealism through primitivism.      I will not engage with this view directly. The best way to undermine the view is to undermine the motivation for primitivism. The chief motivation for irrealist primitivism is the idea that all the other views have fatal flaws leaving only the aforementioned view.13 In the final chapter, I argue that affective relationalism is the best chance we have at giving a good account of O-P. If this proves true, the motivation for irrealist primitivism is deflated and the view becomes a non-starter. So the question is: why think that, after all this, we should embrace affective relationalism?                                                                                                            13 One could claim that motivation is found in attitudes (I approve/disapprove), not objects or experiences. This is a path towards irrealism about pleasantness. If the key characteristic of pleasantness is motivational power and said power is only found in attitudes, then pleasantness properties will only be found in attitudes, not objects or experiences, as well. This is certainly a coherent view. However, I am a (weak) phenomenal realist about sensory affect. I think affective qualities are real and, at least partly, phenomenal qualities. If I am right, the view under consideration would be a non-starter. However, I do agree that it is an avenue to irrealism about the affective qualities. I thank Dom Lopes for proposing this view.       97                                                Chapter 4: Affective Relationalism(s)                   4.1 Affective Relationalism                           In many ways, relational views seem natural. We tend to think of O-P as a relation between objects and subjects. There are at least several variants of the relational view and I’ll get to those a bit later.       There are different motivations for relational views but the ubiquity of variance looms large. There are three variance types. I might find this chocolate cake pleasant but my cat (cats don’t have sweet receptors) does not (inter-species variation). I find the skunk smell, at a particular intensity, unpleasant but you actually find it pleasant (inter-personal variation). I find the feel of sandpaper pleasant at one point in time but come to dislike it over time (intra-personal variation). Jonathan Cohen (2009) gives an argument for color relationalism that is strongly motivated by these variance phenomena. I apply it to the pleasantness case.       The argument is as follows:  Assume that a particular physical type could be the O-P property.  1) There are many different pleasantness effects of a single physical stimulus.  2) None of the variants given to explain (1) are well-motivated.  3) Given these two facts, an ecumenical reconciliation is to be preferred over privileging just one of the variants.   98  4) A relational view is the ideal theory to provide this type of reconciliation.              Clearly, this view is a frontal attack against affective physicalism, a type of objectivism. The latter view is motivated to pick out one physical property type as pleasantness. In the olfactory case, molecular weight is the candidate property. But notice the role of variation. Premise 1 is a classic statement of variation. It captures the three types mentioned in the last paragraph. To simplify, the same stimulus gives rise to different effects. Physicalism cannot explain this in any principled way. The relational view can explain this. Therefore, we should choose the relational view.       However, the case Cohen makes against color physicalism is not strictly analogous with the case of affective physicalism (not that Cohen would disagree). Color physicalism is, along with the relational view, one of the two major players in the philosophy of color. It makes sense that Cohen would seek to knock color physicalism off its perch; it’s the primary rival to his view, though it certainly is a consistent and live option. But the case of pleasantness is different. Affective physicalism is an underdog view, though one that I think should be seriously considered. Given this, an affective relational theory might proceed a bit differently; it might proceed without relying on the argument given above.       To see this, consider the disanalogies between color and O-P. I have already argued that affective physicalism will not ultimately work. But why is color physicalism more likely to be true than affective physicalism? Firstly, O-P is more radically variable than color. Our sensory affective reactions are constantly changing; not so in the case of color. Consider intrapersonal  99  variation. SSS (stimulus specific satiety) is a common phenomenon in smell and taste. I can smell a particular fragrance at a particular intensity and have differing affective reactions within a short period of time. I can get ‘tired’ of the stimulus and its affective tone will change from positive (pleasant) to negative (unpleasant). Color is not as radically variable. There is no strictly analogous case to SSS for colors. You don’t get tired of seeing yellow and then see that stimulus slowly morph into blue. You can, however, see yellow fade so there is some stimulus fatigue. But the fade is not a morph. Of course, you might see a color morph but it’ll be because something physical has changed; different stimulus, different effect, so no SSS.  Despite the variance seen with both O-P and colors, the former is more susceptible than the latter. I suspect this is one of the reasons why affective physicalism cannot get off the ground in the same way that color physicalism can.       But there is a second reason for the disanalogy between color and O-P. Colors are proper sensible qualities. As I discussed at the end of chapter 2, O-P is not. Of course, on the object attribution view we treat pleasantness as if it were a sensible property. And O-P might be a unique property that is, in some sense, a sensible. But we do not have an O-P sensory system. We clearly do have a visual system which transduces wavelengths into color information. Color is the proper sensible of vision. O-P can’t claim that same status with any sensory modality. I think this is another reason for the greater popularity of color physicalism over against affective physicalism. Physicalism will be on the table for colors, sounds, tastes, smells, touch, etc. because each of these is a proper sensible. It could be the case that our sensory systems encode physical properties that just so happen to be color or sound or smell, etc. And this can  100  be true despite the theory’s difficulty with variation. The same cannot be said of O-P. Because it is not a proper sensible property, it is harder to motivate the idea that O-P is a physical property whose information is encoded by the requite sensory system. I suspect this is another reason why affective physicalism cannot get off the ground in the same way that color physicalism can.                                     So I agree that the variance in O-P undermines the affective physicalist. On the latter view, the same stimulus type should give rise to the same pleasantness effects across species, human individuals and the same individual over time. This clearly doesn’t happen. We should jettison the view. However, it takes more work to arrive at the positive conclusion that a relational theory of pleasantness is the way to go. In this chapter, I motivate affective relationalism, discuss several different variants of the view and arrive at my conclusions.  I propose the following argument inspired by the one used above:  1*) There are many different affective effects of a single physical stimulus.  2*) Privileging one variable over others is unmotivated and ad hoc.  3*) The affective relational view explains the facts of 1* better than any competing theory.  4*) Therefore, the affective relational view is to be preferred over any competing theory of pleasantness.            Affective physicalism gets off the ground by claiming that, of all the variants that could be the locus of O-P, there is a unique variant such that that variant is the locus of pleasantness and  101  is a physical property. In other words, there is a unique, physical property that is the locus of O-pleasantness. This contradicts premise 2* above. This premise asserts that there is no unique variant that is both well-motivated and can explain the variance between a physical stimulus and O-pleasantness effects.      In the olfaction case, the affective physicalist can pick out molecular weight as the unique physical property that is the locus of O-P. But, as I argued in chapter 2, that view is both under-supported by the evidence and falsified on the pain of circularity. Affective physicalism, though a coherent view in the logical space, finds few adherents in actuality.       The strategy holds when considering the two biggest rivals to affective relationalism: affective primitivism and affective irrealism. Both primitivists and irrealists reject the physicalist claim that there is a unique physical variant that is the locus of pleasantness. As I discussed, primitivists struggle to explain variation but they certainly dislike the strategy of reducing O-P to physical properties.14 They can meet the relationalist at premise 2* and give it their best shot at undermining that claim. The irrealist agrees with the idea that there is no unique variant that is the locus of O-P. However, unlike the relationalist, they are far more conservative. The relationalist says all the variants matter for pleasantness. The irrealist says none of them do. So like the primitivist, the irrealist can meet the relationalist at premise 2* and give it their best shot at undermining that claim.                                                           14 Affective primitivists do struggle with variation but not in the same way that physicalists do. Given variance, physicalist conclusions are ad hoc and unprincipled. Primitivist conclusions are not these but, given variance, their conclusions force us into unacceptable property proliferation.    102       But why start with relationalism as the default. I think there are good reasons. The best is this: as in the ontology of color, physicalism and relationalism are the conservative, non-revisionary views (see Louge, 2016a, 2016b). When analyzing O-P (and color), there are various physical and relational properties around that are candidates to be the locus of O-P. Both primitivists and irrealists deny that pleasantness can be found among those said property types. It remains to be seen why it should be any other property type or no property type at all. As such, these latter two views are revisionary. Of course, one (or both) of them might just be right. But that doesn’t negate the fact that physicalists and relationalists are working with clearly delineated property types and primitivists and irrealists are not. Given this and given all I’ve said about the implausibility of affective physicalism, affective relationalism should be the default theory of O-P. I will treat it as such.      The relational approach is fairly general. On the view, affective qualities are relations between objects and subjects. There are many different relational properties but the key move here is the relation between objects and subjects. Affective properties are those experience-independent properties (and so objective) that need experience-dependent properties (and so subjective) to exist. However, the relata can be spelled out in different ways. The particular relational views do just that. I spell out each in turn.      4.2 Particular relational approaches         103       The relational view is general. There are different types of relational views (see Cohen, 2009; Matthen, 2005; Thompson, 1995). I will focus on four that loom large in the color debate. I transpose these results to understand the nature of pleasantness. Remember, I am interested in understanding pleasantness as attributed primarily to objects. The four views are: dispositionalism (a historically popular view of the secondary qualities, see McGinn, 1982 for a more recent defense), functionalism, the mixed view and the classification view. There are other relational views on offer—Hatfield (2009), and Chiramuuta (2015) for example—but I focus on the above four approaches.  The views are similar but importantly different. I discuss each in turn.      The Dispositional view has a rich pedigree. Beginning with Locke (see Nolen ed. 2011), philosophers have been tempted by the idea that the secondary qualities--color, taste, smell, etc.--are dispositional. The nature of dispositions is hard to pin down (Choi and Fara, 2012). Examples of dispositions include: fragility, solubility, malleability, human character traits and so on. These are properties that objects have given certain possible conditions. For example, glass has the property of fragility. Certain conditions need to apply in order for the fragility of the glass to be causal; for example, dropping the glass from a certain distance or at a certain speed.       The dispositional view claims that O-P is like fragility. As a counterpoint, note that affective physicalism treats O-P more like mass, say, than it does fragility (pleasant smells are heavier odorant molecules, for example). Take smells. On affective dispositionalism, a pleasant smell is one that disposes a subject to have a positively valenced sensation in a certain circumstance. In other words, X is O-pleasant just in case X disposes a subject S to have a positively valenced  104  sensation in circumstance C. Though this analysis can apply to pleasantness more generally, remember that I am discussing O-P. Take the odor case: an odor O is pleasant just in case O disposes S to have a positively valenced sensation in C; or, O is unpleasant to S in C just in case O disposes S to have a negatively valenced sensation in C.15      Notice how this analysis accounts well for variation. O can smell O-pleasant to S1 in C1 but O-unpleasant to S2 in C2 but also O-unpleasant to S1 in C3. That’s the beauty of dispositions. An object can be disposed to appear differently to different subjects or the same subject over time. Notice how mass is not like that. An object's mass does not vary within or between subjects.      Also note how this fits in to the relational framework. On affective dispositionalism, O-P is analyzed as having both an objective and subjective component. An odor has a certain dispositional property (objective) that becomes O-pleasant when presented in the right way to the right perceiver in the right circumstance (subjective).        Dispositionalism has been the default relational view for some time. More recently, other relational views have been postulated to explain various sensible properties (see ch.8 of Cohen, 2009). The key difference between the views is in how they spell out the relations doing the constituting. We have already seen that dispositions are the requisite relations for affective dispositionalism. Importantly, save the mixed view, these views differ in emphasis and not much, if at all, in theory. I discuss this point at the end of this section.                                                             15 The subject side of the relata does not have to be an experiential property. Any subjective property would do. However, I will you use experiential properties in this section for reasons of consistency.    105       The new view on offer is affective functionalism. At first glance, it is not all that different from affective dispositionalism. The affective functionalist analyzes O-P in the following way (using the olfactory case): O is O-pleasant for S in C if and only if O-P functions to dispose its bearers to cause a positively valenced sensation for S in C; or, O is O-unpleasant to S in C if and only O-P functions to dispose its bearers to cause a negatively valenced sensation for S in C.16      Affective functionalism is quite similar to affective dispositionalism and not only because they are both relational views. The former view claims pleasantness functions to dispose, after all. In fact, the two views collapse into each other if dispositions are taken to be functional types. The relationship between dispositions and their ground could be type-identical, token-identical or functional; if the latter, then affective dispositionalism is not any kind of alternative to affective functionalism. In fact, I follow Cohen in thinking the two views collapse into each other.       Affective functionalism has several advantages. The first is a familiar one. Like all relational views, the view accounts well for variation. O might have the function of causing a positively valenced sensation for S1 in C1 but a negatively valenced sensation for S1 in C2. The benefit of a functional analysis is that it can account for this type of variation.       But there is another, more general, benefit to functional analysis: multiple realizability. For the affective functionalist, O-P could be realized by any number of physical properties. Contrast this with the affective physicalist. On the latter view, pleasant smells are identified with heavier                                                           16 I take affective functionalism to be a type of role functionalism. As Cohen rightly notes, realizer functionalism is not a relational view as it places emphasis on the material properties that realize functional roles. On the latter view, a relation to the subject does not enter the analysis. Given this, I leave realizer functionalism to the side.  106  odorant molecules; these just are the constituents of olfactory O-P. The affective functionalist rejects this move and identifies O-P with the role it plays in causing positively valenced sensations for subjects in certain circumstances. By not identifying O-P with a physical realization, the affective functionalist side-steps concerns I raised in chapter 2; that identifying O-P with a physical property is circular and leaves the inherent motivation of O-P unexplained.       There are two more affective relational theories I’d like to discuss. The first I call the mixed view. This view is similar, in many respects to affective dispositionalism and affective functionalism. Like these latter two views, this affective relational view analyzes O-P with reference to subjects and circumstances/conditions. This marks the view as relational. However, unlike dispositionalism and functionalism, the mixed view also allows non-relational elements into the analyses of O-P (hence, the mixed view). Take smell. On the mixed view, molecular weight—a non-relational property—is part of what it is to smell O-pleasant.       Here is the formal definition of smelling pleasant on the mixed view: O smells O-P to S in C just in case O has a heavy molecular structure that is perceived as belonging to a specific O-pleasantness category by a particular subject in a particular circumstance. Likewise, O smells U-P to S in C just in case O has a lighter molecular structure that is perceived as belonging to a specific unpleasantness category by a particular subject in a particular circumstance.       Essentially, the difference between this view and two previous is what each pick out as relata. On the dispositional view, we are disposed to experience a positively valenced sensation. On the functional view, pleasantness functions to dispose these types of sensations. Notice, there is no explicit reference to a non-relational property here. However, object  107  properties are implied either as ground of the disposition or realizers of a function. On the mixed view, we are related to molecular weight (in the olfactory case) in particular ways in particular circumstances. Object properties are more explicit in the mixed view analysis. They appear first as relata and then as conjuncts in the analysis. This makes the view mixed, though still clearly relational.       The mixed view has a mixed record when it comes to explaining variation. The problem arises with the idea that heavy molecular weight is necessary for being pleasant. Proponents of the mixed view need to account for this given their other relational commitments. The unique contribution of the view is the idea that what is variable includes molecular weight. There is a relation between O-pleasant things and perceivers and circumstances but this holds because O-P smells exemplify a heavier molecular weight that is itself related to perceivers and circumstances (and O-unpleasant smells exemplify lighter molecular weight that…). Variation is accounted for via the reference to perceivers and circumstances. But proponents of the mixed view must reconcile this with their claim that a physical, non-relational property is necessary for pleasantness.         I have two points to make about the mixed view. First, the view brings a physical property back in to the analysis of O-P. To be fair, the mixed view does not identify O-P with heavier molecular weight, a move that would make the view a type of affective physicalism.       However, by making a physical property one of the relata, the mixed view is runs into some trouble. In chapter 2, I argued that physical properties cannot be the grounds of O-P on pain of circularity and empirical underdetermination. However, it is not merely that physical properties  108  cannot be identified with O-P; they are also not the right kind of properties to be used in a relational analysis of O-P either. There are empirical reasons for this. It is simply not clear that molecular weight is the physical property that correlates highly with olfactory O-P. Till such time as a physical property highly correlates with olfactory O-P (or pleasantness found in any other sensory modality), we shouldn’t use a physical property as a relata in our analysis of O-P. Note here how the affective functionalist is not bothered by these concerns. Functional roles are realized by different physical constitutions. This is part of what it means to give a functional analysis.17       Finally, there is one more relational theory of pleasantness I’d like to discuss: the affect classification theory. On this view, O-P is understood as the property classified by the requisite sensory affective system. Pleasantness is classified when it subserves particular actions. So in the olfaction case we have: O is O-P/O-U to S in C just in case S’s sensory-affective olfactory system classifies O in a particular way. The locution ‘proper way’ need not imply that there are norms to affective classification. It is merely a way of saying that our sensory-affective system will classify in some particular way.18          This begs the question: what are some ways that the sensory-affective system classifies? Murat Aydede (2018) has provided a good list. These include (and I quote): -sets interruptible motivational parameters (motivational biasing)                                                           17 I will leave aside the mixed view for the remainder of the essay. The reason: the view might not be a genuine alternative to the views just canvassed. The problem is not that the view might collapse into the others. The problem is that it’s not clear that it does. If the view identifies the ground/realized with heavy molecular weight instead of leaving the ground/realizer open, then the mixed view is not an alternative to dispositionalism or functionalism.   18 This view takes inspiration from Matthen’s (2005) theory of color. Unlike the color view however, the affect classification view I favor is non-normative. I discuss these issues more at the end of this chapter.   109  -prepares the effector or psychomotor systems of the organism, providing action-preparedness (motor biasing) -provides appraisals of the incoming sensory information for its significance for the organism and for its potential for enhancing its behavioral repertoire (epistemic biasing) -influences action preferences on the basis of the sensory stimuli’s informational content for future behavior through associative or cognitive learning, habituation, incentive sensitization, etc.  -provides steady earmarked input to more centralized concept-wielding cognitive, conative and decision-theoretic systems. (P. 258)       These are a seemingly good start in discussing how the sensory affective system classifies. Take the smell of vanilla. You are presented with the stimulus. The following actions occur: you are motivated to get more of the stimulus, you are primed to act to get more, you come to know (even if subconsciously) that this stimulus is good for you and you should get more, you come to learn that this stimulus will be good for you to get in future situations and the information provides input to cognitive, conative and decision making systems such that you form beliefs that the stimulus is good for you to get and you desire and make a decision to get it. If most of these occur, the vanilla stimulus will be O-pleasant. It is an easy exercise to figure out how the stimulus would be classified as O-unpleasant (think of getting less of the stimulus and coming to understand that said stimulus is bad for you). The view is relational because it defines O-P as a relation between objects and the subjects’ sensory affective systems.       Three particular relational views (leaving aside the mixed view) differ in emphasis but not in theory. Let me explain. Affective dispositionalism claims that pleasantness is the disposition of X in C to be pleasant to S. Affective functionalism claims that pleasantness is the function of X  110  to dispose itself in C to be pleasant to S. The affective classification approach says that X is pleasant to S in C iff S’s sensory affective system classifies the stimulus information from X in the right way (the circumstances are especially relevant for the right way of classifying). As Cohen (2009) makes clear, if dispositions are functional, then the dispositional and functional views collapse into each other. If dispositions are functions then clearly they play a functional role. Let’s set aside the functional approach for now and focus on the relationship between the dispositional and classification views.      The dispositional view analyzes pleasantness as a three place relation. There are objects, subjects and circumstances:  x (object property) has a disposition in c (circumstance) to be pleasant to S (subject). The classification approach fits in to this analysis. Essentially, the classification view falls emphasis the subject side of the relation, leaving the object side to be analyzed as a functional role. The object property disposes the subject to classify the stimulus information in the right way. There is no need to say anything else about this particular object property; it simply plays the role just specified. The classification approach emphasizes the subjective side of the relation. But this difference in emphasis does not imply a fundamental difference of theory. The subjective side of the relation is emphasized, among other things, to account for the radical variation found in pleasantness attributions.      But notice that the subjective side of the relata, as I’ve described it, is also functional. The function is to classify stimulus information in the right way. So all I’ve done is specify the subjective function of pleasantness. This shows that the classification approach is not fundamentally different from the affective functional view.  111       Assuming functions are dispositions allows me to claim that the three relational views on mention are all the same fundamental theory; they each differ in emphasis. Focusing on the subjective side of the relation also allows us to answer the projectivist. Unlike the latter view, the classification approach follows all relational views in claiming that objects have a role to play in the analysis of pleasantness.    4.3 Problems for affective relationalism            Now that I’ve gone through four particular relational theories the question arises: are relational theories of O-P successful?        I have discussed affective relationalism and four particular versions of it. The views get a grip by explaining well the variation found in our attributions of sensory pleasantness to objects. By relativizing O-P qualities to subjects and conditions, these views are in a good position to explain how you and I, for example, can differ on how O-pleasant (me) or O-unpleasant  (you) we find the same skunk smell to be. In this section, I raise two potential problems for affective relationalism. The problems are: dispositions and radical variance, objections from phenomenology. I discuss each in turn.  Complexity         112       Assume that colors are identified with dispositions. On this view, blue is disposed to look blue to S in C. There are several reasons one could hold this view, including, as in the O-P case, that it explains variation well. One of the reasons you might hold color dispositionalism is that it explains an important fact: the blue bike does not cease being blue in the dark or when no one is seeing it. All that is needed is the right condition to obtain--the proper lighting condition or the condition of being seen by a subject--and the blue is perceived. This is because our attributions of color dispositions are relatively stable. We attribute them to dispositional properties of objects; dispositions that don’t simply ‘disappear’ willy nilly. Of course, given the relativizing to conditions and subjects, you might see the bike as purple. But notice that you too would see this purple as a stable disposition of the object. You don’t see the bike as purple but then blue and now somehow suddenly black. The reason is straightforward: we take it that color dispositions exist even if they are not a current part of your sensory experience.      The same is not always true for attributions of O-P, especially in the taste and smell modalities. Consider oysters. The O-pleasantness attributed to oysters is highly variable. Much depends on when the catch came in, how the chef prepared and seasoned it, where the oysters were caught, etc. Given this high variability, the O-P dispositions will be, contra color, attributed in a more complex way. Consider the skunk smell. It might be O-pleasant to me at lower intensity (in this case, further away from the source) but O-unpleasant at higher intensity (closer to the source). The olfactory disposition is attributed in a more complex way to its source (whether that is an object or region of space; see Batty, 2010; Young, 2016).   113       Consider how complex our pleasantness attributions are. Take color and shape. I attribute the color gray to my bike and I also attribute the shape triangle to the center portion of my bike frame. The latter attribution will, barring a non-veridical sensory experience, rarely, if ever, change; attributing shape is, in this sense, a simple attribution. The former attribution will not be quite as simple but color attribution is, on the whole, also a very simple type of attribution. I am more likely to attribute a different color to my bike than I am to attribute a different shape to my bike frame. Perhaps the lighting conditions are such that I attribute brown to my bike or maybe a friend attributes an earthy green color in other lightning conditions. Still, this will not be all that common.       Attributing O-P is more complex in comparison. Consider again the case of tasting oysters. I will attribute pleasant to them if, say, they were caught in late spring in the North Atlantic and the chef seasoned them with lemon and paired them with a white wine that goes well with the texture and temperature of the oysters. Clearly, there are many contextual factors here. If any one of them were changed, it is likely that I would attribute unpleasant to the oysters. This is what I mean when I say that O-P attributions are complex. Change any aspect of the relevant context and you might easily change the nature of the attribution.       In the shape case, assuming veridicality, you may attribute a different shape than I to my bike frame. Perhaps the viewing conditions cause this difference of attribution. However, once you come to see the bike frame in the proper viewing conditions it is unlikely you will ever misattribute its shape again. You may misattribute the color if the lighting conditions are different from the last time you saw the bike. This is true even if you know the bike is gray.  114  There is an additional contextual factor to the color attribution that the shape attribution is not constrained by. However, O-P attribution is constrained by many more contextual factors than the attributions of shape and color are. And this is why said attributions are more complex. Sensory quality attributions (and those that function like these attributions, like O-P) are simple or complex in proportion to how many contextual factors are in play when making the attribution.       The same fact does not hold for color. When presented with the color, you are presented with certain dispositions. This is true on any theory of color not just dispositional views (the key question of color ontology is whether color is to be identified with dispositions or something else). Even a novel color will present you with various dispositions. Importantly, we attribute these color dispositions to the object in absence of perceivers and circumstances. The novel color, whatever it is, remains that color for me in this particular circumstance but appears different to you in another circumstance. The novel color contains both of these dispositions and potentially many more. This is just another way of saying that colors have simple disposition attributions.        The worry I am driving at is this: we need to simply attribute O-P dispositions to objects if any form of affective dispositionalism is true. The reason: we are discussing attribution; given this, it is unlikely that we can identify O-P with a property that is not always attributed to objects (or anything else for that matter). Notice that this holds even if we do sometimes simply attribute O-P dispositions to certain objects. I can attribute O-pleasantness to chocolate ice  115  cream in the absence of occurent experience. Sometimes O-P can have a simple disposition attribution. However, there are many cases of O-P where this does not occur.       Complex dispositions are problematic. Of course, at times dispositions can be very complex. The complexity objection is not meant as a knock-down argument against affective relationalism. Rather, the objection is meant to challenge the idea that said view best describes pleasantness. Human character brings this point out. Human character is dispositional. My mother is not always kind but she is disposed to be under certain conditions (fortunately, there are many of these). It makes little sense to say my mother is kind or disposed to be kind if that disposition is too complex, if it is not a stable property of her.19 If someone you know does not exhibit kindness, or any other character trait, in a stable way, i.e. they are kind in just those circumstances you would expect of them, then we would question whether we should attribute kindness to them at all.       The same can be said for O-P. Oysters do not exhibit (gustatory) O-pleasantness in just those circumstances you would expect of them. We should question whether to attribute O-pleasantness to them at all. My point is this. If affective dispositionalism, or functionalism, is true then we would simply attribute O-P dispositions to objects. But we don’t always simply attribute O-P dispositions to objects. Therefore, O-P cannot be identified with dispositions. If this is true then perhaps we should look for the O-P property elsewhere. The question to be                                                           19 There is an interesting question here. What is the relationship between stable properties and simple attributions, especially when applied to relational properties? Olfactory and gustatory properties are less stable than other sensible property types. O-P properties are not especially stable. This fact could be what’s driving the complex O-P attributions we make.    116  answered: how can O-P be dispositional if certain O-P dispositions are attributed in such a complex manner?       Complex O-P dispositions seemingly weaken the motivation for affective relationalism, especially if the affective classification view and affective functionalism collapse into affective dispositionalism. However, there is another problem I’d like to raise that presents problems for all four view affective relational views: the problem of phenomenology.   Phenomenology            One of the most prominent objections against relationalism of any kind, at least when applied to sensible properties, is the idea that we do not perceive relational properties as relational properties. That is, you do not see the dispositional property relevant to red, you see red; the manifestation of said disposition. The same can be said for any relational property. Relational properties are not part of the content of our phenomenology, or at least that’s how the objection goes (for arguments along these lines see Stroud, 2000; McGinn, 1996 and Johnston, 1992; for counter-arguments see Cohen, 2009 and Shoemaker, 1997).       The argument from phenomenology leads to the claim that colors are experienced as intrinsic and/or monadic properties (as opposed to relational, dyadic properties) of the surface of objects. We don’t, on this line of thought, see colors as properties that straddle the boundary between subjects and objects. If color phenomenology works this way but color relationalism is true, then color appearances would misrepresent the nature of color (see McGinn, 1996). In a  117  way, the charge here would be that on the relational view color appearances are always in error and we should reject this conclusion.20       The above line of thought can be easily transported to O-P. The objection is as follows. Affective relationalism argues that O-P is a relational property, one that straddles the boundary between subjects and objects. But affective phenomenology does not present O-P as a relational property of any kind. Take the pleasant smell of lavender. The O-P is presented as a monadic, intrinsic property of a region of space or the flowers themselves (depending on the context). Take an easier example: the flavor of chocolate ice cream. The O-P is presented as a monadic property of the object (the ice cream). In undergoing an O-pleasant olfactory experience of the lavender we are not presented with something like the following content: these properties of the lavender smell ground why this smells good to me in this circumstance. We are presented with something like: this lavender smells good. The former content introduces a relation to a subject, while relational properties do not enter into the latter content.       A relationalist can push back against this view. For example, Shoemaker (1997) claims the following (also see chapter 6 of Cohen, 2009):  We do not, at least ordinarily, experience things as affecting our experience in certain ways, or as being apt to affect our experience in certain ways. The content of our                                                           20 Johnston (1992) claims that we are directly acquainted with color dispositions via their manifestations. In seeing red, I see the dispositional property to look red by seeing the manifestation of that property. That is, relational properties can be part of our phenomenological color content. On this view, we get a relational theory of color that does not presumably violate our phenomenology. One can, of course, push back against the idea that dispositional properties could ever be a part of our phenomenological color content. I won’t go into the details here but I do want to highlight a relational theory that is sensitive to the phenomenological objection and attempts to meet it.   118  experience is not relational in this way [the objection from phenomenology]…But the way properties are represented in our experience is no reliable guide to what the status—as monadic, dyadic, etc.—of those properties is…when something feels heavy to me [a relational property], no explicit reference to myself, or to my build and strength, enters into the content of my experience. Indeed, just because one is not oneself among the objects of one’s perception, it is not surprising that where one is perceiving what is in fact the instantiation of a relational property involving a relational to oneself, one does not, pre-reflectively, represent the property as involving such a relation (p.234).        Shoemaker argues that the objection from phenomenology does not imperil relational views. There is a gap, he claims, between what our experience represents and the ontological nature of the represented properties. His example of ‘weight’ is instructive. We represent ‘heavy’ as monadic: ‘this couch is too heavy to move’, ‘I can’t carry all the groceries up the stairs, they are too heavy’. In these instances, the heaviness of the couch and groceries is represented in experience as a property of the couch and groceries respectively. But this is wrong. Heaviness is a relational property. If you have any doubts, think about whether my body-builder neighbor would ascribe heaviness to the said objects or weakness to me. The point is this: sensory experience can represent relational properties as intrinsic; therefore, we should not look to phenomenology to tell us the essential nature of represented properties.       The point is well taken. However, Shoemaker is arguing this in the color case. He says that relational theories of color are not imperiled by the objection from phenomenology. That may be true. But does the same objection hold in the O-P debate? I argue that it does.   119       The most common objection to the relational view of O-P focuses on phenomenology. I argue it won’t work. The objection runs like this. When we introspect our phenomenology of O-P, we find no relational properties. So any relational theory of O-P violates our phenomenology. If any theory violates our phenomenology, then we should look for a theory that better explains it. Therefore, we should jettison the relational theory.       The argument is this: O-P phenomenology only presents non-relational properties. On this view, O-P phenomenology does not present relational properties because it presents O-P as an intrinsic property type. In other words, O-P phenomenology is far from mute on the ontological status of the O-P property type; it loudly proclaims said property type to be intrinsic and non-relational (see McGinn, 1997 for an explicit argument along these lines against the color relationalist).       My claim is this: the relational theory of O-P is not threatened by the argument from phenomenology; said argument is simply false on its face. The argument from phenomenology would only work if the following assumption is made: O-P phenomenology presents the ontological status of O-P properties.      In the argument from phenomenology, O-P phenomenology presents O-P properties as intrinsic. Indeed, in the color ontology debate something like this idea motivated McGinn (1997) to abandon a relational theory of color for color primitivism. However, this argument falters for two reasons: 1) it gets the facts wrong and 2) it too relies on the above unstated assumption. I discuss the first point later in the chapter when I argue for my view of O-P. Essentially, I argue that, in fact, O-P phenomenology presents O-P properties as relational. The  120  argument from phenomenology is simply false on its face. In the rest of this section, I want to discuss (2).       Assume the argument from phenomenology. This is not enough to guarantee the falsity of relationalism. There is, yet again, the following assumption: O-P phenomenology presents the ontological status of O-P properties. The following view is still in play, sans the assumption: O-P phenomenology presents intrinsic properties but said properties are in fact relational. Consider the above Shoemaker quote. Shoemaker makes clear that just because X is presented as intrinsic it does not follow that X is in fact intrinsic. The pleasant smell of the lavender could be a relational property presented as intrinsic. The argument from phenomenology relies heavily, then, on the unstated assumption discussed above. And as we’ve seen, phenomenology can mis-present the ontological status of property types.21 I claim that the above assumption is not true and should be discarded: relational properties can be presented as intrinsic.       I do not think this is how O-P phenomenology works but that is not crucial for my point here. The point is this: even if the argument from phenomenology gets the phenomenology right, which I claim it doesn’t, said argument still rests on a false assumption. Take away the assumption and the argument from phenomenology against relationalism loses much of its motivation. In my defense of relationalism, I argue that, in fact, relational properties are what is presented in O-P phenomenology. Not only is O-P phenomenology not an enemy of the relational approach to O-P, it is a strong ally.                                                                      21 There are echoes of this claim in chapter 3. In that chapter, I discuss naive phenomenology, the immediate, pre-reflective aspect of sensory experience. Primitivists, I argue, rely on the idea of naive phenomenology to claim that O-P is a primitive property. However, primitivists fall into the trap of assuming that naive phenomenology directly presents the ontological status of sensible property types. In the next chapter, I look at how relational theories can be true to the phenomenological facts. All we need to do is de-privilege naive phenomenology.           121  4.4 What makes objects pleasant?        The story so far: objectivist (physicalist, primitivist) theories of O-P don’t work, affective relationalism is a better approach; why: the theory accounts well for variation and has fewer shortcomings than its objectivist rivals. I canvassed four particular relational theories of O-P: dispositionalism, functionalism, the mixed view and the classification approach. This last approach is the most promising and the one I will argue for. This section is, in essence, the positive conclusion to my study. So why think O-pleasantness is a type of classification?       The affective classification view has the following structure: O-P is the property classified in a particular way by the requisite sensory affective system.        Relations             There is much to unpack here. The first is this: what makes this a relational view? Relational view partisans argue that O-P is a relation between an object and subject (and circumstance/context). Consider the affective dispositional view that I canvased above: O-P is the power to cause sensations of a certain kind given certain circumstances (relata emphasized). The power is a property of the object, sensations a property of the subject and the circumstances are built into the definition; a classic relational view. But it is not immediately clear how the affect classification view is also relational.   122       A more explicit formulation is needed: stimulus S is O-P to organism O in context C just in case O’s sensory-affective system classifies S in a particular way. The stimulus is the property of objects that, whatever that property is, feeds information to the sensory-affective system. This is the objective relata. The sensory–affective system then classifies said information in a particular way. By doing so, the organism undergoes a particular subjective ‘feeling’ sensation.22 This is the subjective relata. The context/circumstances are more implicit in the definition of the affective classification than they are with affective dispositionalism. These enter into the analysis through the importance of the organism acting in particular ways. The context/circumstance matters for the organism. In particular, the circumstances of the organism and the context they find themselves in enter into the analysis of O-P. The affect classification view is clearly relational.   The view       Relativizing pleasantness to subjects and circumstances does wonders. In particular, it shows strength where physicalism and primitivism are weak. Physicalism suffers from an inability to explain two things, the biological significance of pleasantness and the same stimulus/different effect phenomenon. Primitivism also suffers from an inability to explain two things, E-P and the strange consequences following from the postulation of non-natural properties. I discuss how                                                           22 This underscores how the view could explain E-P. I will not give a detailed account of E-P in this work. However, I think the affective classification view on offer fits well with hedonic tone theories of sensory affect or any view that takes phenomenal realism about sensory affect seriously (see Smuts, 2011). Our affective sensations are hedonically toned and this depends on how information is classified.   123  the relational approach explains these four problems before I go on to discuss my own preferred take on relationalism.       As I showed in chapter 2, physicalism can only give a circular explanation of the biological significance of pleasantness. In the olfaction case, pleasantness is what motivates us to seek out heavy odor molecules but the physicalist must say that the heavy odor molecules are the pleasantness. There is no principled way for the physicalist to claim that olfactory pleasantness is motivational. They must rely on the brute physical fact that heavy odor molecules are pleasant but this is just what we are trying to explain.         The relational view can explain the motivational characteristics of pleasantness. It is still true that pleasantness is what motivates us to seek out heavy odor molecules or whichever other properties of the odorant play a similar role. However, we can explain why we seek out these object properties by appeal to subjects and circumstances. It is the subjective and contextual properties working in conjunction with the requisite odorant properties that explain why I would be motivated to get more of a particular smell. Relativizing to subjects and circumstances blocks any circularity.       More importantly, the relational view is well suited to explain the same stimulus/different effect phenomenon. As I mentioned earlier in the chapter, this is the prime motivation for accepting the relational view. I find the skunk smell unpleasant but you don’t. Assume the same token stimulus. We can explain this in the following way. You, the subject, are from a skunk free region of the world, the context, and so do not classify the smell as unpleasant. I, the other subject, live in a region with many skunks, my context, and so classify the smell as unpleasant.  124  Relativizing to subjects and circumstances has a built-in answer to the same stimulus/different effect phenomenon. The stimulus in question is presented to different subjects in different circumstances; this is what explains the different perceptual effects.       I have shown how the relational approach is superior to physicalism. I want to say something about primitivism. I will discuss E-P more when I discuss my own view so I’ll challenge primitivism on this point a bit later. In chapter 3, I discuss the idea that primitivism leads to the consequence that every object instanties every primitive O-P property. This primitive property proliferation is problematic. The reason: primitive properties are intrinsic. It is problematic to claim that objects have many intrinsic properties, primitive or otherwise. The relational view has no trouble here. Relational property proliferation is not problematic because there can be as many relational properties as there are subjects and circumstances. Relational properties are not intrinsic so proliferation, even a high amount, is not an issue. The relational view is clearly superior to the two objectivist views.       I want to argue for the classification approach. Before I do, I want to clarify how this view fits in with the other two relational views. Affective dispositionalism claims that pleasantness is the disposition of X in C to be pleasant to S. Affective functionalism claims that pleasantness is the function of X to dispose itself in C to be pleasant to S. The affective classification approach says that X is pleasant to S in C iff S’s sensory affective system classifies the stimulus information from X in a particular way. As Cohen (2009) makes clear, if dispositions are functional, then the dispositional and functional views collapse into each other. If dispositions are functions then  125  clearly they play a functional role. Let’s set aside the functional approach for now and focus on the relationship between the dispositional and classification views.       The dispositional view analyzes pleasantness as a three place relation. There are objects, subjects and circumstances:  x (object property) has a disposition in c (circumstance) to be pleasant to S (subject). The classification approach fits in to this analysis. Essentially, the classification view falls on the subject side of the relation, leaving the object side to be analyzed as a functional role. The object property disposes the subject to classify the stimulus information in a particular way. There is no need to say anything else about this particular object property; it simply plays the role just specified. The classification approach emphasizes the subjective side of the relation. But this difference in emphasis does not imply a fundamental difference of theory. The subjective side of the relation is emphasized, among other things, to account for the radical variation found in pleasantness attributions.      But notice that the subjective side of the relata, as I’ve described it, is also functional. The function is to classify stimulus information in a particular way. So all I’ve done differently from other relational views is specify the subjective functional role of pleasantness. This shows that the classification approach is not fundamentally different from the affective functional view.      The classification approach emphasizes the subjective relata. The idea is this: O-P is whichever object property disposes the subjects’ sensory-affective system to classify the stimulus in a particular way. Notice that this is a version of the dispositional/functional view; the only difference is one of emphasis. Like the dispositional view, the classification approach emphasizes the disposition of an object to X. The latter view simply fills in X.   126       Like the functional view, the classification approach postulates functional roles for the object and subject. The requisite object property is whichever plays the particular role of disposing a particular classification. The requisite subject property is the role played by our sensory-affective system in classifying the stimulus information. On both sides of the relata, functional role properties are what constitute O-P. The realizers of these roles will perhaps one day be filled in by cognitive scientists. As for now, we need say no more.       But why fill out the subjective side of the relata as classification? I discuss why later in this chapter. For now, I want to spell out the view a bit more. In particular, I want to spell out the affective classification view in contrast to the color classification approach developed by Matthen (2005).        Matthen argues that what it is to be red/blue/green, etc. is constituted by how our color system classifies certain information about light. Interestingly, different classificatory systems will allow for different classifications. So a pigeons’ color system will classify a particular light stimulus differently then we will. This allows Matthen to explain inter-species variation quite well and this is a strength of the color classification approach.       But mere classification is not enough on Matthen’s picture. The color system must classify in the right way. There is a normative element to the theory. What is to classify in the right way? The classification must be tied to teleologically apt epistemic and motor actions. Roughly, colors are the properties of objects that are properly classified by the color perception system. Proper classification occurs when particular teleologically apt actions have been subserved. These  127  actions include but are not limited to: 1) re-identification of objects and properties, 2) visual scene segmentation, 3) visual search, 4) matching and differentiation of objects.       Specifying particular correct classifications works in the color case. However, it presents a potential problem case for the affective classification approach. The case is instructive in elucidating why the classification approach is best developed as a functional role theory in the case of O-P. In the color case, the correct classifications function to help the agent with various cognitive and conative tasks. Importantly, biological properties are not found in this list. Correct color classifications rely on various skills and do not rely on things like benefit or harm to the organism.To classify a stimulus as red is, in part, to help the agent discriminate this stimuli from its surround.      No doubt this aids in survival for the organism but there is no need for the organism to rely directly on benefit and harm in its correct color classifications. Classifying the apple as red does benefit the organism but the redness of the apple is a signal of benefit. Re-identification and matching/differentiation, for example, are what the color classification business it all about. That these skills aid in survival is not a sign that benefit and harm are built directly into their classifications.      But the affective classification approach has less dexterity here than Matthen’s theory. Assume that norms are required in the O-P case, that correct classifications partly constitute O-P. Following Matthen, we can also assume that teleologically apt actions are what make for a correct classification of O-P. But what types of actions would be teleologically apt in the case of O-P? There is a clear answer: biology enhancing properties; benefit and harm in particular.   128       Suppose the smell of freshly baked bread is pleasant to me. On the correct classification picture, there is a clear reason: consuming the bread will provide nutrients, it will benefit me. So the reason the bread smells pleasant is that it benefits me. Why is the smell of ammonia at high intensity unpleasant? This cleaning chemical is harmful to us at a high intensity levels. On the correct classification view, the ammonia is unpleasant because harmful.       Unfortunately, the correct classification approach will not work for O-P. The reason is this: pleasantness cannot be identified with benefit and unpleasantness cannot be identified with harm. Consider the latter point first. If a doctor is stitching up a gash in your head, this will hurt; you will feel pain. But of course, the stitches are benefitting you. So we can attribute pain to things that benefit us. Likewise, certain things might smell unpleasant yet still benefit us. Smelling salts are experienced at a high intensity and the smell is unpleasant but they are used to benefit you by alerting you enough to go through concussion tests.       The same is true in the other direction. Smells can be pleasant even though harmful. Many find the smell of gasoline to be pleasant but clearly the stimuli is harmful and potentially deadly. Given that pleasantness cannot be identified with benefit and unpleasantness with harm, beneficence cannot be what makes a classification of O-P correct or incorrect. So that leaves two options: either find another category in which to ground correct classifications or ditch the idea of there being correct classifications altogether.      The first option is not promising. There is, perhaps, one option. According to Lisa Shapiro (2014), certain early modern philosophers thought that pleasantness functioned, in part, to tell us about external world properties. On this view, O-P would function as a knowledge provider.  129  So maybe the correct sensory-affective classification of stimuli is grounded in whether or not the stimuli information provides a truthful representation of the corresponding external world property. If so, the correct classification view need not fall back on beneficiance to ground the norms of classification.      We have reason to reject this view. In chapter 2, I discussed the function of sensory perception. I showed how certain philosophers have argued that the primary function of sensory perception is to ‘guard the body.’ This is in contrast to the historically popular view that perception functions to give us epistemic access to the external world. The idea I canvassed in chapter 2 would allow this as a secondary but not the primary perceptual function. If epistemic access is not even the primary function of sensory perception, then it is very unlikely to be so for sensory-affective states. Of course, it could still be a function but then we are left claiming that it is not the primary function of O-P but the secondary one which grounds correct classifications. This is a strange result. I think the classification approach should not look to cognitive states for help in grounding normative classifications.      The right move for the classification view of O-P is clear: there is no correct classification, only mere classification. Unlike Matthen’s color classification view, in the O-P case there are no norms of classification. This allows us to block the idea that O-P must be identified with benefit and harm. If there are no correct classifications, then we can leave the subject side of the O-P relata as a functional role. Affective neuroscience and cognitive science will fill in the details of these classifications. But we can rest assured that beneficence will not be there. Sensory  130  affective classifications function to partly constitute O-P. This can be preserved without any reference to correctness conditions.   E-P           There is one final and important motivation for the classification view that I wish to discuss. It concerns E-P.  I’ve discussed how primitivists must reject the idea of E-P if they are to preserve the motivations for their view. I argued that we should reject any view that itself rejects E-P.       Unlike the primitivist, the O-P physicalist can accept E-P without undermining the theory. However, a correspondence account must be given between E-P and the requisite O-P physical property. So, for example, a physicalist could say that E-P is the property of feeling good, or a pro-attitude directed on a sensation, or a type of evaluative or imperative content. No matter their theory of E-P, they will have to give an account of how that corresponds with, if using the olfaction case, molecular weight. This is by no means an impossible task and the physicalist is in better shape here than the primitivist; the former theory is not forced to reject E-P in order to save their version of O-P.       Be that as it may, the classification view has a built in advantage: E-P is built into the theory. There is no need to reject E-P, as per primitivism, or seek a method of correspondence between the two types of pleasantness properties, as per physicalism. On the classification approach, how the sensory-affective system classifies certain dispositional properties of an object determines O-P. E-P is also determined by classification, though in a different way than O-P.  131  Once the requisite lavender olfactory properties are detected and classified as O-P pleasant, we undergo an E-P pleasant experience. E-P accompanies the classification of O-P. The pleasant O-P smell of the lavender object is accompanied by the pleasant E-P experience. Both result from the way the stimulus is classified.       Of course, object properties are not constituents of E-P like they are of O-P. In my view, E-P is a subjective property constituted by ways stimulus information has been classified (see Aydede, 2018). This is another indicator of the fact that E and O-P are indeed different properties. E-P does not need to jump an ontological objective-subjective gap, a gap between a physical and subjective property, as happens on the physicalist account. When O-P is classified as pleasant, E-P will be as well. Classification is doing the work in both property instances. E-P and O-P are different properties but both are subjective; no need to reconcile the objective and subjective.       That’s what I mean when I say that E-P is built into the classification approach. This is an advantage the physicalist just doesn’t have; the relative simplicity of the E-P explanation on the classification approach compared to the relatively more complicated physicalist explanation is a motivation to hold the former theory over the latter.       Of course, the literal projectivist can also run a classification view of E-P. That’s all fine and good. As we saw above, they must reject O-P; this is the heart of their theory. And as I argued above, rejecting O-P is the wrong way to go. The classification approach is to be preferred because the view naturally explains both O-P and E-P and how they are closely connected without being identical. We went to be able to explain how similar E-P and O-P are without  132  denying that (at least) one exists. The classification approach does this, the other three theories do not.       To sum up: O-P is the property of objects to dispose a subjects’ sensory-affective system to classify the stimulus information in a particular way. There is no need to specify the realziers of the object properties or the classification types. Both are functional roles. E-P is explained as a way of classifying that does not rely on object properties.                                                                                           133                                                                           Conclusion       I have presented four views on the nature of O-P: physicalism, primitivism, projectivism and relationalism. I have argued that the first three—the three p’s, so to speak—do not work. Relationalism is the most promising approach to O-P. In particular, I argued that the affect classification view best explains the structure of O-P.       The biggest question not addressed: what is the structure of E-P? I gestured at an answer in the subsection preceding. I am a phenomenal realist about E-P. I believe that E-P is, at least in part, a phenomenal property and not, say, a type of attitude. This is compatible with several different theories of sensory affect: the felt quality view (Bramble, 2013), the hedonic tone approach (Smuts, 2011) and the adverbial view (Aydede, 2018). Phenomenal realism about E-P is denied by attitude views (Robinson, 2006; Heathwood, 2007) and content theories (Bain, 2013; Klein, 2015; Schroeder, 2004). As such, my view is less compatible with these latter theories.       A theory of E-P opens the door to a unified theory of both pleasantness property types. I think the affect classification view is well positioned here. I sketched a nascent view in the last subsection. Roughly, a positive sensation results from a positive classification and a negative sensation from a negative classification. What counts as positive and negative classification in this case will be filled in by cognitive scientists working in this area. There is, of course, far more to say. In particular, 1) which theory best explains the phenomenally real E-P property, 2) why  134  think of E-P as phenomenally real to begin with and 3) can O-P and E-P be unified into a single theory of sensory affect (see Aydede, 2014, 2018 for an attempt)?       There is also another interesting question: what is the relationship, if any, between pleasantness and higher order pleasures? I have in mind here not so much propositional pleasures—like the pleasure I take in the fact that the Golden State Warriors have just won their third NBA championship in 4 years—but those sensed pleasures that Johnston (2001) thinks bolster our moral and aesthetic judgments—for example, the attraction I feel when viewing a Vancouver sunset from the beach, which allows me to form the judgment that the sunset is beautiful (for the idea of moral perception see Cullison, 2009). There may be no relationship but I think Johnston raises an intriguing line of inquiry even though I think his account doesn’t ultimately succeed.           I close with a final point. In my view, though relational theories generally transcend the objective-subjective divide (see Hatfield, 2009)—an attractive characteristic in my view, the affect classification view leans more strongly on the subjective side of the ledger. The act of classifying is done by sensory-affective systems responding to objective information from the environment. This is why I claimed the view is closer to projectivism than it is to either physicalism or primitivism. You can see how the dispositional view, say, remains in the middle here; the powers of objects to cause pleasant sensations is the trigger, whereas on the view I favor the act of classifying largely plays that role. In that sense, the affect classification view can be seen as the sophisticated subjectivist cousin of projectivism. Given our pre-theoretical  135  notion that pleasantness is in the eye (or nose, or tongue, or ear, or skin) of the beholder, I take my view as vindication for a more sophisticated variant of our pre-theoretical intuitions.23                                                                                                                                                                                                                                                                                                                                             23 Note that I didn’t say ‘pleasure’ is in the eye of the beholder. For example, it’s possible that matters of taste—commonly thought of as being subjective—might be more amenable to an objectivist account than a theory of sensory affect (pleasantness). For a good discussion of the concept of aesthetic taste see Shelley (2017) and for an argument against subjectivism about taste see Kivy (2016).    136                                                             Bibliography    Akins, K. (1996) “Of sensory systems and the "aboutness" of mental states”, Journal of Philosophy, 93 (7), pgs. 337-372 Andrews, K. (2014) Animal Minds; An Introduction to the Philosophy of Animal Cognition (Routledge Press; London, UK) Armstrong, D. (1968) A Materialist Theory of Mind (Routledge Press: London, UK) Averill, E.W. (2005) “Toward a Projectivist Account of Color”, The Journal of Philosophy, CII: 5, pgs. 217-234 Ayabe-Kanamura, S., Schicker, I., Laska, M., Hudson, R., Distel, H., (1998) “Differences in perception of everyday odors: a Japanese-German cross-cultural study”, Chemical Senses, 23, pgs. 31-38  Aydede, M. (2014) "How to Unify Theories of Sensory Pleasure: An Adverbialist Proposal," in Review of Philosophy and Psychology, 5(1), pgs. 119-133 Aydede, M. (2018) “A Contemporary Account of Sensory Pleasure,” in Pleasure: A History, ed. L. Shapiro (Oxford University Press: Oxford, UK)  Aydede, M. and Fulkerson, M. (2014) “Representationalism’s Headache”, Philosophical Studies, 170, pgs. 175-198 Aydede, M. and Fulkerson, M. (unpublished ms.) “Affective Qualities” Bain, D. (2013) “What Makes Pains Unpleasant”, Philosophical Studies, 166:1, pgs. 69-89  Barwich, A.S. (2014) “A Sense So Rare: Measuring Olfactory Experiences and Making a Case for a Process Perspective on Sensory Perception”, Biological Theory, 9:3, pgs. 258-268 Barwich, A.S. (2016) "Making Sense of Smell", The Philosophers' Magazine, 73, pgs. 41-47 Batty, C. (2010) “Scents and Sensibilia,” in American Philosophical Quarterly, 47, pgs. 103-118 Batty, C. (2011) “Smelling Lessons,” in Philosophical Studies, 153, pgs. 161-174 Berridge, K. (2003) “Pleasures of the Brain”, Brain and Cognition, 52, pgs. 106-128  137  Berridge, K. (2004) “Motivation Concepts in Behavioral Neuroscience”, Physiology and Behavior, 81, pgs. 179-209 Boghossian, P. and Velleman, D. (1997) “Colour as a Secondary Quality” in Readings on Color, Vol. 1: The Philosophy of Color, eds. A. Bryne and D. Hilbert. (1997), pgs. 81-103 (MIT Press: Cambridge, MA) Bramble, B. (2013) “The Distinctive Feeling Theory of Pleasure”, Philosophical Studies, 162, pgs. 201-217  Bro, R. and Smilde, A., 2014 “Principal Component Analysis”, Anal. Methods, 6, pgs. 2812-2831   Byrne, A. and Hilbert, D. (ed.) (1997) Readings on Color, volume 1: The Philosophy of Color (MIT press: Cambridge, MA) Byrne, A. and Hilbert, D (2007) “Color Primitivism”, Erkenntnis, 66, pgs. 73-105 Campbell, J. (2002) “Berkeley’s Puzzle” in Conceivability and Possibility eds. T. Gendler and J. Hawthorne, (MIT Press: Cambridge, MA) Campbell, J. (1997) “A Simple View of Color” in Readings on Color, Vol. 1: The Philosophy of Color, eds. A. Bryne and D. Hilbert. (1997), pgs. 177-190 (MIT Press: Cambridge, MA) Campbell, J. (2005) “Transparency vs. Revelation in Color Perception” Philosophical Topics, 33:1 pgs. 105-115 Campbell, J. and Cassam, C. (2014) Berkeley’s Puzzle: What Experience Teaches Us (Oxford University Press: Oxford, UK) Carruthers, P. (2018, forthcoming) “Valance and Value”, Philosophy and Phenomenological Research  Castro, J. and Seeley, W. (2014) “Olfaction, valuation, and action: reorienting perception”, Frontiers in Psychology, 5, pg. 299 Castro, J., Ramanathan, A., Chennubhotla, C., (2013) “Categorical Dimensions of Human Odor Descriptor Space Revealed by Non-Negative Matrix Factorization”, PLOS ONE 8:9, Open Access: (https://doi.org/10.1371/journal.pone.0073289), accessed on September 1, 2018 Chalmers, D. (2006) “Perception and the Fall from Eden”, in Perceptual Experience, eds. Tamar Gendler and John Hawthorne, pgs. 49-125 (Oxford University Press: Oxford: UK)  138  Chirimuuta. M. (2016) Outside Color (MIT Press: Cambridge, MA) Choi, S. and Fara, M. (2012) “Dispositions” in The Stanford Encyclopedia of Philosophy (https://plato.stanford.edu/entries/dispositions/), accessed on September 1, 2018 Clark, A. (2005) ‘Painfulness is Not a Quale” in Pain: New Essays on its Nature and the Methodology of its Study, ed. M. Aydede, pgs. 177-198 (MIT Press: Cambridge, MA) Cohen, J. (2009) The Red and the Real: An Essay on Color Ontology (Oxford University Press: Oxford, UK) Copenhaver, R. (2013) “Perception and the Language of Nature” In The Oxford Handbook of British Philosophy in the Eighteenth Century, ed. J. A. Harris, pgs. 107-127 (Oxford University Press: Oxford, UK) Crane, T. (2006) “Is There a Perceptual Relation” in Perceptual Experience, eds. Tamar Gendler and John Hawthorne, pgs. 126-146 (Oxford University Press: Oxford: UK) Cullison, A. (2009) “Moral Perception”, European Journal of Philosophy, 18:2, pgs.159-175 Cutter, B. and Tye, M. (2011) “Tracking Representationalism and the Painfullness of Pain”, Philosophical Issues, 21:1, pgs. 90-109   Cutter, B. (2018) “Paradise Regained: A Non-Reductive Realist Account of the Sensible Qualities”, Australasian Journal of Philosophy, 96:1, pgs. 38-52  Delplanque, S. Grandjean, D., Chrea, C., Aymard, L., Cayeaux, I. (2008) “Emotional processing of odors: evidence for a nonlinear relation between pleasantness and familiarity evaluations”, Chemical Senses, 33, pgs. 469-479 Doty, R. (1975) “Examination of relationships between pleasantness, intensity, and concentration of 10 odorous stimuli”, Perceptual Psychophysics, 17, pgs. 492-496   Dravnieks, A. (1985) Atlas of odor character profiles (ASTM: Philadelphia: PA) Fish, W. (2010) Philosophy of Perception: A Contemporary Introduction (Routledge: New York, NY) Fulkerson, M. (2014) The First Sense: A Philosophical Study of Human Touch, (MIT Press: Cambridge, MA)   139   Gottfried, J. (2010) “Central mechanisms of odour object perception” Nature Reviews Neuroscience, 11, pgs 628-641  Haddad. R, Medhaine, A., Roth, Y., Harel, D., Sobel, D., (2010) “Predicting Odor Pleasantness with an Electronic Nose”, PLoS Computational Biology, 6:4, Open Access: (http://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1000740), accessed on September 1, 2018 Hatfield, G. (2009) “Objectivity and Subjectivity Revisited: Color As A Psychobiological Property,” in Perception and Cognition: Essays in the Philosophy of Psychology, pgs. 281-296 (Oxford University Press: Oxford, UK) Heathwood, C. (2007) “The Reduction of Sensory Pleasure to Desire”, Philosophical Studies, 133, pgs. 23-44 Helm, B. (2002) “Felt Evaluations: A Theory of Pleasure and Pain”, American Philosophical Quarterly, 39, pgs. 13-30 Herz, R. (2008) The Scent of Desire: Discovering Our Enigmatic Sense of Smell (Harper Perennial: New York, NY)  Hilbert, D. (1987) Color and Color Perception: A Study in Anthropocentric Realism (CSLI: Stanford) Hudson, R. (1999) “From molecule to mind: the role of experience in shaping olfactory function”, J. Comp. Physiol. A, 185: 297-304 Johnston, M. (1997) “How to Speak of the Colors,” in Readings on Color, Vol. 1: The Philosophy of Color, eds. A. Bryne and D. Hilbert. (1997), pgs. 137-172, (MIT Press: Cambridge, MA) Johnston, M. (1998) “Are Manifest Qualities Response-Dependent”, The Monist, 8:1, pgs. 3-43 Johnston, M. (2001) “The Authority of Affect”, Philosophy and Phenomenological Research, LXIII: 1, pgs. 181-214   Johntson, M. (2006) “Better Than Mere Knowledge?: The Function of Sensory Awareness” in Perceptual Experience, eds. Tamar Gendler and John Hawthorne, pgs. 260-290 (Oxford University Press: Oxford: UK) Johnston, M. (unpublished ms.) “The Manifest”  Keller, A. (2016) Philosophy of Olfactory Perception (Palgrave Macmillan: London, UK)  140  Khan, R., Luk, C., Flinker, A., Aggarwal, A., Lapid, H., Sobel, N., (2007) “Predicting odor pleasantness from odorant structure: pleasantness as a reflection of the physical world”, Journal of Neuroscience, 27, pgs. 10015-10023    Kivy, P. (2015) De Gustibus: Arguing About Taste and Why We Do It (Oxford University Press: Oxford: UK) Klein, C. (2015) What the Body Commands: An Imperative Theory of Pain (MIT Press: Cambridge, MA)  Kringelbach. M. and Berridge, K. (2015) “Motivation and pleasure in the brain” in The Psychology of Desire, eds. W. Hofmann and L.F. Jordgren, pgs. 129-145 (Guilford Press: New York, NY) Korsmeyer, C. (1999) Making Sense of Taste: Food and Philosophy, (Cornell University Press: Ithaca, NY)  Lapid, H., Shushan, S., Plotkin. A., Voet, H., Roth, Y., Hummel, T., Schneidman, E., Sobel, N., (2011) “Neural activity at the human olfactory epithelium reflects olfactory perception”, Nature Neuroscience, 14:11, pgs. 1455-61 Louge H. (2016a) “Metaphysics of Color 1: Physicalist Theories of Color”, Philosophy Compass, 11: 4, pgs. 211-219 Louge, H. (2016b) “Metaphysics of Color 2: Non-Physicalist Theories of Color”, Philosophy Compass, 11: 4, pgs. 220-231  Mackie, J.L. (1976) Problems from Locke (Oxford University Press: Oxford, UK) MacPherson, F. (2011) “Introduction: Individuating the Senses”, The Senses: Classic and Contemporary Readings, pgs. 3-45 (Oxford University Press: Oxford, UK) Martinez, M. (2015) “Disgusting Smells and Imperativism”, Journal of Consciousness Studies, 22:5-6, pgs. 191-200 Matthen, M. (2005) Seeing, Doing and Knowing: A Theory of Sense Perception (Oxford University Press) Matthen. M. (2017) “Realism, Relativism, Adverbialism: How Different are they? Comments on Mazviita Chirimuuta’s Outside Color”, Philosophy and Phenomenological Research, XCV: 1, pgs. 236-243   141  McDowell, J. (1994) Mind and World (Harvard University Press: Cambridge, MA) McGinn, C. (1982) The Subjective View (Oxford University Press: Oxford, UK) McGinn, C. (1996) “Another Look at Color”, Journal of Philosophy, 93: 11, pgs. 537-53 Merleau-Ponty, M. (1945) Phenomenology of Perception (Routledge Press: London, UK) Nolan, L. (ed.) (2011) Primary and Secondary Qualities (Oxford University Press: Oxford, UK)  O’Callaghan, C. (2007) Sounds: A Philosophical Theory (Oxford University Press: Oxford, UK) Panskeep, J. and Biven, L. (2012) The Archaeology of Mind: Neural Origins Of Human Emotion (WW Norton: New York, NY) Panskeep, J. (1998) Affective Neuroscience: The Foundations of Human and Animal Emotions (Oxford University Press: Oxford, UK) Pautz, A. (unpublished ms.) “Color Eliminativism” found at (http://www.i-m.mx/apautz/adampautz/), accessed on September 1, 2018 Pitcher, G. (1970) “Pain Perception”, Philosophical Review, 79:3, p.368 Reid, T. (1764) An Inquiry Into the Human Mind (Public Domain) Rickles, S. (2013) Berkeley’s Argument for Idealism (Oxford University Press: Oxford, UK) Robinson, W. (2006) “What it’s Like to Like”, Philosophical Psychology, 19:6, pgs. 743-765 Robinson, W. (unpublished ms.) “Pleasure and Attending”  Rolls, E. (2013) Emotion and decision-making explained (Oxford University Press: Oxford, UK) Rolls, E., Kringelbach, M., de Araujo, I. (2003) “Different representations of pleasant and unpleasant odours in the human brain”, European Journal of Neuroscience, 18, pgs. 695-703    Rozin, P. (1982) “’Taste-Smell Confusions’ and the Duality of the Olfactory Sense,’ in Perception and Psychophysics, 31(4), pgs. 397 Schroeder, T. (2004) Three Faces of Desire (Oxford University Press: Oxford, UK)  Shapiro, L. (2014) “Pleasure, Pain and Sense Perception” in Routledge Companion to 18th Century Philosophy, ed. A. Garrett, pgs. 400-418 (Routledge Press: London, UK) Shelley, J. (2017) “The Concept of the Aesthetic”, Stanford Encyclopedia of Philosophy (https://plato.stanford.edu/entries/aesthetic-concept/), accessed on September 1, 2018  142  Shoemaker, S. (1997) “Phenomenal Character,” in Readings on Color, Vol. 1: The Philosophy of Color, eds. A. Bryne and D. Hilbert. pgs. 227-245, (MIT Press: Cambridge, MA) Shoemaker, S. (2000) “Introspection and Phenomenal Character”, Philosophical Topics: 28 (2) Shoemaker, S. (2006) “On the Way Things Appear” in Perceptual Experience, eds. T. Gendler and J. Hawthorne, pgs. 461-481 (Oxford University Press: Oxford: UK) Siegel, S. (2010) The Contents of Visual Experience (Oxford University Press: Oxford, UK) Simmons, A. (2015) “Perception in Early Modern Philosophy” in The Oxford Handbook of Philosophy of Perception, ed. M. Matthen, pgs. 81-99 (Oxford University Press: Oxford: UK)  Smith, A.D. (2002) The Problem of Perception (Harvard University Press: Cambridge, MA)  Smith, B. (2007) “The Objectivity of Tastes and Tasting” in Questions of Taste: The Philosophy of Wine, Barry Smith (ed.), Oxford University Press, pgs. 41-77  Smith, B. (2013) “Taste, Philosophical Perspectives” in The Encyclopedia of Mind, Hal Pashler (ed.), Sage Publications   Smith, B. (2015) “The Chemical Senses”, in The Oxford Handbook of Philosophy of Perception, ed. M. Matthen, pgs. 314-352 (Oxford University Press: Oxford: UK) Smuts, A. (2011) “The Feels Good Theory of Pleasure”, Philosophical Studies, 155, pgs. 241-265  Stevenson, R. and Wilson, D. (2007) “Odour Perception: an object-recognition approach”, Perception, 36, pgs. 1821-1833 Stroud, B. (1997) Hume (Routledge Press: London, UK) Stroud, B. (2000) The Quest for Reality (Oxford University Press: Oxford: UK) Tappolet, C. (2016) Emotions, Values, and Agency (Oxford University Press: Oxford, UK) Thompson, E. (1995) Colour Vision (Routledge Press: London, UK) Tye, M. (1997) “A Representational theory of Pains and their Phenomenal Character” in The Nature of Consciousness: Philosophical Debates, eds. N. Block, O. Flanagan, G. Guzeldere, pgs. 329-340 (MIT Press: Cambridge, MA)   Watkins, M. (2005) “Seeing Red: the Metaphysics of Color without the Physics”, Australasian Journal of Philosophy, 83: 1, pgs.33-52 Watkins, M. (2010) “A Posteriori Primitivism”, Philosophical Studies, 150, pgs.123-137   143  Young, B. (2016) “Smelling Matter”, Philosophical Psychology, 29:4, pgs. 1-18 Yeshurun, Y. and Sobel, N. (2010) “An Order is Not Worth a Thousand Words: From Multidimensional Odors to Unidimensional Odor Objects”, Annual Review of Psychology, 61, pgs. 219-241 Zahavi, D. (2005) Subjectivity and Selfhood: Investigating the first-person perspective (MIT Press: Boston, MA)  Zarzo, M. (2008) “Psychologic dimensions in the perception of everyday odors: pleasantness and edibility”, Journal of Sensory Studies, 23, pgs. 354-376  Zarzo, M. (2011) “Hedonic Judgments of Chemical Compounds Are Correlated with Molecular Size”, Sensors, 11, pgs. 3667-3686                                                                                                                                           


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