University of Southern California Dissertations and Theses

A perceptual model of evaluative knowledge

(USC Thesis Other)

A perceptual model of evaluative knowledge

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content 1 A Perceptual Model of Evaluative Knowledge by Michael Milona A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (PHILOSOPHY) August 2016 2 Table of Contents Acknowledgments 3 Introduction 5 1. On The Epistemological Significance of Value Perception 14 2. Intellect vs. Affect: Finding Leverage in an Old Debate 45 3. Toward the Best Version of Sentimental Perceptualism 86 4. Answering the Content Challenge, Part 1 99 5. Answering the Content Challenge, Part 2 130 6. Perceptualism and the Puzzle of Armchair Evaluative Knowledge 149 7. Taking the Perceptual Analogy Seriously 165 References 194 3 Acknowledgments When I decided to attend USC, I believed that I had made a good choice, but I did not yet realize just how good. I cannot imagine a faculty that collectively devotes more time and effort to its graduate students. In particular, I am very grateful to the members of my dissertation committee: Stephen Finlay, Richard Johns, Janet Levin, Mark Schroeder, and Ralph Wedgwood. Their sage advice and penetrating criticisms (even the ones that kept me up at night!) have been invaluable. Most of all, I would like to thank Mark Schroeder, whose dedication as chair has surely gone beyond anything I deserve. Time and again, Mark has demonstrated his incredible philosophical perception, helping me to see the proverbial forest through the trees. I also owe a debt of gratitude to many others at USC and elsewhere who have given feedback on various portions of my dissertation: Greg Ackerman, Robert Cowan, Nathan Howard, Nicholas Laskowski, Alida Liberman, Tanya Kostochka, Caleb Perl, Abelard Podgorski, and Preston Werner. Additionally, I have had the good fortune to present much of the material to helpful audiences at the University of Glasgow, the University of Colorado at Boulder (on two occasions), the University of Edinburgh, and USC (on numerous occasions). Finally, I would like to thank my parents. I can still remember telling them on the phone that I wanted to pursue a PhD in philosophy, and despite not really having a clue what philosophy was, they provided the love, support, and encouragement that I needed to 4 pursue my passion. Thanks also to Lauren, whose faith in me (and equal passion for tasty food and scotch) has been crucial to completing this dissertation. 5 Introduction We know a great many things about what is right or wrong, good or bad. For example, we know that torturing animals is wrong, that helping the needy is good, and that breaking promises is bad. We also know particular truths such as that Sasha was right to help the elderly gentleman cross the street and that Bush made a bad decision to invade Iraq. Skeptics, of course, will question whether we really have such knowledge, but I take it for granted that we do. The key question for me is how we get knowledge of what is right or wrong, good or bad (henceforth, value). My plan is to explore the prospects for the view that evaluative knowledge is rooted in something closely analogous to the way that empirical knowledge is rooted in perception. This statement of perceptualism, as I call it, can seem radical, leading many skeptics to dismiss it straightaway. But I believe that the best way to evaluate this claim is to understand the most promising way that it can be developed. The main tasks for this dissertation are (i) to get the most promising perceptualist theory on the table, and (ii) to explain how its proponents should address key objections. 1 What Perceptualism Is It will be helpful to state in more detail what makes a value epistemology perceptualist. The core commitment is as follows: just as an ordinary perceptual experience (e.g., of a dog running) provides some basic justification for believing that the perceptual experience’s content is true, so too does an evaluative experience (e.g., of an act being wrong) provide some basic justification for believing that the evaluative 6 experience’s content is true. 1 For perceptualists, to say that a perceptual or evaluative experience represents some content, C, entails that accurately describing the experience’s phenomenology, or what the experience is like, requires reference to C. 2 To illustrate, a person can describe her experience of a green cube only by describing a green cube. Similarly, perceptualists insist that we have evaluative experiences that can only be described with evaluative language. To avoid confusion with ‘representation’, which is used in many ways by philosophers, contemporary perceptualists often talk of perceptual and evaluative experiences as presenting. 3 ‘Presentation’ also captures the idea that in perceptual and evaluative experience we present things as being some way (in contrast with say, visual imagining, which involves no such presentation). The central epistemological component of perceptualism is that value experiences are our source of basic justification (alternatively, immediate or independent justification). The definition that I work with is the following: a state, E, is a basic source of justification for some content, C, just in case E does not confer justification for believing C because some other state confers justification for believing C. 4 States which 1 The list of philosophers who accept that there are evaluative experiences which are epistemically analogous to perceptual experiences is incredibly large. Recent proponents include, among others, Stampe [1987], Greco [2000], Johnston [2001], McGrath [2005], Huemer [2005], Oddie [2005], Tenenbaum [2007], Chappell [2008], Döring [2007], Brewer [2009], Cullison [2009], McBrayer [2010a] and [2010b], Church [2013], Roberts [2013], Kauppinen [2013], Stratton-Lake [2015], Werner [2015], and Chudnoff [forthcoming]. 2 Most value epistemologists who appeal to the perceptual analogy accept this phenomenological claim, but some arguably do not (e.g., Stampe [1987]). In this dissertation, the phenomenological claim only plays a central role in Chapter 2, but as the reader will see, it still may not be essential to the argument of that chapter. 3 Talk of presentation is increasingly common in philosophy of mind. See Silins [2015: 24]. For examples of perceptualists who explicitly use presentation-talk, see Johnston [2001] and Kauppinen [2013]. Huemer [2005] talks of a sui generis seeming state to get at basically the same point. Peter Goldie [2000], who accepts the psychological framework of sentimental perceptualism (though perhaps not its epistemology), talks of emotions as involving a feeling-towards component, i.e., a feeling which can only be described by describing the world in a value-laden way. 4 Here I follow Robert Cowan [2015a]. 7 are sources of basic justification may be able to produce true beliefs which are basic knowledge, but more is required for knowledge than justified true belief. 5 For example, if due to a brain lesion a person veridically hallucinates a red apple and believes (justifiably) on the basis of that experience that there really is a red apple, then it would not follow that the person knows that there is a red apple, for it is an accident that they got it right. 6 So even if perceptualists have an attractive model of evaluative experience and justification, they still owe us an account of how such experiences can ground basic knowledge. 2 Three Perceptualist Theories Before we can gauge the prospects for perceptualism, we need to understand what these evaluative perceptual-like experiences are even supposed to be. How perceptualists answer this question makes a big difference, as we shall see. Historically, theories of value experience have been either rationalist or sentimentalist in nature. Rationalist proponents of perceptualism invoke intellectual experiences (intellectual perceptualism). 7 To illustrate, here is a representative remark from Richard Price: As bodily sight discovers to us visible objects; so does the understanding, (the eye of the mind, and infinitely more penetrating) discover to us intelligible objects… 5 Gettier [1963] famously made this point. 6 Armstrong [1973] 7 Many of the great Early Modern rationalists (e.g., Richard Price, John Balguy, and Ralph Cudworth) can be read as intellectual perceptualists, or so I would argue. For a major contemporary defense of intellectual perceptualism, see Huemer [2005]. 8 [W]e have a power immediately perceiving right and wrong: the point I am now to endeavor to prove, is, that this power is the Understanding… [1991, 141 – 42] In contrast, sentimentalist proponents invoke affective experiences (sentimental perceptualism). 8 Here is a representative remark from Francis Hutcheson: That some actions have to men an immediate goodness; or, that by a superior sense, which I call a moral one, we approve the actions of others, and perceive them to be their perfection and dignity, and are determined to love the agent; a like perception we have in reflecting on such actions of our own, without any view of natural advantage from them. [1991: 263] The core idea of the sentimental perceptualist theory is that some subset of our emotions and/or desires ground our evaluative knowledge. Recent years have seen the rise of a new model of value experience, one which insists that value experiences are not merely analogues of perceptual experiences. This third model of value experience, which I call the high-level theory, says that at least some ordinary perceptual experiences can have evaluative content. 9 By ‘ordinary’ I mean to denote the traditional five senses. High-levelists invariably hold that we can see values, but some also allow that we can hear and touch values, too. In sum, these are the three models of perceptualism that I will appraise: (i) high- level perceptualism, (ii) intellectual perceptualism, and (iii) sentimental perceptualism. 10 The aim of the first two chapters is to make the case that sentimental perceptualism offers 8 Many of the great Early Modern sentimentalists (e.g., Anthony Ashley Cooper (Shaftesbury), Francis Hutcheson, and David Hume) defend sentimental perceptualism. For a major contemporary defense of sentimental perceptualism, see Oddie [2005]. 9 Proponents of the high-level theory include, among others, Audi [2009] and [2013], Cullison [2009], McBrayer [2010a] and [2010b], and Werner [2014]. 10 In Chapter 2, I define intellectual perceptualism in such a way that any theory on which value experiences are not affective counts as intellectual. This has a useful dialectical consequence: it ensures that no form of perceptualism is left out of the taxonomy. 9 the most promising model of what our evaluative experiences are. In Chapters 3 through 7 I develop sentimental perceptualism by answering some of the key objections to the theory. 3 The Plan I will briefly outline the aims of each chapter. The first sets aside the high-level theory of value perception. Most theorists who evaluate the high-level theory focus on whether there are such experiences, and if there are, whether they can be a basic source of justification. I take a different tack, arguing, perhaps surprisingly, that it just is not an important value-epistemological question whether high-level value perception is possible. In our quest to discover the best form of a perceptualist theory, we need not bother asking whether there is high-level value perception. The task for Chapter 2 is to adjudicate between intellectual and sentimental perceptualism. The debate between these two theories is very old, and so it comes as no great surprise that contemporary proponents/opponents of both approaches typically tinker with familiar arguments in an effort to get leverage one way or the other. For example, much of the discussion continues to be about the nature of the connection between evaluative judgment and motivation (and which kinds of theories can explain it) or the phenomenology of evaluative experience. In contrast, I introduce a novel way of getting leverage in this debate. The perceptualist’s hand will be forced either in the direction of intellectual or sentimental perceptualism once she decides between two views about the modal status of our basic evaluative knowledge. My claim is that intellectual perceptualism pairs best with a view on which our basic evaluative knowledge must be 10 of necessary evaluative truths (e.g., that we have a prima facie duty to keep promises), while sentimental perceptualism pairs best with a view on which our basic evaluative knowledge may be of contingencies (e.g., Cindy ought to attend her son’s ballgame). I then argue that much of our basic knowledge is of contingencies. The conclusion of this line of thought is that perceptualists ought to be sentimentalists. But sentimental perceptualism faces important challenges, both to its psychological and normative dimensions. I start with the main psychological challenge. Sentimental perceptualism, recall, says that affect involves perceptual-like representations of value, but some critics doubt its defenders can explain how such states could have come to have such content. 11 This is a serious challenge, one which needs to be confronted head on. I begin my response in Chapter 3 by arguing for a conditional thesis: if affective experiences (or underlying affective systems) have the biological function of standing in causal-covariational relations with value, then this provides great evidence that affect is about value. The next two chapters make the case for the antecedent of this conditional. In Chapter 4, I argue that it is reasonable to think that certain affective experiences, namely desires, have the biological function of covarying with value. 12 But not just any kind of value. 13 It turns out to be difficult to maintain that desires have the function of covarying with goodness, which means that sentimental perceptualists should be wary of popular “guise of the good” theories of desire. But desires do plausibly have the function of covarying with practical reasons (namely, those of the agent having the 11 See Schroeder [2008] and Schafer [2013]. 12 The defense of the functional claim will not, however, be complete until the end of Chapter 5. 13 It may be helpful to reiterate that I use ‘value’ broadly to refer not only to narrowly evaluative notions such as goodness but also normative properties/relations which concern action. 11 experience). 14 These observations about desire should make us optimistic that the content challenge can be addressed for other types of affect (e.g., anger, fear, joy), but I leave detailed analyses for a later date. (An interesting question going forward is whether sentimentalists should think that all forms of affect represent reasons.) In sum, then, Chapter 4 provides the beginnings of an attractive response to the content challenge; but sentimental perceptualist still need to make it plausible that desires are caused by reasons. In Chapter 5, I argue (leaning heavily on some arguments of Graham Oddie’s) that desires are often caused by reasons. Additionally, I argue that we should be optimistic that desires, and affect more generally, have the function of detecting value (though admittedly the matter ultimately depends on difficult and unsettled empirical questions). But before concluding that sentimental perceptualists have a reasonable response to the content challenge, I consider two important objections. The first objection targets the defense of the causal claim, in particular. Ralph Wedgwood argues that affective experiences are causally sensitive not to particular values but rather to other mental states (paradigmatically perceptions and beliefs). In this way affect is unlike perception, since perceptions are causally responsive to the empirical properties that they represent. 15 The second objection is a broader attack on my response to the content challenge; it aims to show how all of the data I have presented in responding to the content challenge can be interpreted in a way that is not friendly to the sentimental perceptualist’s psychological framework. Neither of these objections work, but in answering them we will gain a deeper appreciation for the sentimental perceptualism’s psychology. 14 The view that desires involve (some kind of) representation of a reason is defended by Scanlon [1998], Schroeder [2007a], and Gregory [2013]. 15 Wedgwood [2001] 12 Chapter 6 takes on what is arguably the biggest epistemological challenge to sentimental perceptualism. The challenge is to explain how we get evaluative knowledge by mere reflection, or “from the armchair.” We can recognize the force of this challenge by noticing that (i) it seems as if we can get evaluative knowledge by mere reflection, but (ii) it does not seem as if we can get empirical knowledge in that same way. 16 The perceptual analogy thus appears to break down in any important way. Furthermore, sentimental perceptualists cannot simply borrow familiar rationalist (or, more narrowly, intellectual perceptualist) stories for how we get armchair evaluative knowledge, since such stories are only good for explaining knowledge of necessities. Affective experiences, however (as I argue in Chapter 2), are suited for giving us knowledge of contingencies. I offer a novel explanation of how we get evaluative knowledge of contingencies from the armchair, one that draws on key observations made in the response to the content challenge. The explanation I offer shows how we can actually use causal relations between affect and value to explain how we get evaluative knowledge by mere reflection, though the causal relations do not obtain between the armchair responses and the values that they are about. In Chapter 7, I take on some recent objections to sentimental perceptualism developed by András Szigeti [2013] and Michael Brady [2013]. These objections are meant to show that while affect is often epistemically useful in helping us to recognize value, affect cannot be the foundation of our evaluative knowledge. But the objections, if they are any good, can at least initially seem to apply to ordinary perception and the empirical domain. By noticing how analogous objections would fail in the case of 16 As we shall see, however, this asymmetry is in no way essential to the challenge. 13 perception, we see the sorts of commitments we would expect sentimental perceptualists to adopt. I list these commitments. Perceptualism has, I think, been a seductive theory throughout the history of moral philosophy in large part because it seeks to demystify an initially very puzzling species of knowledge, evaluative knowledge, by showing that it is really not so different from a comparatively much less puzzling species of knowledge, ordinary empirical knowledge. An attractive perceptualist theory would provide us with all that we could ever hope for from a theory of evaluative knowledge. It is the hope for the ideal which motivates this project. The seven chapters to come aim to reveal and develop the best form of a perceptualist theory. 14 Chapter 1: On The Epistemological Significance of Value Perception 0 Introduction This chapter critically evaluates what I referred to above as the high-level theory of value perception, an increasingly popular theory (or, rather, family of theories). A high-level theory, recall, says that at least some ordinary perceptual experiences – whereby ‘ordinary’ denotes the traditional five-senses – can in certain instances have veridical evaluative content. 17 My central claim is that value epistemologists need not take sides in difficult debates about high-level value perception, for the question of high level value perception is not of epistemological significance. 18 But I close the chapter by noting that it may matter a great deal for value epistemology whether certain other theories of value perception are true. This chapter begins with a description of an ambitious theory of value perception, which plays an important dialectical role in my central argument. An ambitious theory says (roughly) that any justified belief about whether something is valuable (e.g., good, bad, right, wrong) epistemically depends on value perceptual experiences, however exactly we understand value perceptual experiences to be. In the next two sections, I 17 Readers of this dissertation will notice that I regularly shift between ‘perception’ and ‘perceptual experience’. The former, of course, is factive; I can only perceive that p if it really is the case that p. There can thus only be value perception if there are values. In this dissertation, I simply assume there are. More specifically, I assume that if there are value perceptual experiences, then there are value perceptions. One who disagrees could substitute any instance of ‘perception’ with ‘perceptual experience’. 18 Numerous philosophers hold that high-level value perception is possible. My dispute is not with the view itself but, rather, with its significance for value epistemology. (For a different way of attacking the import of high-level value perception, see Robert Cowan [2015a].) For a sampling of philosophers who not only defend high-level value perception but also seem to assume that the debate about high-level value perception matters for value epistemology, see Greco [2000], Audi [2010] and [2013], Cullison [2009], McBrayer [2010a] and [2010b], and Werner [2015]. Matters are a bit complicated when it comes to Cullison, however (see section 4). 15 describe the high-level theory in more detail and then explain why it is not a plausible basis for an ambitious theory. But as I go on to note, it still appears to matter quite a bit for value epistemology whether there is such a thing as high-level value perception. Appearances, however, are misleading. Reflection on how one of the alternative ways of justifying evaluative beliefs – a way that the defender of high-level value perception in particular needs to allow – relates to high-level value perception leads us to the conclusion that the question of whether there is any high-level value perception is not so significant, after all. 1 Ambitious Theories This section sketches the outlines of what I call an ambitious theory of value perception. The crucial point that I want to make is that even an ambitious theory should allow for some evaluative knowledge not grounded in value perception, namely what I call below non-substantive evaluative knowledge. Placing such a limitation not only helps to avoid certain objections, but is also natural for the theory. Let me explain. Some philosophers have argued that it is a problem or theories of value perception that they cannot explain the full range of evaluative knowledge. 19 Consider the following remark from Simon Blackburn 20 : Literal talk of perception runs into many problems. One is that the ethical very commonly, and given its function in guiding choice, even typically, 19 Pekka Väyrynen [2008] suggests that defenders of perceptual views often seem to commit themselves to the bold claim that all of our evaluative knowledge is grounded in value perception, or at least the claim that no ethical knowledge is a priori. Many of the “Cornell Realists,” for instance, seem to suggest that all of our evaluative knowledge is acquired by observation, in a way analogous to how we acquire knowledge in science. See especially Richard Boyd [1988]. I won’t engage in the lengthy exegetical task to try to pin the bold view on these thinkers, however. 20 For a similar sentiment (although about a different theory of value perception not of immediate concern here), see Michael Smith [1994: 22]. 16 concerns imagined or described situations, not perceived ones. We reach ethical verdicts about the behavior of described agents or actions in the light of general standards. And it is stretching things to see these general standards as perceptually formed or maintained. Do I see that ingratitude is base only on occasions when I see an example of ingratitude? How can I be sure of the generalization to examples that I did not see (I could not do that for color, for instance. Absent pillar-boxes may be a different color from present ones; only an inductive step allows us to guess at whether they are). Or, do I see the timeless connection—but how? Do I have an antenna for detecting timeless property-to-value connections? [1988: 364 – 65] 21 There is a lot going on in this quote, but I simply want to highlight the thought that a perceptualist view can’t be squared with our knowledge of supervenience. 22 Blackburn is baffled by how we could ever (literally) perceive a supervenience relation. Do we generalize from cases, or do we (as Blackburn mockingly suggests) have an antenna for detecting eternal property-to-value connections? The response should just be that perception is not the basis for our knowledge of the supervenience of evaluative properties on non-evaluative properties, which seems to be a non-substantive, or conceptual, evaluative truth. 23 Another candidate example of a non-substantive true is the transitivity of value: if x is better than y, and y is better than z, then x is better than z. It is presumably acceptable to have different standards for 21 After listing these questions which he takes to indicate problems for the target theories, Blackburn [1988: 365] says, “Perhaps these questions can be brushed aside.” I’ll point out in a moment that they clearly can be. That said, my best guess is that Blackburn is lead to pose these “challenging” questions because defenders of value perception often seem to be defending the view that all of our evaluative knowledge can be grounded in value perceptions. 22 I believe that Blackburn is referring to a supervenience relation when he talks about “timeless connections,” but if he really means to refer just to a temporal relation then my point in this section could be reframed accordingly. 23 Some philosophers argue that conceptual truths can be substantive. See Terence Cuneo and Russ Schafer-Landau [2014]. Although I think it is a mistake to allow for non-substantive conceptual truths, nothing important turns on this disagreement here. However, allowing substantive conceptual truths leaves less room for value perception to play an important role, unless we allow that value perception helps us to learn conceptual truths. 17 answering different kinds of questions. Blackburn himself makes this point. For Blackburn, a truth is conceptual just in case “we cannot imagine it otherwise; we could make nothing of a way of thought which denied it.” 24 But in both ethical and non-ethical domains, it may turn out that some truths aren’t conceptual, and so must be discovered by other means. Blackburn tells us: In particular in the moral case it seems conceptually or logically necessary that if two things share a total basis of natural properties, then they have the same moral qualities. But it does not seem a matter of conceptual or logical necessity that any given total natural state of a thing gives it some particular moral quality. For to tell which moral quality results from a given natural state means using standards whose correctness cannot be shown by conceptual means alone. It means moralizing… [1984: 184] So some ethical questions can be settled by conceptual standards alone, while others need to be settled by other standards. Blackburn places an unreasonable constraint on the ethicist who has value perception playing a central role in value epistemology; we would never place an analogous constraint on paradigmatic forms of perception, as if visual perception couldn’t play a central role in acquiring empirical knowledge unless it informed us of certain conceptual truths. 25 I propose, then, that we can usefully define an ambitious theory of value perception as a theory that accepts the following: for any value property, V, perceptions of value are epistemically indispensable for knowing that something (an object, way of acting, particular action, etc.) is V, assuming the something in question is not V by 24 See Blackburn [1984: 217]. Alternative ways of characterizing conceptual knowledge are available, and I won’t try to adjudicate between them here. See, for instance, Boghossian [1996]. 25 That said, as I mention above (n. 21), Blackburn remarks at the end of his list of challenging questions that the defender of value perception may be able to set the questions aside. Blackburn may think his questions are only a problem if his targets accept a very bold thesis about the extent of our evaluative knowledge that value perception can explain. The text is unclear on this point. 18 definition. 26, 27 A natural thought for a defender of this ambitious view to have is that without value perceptions, we wouldn’t have any input to begin to form justified (substantive) ethical beliefs, for even if there were valuable things, nothing would seem good or bad, right or wrong. For such an ambitious theorist, even the question of whether an agent’s perceptual faculties are working well is probably going to be a question for value perception, unless one believes such a question can be answered conceptually. 28 2 High-Level Value Perception An ambitious theory of value perception would be of great interest, since it would provide an answer to the question of how we know substantive truths about value. This is a question with relevance not only to the epistemology of value but also the metaphysics of value. After all, if we can find no plausible account for how we know about value properties, this is a step toward calling into question whether there really are any. 29 An unambitious theory may still be interesting, of course, but even if an unambitious theory were defensible, we would still be left the task of defending some other account (or accounts) of what grounds the other substantive evaluative knowledge we presume ourselves to have. And, furthermore, there may also be a desire for a unified 26 I do not require that the perceptual experiences be of an object’s being V. An ambitious theorist can allow that we need value experiences to know about the presence of any value (goodness, badness, rightness, etc.), even though we only have perceptual experiences of certain values. Suppose, for instance, that we can perceive goodness but not rightness. If the concept of rightness is analyzed in terms of that of goodness, then value perceptions may still be essential for knowing what is right, even though we never perceptually experience anything as being right. See my [forthcoming b]. 27 Testimony, for instance, is plausibly a way of justifying evaluative beliefs, too, but testimony is not plausibly an epistemically independent route. 28 But this is not a bad result, for maters are much the same with ordinary perception. We can only learn when our perceptual faculties are function well by relying on perception. For more on this, see Chapter 7 and my [unpublished b]. 29 Mackie [1977] offers an epistemological objection to moral realism. 19 theory of what grounds substantive evaluative knowledge, in which case an ambitious theory would be a desideratum (though perhaps one we could eventually be talked out of). Before we can consider whether any theory of value perception can be ambitious, however, we need to know more about the theories on offer. For this chapter, I focus on high-level value perception, which in recent years has been growing in popularity. This section explains what a high-level theory of value perception is. The ensuing sections explain why such theories not only fail to be ambitious but fail to be interesting at all. It is uncontroversial that familiar types of perceptual experience can have low- level content. I take the content of any perceptual experience to be the accuracy conditions of an experience that are conveyed (or presented) to the subject. 30 For example, a visual experience as of a red sailboat in the distance is accurate just in case there is a red sailboat in the distance; and the phenomenal character of the visual experience consists (at least partly) in those accuracy conditions appearing to observer to be the case (i.e., being conveyed to her). Low-level visual content includes (inter alia) color and shape. (Or, in the case of audition, low-level content includes pitch and tone. But to keep things simple, I stick with vision.) While some theorists contend we can only perceive such low-level content, others are drawn to more liberal views. 31 At the least, we often talk as if we can perceive other things. For example, we often speak of seeing that there is an apple, that one event caused another, etc. Following Nicholas Silins, I define high-level (visual) perception as the 30 On this way of thinking about content, see Suzanna Siegel [2010: 28]. 31 Those who reject the possibility of any high-level perception include Fred Dretske [1995] and Austin Clarke [2000]. For an outline of different possible views, see Siegel [2010]. Siegel herself defends a more liberal account of perception. 20 perception of content other than shape, color, and location (low-level content). 32 The defender of high-level perception claims that familiar kinds of perceptual experience can have high-level content. The defender of high-level value perception defends this claim for values in particular. Finally, I use the words ‘familiar’ and ‘ordinary’ to denote the traditional five-senses, i.e., vision, audition, gustation, tactility, and olfaction. 33 The high- level view, as I discuss it, is a view about those familiar sense modalities. 34 If we drop the word ‘familiar’ (or some word used in the way I have defined ‘familiar’) from the definition of high-level value perception, then in our discussion of theories of value perception, we risk collapsing the distinction between importantly different kinds of views. 35 The plan for getting a better handle on what high-level theories of value perception essentially are is by way of considering an exemplar of such a view. 36 Robert Audi (in his recent Moral Perception) develops the most detailed account of high-level value perception of which I am aware, and so the aim is to take advantage of his extensive 32 See Nicholas Silins [2013]. One might worry about such a definition by list. The reason for defining low-level content in this way is, at least in part, to denote the problem-space in a theory-neutral way. More illuminating definitions are on offer, but those who defend such definitions, as far as I am aware, always begin with the standard list definitions. See Alva Noë [2009] and Cowan [2015a]. 33 There are many other types of perception. For example, kinesthetic perception is our sense of the movement of our bodies. I set these other forms of perception aside. 34 If we allow for such high-level perceptual content, then there will plausibly be cases in which our experiences in one sense-modality affect our experiences in another modality. My auditory experiences may affect what I see, e.g., I see that a person is saying such-and-such, and that experience is intimately related somehow to my auditory experience of the sounds of the speech. In any case, for the purposes of this chapter, I do not try to answer whether or not such experiences are irreducibly intermodal. Nothing important for my arguments turns on the possibility of intermodal experiences. For more discussion on these issues, see Siegel [2010: 24 – 26]. 35 If we define the high-level theory of value perception as the theory that we can have perceptual experiences with evaluative content, then we risk counting the theory that emotions have evaluative content as a theory of high-level value perception. But the view that emotions are perceptual experiences that can have evaluative content is prima facie very different from the view that some visual experiences can have evaluative content. 36 One can of course define ‘high-level value perception’ however one likes. It’s a technical term, after all. My aim is to come to define it so that it picks out a view worthy of philosophical attention, if only for the reason that it has been so often discussed and criticized in the recent literature. 21 treatment of the topic to help in getting a handle on the target view. The criticisms I develop in the ensuing sections turn not on any idiosyncrasies of Audi’s position but, instead, only on the features that make his view, and indeed any view, a high-level theory of value perception. To begin, Audi [2013: 37 – 8] points out that a thought inimical to the possibility of value perception is the thought that all visual perception has to be cartographic. Cartographic perception involves a mapping from phenomenal properties (low level properties like shape and color) to properties perceived. Audi’s example of cartographic perception involves going from the impression of four squares to the property of being divided into four squares. It is not plausible, Audi argues, that value perception could work on a cartographic model, since there is no mapping of phenomenal properties to evaluative properties. But Audi, at least, believes this mapmaking model of perception is limited and does not square-well with the phenomenology of actual perception. 37 We seem to have non-doxastic visual experiences of properties that we cannot easily have arrived at in a way consistent with the cartographic model. 38 For example, when a trained botanist and a lay person walk through a forest together, it is natural to expect that the surroundings will look different to the botanist, for her expertise would seem to help her to see the different kinds of foliage. 39 37 See Audi [2013: 38 – 9]. 38 I find much of what Audi says about the cartographic model underdeveloped. It perhaps would have been wise for Audi to appeal to Susanna Siegel’s much discussed ‘phenomenal contrast’ argument for why we should believe visual experiences with high-level properties is possible. Siegel uses the argument specifically to argue we can perceive natural kinds, but a parallel argument could be developed, I think, for values. Siegel’s argument, like Audi’s, appeals to phenomenology, but it is a far more robust abductive argument that attempts to head off alternative explanations of the phenomenological data. See Siegel [2010]. 39 An opponent of high-level perception will want to try to explain the different way in which the botanist and lay person experience their surroundings without appealing to differences in how things look. Susanna Siegel [2010] argues in much more detail than Audi against many of these alternative 22 If there is such a thing as visual value perception, then there must be “phenomenal sensings” of wrongdoing/badness/goodness/etc., since for anything we see (e.g., a lemon) there is something it is like to see it. 40 To get a grip on what it might be like to see value, consider an example. A patron in a bar is casually watching a married couple. 41 The husband is quite intoxicated, yet he nonetheless requests another whisky. His wife kindly asks him to stop drinking, and he responds by slapping her across the face. The observer can see the wrong act, but can he see its wrongness? Well, Audi finds that the observer’s perception of the wrong-making features may be coupled with a distinctive kind of phenomenology, which we would naturally describe as a recognition (not necessarily involving any belief) of wrongdoing; this phenomena is plausibly a phenomenal sensing of wrongdoing, exactly what we are after. 42 This is a kind of “felt connection” between the base (wrong-making) properties and the property of being wrong. Audi’s claim is that the relationship (felt connection) between the ordinary perception (of the wrong- making features) and the phenomenal sensing of an evaluative property is such that it is correct to speak of value perception. Speaking of a sensing of injustice, in particular, Audi says: The sense of injustice, then, a kind of impression of it, one might say, as based on, and as phenomenally integrated with, a suitable ordinary perception of the properties on which injustice is consequential— grounded, to use another term for the same relation—might serve as the experiential element in moral perception. explanations (although Siegel does not consider value perception, in particular), but for our purposes, we need not go into so much detail. 40 Sometimes philosophers say, which comes to the same thing, that all perception is “experiential.” 41 This example is from Audi [2013: 61 – 2]. 42 It is helpful to notice that high-level value perception is probably going to be one of the most controversial forms of high-level perception. This is because it depends on our ability to have high-level perceptual experiences of the properties (e.g., persons, mental states, causes) that (metaphysically) ground the values. 23 An important constituent in this phenomenal integration is the perceiver’s felt sense of connection between, on the one hand, the impression of, say, injustice or (on the positive side) beneficence and, on the other hand, the properties that ground the moral phenomena. [2013: 38 – 9] As I interpret Audi, it is best to separate his view into two key commitments: (i) there are a variety of different kinds of phenomenal sensings of evaluative properties, which can apparently occur independently of value perception (see section 4), and (ii) these phenomenal sensings are often integrated into the content (i.e., the accuracy conditions conveyed to the subject in her experience) of familiar perceptual experiences so as to generate a value perception. 43 It is a delicate task to characterize in greater detail how Audi understands “what it is like” to experience an evaluative property. If I understand the view correctly, phenomenal sensings of evaluative properties are manifold; in fact, phenomenal sensings of the same evaluative properties can vary. Audi [2013: 39] tells us that felt sense of connection may be partly constituted by an emotion, though it need not be. (So, we should not take the word ‘feel’ to necessarily indicate an emotional experience.) It may also be partly constituted by an intuition, though, again, it need not be. 44 Audi tries to 43 Audi is not explicit that by ‘integration’ he means integration into the content. But I think this is what he intends and it is any case hard to see how he can avoid it. Consider that Audi thinks that we can literally see wrongness. But then if he does not think the phenomenal representation of wrongness is part of the content of the visual experience, then he would have to think a visual experience as of something's being wrong could be correct (assuming a visual experience is correct just in case its content is true) even if there is no wrongness, because the wrongness is not part of the content. But then it's hard to see how talk of seeing wrongness could be anything other than metaphorical, contrary to what Audi says. Furthermore, if by ‘integration’ Audi does not mean integration into the content, then his view may actually be interpretable as a counter-hypothesis to the common way of formulating the thesis that there is high-level value perception. On this last point, see Robert Cowan [2014]. 44 See Audi [2013: 134 – 36]. For Audi, intuitions can come in a variety of forms. We can have an intuition that p, which is a kind of belief. We can also have objectual intuitions, which are direct apprehensions of concepts, properties, or relations (see Audi [2013: 85 – 88]). Both kinds of intuitions can presumably constitute phenomenal sensings of an evaluative perception. Though, as far as I can tell, Audi only mentions the possibility that an intuition that p can be part of an evaluative perception. 24 capture the rich variety of phenomenal sensings by engaging in phenomenological inquiry, drawing distinctions on the basis of an examination of different cases. 45 For instance, in one case, we may “feel” disapproval when a man deliberately spills hot liquid on his friend’s hand. Or there may be a “felt unfittingness between the deed and the context, as where we see a male and female treated unequally in a distribution of bonuses for the same work” (Audi [2013: 39 – 40]). But a “felt unfittingness” can manifest in different ways; it can be conceptual, as it often is with an adult who sees that an act is unjust, or can be non-conceptual, as it may be with a young child who is uncomfortable at seeing peers treated unequally. 46 More could be said about how Audi analyzes various different types of phenomenal sensings, but this basic characterization of Audi’s high- level theory of value perception is sufficient for my purposes. Generally speaking, then, I understand any high-level view to involve both a phenomenological thesis and an integration thesis. The first step is to identify certain evaluative experiences (I will follow Audi in speaking of sensing value properties) and the second is to argue that the experiences can be integrated into the content of familiar kinds of perceptual experiences to produce a value perception. (Integration secures the possibility of literal visual, auditory, etc. experiences of evaluative properties; the high- level view is not the far more familiar and less controversial view that we can have evaluative responses to visual, auditory, etc. experiences but that are external to the visual, auditory, etc. experiences’ content.) 45 After identifying the different ways we can have a phenomenal sensing of an evaluative property, he goes on to sketch theoretical accounts of each of those ways. It would be tangential to my purposes to spell all this out in detail. 46 See Audi [2013: 45 – 49]. Examples of non-conceptual phenomenal sensings involving children are only used to help us see the relevant phenomena. Audi would agree adults often have non-conceptual phenomenal senses, too, presumably. 25 3 The Need for an Alternative Way Defenders of high-level value perception claim that the perception of values is tied to one or more of the ordinary five senses. 47 (Audi discusses visual, tactile, and auditory value perception.) In the case of vision, we see high-level properties by seeing other, low-level properties. This means that in imagining a scenario, an agent will never literally see any values, for she is not having any visual experience. 48 The following thesis seems rather obviously true: Limits In imagining something, we do not perceive it with any of our five- senses. Of course, it is almost surely true that our ability to reflect, whether imaginatively or not, causally depends on our having certain experiences. (Even mathematical reflection likely depends on having some experiences.) But the point is that the ordinary sensory experiences are not part of what constitutes such reflection. 49 There are manifold ways in which we engage the imagination in evaluative reflection. The paradigm is when sensorily imagine various sights, sounds, smells, etc. For example, in reflecting on the moral status of torture, we may conjure up images of the sights and sounds of a suffering prisoner. 50 In other cases, we imagine in a way that 47 I assume defenders of the high-level view do not believe that there is any other kind of value perception. But they actually could; the view that desires and/or emotions often involve perceptions of values is compatible with the high-level view. At certain points, Audi seems to suggest that emotions can be a kind of value experience (perhaps similar in some ways to perceptual experience) contained within a visual experience, making that visual experience evaluative. 48 As Audi points out, visual imagination is “possible even given blindness.” And so “It is not perceptual” [2013: 9]. 49 For helpful discussion on the relationship of experience to ethical and mathematical reasoning, see Sarah McGrath [2011]. 50 There is a great deal of complexity that I am glossing over here. For example, in visually imagining a suffering prisoner, we might imagine having a visual experience of the suffering or simply imagine it 26 does not seem to be straightforwardly sensory, if even sensory at all. 51 Most obvious here are the cases when we imagine something that is not perceivable, e.g., that Athens wins the Peloponnesian War or that McCain wins the 2008 U.S. election. 52 Even if we conjure up images to facilitate these imaginings, the images are not sensory-like images of the thing imagined. The crucial point for us is that the imagination, whether sensory or not, can sometimes be a route to evaluative knowledge, and this is a problem for any high-levelist who advocates what I have labeled an ambitious theory of value perception (at least insofar as she thinks high-level value perception is the only kind of value perception that there is). Much of ethical inquiry is deliberation about what to do in the future; and when we engage in such reflection, there is nothing yet to perceive, at least in so far as our five senses are concerned. 53 Although one could hold that substantive (i.e., non-conceptual) evaluative knowledge or justification for beliefs about what is valuable can only be acquired through high-level value perception, and not by, say, imaginatively simulating how events would unfold if we performed a given action, such a view seems difficult to motivate on theory-independent grounds. 54 from some viewpoint. On this distinction, see, for instance, Richard Wollheim [1987] and Dominic Gregory [2010]. 51 I think it is a mistake to suppose that the imagination is essentially sensory. But if one thinks it is, then I am happy to call the phenomena I am about to describe ‘reflection’. Nothing turns on the disagreement. 52 See Stephen Yablo [1993] and M. Oreste Fiocco [2007]. 53 Dancy makes this point, but he seems to have in mind value perception as such. See Dancy [2010: 115 – 16]. In my view, his claim is correct if we are considering a high-level view of value perception, but it may be overstated if we are talking about desiderative or emotional evaluative perception (see section 6). 54 Richard Swinburne [2015: 620] articulates the core idea as follows: “When examples of particular situations (e.g., the trolley problem) are adduced in order to persuade us that some general moral principle is or is not true, it is quite irrelevant whether the examples are examples of an actual event or of an imagined event. What matters is what it would be right to conclude about which actions in that situation would be good or bad; whether or not the situation actually occurred is irrelevant.” 27 One intuitive way to push this thought is to notice an apparently important asymmetry between evaluative and empirical inquiry. While with the latter we rely on actual experiments, evaluative inquiry only seems to require thought experiments. 55 It is one thing to deny that any of our evaluative beliefs count as knowledge, or (less plausibly) that they are ever justified; it is another thing altogether to deny the following premise: if ordinary deliberators can gain evaluative knowledge through high-level value perception (e.g., that her helping the man across the street is good), then in most cases, they could have gained similar evaluative knowledge through imagining (e.g., that her helping the man across the street would be good). 56 A theory which denies the possibility of evaluative knowledge by mere reflection is going to be highly revisionary; and many would rightly count such a commitment as a serious strike against the theory. We can push the point a bit more with an example. Consider a toddler who has never had any high-level perceptual experiences of wrongness. To be sure, this toddler has perceptually experienced things which are wrong (e.g., lying), but she has never perceptually experienced anything as wrong. Now imagine our toddler cackling to herself as she remembers, say, pushing another toddler into the mud. But after a moment or two of imagining the scenario, as her attention is drawn in a special way to her classmate’s pain and embarrassment, she suddenly experiences her act as wrong. It seems to me that the high-levelist should allow that this is perfectly possible, and that this experience is a source of justification similar to as if she had actually visually experienced the wrongness. The high-levelist should allow that we can acquire justification for an 55 McGrath [2004] makes this point. 56 There are other ways which we can try to argue the point. Dancy, for instance, claims that when we deliver a verdict about a hypothetical case, we do not think we need to actually observe the case to “check” our judgment [2010: 115 – 16]. I will return to this thought below. 28 evaluative belief through imagining some possibility; and that justification will in many cases be independent from any evaluative perceptions. The following thesis is difficult to deny: Liberality justification for evaluative beliefs can be gained, independently of high-level value perceptions, by imaginatively reflecting (whether sensorily or non- sensorily) on cases. 57 Liberality is silent about the exact way in which we acquire evaluative justification through the imagination, and, more specifically, what high-levelists should say about this. But if what I have argued in this section is correct, we can at least say that high-level value perception is not an attractive basis for an ambitious theory of value perception. (And it will become even clearer in the next section why that is.) The next section defends a much stronger claim: building off Limits and Liberality, I claim that so long as we accept a few more attractive theses, whether there is high-level value perception is simply not an important issue at all for value epistemologists. 4. High-Level Value Perception and Imagination 4.1 Isn’t It Obvious? The argument in the previous section hardly shows that the question of whether there is high-level value perception is not important for value epistemology. A natural 57 For the central argument of this chapter, it is only important that we can gain evaluative justification independently of high-level value perception by way of either the sensory imagination or non-sensory imagination. It is not required that we be able to acquire evaluative justification in both ways. But because our ability to acquire knowledge in both ways becomes relevant when we consider a revised version of the high-level view (section 5), I simply work with the strong principle for the sake of simplicity. 29 thought is that if we discover that there are such perceptions, we will have discovered a unique and distinctive way in which we can acquire evaluative knowledge. Consider why we might care about any brand of high-level perception, whether evaluative or otherwise. 58 Many philosophers are inclined to believe that our perceptions can give us immediate (basic or non-inferential) justification for relevant beliefs. Such philosophers, often called foundationalists, have tended to focus on the perception of low-level properties like color and shape, assuming that other things, e.g., kinds, causation, emotion, are outside the reach of perception. To account for our knowledge of such phenomena, the foundationalist has to explain how we can rationally transition from our immediately justified perceptual beliefs to our beliefs about kinds, causation, emotions, etc. However, if we can literally perceive such things, then perhaps a foundationalist can plausibly argue that the range of immediately justified beliefs is much greater than initially realized. This observation might lead us to thinks that whether there is high-level value perception is of deep importance to value epistemology. But that would be a mistake. My claim is that the defender of high-level value perception should adopt three theses that jointly (along with Limits and Liberality) mute the epistemological significance of whether there is high-level value perception. (The question of high-level value perception is still of great importance for understanding the nature of perception, of course.) 58 My discussion in this paragraph of why we might care about high-level perception is inspired by Nicholas Silins’s [2013] recent discussion of the significance of high-level value perception. I say ‘inspired’ because his discussion is far more detailed and subtle than I can reproduce here. (Silins ultimately has his own reasons for wanting to temper excitement about high-level perception in general.) 30 4.2 Normative Similarity I noted in the previous section that a defender of high-level value perception should allow that in addition to acquiring substantive evaluative knowledge through perception, we can also acquire it through imaginative reflection (Liberality). I suggest that the defender of high-level value perception should probably accept a further normative similarity between imagination and value perception: Normative Similarity imaginative reflection can in principle (though perhaps with one exception to be explained below) supply just as good evidential support for a normative belief that something would be good/bad/wrong/etc. as any value perception of the relevant evaluative property can supply for the belief that the relevant thing is good/bad/wrong/etc. Just as the patron in the bar can literally see that the husband behaves badly when he slaps his wife, he could also see “in the mind’s eye” that such behavior would be bad. And, of course, seeing “in the mind’s eye” is not literally seeing; it doesn’t actually engage the visual faculties and so is not a high-level perception. But the rational support for the belief would apparently be the same. When we engage in imaginative reflection, we do not ordinarily think that we need to actually observe the case to be sure that the verdict about the imaginary case is correct. We do not think we need to use perception as a “check” on our value judgments about hypothetical cases. 59 In the statement of Normative Similarity, I noted that we may want to complicate the principle by allowing for an exception. Here is the candidate exception: 59 See Dancy [2010]. 31 Normative Similarity-Exception the “vividness” of a value perceptual experience is often revelatory of something’s evaluative significance in a way that cannot be replicated imaginatively, at least not by ordinary human agents. Consider an agent, Cindy, who imagines killing her rival, Lenny. When she imagines what it would be like, killing him appears very good, especially given all the awful things Lenny has done. Now suppose she carries out the act by driving a knife through Lenny’s heart. After carving into Lenny’s chest, well past the point of no return, the killing starts to appear differently. The apparent disvalue of her act intensifies as she sees Lenny wriggling on the floor, gasping for his last few breaths. Because the action is now presented to her in a more “vivid” way, she is able to see the inhumanity and badness of her act. Perhaps the vividness of this experience could never (for a human being, at least) be fully replicated imaginatively. The term ‘vividness’ may be interpreted in different ways. It might have to do, for instance, with our inability to accurately imagine in full detail certain actual experiences. Or, it might have to do with the greater reliability (at least sometimes) of our evaluative responses to actually perceived scenarios. So, maybe perception is in many cases better for revealing degrees of value (due to the “vividness” of perception). For my purposes, I will assume that value perception is, at least in some cases, superior to imagination for revealing degrees of value. (However, merely imagining a possibility, as opposed to actually experiencing it, occasionally provides us with the “reflective distance” needed to make a soberer and accurate evaluative judgment. This will especially be the case when we do not enjoy doing what is best or right.) But as we shall eventually see, I do not think that there is any plausible way of filling in what 32 ‘vividness’ means that salvages the epistemological significance of high-level value perception. As we will see over the next two sub-sections, granting a certain descriptive similarity between imagination and high-level value perception means that Normative Similarity-Exception cannot salvage the significance of the question of whether there is high-level value perception. This descriptive similarity helps to drive home the epistemological unimportance of high-level value perception. 60 4.3 Descriptive Similarity Audi emphasizes the parallels between value perception and moral imagination. 61 He says: [T]he exercise of moral imagination can, through vivid imaging of morally significant events, and through envisaging diverse possibilities, produce an experience significantly like a moral perception. [2013: 160] This claim is not surprising. Our imaginations allow us “replay” and “preplay” perceptual experiences; the different components of the perceptual experience will have analogues that are “in the mind’s eye.” 62 In some cases, those “analogues” turn out to be the very same kinds of phenomena involved in actual experiences. Intuitions (of whatever kind, if one believes like Audi that there are many kinds) had in response to actual cases are plausibly the same kind of thing as intuitions had in response to imaginary ones. Likewise for the emotions. Recall now the notion of a sensing, or experiencing, an evaluative 60 One can also read the next section as providing further support for Normative-Similarity. 61 Ethical imagination is not a special type of imagination; it is just imagining about ethical matters. 62 See Audi [2013: 47]. 33 property, which is necessary for the possibility of high-level value perception. (High- level value perception involves the integration of one of these sensings with a suitable ordinary perception.) For any sensing of an evaluative property involved in value perception, there will be an analogue of that sensing that could have occurred had the agent imagined the same scenario. But should we go further, claiming that the sensings had in response to imaginative cases are the very same kinds of phenomena? To make the question vivid, return to the case of the husband who wrongfully slaps his wife when she asks him not to have another whisky. As we saw, the observer’s perception of the wrongness (assuming he does see the wrongness) is partly constituted by a sensing of wrongdoing. But had the observer imagined the case, rather than perceived it, could he have had the same kind of experience (sensing) of wrongdoing? 63 I claim the following: Descriptive Similarity For any sensing of an evaluative property suitably integrated with an ordinary perception to create a value perception, the same kind of sensing of an evaluative property can occur through imagining such a case. If Descriptive Similarity is true, then it can help us see why Normative Similarity seems so natural. This is because in both perceptual and imaginary cases, it is plausibly the sensings that are doing the important justificatory work (see also section 4.4 below). But why think Descriptive Similarity is true? As I have already suggested, there are phenomenological grounds for believing this thesis. To begin, consider emotions and intuitions. Just as we can experience an 63 Of course, we should be careful about generalizing from one case, but I think matters will be much the same for other examples. 34 emotion, say, revulsion, at some actual event, we can experience revulsion when reading fiction or the newspaper, imagining possibilities, etc. Furthermore, the same kind of intuitive response might be had to a given scenario, whether actual or possible. Part of the reason such claims about emotions and intuitions go mostly unquestioned in the literature is, I believe, because the denial of these claims would fly in the face of our actual experiences. 64 My view is that matters are much the same with sensings of evaluative properties. Audi assumes without much ado that the myriad of ways we can sense an evaluative property have analogues in imagination, I am going one step beyond: the reason this seems obvious is that the very same kinds of experiences can occur when we are considering imaginary cases. 65 Although I am open to counterexamples, I cannot conceive of any way of sensing of an evaluative property that cannot be reproduced in imagination. 66 And if I have gotten the phenomenology correct, that is outstanding evidence for Descriptive Similarity, since after all, sensings of evaluative properties just are certain kinds of experiences. 64 Walton [1978] and Doggett and Egan [2012] argue that emotions in response to imaginings and fictions (or what we take to be imaginings and fictions) are different in type from our responses to actual cases (or what we take to be actual). But the reasons why they believe this do not provide an attractive model for resisting Descriptive Similarity (see n. 66 below). 65 But notice that the analogue claim may be enough for the main point. Suppose the experiences are different, e.g., in virtue of different functional roles. Well, even if that’s defensible, it may still be that what is epistemically significant about the sensings remains constant in responses to imagined cases and actual ones. In that case, I would simply have to rephrase Descriptive Similarity to be about that feature rather than the sensings. 66 There is a complication worth flagging. As I mentioned in n. 64, some philosophers (e.g., Walton [1978] and Doggett and Egan [2012]) believe that emotions in response to imaginings and fictions really are different in type than their online counterparts. The reason why they believe this has to do with the supposedly different functional and motivational profiles of our offline responses. But this does not provide a model for resisting Descriptive Similarity, at least not in a way that matters. For one, it is not clear that we should be typing phenomenal sensings by their motivational profiles. And, furthermore, even if we do, an “online” sensing and its “offline” analogue will still be the same in that they present the same evaluative properties or relations; and it is that psychological constant which matters. 35 In sum, from a phenomenological angle, it seems Audi should allow that what is distinctive about value perception is never the sensing of the evaluative property, as such, which could occur in merely imagining the same case. What is special about high-level value perception is the sensing’s integration with the ordinary perception, which secures the possibility of seeing, hearing, etc. values. 4.4 Immediacy Descriptive Similarity and Liberality (the claim that imagination is a source of justification for evaluative beliefs) make it highly attractive to adopt one last thesis: Immediacy the various sensings of evaluative properties are sources of immediate justification for relevant evaluative beliefs, whether in response to real or imagined cases. 67 To deny this claim, while accepting each of Descriptive Similarity and Liberality would generate a peculiar asymmetry in the way our evaluative beliefs are justified. 68 It is not clear what could warrant positing the asymmetry, and I suspect positing such an asymmetry would seem unattractive to a defender of high-level value perception. To get a grip on Immediacy, a bit more needs to be said about what the relevant evaluative beliefs are in imagined cases. Consider a concrete case. Sandra and Ronaldo are enjoying themselves at a party, until Ronaldo suddenly, and unexpectedly, finds 67 Arguably this thesis is entailed by normative similarity. 68 As far as I am aware, those who offer a lengthy defense of value perception hold that value perceptions are in standard cases sources of immediate justification for relevant evaluative beliefs. One can imagine a coherentist who claims value perceptions are uniquely important but denies they are a source of immediate justification. I do not believe that this is a way of salvaging this significance of high-level value perception for value epistemology, but do not argue the point here. 36 himself embroiled in an uncomfortable conversation about his political beliefs. Sandra, who is not directly involved in the conversation, is looking for a way to diffuse the situation. Sandra quickly imagines several possible courses of action: (i) stepping in to defend Ronaldo’s views, (ii) telling Ronaldo’s questioner to pipe down on account of Ronaldo’s being uncomfortable, or (iii) subtly trying to change the subject by bringing up an interesting but apolitical news story. She imagines events unfolding in different ways, depending on the course of action she has in mind. Her evaluative responses to the different imaginings give her some immediate justification for an evaluative belief about the possibility as she imagines it. 69 As it turns out, when Sandra imagines subtly changing the subject (and the subsequent unfolding of events), she experiences that choice and subsequent unfolding of events (however she imagines those things as unfolding) as especially good. She has non-immediate justification for believing that right now it would be good to attempt to subtly change the conversation insofar if she is justified in believing that her situation is similar (in non-normative respects), and would unfold in similar ways, to the situation she just imagined. There is much more that would need to be said to give a complete account of how we get justification for evaluative beliefs through imaginative reflection, but these brief remarks should suffice to illustrate the only point needed for my purposes: the high-levelist should allow that evaluative experiences had in response to imagined scenarios supply us with immediate justification for relevant evaluative beliefs. 69 Our evaluative responses to imaginings that are vague and undetailed will presumably not be a particularly good source of justification for evaluative beliefs, although the high-levelist should allow that even in those cases, we get some justification. Similarly, we may have high-level value perceptual experiences on the basis of very limited information. Imagine an agent having a high-level visual experience as of S’s taking money out of the wallet’s being wrong. Well, if for all the agent knows the wallet is S’s, then this experience will not be a great source of justification for believing S acts wrongly, even though it presumably provides some evidence (which is sure to be defeated). 37 4.5 Summing Things Up We are now in a position to recognize why accepting the three theses I have put forward makes the question of whether there is high-level value perception unimportant for value epistemology. If the defender of high-level value perception accepts each thesis, then she is committed to the following: whether or not sensings of value are ever suitably integrated with ordinary perception to generate a high-level value perception, those sensings (or experiences) of value are already a source of immediate justification that is just as good. And, furthermore, although Normative Significance-Exception seemed to leave open the possibility that value perception is a crucially important way of gaining insight about something’s degree of value, this is not so. The myriad of ways in which we might sense value are already equipped to supply the same insights, whether or not they are integrated with an ordinary perception. To recognize this, return to Cindy’s murder of Lenny. Consider Cindy’s imagining murdering Lenny. As her imagination of the deed becomes more vivid, she is poised to gain more insight into the degrees of the act’s disvalue. But actually carrying out the act, and so seeing, hearing, and feeling it, supplies her with maximal vividness, which for ordinary humans is only going to be achievable by observing the actual deed. The actual experience may make possible greater insight, but the possibility of the greater insight is not dependent on the phenomenon of high-level value perception; it is made possible, rather, by the degree of vividness with which the bad-making properties are presented (in imagination or perception). Defenders of the high-level view are forced to defend difficult, controversial theses in the philosophy of perception. It is very important, for example, that all visual 38 perception is not “cartographic,” that it be possible for sensings of values to be suitably integrated with ordinary perceptions, and so on. High-level value perception, moreover, is plausibly going to be one of the most controversial kinds of high-level perception, since high-level value perception depends on our ability to perceive the properties that ground the value properties. For instance, value properties are in some cases partly grounded in natural kind properties, but even the latest arguments for natural kind perception have been called into doubt. 70 However important this debate might be for understanding perception, it is not important for value epistemology. Whether or not the sensings of values are ever “appropriately integrated” with ordinary perceptions, the high-level theorist should already think such experiences are a source of justification that need not itself be justified. 71 Four points of clarification. First, I concede that value perception may be a common way in which we come to have evaluative knowledge. What I have claimed is 70 See Brogaard [2013]. 71 One might try to salvage the epistemological significance of high-level value perception as follows. It is a common thought that in order to imagine, say, colors, shapes, flavors, sounds, etc. we must first actually perceive those colors, shapes, flavors, sounds, etc. So, one might argue, in order to be able to imaginatively experience value properties, we first have to perceptually experience those properties. In that way, the ability to acquire substantive evaluative knowledge perceptually is fundamental. (Audi [2013: 173] may be making this claim about the fundamentality of value perception, although it is not clear.) In response, I say the following. (See also my example above of an agent experiencing an evaluative property for the first time through the imagination.) First, if this is what the perceptual theorists needs to argue to salvage the significance of her view, then she has a serious argumentative burden, for the empirical claim about what is required for us to be able to imaginatively experience values needs defense. In particular, we need to know why the case of values is like color rather than like, say, mathematics. While humans might need perceptual experiences of some kind to be able to engage in mathematical thinking, there is no need to invoke mathematical perceptual knowledge or experiences in order to explain how we are able to justify mathematical beliefs (see McGrath [2011]). We need to know, then, why the phenomenal sensing’s being integrated into the content of a familiar kind of perceptual experience is at all important. Second, there is some intuitive evidence against the thought that we need perceptions of values to imagine values. The best cases for the primacy of perceptual knowledge involve low-level properties. But high-level properties seem different. It seems an agent can imagine a pine-tree without having ever come into perceptual contact with any pine-trees. (The easiest way to see this is to imagine some species of tree which does not yet exist but could.) This asymmetry is probably explained by the different ways in which low-level properties and high-level properties are represented in experience. 39 that the defender of high-level value perception should accept certain theses that ensure any knowledge or justification we get by value perception is in an important sense also had on non-perceptual grounds (since integration is not doing epistemological work). And this mutes the significance of whether there is value perception, at least for value epistemologists. Furthermore, there are questions closely related to the question of whether there is high-level value perception that are clearly important for value epistemology, but we should distinguish the significance of such questions from the question of high-level value perception. For example, some philosophers may want to claim that evaluative perception is not possible unless value properties are causally networked, or are reducible to properties that are. And the question of causation is (however one comes down on it) hugely important in value epistemology (and metaethics as a whole). 72 But it is causation that is important, strictly speaking, not perception. 73 Third, I am not arguing that the question of whether there is perceptual knowledge of value is insignificant, at least given a liberal understanding of what can count as perceptual knowledge. In a recent paper, Andrew Cullison [2009] argues for high-level value perception as a way of securing perceptual evaluative knowledge. 74 But when considering a certain objection to high-level value perception, he argues we may be able to get perceptual knowledge of value without high-level evaluative perception. His thought is that by perceiving stuff regularly correlated with some value property, V, perhaps we can know on that basis that V is instantiated. I suspect he would also say we 72 See Oddie [2005] and Enoch [2011]. 73 Most defenders of value perception claim that the question of value perception doesn’t turn on whether values are causally networked. See Audi [2013] and McBrayer [2010b]. 74 Cullison speaks of ‘moral perception’ rather than ‘high-level value perception’, but I take the differences to be merely terminological. 40 need to have some epistemically significant evaluative response, or intuition, to what we perceive (see Cullison [2009: 17 – 19]). But even if many philosophers would balk at calling knowledge acquired by such means perceptual, I suspect they would find the description of the view, at least in the very abstract way Cullison describes it, a familiar and not terribly implausible picture of how we acquire much of our evaluative knowledge. The final point of clarification is that, strictly speaking, the theses I offer are not necessary to generate the result that the question of high-level value perception does not matter for value epistemology (although they are sufficient). 75 We get the desired result so long as one accepts that it does not matter whether phenomenal sensings of values are integrated into the content of ordinary perceptual experiences. What the theses do is draw attention to our ability to acquire evaluative knowledge through the imagination, independently of familiar perceptual experiences (see Limits and Liberality), and then to emphasize the descriptive and epistemological similarities between the knowledge we get by imagining cases and the knowledge we get by perceiving actual cases (see Normative- Similarity, Descriptive Similarity, and Immediacy). The theses help us to see why the integration thesis – a central component of high-level value perception – simply does not matter for value epistemology. But even if it turns out, say, that Descriptive Similarity is false, because value experiences integrated with ordinary perceptions somehow have a different nature from those not so integrated, the high-levelist who maintains the import of her view still has a serious challenge: explain the epistemological import of integration. 75 Thanks to Pekka Väyrynen for helpful discussion on the points I make in this paragraph. 41 5 A Revised High-Level View As we know, an essential component of the high-level view is a claim about integration: value experiences can literally be part the content of visual, auditory, and tactile experiences. I have been arguing that whether these experiences are integrated just is not an epistemologically significant question. I have pushed this thought by appealing to our ability to acquire evaluative justification through the imagination. Suppose now that a high-levelist contends, plausibly enough, that when we imagine scenarios, we often (passively) imagine seeing, hearing, and feeling evaluative properties. So, for example, when I imagine Wesley copying answers off Jackie’s test, I also (let’s suppose) passively imagine seeing Wesley’s action as wrong. 76 The revised high-level view says this: high- level perceptual experiences of value occur when we have visual and other familiar perceptual experiences of value properties, or when we passively imagine having such experiences. 77 Although when we imagine have such experiences, we are not having genuine perceptual experiences, the defender of the revised high-level view claims that they are on a par with literal high-level value experiences. This revised high-level view fails to secure the view’s epistemological importance. The first point to notice is that if there are value sensings, or experiences, of the sort the high-levelist imagines, it is unlikely they are tied essentially to actual or imagined visual, auditory, and tactile modes. For example, suppose we read a newspaper story about a complicated but sketchy Wall-Street business deal. It seems that we can reflect on the various propositions comprising the story without conjuring any sensory- 76 It is important that the imagining be passive, or involuntary. If someone actively decides to imagine that they see some action as wrong, then it is hard to see how that experience could supply them with evidence of anything. 77 Janet Levin pointed out to me the possibility of revising the high-level view in this way. 42 like images of the events taking place. Nevertheless, when engaging in such reflection, we might have a phenomenal experience as of the deal’s being bad. Take another case. Suppose we imagine that John McCain defeats Obama in the 2008 U.S. election. Even if we conjure some images into our head when we imagine such a scenario, it is not clear that those sensory images are of the imagined event, since it does not seem as if someone’s winning an election is the kind of thing that can be perceived. 78 Nevertheless, it seems that when we imagine McCain defeating Obama in 2008, we may well have a passive experience as of its being bad. There is another kind of case worth mentioning. Even when we imaginatively visualize something, we may not (and I suspect we normally do not) imagine having a visual experience of it. Dominic Gregory has a helpful analogy for illustrating the point: We naturally interpret photographs as showing how things looked. We usually treat them as merely displaying the layout of a past scene, as showing how things once looked from a viewpoint within a previous situation…But we can also treat the same photos as showing how things looked in the course of past visual sensations, that is, in the course of visual sensations which occurred at viewpoints in past scenes. (Photos are sometimes used in that way in recounting first-personal narratives in comics, for example.) [2010: 744] Gregory’s view is that the same goes for visual images. A given visualization may be an imagining of how things looked from some viewpoint, or it may be an imagining of a visual sensation from that viewpoint. But when we visualize something in the former way and have an evaluative experience in response to the imagining, the response will not be imagined as integrated with any imagined visual experience, since ex hypothesi 78 See Yablo [1993] and Fiocco [2007]. 43 there is not any imagined visual experience. So it seems to me a stretch to maintain that evaluative experiences are in some way bound to ordinary perceptual experiences or imaginings of ordinary perceptual experiences. The high-levelist should allow that while an experience of goodness, badness, rightness, wrongness, etc. can be integrated into sensory experience, or imagined sensory experience, it could also occur independently. Now we can see that the revised high-level theory does not salvage the view’s import. When we have evaluative experiences detached from ordinary perceptual experiences (real or imagined), as the high-levelist should allow is possible, it strikes me as dogmatic and ad hoc to insist that in such cases, the experiences are not a source of immediate justification for evaluative beliefs. There does not seem to be anything special about real or imagined evaluative experiences that occur in visual, tactile, or auditory modes. (As another example, suppose that we sensorily imagine something but then as it fades, we are suddenly struck by the badness of what we just imagined. We do not imagine seeing/hearing/touching badness, but it is hard to see how a high-levelist can avoid allowing that the experience is a source of justification just the same.) But then this suggests that if we come to find out that the high-levelist is wrong, and value experiences are never integrated into sensory experience or imagined sensory experience, but are merely often responses to sensory experience or imagined sensory experience, then it hard to see how that finding could be of much epistemological import. 6 Conclusion: More Epistemologically Exciting Models of Value Perception? The high-level theory of value perception is not the only kind of value perceptual theory on offer. As I mentioned in the Introduction, there are two important alternative 44 approaches to value perception. The view that I call intellectual perceptualism says that we are able to intellectually experience evaluative propositions as being true in much the same way that we sometimes (at least according to these theorists) intellectually experience mathematical propositions as being true. And another view, what I call sentimental perceptualism, says that desires and/or emotions involve perceptual (or perceptual-like) experiences of value. I leave a detailed study of these two views to the next chapter, but for now it is important to notice that, in contrast with high-level value perception, these two theories do not fall prey to the arguments I have pressed in this chapter against the importance of the high-level theory. This is for the reason that, if indeed there are intellectual or sentimental evaluative experiences, then they can occur when we are engaged in reflective or imaginative evaluative thought. As I noted above, we have desiderative and emotional responses not only to actual cases but also imaginative and fictional ones; and, additionally, intellectual experiences paradigmatically occur in the context of reflection. So intellectual and sentimental perceptualism are in an important respect more promising approaches to value epistemology. The task for the next chapter is to adjudicate intellectual and sentimental perceptualism. 45 Chapter 2: Intellect vs. Affect: Finding Leverage in an Old Debate 0 Introduction A historically apt way to situate the debate between intellectual and sentimental perceptualism is as a version of the clash between rationalist and sentimentalist approaches to value epistemology. Rationalist proponents of perceptualism invoke intellectual experiences (intellectual perceptualism), while sentimentalist proponents invoke affective experiences (sentimental perceptualism). The goal of this chapter is to offer a fresh strategy for adjudicating between intellectual and sentimental perceptualism. I argue that the perceptualist’s hand will be forced either in the direction of intellectual or sentimental perceptualism once she decides between two views about the modal status of our basic evaluative knowledge. Intellectual perceptualists, I argue, are pressured to say that our basic evaluative knowledge must be of (metaphysically) necessary evaluative propositions. Sentimental perceptualists, however, are pressured to say that our basic knowledge may be of contingent evaluative propositions. This asymmetry is exciting news, for questions about the modal status of our basic evaluative knowledge are ones we are positioned to make progress on, at least given the assumption of perceptualism. I close the chapter by making a preliminary case for two theses: (i) perceptualists of all stripes need to invoke evaluative experiences suited for tracking contingencies, and (ii) those are the only kinds of experiences perceptualists need. The argument, then, is that perceptualists ought to be sentimentalists. 46 1 Preliminary Remarks Perceptualists – whether intellectual or sentimental – need to convince us that the relevant evaluative experiences actually exist. In recent years, sentimental perceptualists have largely converged around a particular strategy for making sense of affective evaluative experience. The strategy is to argue that proper analyses of ordinary desiderative and/or emotional experiences reveal some subset of those experiences to involve experiences of value. 79 (There is thus no special “moral sense” which stands apart from our more familiar emotions and desires.) So, for example, one view about anger, tracing back to Aristotle, is that it involves a presentation of having been wronged. 80 Another historically popular view about desiderative experience is that it involves a presentation of its object as good. 81 Intellectual perceptualists typically begin to make sense of intellectual experiences by arguing that they play a role in producing beliefs in a priori inquiry, e.g., mathematical inquiry. 82 When we contemplate, say, the proposition that two and three make five, the proposition strikes us as true. That is, when carefully attending to the proposition, we often have an experience which can only be described as an experience as of the truth of the proposition, or so the story goes. The next move is to argue that these experiences are not specific to certain non-evaluative domains such as mathematics. 79 See, for example, Oddie [2005], Döring [2007], and Roberts [2003]. Some theorists who ascribe to this view about ordinary desires and emotions do not write systematically enough about value epistemology to make it clear whether they think of themselves as sentimental perceptualists. 80 Aristotle tells us that anger is in part a response to “a conspicuous slight directed without justification towards what concerns oneself or towards what concerns one’s friends.” See Rhetoric 1378b. 81 This view also has roots in Aristotle. See Moss [2013] for careful historical discussion of Aristotle’s position. 82 See, for instance, Huemer [2005]. 47 Just as we can have intellectual experiences with, say, mathematical content, so too can we have such experiences with evaluative content. How do perceptualists typically attempt to adjudicate between the intellectual or sentimental varieties of the view? I will mention three of the most common strategies, all of which trace in some form back to the 17 th and 18 th centuries. First, there are arguments from motivation. Sentimental perceptualists often insist that their view is better equipped to account for the connection between motivation and evaluative judgment. 83 However, as is well-known, intellectual perceptualists, and rationalists more generally, have a variety of strategies for trying to explain why we are typically motivated by our evaluative judgments, though not in a way that conflicts with a rationalist answer to the question of how we get evaluative knowledge. 84 Second, there are arguments from psychology. Sentimental perceptualists often argue that the intellectual evaluative experiences her opponent posits do not exist; and similarly, intellectual perceptualist often press an analogous complaint against sentimental perceptualists. Historically, these psychological arguments are made on phenomenological grounds. Sentimental perceptualists often insist that evaluative experiences, or “intuitions,” seem to have a feeling component that makes them different from intuitions in mathematics. 85 83 Many early modern sentimental perceptualists offer such arguments. See Gill [2006] for discussion. For an updated version of a sentimental perceptualist argument from motivation, see Oddie [2005]. 84 See Huemer [2005]. Intellectual perceptualists can also borrow lines of argument offered by rationalists who are not themselves intellectual perceptualists, e.g., those in Enoch [2011]. 85 For a discussion of the history of such phenomenological arguments, see Gill [2006] and [2009]. Similar arguments are offered today. For example, Kauppinen [2013] and Oddie [2005] use phenomenological arguments to defend sentimental perceptualism, and Huemer [2005] does the same in defending intellectual perceptualism. (Huemer, however, does not consider sentimental perceptualism as an alternative to intellectual perceptualism, and simply argues for perceptualism by arguing for the intellectual version of the view.) In many cases, phenomenological objections to a form of perceptualism are raised by those who reject perceptualism altogether. Williamson [2007] and Sosa [2007] press phenomenological arguments against the existence of intellectual presentations, whether about mathematics, value, or whatever. Analogously, Whiting [2012] and Dokic and Lemaire [2013], to note 48 Intellectual perceptualists push back with phenomenological arguments of their own. For example, Balguy argues that in some cases, there just is not an affective experience present to justify a person’s well-formed moral judgment. 86 A third brand of argument is directly epistemological. The idea here is to allow the psychological reality of both affective and intellectual evaluative experiences (at least for the sake of argument), but then to argue that only one is capable of grounding evaluative knowledge. Perhaps affect fluctuates too much to ground knowledge of eternal evaluative truths, as many intellectual perceptualists have thought. 87 Despite numerous attempts to update the traditional lines of argument from motivation, psychology, and epistemology, whether in favor of intellectual or sentimental perceptualism, there is still as much disagreement as ever about which version of perceptualism is best. 88 In this chapter, I spell out a different strategy for getting leverage in the debate between intellectual and sentimental perceptualism. just a couple examples, argue on phenomenological grounds against sentimental perceptualism’s psychology without themselves adopting an alternative perceptualist framework. 86 For discussion of Balguy’s argument, see Irwin [2008: 446 – 47]. 87 Gilbert Burnet, in a letter to Hutcheson, argues that affect is too variable and uncertain to be the foundation of evaluative knowledge. See Burnet and Hutcheson [1971]. For updated epistemological arguments against sentimental perceptualism, see Brady [2013] and Szigeti [2013]. I respond to Brady and Szigeti in Milona [forthcoming b]. 88 I do not mean to suggest that these are the only kinds of arguments. Many, for instance, have thought that how we acquire evaluative concepts has implications for the debate between intellectual and sentimental perceptualism. Arguments from the origins of evaluative thought were popular in the early modern period. Geoffrey Sayre-McCord [2001] revives such an argument in support of sentimental perceptualism. 49 2 A Strategy for Adjudicating Intellectual and Sentimental Perceptualism 2.1 Necessary and contingent evaluative truths As I have already indicated, intellectual perceptualists have tended to argue that our basic evaluative knowledge is of metaphysical necessities, i.e., propositions which are true in all metaphysically possible worlds. There are different views about the proper way to formulate such propositions. They might be thought to take the form of a conditional, “if A, then B,” whereby the antecedent is non-evaluative and the consequent evaluative. Here is a candidate such principle: if an agent, S, intentionally kills another agent, then S acts wrongly. Others may opt for necessary principles ascribing evaluative properties to act-types (in addition, perhaps, to other types of things), as in the following: intentional killings are wrong. The basic form that these necessities take may ultimately matter a great deal, though nothing central to this chapter turns on the matter. 89 Although intellectual perceptualists tend to focus on necessities, much of our evaluative knowledge is of contingencies, i.e., propositions which are not true in all metaphysically possible worlds. Here are a couple examples: (1) Cindy ought to attend her daughter’s ballgame, and (2) If Lindsey borrows Ronald’s carving knife, it would be best for her to return it when next she sees him. Concerning (1), even though it is important for Cindy to support her daughter, her other child might suddenly become ill, in which case she should not attend the game, after all. And regarding (2), it is possible that if Lindsey borrows Ronald’s carving knife it would not be best for her to return it when next she sees him, if, say Ronald’s partner has just broken up with him on the phone and Ronald is threatening suicide. 89 On some ways it may matter how these necessities are formulated, see Schroeder [2014]. 50 A perceptualist may adopt what I call the necessity-requiring theory. The necessity-requiring theory says that our perceptual-like evaluative experiences track, when all is going well, not merely how things in fact are evaluatively but how they must be. (The theory does not require that we present as necessary the evaluative propositions we have basic knowledge of.) Our knowledge of evaluative contingencies is standardly achieved by combining our basic evaluative knowledge of necessities with our knowledge of contingent, non-evaluative facts. We will spend time later (section 4) on the question of how this combining works; what is important for now is just the very idea of a necessity-requiring theory. Alternatively, a perceptualist may be attracted to the contingency-allowing theory, according to which our perceptual-like evaluative experiences track, when all is going well, how things in fact are evaluatively but not how they must be. The contingency- allowing theorist holds that we can have basic knowledge of evaluative contingencies. We could know, say, that Jesse was wrong to kill Shannon without relying on our knowledge of any necessary evaluative truths. It is contingent that Jesse’s killing of Shannon was wrong, since, for example, it is possible that Jesse kills Shannon yet does not thereby act wrongly. The contingency-allowing theorist also believes that our knowledge of necessities, to the extent we have such knowledge, is typically going to be grounded in our knowledge of contingencies. I will return below (section 4) to the question of how a contingency-allowing theorist thinks we can go from contingencies to necessities. For now, what is crucial is just the very idea of a contingency-allowing view, namely that our perceptual-like evaluative experiences are sensitive not to how things are evaluatively in any world but rather how things are evaluatively in this world. 51 Any contingency-allowing theorist needs to allow that we could stumble into basic knowledge of a necessary evaluative truth. (A contingency-requiring theory is a non- starter.) To illustrate, suppose that Brenda imagines Randy torturing a baby just for fun, has a perceptual-like response to her imagining that so-acting would be wrong, and concludes that it would be wrong of him to torture the baby for fun. As it happens, Brenda’s basic evaluative knowledge here is of a necessary truth, since it is not metaphysically possible for Randy to torture the baby for fun and for that not to be wrong. This is not a problem for the contingency-allowing theorist, however. Brenda has stumbled into basic knowledge of an evaluative necessity, but the type of experience she relies on is sensitive to what things are like in the actual world, even when it happens to have content that is true in any world. Put metaphorically, a contingency-allowing theorist says that her evaluative experiences don’t care about how things must be in all worlds but merely how they are in this world. 90 One way to get adjudicate the longstanding dispute between intellectual and sentimental perceptualists is to adjudicate the (much underexplored) dispute about whether to be necessity-requiring or contingency-allowing. This is because, as I will argue, intellectual perceptualism fits comfortably with a necessity-requiring approach but not a contingency-allowing approach. Sentimental perceptualism is just the opposite. 90 The distinction between necessity-requiring and contingency-allowing theories is orthogonal to the more familiar distinction between general and specific evaluative truths. An example of a general evaluative truth is that killing is wrong. And we can make this more specific by building in further conditions, e.g., intentional killing is wrong. The proposition that (intentional) killing is wrong is true but not necessarily so; it is a principle that holds all else being equal in worlds like ours. In a much harsher world with only a few innocent people among a population of irredeemably evil villains, it may not be true that (ceteris paribus) killing is wrong. 52 2.2 Intellectual perceptualism, necessity-requiring, and contingency-allowing 2.2.1 How necessity-requiring intellectual perceptualists demystify evaluative knowledge Intellectual perceptualists, as I have noted, typically use a comparison with mathematics in their effort to make sense of intellectual experiences. The most popular such strategy says that an understanding of mathematical propositions often gives rise to intellectual experiences which serve as a basic source of justification for mathematical beliefs. 91 Typically, understanding is taken to consist in a grasp of the concepts involved in the target proposition. So long as such an experience in fact arises on the basis of our understanding alone and is accurate, it can generate knowledge. 92 Intellectual perceptualists who adopt this picture of mathematical knowledge argue that such experiences can also arise when thinking about propositions in other domains, e.g., the evaluative. The understanding-based strategy of illuminating intellectual experience and evaluative knowledge works well for necessity-requiring intellectual perceptualists, for a simple reason. Because these intellectual experiences arise on the basis of our understanding alone, rather than on the basis of what the actual world is like, they are naturally taken to justify beliefs in necessary propositions. A contingently true 91 Almost all contemporary intellectual perceptualists accept this view. (Elijah Chudnoff, whose view I discuss below, is a rare intellectual perceptualist who resists this model.) For a detailed discussion of the history of the understanding based view and a detailed overview of how it works – albeit with a focus on mathematics – see Chudnoff [2014]. 92 Some rationalists argue that there is no need to appeal to intellectual experiences to explain how understanding gives rise to knowledge. Understanding is sufficient all on its own. Audi [1998] is a major proponent of this kind of view. For recent discussions of this non-perceptualist form of rationalism, see Bedke [2010] and Tropman [2014]. 53 proposition is not true in all worlds, so it seems as if we would need to know something of what the actual world is like before we could know the proposition. 93 (There is a complication worth noting. Consider the following: that Randy tortures the baby just for fun is wrong. This proposition is contingent, for if Randy never tortures the baby just for fun, then it will not be true that his torturing the baby just for fun is wrong. (Any proposition which entails a contingent proposition must itself be contingent.) In such worlds, the proposition will be either false or lacking in truth value. However, if Randy does torture the baby just for fun, then what he does is guaranteed to be wrong; it is a necessary truth that it would be wrong for Randy to torture the baby just for fun. We can label evaluative truths which become necessary when modalized in this way weakly contingent. An understanding-based intellectual perceptualist may argue that knowledge of Randy’s behavior could lead to a direct apprehension that Randy’s torturing the baby just for fun is wrong, even though the truth is contingent. The thought is that the knower is responding to conditions which guarantee the presence of the relevant property, just as she does when reflecting on the question of whether such behavior would be wrong. Such a view is still necessity-requiring, however, since a necessity-requiring perceptualist view says that our evaluative knowledge is grounded in experiences which respond to conditions which guarantee the presence of the relevant evaluative property. For the sake of simplicity, I set aside weak contingencies; when I talk about contingencies I always mean to refer to non-weak contingencies.) 93 The reader may be wondering about the contingent a priori. I will consider in the next section the prospects for an intellectual perceptualism modeled on the contingent a priori (assuming there is any such knowledge). 54 Elijah Chudnoff is a contemporary intellectual perceptualist who resists the understanding-based model, but his alternative view also goes best with the necessity- requiring picture. Chudnoff develops his theory (which he traces back to Plato and Husserl, among others) with mathematical knowledge principally in mind, but he believes it can be extended to other domains, notably the evaluative. 94 According to this alternative, we are capable of intellectual experiences that are not based in understanding but rather on a direct awareness of abstracta, i.e., objects which are non-spatiotemporal and causally inert. (This model takes its cue from direct realist views about perception.) On this view, an intellectual experience, say, that circles are symmetrical about their diameters is an experience whereby circularity figures in the explanation of what the experience is. That is, circularity – an abstract object – figures into the mereological explanation of the experience. Chudnoff argues that such an experience provides us with basic justification for the belief that circles are symmetrical about their diameter, since in having the experience, we are directly aware of the truth-maker for the proposition. This requirement that we be aware of a truth-maker pushes us to say that intellectual experiences, so understood, can only give us knowledge of necessities. 95 Even though a contingently true proposition may itself be an abstract object, the truth-makers for contingent propositions will not themselves be abstracta (i.e., non-spatiotemporal and causally inert). Necessity-requiring intellectual perceptualists, as we have seen, lean on a comparison with mathematics to make sense of evaluative knowledge. But is there hope 94 Chudnoff [2013] 95 We could give up this requirement, but then it would be puzzling how these experiences could generate knowledge. For discussion, see Chudnoff [2013]. 55 for a novel intellectual perceptualism that turns away from the mathematical analogy? In other words, what are the prospects for a contingency-allowing intellectual perceptualism? 2.2.2 Contingency-allowing intellectual perceptualism? My contention is that contingency-allowing intellectual perceptualists lack a comparatively attractive alternative for demystifying evaluative knowledge. Proponents of such a view would need to do three things: (i) identify some type of intellectual evaluative experience – one which is featured in actual human psychology – suited for tracking evaluative contingencies, (ii) argue that such experiences can be a source of basic evaluative knowledge, and (iii) show that they can be the basis for a complete perceptualist value epistemology. I am highly skeptical that there is any experience which does all three of these things. To motivate why, I explore the domain of the a posteriori for an attractive kind of intellectual experience, and thereafter I turn to the contingent a priori. Before we can embark on the exploration, we need to answer the following question: which kinds of experiences which might provide us with knowledge of contingencies are even candidates for being intellectual experiences? In this chapter, I adopt a very loose understanding of intellectual experiences: any experience which is not affective counts as intellectual. This loose understanding ensures that no perceptualist theories are left out of our taxonomy. It also means that some theories which are not typically thought of as forms of intellectual perceptualism count as versions of the theory. For example, a high-level theory of value perception might (see below for why I say 56 ‘might’) count as a version of intellectual perceptualism. I believe that this is a fair price to pay for completeness. In what follows, I revisit the high-level theory, deepening my critique of that approach, and then consider an alternative theory which says that humans are “hard-wired” for certain intellectual evaluative experiences. As I mentioned in the previous chapter, there has been growing interest in the possibility of what is often called high-level perception, and high-level value perception, in particular; and contingency-allowing intellectual perceptualists may look to such experiences to develop an a posteriori value epistemology. 96 To illustrate high-level perception (evaluative or otherwise), I will use visual experience as the paradigm. Contemporary philosophers of mind mostly agree that there are visual experiences with low-level contents, namely experiences of shape, color, motion, and body. But some argue that visual experiences can have high-level content, e.g., natural kinds, causation, emotions, and values. The distinction between high-level and low-level content is initially defined as that which is uncontroversially part of visual experience (low-level) and that which is not (high-level), but defenders of high-level content normally take the high-level experiences to get their content in a certain distinctive way. Defenders of high- level value perception standardly argue that such experiences are made possible by the cognitive penetration of other mental states such as beliefs, desires, or emotions on our perceptual faculties. 97 A visual experience is cognitively penetrated when some non- 96 On high-level perception in general, see Siegel [2011]. On high-level value perception, see, for instance, Cowan [2015a] and Werner [2014]. Many intellectual perceptualist will, I suspect, not see the high-level perceptualist view of evaluative experience as a brand of intellectual perceptualism. But, remember, I count as intellectual any evaluative experiences which are not affective (see the Introduction to this dissertation). 97 Cowan [2015a] explains in detail why cognitive penetration of this sort is the best explanation for how high-level value perception is possible. Another possibility, as Cowan notes, is that dispositions to experience visual evaluative representations in response to certain stimuli are innate. I consider this possibility below. 57 visual mental state (it need not be “cognitive” in any traditional sense) alters the representational content of the visual experience, and not by merely altering attention. Suppose that a person observes a group of kids torturing a small animal. A high-levelist might say that the observer’s outrage at the treatment of the animal may alter the content of the visual experience such that she counts as seeing the torture as wrong. Contingency-allowing intellectual perceptualists should not appeal to high-level value perception. First, desires and emotions are some of the most popular candidates for the mental states that cognitively penetrate visual experience to produce visual experiences of value. 98 If that is the right picture of high-level value perception, though, then the view is better characterized as a version of sentimental perceptualism. But even if we allow that non-affective evaluative beliefs are doing the penetrating, we still run into a problem. If evaluative beliefs are the sine qua non for any such experiences, then the justification the experiences supply does not seem to be basic, i.e., it does not seem to be independent of the justification for the relevant beliefs, as others have already pointed out. 99 To illustrate, suppose Francis has the irrational belief that it is always bad to pursue what is not in one’s self-interest. This belief in turn often causes him to visually experience selfless acts as bad. As Pekka Väyrynen rightly points out, a “garbage in, garbage out” principle seems to apply in these cases. Because the belief that plays a crucial role in the etiology of the high-level value experiences is irrational, the high-level value experiences cannot then be used to justify an evaluative belief. 100 98 See Cowan [2015a] and Werner [2014]. 99 See Cowan [2015a] and Väyrynen [forthcoming]. 100 Even if the belief in the example above were rational, the justification supplied by the experience would still intuitively fail to be basic. See Väyrynen [forthcoming]. (Väyrynen’s point is hypothetical, though, since he ultimately denies the existence of high-level value experiences.) Cowan [2015] makes a similar argument. 58 It is important to notice that the claim here is not that high-level experiences cannot justify evaluative beliefs or that they can play no important epistemic role. The claim, rather, is that they are not good candidates for being a basic source of justification. This is for the simple reason that high-level experiences are sophisticated in that they are made possible by more primitive evaluative experiences or beliefs; and it is very natural to think that any sophisticated experience is going to be largely epistemically dependent on the evaluative beliefs or experiences that make them possible. But even if you do not buy this, there is still trouble for a contingency-allowing intellectual perceptualist who wishes to appeal to high-level experiences. Namely, such sophisticated experiences will not be able to account for the full scope of our evaluative knowledge. If the sophisticated experience is made possible by a primitive kind of evaluative experience (e.g., an emotion or intellectual perception of the sort outlined in section 3.2.1), then we would expect the more primitive experience to be a source of justification independently of its impact on perception. But if the penetrating state is a belief, then presumably the belief will often already be justified, independently of high-level experience. 101 A contingency-allowing intellectual perceptualist may, then, seek to understand intellectual evaluative experiences as primitive in nature, i.e., not made possible by other evaluative experiences or beliefs. I am not aware of anyone who has developed such a view, but it is worth reflecting on how such a theory might be developed. The most obvious candidate experiences are low-level perceptual experiences (e.g., visual experiences of shape, body, and color); but to my knowledge, no psychologists or philosophers have ever thought that any of these more primitive perceptual experiences 101 I defend this claim at length in Chapter 1 and in Milona [forthcoming a]. 59 can sometimes have evaluative content. But some psychologists and philosophers do think that some evaluative principles are innate, and perhaps (although this is a major additional commitment) our innate evaluative structure plays a role in generating primitive presentations of evaluative contingencies. 102 Before evaluating the prospects for this view, we need to be sure we are zeroing in on a class of theories friendly to intellectual perceptualism. Some psychologists and philosophers, including Jonathan Haidt, Craig Joseph, Shaun Nichols, and Chandra Sripada – propose theories which posit innate affective structure in order to explain the emergence of evaluative life. 103 Let’s set these theories aside. Friendlier to intellectual perceptualism are the views of Susan Dwyer, John Mikhail, and Gilbert Harman. 104 These theorists argue that there is an innate, non-affective “moral grammar,” characterizable by a set of principles. (I will use ‘evaluative’ rather than ‘moral’, but nothing I argue here turns on the matter.) The principles which characterize the grammar are ones which may be different from the norms we invoke in actual evaluative discourse, but just as it is difficult for competent speakers to articulate linguistic principles, so too may it be difficult for competent evaluators to articulate these innate evaluative principles. The appeal to a non-affective innate evaluative structure is, I think, the best hope for a contingency-allowing intellectual perceptualism, but it is nevertheless not very promising. One concern is that intellectual perceptualists who choose this path have to argue that this structure gives rise to non-affective experiences, but the best way of 102 Perhaps our innate evaluative structure consists in innate evaluative beliefs, in which case the experiences are not really primitive (i.e., because they arise out of prior evaluative beliefs). 103 Haidt and Joseph [2004], Nichols [2004], and Sripada [2008] 104 Dwyer [2008] and [2009], Harman [2008], Mikhail [2007] and [2011] 60 thinking about the view may be that the innate structure just directly gives rise to evaluative beliefs without any mediation by experiences. But even if that worry can be overcome, I am skeptical that such a theory could never be the basis for a complete value epistemology. Psychologists and philosophers who defend innate evaluative structure generally take the innate structure to place boundaries on evaluative belief. For example, Susan Dwyer suggests that our innate evaluative principles may prevent us from ever believing that it would be good to torture babies just for fun. 105 But the innate structure is consistent with a variety of different evaluative systems. To illustrate, our innate evaluative structure is, according to most accounts, consistent with both accepting or rejecting a culture of honor, and so if we learn that such a culture is misguided, it presumably will not be on the basis of our innate principles. To be sure, a committed contingency-allowing intellectual perceptualist could make the bet that our non-affective, innate evaluative structure is far more robust (and internally inconsistent) than we now realize, and that that structure is the origin of all non-affective intellectual experiences of value which immediately justify evaluative beliefs; but until we get serious empirical evidence for such a view, such a bet seems to me unwise. 106 105 See Dwyer [2008: 414] for the claim that innate evaluative (or moral) principles would rule out this belief. 106 Proponents of non-affective, innate moral grammars have been very hesitant to attempt characterizing that structure in a precise way. Dwyer [2008: 414] even says “To be frank, the form and content of the principles that I claim characterize the moral faculty remain a mystery.” John Mikhail [2007: 143 – 52], however, proposes several principles, but many of the principles he proposes are not characterized in non- evaluative language. For example, one says that negating a good effect is bad; but that principle won’t do us any good until we have some way of figuring out what is good. One substantive principle he proposes is “an effect that consists of the death of a person is bad.” But it seems to me that this is probably only true all else equal, and in any case, it is only one principle. For some interesting remarks about Mikhail’s view, see Dancy [2014]. My aim here is not to criticize Dwyer, Mikhail, or anyone else; it is only to say that intellectual perceptualists should not think that they can appeal to the work of these theorists in order to have a plausible basis for a contingency-allowing theory. 61 We are now in a good position to see that appealing to a posteriori intellectual experiences is not so promising for the contingency-allowing intellectual perceptualist. 107 Such evaluative experiences will be sophisticated (made possible only by other evaluative experiences/beliefs) or not. We can reasonably expect the sophisticated candidates to either be not a plausible source of basic evaluative knowledge or insufficient for explaining the full scope of our evaluative knowledge, and we can expect the primitive candidates to either be psychologically dubious or similarly insufficient. Assuming all basic knowledge is a posteriori or a priori, the only avenue remaining for the contingency- allowing intellectual perceptualist appeals to the contingent a priori. 108 But this is not a promising avenue, either. 109 Saul Kripke offers the most well- known (purported) example of the contingent a priori. 110 He argues that we can know a priori that the standard meter stick is one meter long. We know this a priori because the referent of ‘meter’ is just whatever length the standard meter stick happens to be. But the length of the stick could have been other than it is and so our a priori knowledge is of a contingency. Even if we grant that this is an example of contingent a priori knowledge, however, it does not look like a good model for a contingency-allowing intellectual perceptualist. There are several reasons why. First, it is not as if there is a standard action – say, whatever type of action Adam does after time t – which fixes the reference of an 107 A reader may wish to classify knowledge of evaluative contingencies grounded in innate evaluative experiences as contingent a priori. We need not dwell on classificatory questions, though. 108 The possibility of contingent a priori knowledge is highly contentious. Both Hume [1993] and Kant [1996] believe all a priori knowledge is necessary. George Bealer [1999] and Lawrence Bonjour [1998] are among the many contemporary theorists who agree. 109 Dancy [2004] argues that that our basic evaluative knowledge is a priori and contingent. McKeever and Ridge [2006] develop some powerful objections to this position. 110 Kripke [1980] 62 evaluative term – say, ‘good’. Furthermore, the knowledge in Kripke’s case is derived from our understanding of the meaning of ‘meter’. 111 But while some of our evaluative knowledge may be semantic (e.g., that the evaluative supervenes on the descriptive), much of it is not. Finally, we do not seem to need to invoke intellectual presentations to explain the knowledge acquired in Kripke’s example. And this points to a general difficulty for a contingency-allowing intellectual perceptualist who wants to appeal to the contingent a priori: such a theorist needs to find not only a reasonably plausible class of contingent a priori truths (a controversial enough commitment) but, moreover, those truths need to be plausibly grounded in intellectual presentations. 112 The aim in this section has been to vindicate the wisdom of the common tendency among intellectual perceptualists to posit evaluative experiences which track how things must be evaluatively. A necessity-requiring approach allows the intellectual perceptualist to demystify evaluative knowledge by arguing that it really is not so different from mathematical knowledge. (Or at least to demystify it for those who find it plausible that intellectual presentations justify our mathematical beliefs.) But there is no comparatively attractive approach for contingency-allowing intellectual perceptualists. 111 McKeever and Ridge [2006] make essentially the same point against any rationalists who wish to appeal to Kripke-style examples of the contingent a priori to make sense of ethical knowledge. 112 There are other purported examples of the contingent a priori. McKeever and Ridge [2006] discuss some of them and explain why they are not good models for any rationalists. It is worth noting that some theorists who invoke the contingent a priori do so in order to explain how we can be justified in taking our sense experiences to be reliable. See, for instance, Cohen [2010]. Our a priori knowledge of the reliability of sense experience, if we have it, would be contingent, since sense experience might not be reliable. But theorists who believe that we can have a priori contingent knowledge for the reliability of sense experience do not think that we achieve such knowledge by way of intellectual presentations of the reliability of sense experience. This is (inter alia) presumably because it would force us to ask how we know intellectual presentations of contingencies (e.g., that sense experience is reliable) are reliable. Thus whatever model of the contingent a priori these theorists defend won’t be a good model for intellectual perceptualists. 63 2.3 Why sentimental perceptualists should be contingency-allowing As I noted above, the view that some of our affective experiences present their objects as valuable is a popular view, both historically and today. I will not attempt to argue for this psychological picture, however. Rather, I take it for granted that at least some desires and/or emotions present value and then argue the following: it is more plausible that well-functioning affective experiences of value present, at least typically, evaluative propositions which are only contingently true than it is that they must present evaluative propositions which are necessarily true. This means that sentimental perceptualism fits comfortably with a contingency-allowing, but not a necessity- requiring, approach to value epistemology. There are at least a couple ways to motivate the claim that sentimental perceptualists should be contingency-allowing. 2.3.1 The way of examples The first line of argument involves working through examples. I will offer two illustrations, one involving a familiar type of emotion (anger) and another involving a familiar type of desire (parental concern); and I hypothesize, reasonably in my view, that similar examples could be constructed for other candidate affective experiences. Here is the first case. Jacqueline is walking down one of the (fairly uncrowded, at this point) corridors of a stadium. To her surprise, her favorite athlete walks past. He does not initially make eye-contact with her, but the large athlete does manage to step firmly onto her foot, offering in response no more than an apathetic glance. Jacqueline is likely to be angry about his lack of concern. As mentioned above, a number of theorists, including Aristotle, take anger to present a slight, or wronging. Let’s work with 64 that assumption. (Some may prefer to label Jacqueline’s emotion as outrage or indignation. It makes no real difference for the example.) There are numerous possibilities for how precisely to characterize the content of Jacqueline’s anger, each of which a sentimental perceptualist can allow is possible. One possibility is that the anger involves a presentation of (perhaps among other things) the athlete as wronging her by uncaringly stomping on her foot. It may also be that she experiences the athlete as wronging her by stomping on her foot, without also experiencing the stomping as uncaring. There are presumably other possible ways the content of her anger might be filled-out. In each case, the anger presents an evaluative contingency. For example, all stompings are not wrong, since some stompers take care not to step on others but do so accidently. And even some uncaring stompings might not constitute a wrong, since uncaring stompers could be delirious from head injuries. The key observations I want to make about the case are these: (i) by stipulation, Jacqueline’s anger response involves an accurate presentation of an evaluative contingency, and (ii) her response is, at least for all we can tell, the product of her well- functioning anger-system. But then this indicates that sentimental perceptualists should be contingency-allowing, since it is precisely our well-functioning, accurate evaluative presentations that we would expect to play the foundational role in a sentimental perceptualist theory. Now consider a second case. A mother is driving to her son’s school to pick him up. When she pulls up, she observes several other kids mocking him. She is averse to the treatment of her child. On a traditional sentimental perceptualist model, aversions 65 involve presentations of their object as bad. 113 Let’s assume this view. On this picture, her aversion involves a presentation of the other children’s mocking her child as bad. This is a presentation of an evaluative contingency, since, for example, her child may encourage and even enjoy the mocking, perhaps in the context of some kind of game, or it may be that he has done something very bad to merit such treatment. But these are rather unlikely possibilities. The mother’s desire seems from the description of the case to be the product of her well-functioning desiderative system, and, moreover, her aversion involves, by stipulation, an accurate presentation of an evaluative contingency. Once again, it seems to me that this is just the sort of experience we would expect to play a foundational role in a sentimental perceptualist’s theory. 2.3.2 Explaining affective content Now consider a second way of motivating why sentimental perceptualists ought to be contingency allowing. As noted above, some resist sentimental perceptualism because they doubt its psychological framework. The resistance often takes the form of what I will call the content challenge: it needs to be explained how emotions and/or desires could have come to present evaluative content (and this needs to be done in a way that preserves the epistemological ambitions of the view). 114 Surprisingly, sentimental perceptualists have paid little attention to this important challenge. 115 However, the most 113 See Oddie [2005]. 114 See Schroeder [2008] and Schafer [2013]. 115 One exception, as Schafer [2013] points out, is Stampe [1987]. Stampe argues that desires get to be about goodness by co-varying with goodness. But Schafer rightly points out that mere covariation is insufficient for explaining content. Here is a simple reason: if co-variation were sufficient, then we would expect visual experiences of greenness to also be about patterns of light arrays on the retina which covary with such experiences, but that is entirely the wrong result. See Neander [2012]. That sentimental perceptualists, with the exception of Stampe, have largely ignored this challenge is surprising, especially in light of the fact that defenders have high-level value perception have 66 natural strategy for addressing it pushes the sentimental perceptualist toward the contingency-allowing view. 116 A natural assumption I will make here is that sentimental perceptualists should stick close to their perceptual analogy and model their content-determination story on that of (low-level) perceptual experience. Philosophers of perception, at least those who think perception has content, often seek to explain how, for example, we have come to experience (in the case of vision) greenness, squareness, motion, and so on. Typical stories for how perceptual content-determination works appeal to biological functions. The basic idea is that our perceptual systems have the function of producing experiences which carry information (whereby ‘information’ is usually given a covariational and/or causal analysis) about whatever those experiences are about. 117 For our purposes, we need not delve into the details of such theories, which are often quite complex. But to illustrate, here is a more precise statement of the shape such a theory might take. According to one precisification, we explain how we have, say, visual experiences of greenness by appealing to the biological function of our visual system to produce such experiences in response to proximal stimulations (notably, light arrays on the retina) that tend to covary with the presence of green objects. paid such significant attention to the analogous content challenge that they face. High-levelists, as we have seen, typically answer their content challenge by appealing to cognitive penetration. However, we have also seen that this way of addressing the challenge seems to undermine the view’s epistemological import. 116 For a more detailed response to the content challenge, see my [unpublished]. 117 For a discussion of this view and its history by one of its proponents, see Neander [2012]. For other defenses, see Prinz [2004] and Burge [2010]. Burge in particular appeals to bio-functions not in order to reduce content, as some others do, but merely to explain how perceptual experiences end up with certain content. I am suggesting sentimental perceptualists do something similar. The content challenge, as I conceive it, is not a challenge to give a reductive analysis of affective evaluative experiences but rather to explain how there are affective experiences with such content. 67 Sentimental perceptualists could give a natural and reasonable answer to the content challenge, if they can make the case that desiderative and/or emotional experiences (or underlying affective systems) have the biological function of carrying information about value. The important point for our purposes is that a necessity- requiring sentimental perceptualist is in a poor position to take advantage of this functional story about content determination. The necessity-requiring sentimental perceptualist would have to say that our desires and emotions have the function of responding only when there are stimulations (perceptions, beliefs, imaginings, etc.) which, if true, would guarantee the presence of the relevant value. But why would they have such a function? There is no precedent for this with ordinary perception. Ordinary perceptual experiences do not have the function of tracking necessary truths, nor do they have the function of responding to stimulations which guarantee the accuracy of the experience. For example, an agent’s visual representations of edges and surfaces may trigger a visual representation of solidity, but it is possible for a non-solid object to have edges and surfaces. Beyond generating a peculiar disanalogy with ordinary perception, the view would have other problems. For one, it does not seem to fit with our case judgments about when our desires and/or emotions are well-functioning (see the examples above). We often judge that our emotions are well-functioning, accurate responses to value, even though the agent has not ruled out that some defeating condition has not obtained. Setting aside comparisons with perception and case judgments, the view that affect has the function of responding to evaluative necessities just seems false. What would seem to be important from an evolutionary perspective is that affective experiences 68 respond to conditions which indicate the presence of the relevant values in this world. To give a concrete illustration, fear will typically trigger on the basis of limited information that far from guarantees the presence of a threat to our well-being (or whatever value fear presents). 118 And this is presumably because if we required great evidence of a threat before being afraid, we’d often end up dead. 119 If some emotional or desiderative experience (much less the full suite of such experiences) ever ended up being such as to respond (when well-functioning) only to conditions which guarantee the presence of the property the experience is about, this would be a surprising bit of happenstance rather than something our theory of content determination would lead us to predict. Barring empirical evidence of systematic happenstance, then, sentimental perceptualists who wish to model their answer to the content challenge on ordinary perception should leave open the possibility of basic knowledge of evaluative contingencies, i.e., they should be contingency-allowing. 120 2.4 Taking stock I started out this section by distinguishing necessity-requiring and contingency- allowing perceptualist theories. Necessity-requiring theories posit perceptual-like evaluative experiences that track how things must be evaluatively. Such experiences are supposed to provide us with basic knowledge of necessary evaluative truths. Contingency 118 If a sentimental perceptualist doubts that an emotion like fear presents any sort of value, then some other example could be chosen. For arguments that nearly all emotions are like fear in that they tend to trigger on the basis of limited information, see Brady [2013]. 119 See Ellsworth [1994]. Note that the justification we get from affect that triggers on the basis of highly limited non-evaluative information need not be knowledge-level justification. 120 Even if a sentimental perceptualist favors some alternative answer to the content challenge, the argument I give in this section can still be recovered. Appealing to such bio-functions is the most straightforward way for sentimental perceptualists to address evolutionary debunking arguments (e.g., Street [2006]). 69 allowing theories, in contrast, posit perceptual-like evaluative experiences sensitive to how things are in this world. Such experiences paradigmatically provide us with basic knowledge of evaluative contingencies. I then argued that intellectual perceptualism pairs well with a necessity-requiring theory but not a contingency-allowing one. Sentimental perceptualism is just the opposite. This result is encouraging, for it tells us that if we can settle the question of whether perceptualists should be necessity-requiring or contingency-allowing, then we can get leverage in the old debate between intellectual and sentimental perceptualism. The remainder of this chapter aims to show that we can at least make significant progress on the question of whether perceptualists should be necessity-requiring or contingency-allowing. I argue, in a preliminary way, that perceptualists need to allow basic knowledge of contingencies (section 4), and that, moreover, for the most part, perceptualists only need to allow basic knowledge of contingencies (section 5). (In fact, I think any foundationalist about evaluative knowledge should be contingency-allowing, but as we will see, the reasons are especially compelling for perceptualists.) I say “for the most part,” since, as I have already pointed out, contingency-allowing perceptualists should allow that we can “stumble into” basic knowledge of evaluative necessities. (Or, in other words, a contingency-requiring view is untenable.) Whether one finds the ensuing arguments compelling, I hope it can at least be agreed that the question of whether to be necessity-requiring or contingency-allowing is an important one we can make progress on. 70 3 Why Perceptualists Need Basic Knowledge of Contingencies 3.1 How perceptualists might try to avoid basic knowledge of contingencies The aim of this section is to argue that perceptualists should allow for basic knowledge of evaluative contingencies. I will address this question by first considering how a necessity-requiring perceptualist might think we can come to have non-basic knowledge of such contingencies. A necessity-requiring perceptualist needs to argue that we derive evaluative contingencies from evaluative necessities along with contingent non-evaluative truths, since no contingencies follow from necessities alone. Invoking what I call the standard model is the most obvious strategy for a perceptualist hoping to make sense a necessity- requiring view. On this picture, we deduce our knowledge of evaluative contingencies from our a priori knowledge of necessary evaluative principles and our a posteriori knowledge of the non-evaluative facts. 121 (Such deductions need not be transparent to the one who makes them.) These principles might take the form “if A, then B,” whereby the antecedent is non-evaluative and the consequent evaluative. For example, a candidate principle might look like this: if an agent, S, intentionally kills another person, then S acts wrongly. Assuming this principle is true, we can derive that the contingent evaluative truth that Jesse was wrong to kill Shannon (a contingent evaluative truth), if we know the principle (a necessary truth) and we know that Jesse intentionally killed Shannon (a contingent non-evaluative truth). Despite the standard model’s seductiveness, it is easy to see why some necessity- requiring theorists will be uneasy about it. Take the principle just mentioned: if an agent, 121 Schroeder [2005] uses ‘the standard model’ to denote a view about moral explanation. I am using it to denote a similar view about the order of knowing in evaluative inquiry. 71 S, intentionally kills another person, then S acts wrongly. This principle seems susceptible to clear counterexamples, e.g., a case in which an agent intentionally kills a well-known villain because that is the only means to prevent the villain from poisoning a city’s water source. It might be very hard to formulate an indefeasible principle about when it is wrong to kill, and even if we could, it might seem a stretch to say that ordinary agents use (or must use) such complicated principles to discover which killings are wrong. Indeed, it might be rare that we ever acquire knowledge of indefeasible principles about what we ought to do (e.g., about when we ought to keep promises, tell the truth, and so on). 122 A necessity-requiring perceptualist may argue that the only necessary evaluative truths we know are about which considerations count in favor of which “all- things-considered” verdicts. For example, we might know principles about normative reasons – say, if an agent promises to φ, then the agent has a reason to φ – which help us to discover facts about what agents ought to do. “All-things-considered” contingent evaluative truths will not be entailed by the necessary evaluative principles we know along with the contingent non-evaluative truths. However, the necessary evaluative principles we know do in some way support (e.g., render more probable) certain all- things-considered judgments. Call this the support model. I will not explore support models in more detail here, since the problem such models run into does not turn on details about how it is developed. There is a common problem for both standard and support models. Each requires that whenever we know a contingent evaluative truth, we know some necessary evaluative 122 This thought is familiar from W.D. Ross [2002]. 72 truth which we use to support it (deductively or non-deductively). 123 But I will argue that it is implausible we always arrive at knowledge of contingencies in this way. 3.2 A problem for necessity-requiring perceptualism When coupled with perceptualism, standard and support models run into a simple problem. The problem is one of extensional inadequacy; they predict that people lack evaluative knowledge when they clearly have it. Let’s begin with an example. Suppose Sasha sees her elderly neighbor, Hans, struggling to cross the street. Sasha is very confident that it would be good to help struggling Hans to cross the street. And she is right to be confident about this evaluative contingency. If we know any evaluative truths, it seems we should allow we know ones like this. The standard and support models require that Sasha know some necessary evaluative truth which she uses to help justify her particular judgment. But she might be deeply unsure about any such truths. One candidate necessary truth, friendly to the standard model, is this one: if an agent is struggling to achieve her ends and another agent, S, is able to help, then it would be good for S to do so. 124 But Sasha might never have thought about this principle. And, in any 123 The argument I give works against other models necessity-requiring theorists might adopt, too. Elijah Chudnoff [forthcoming] defends what he calls low-level intuitions, which consist in seeing a general evaluative (or mathematical, etc.) truth by seeing a particular one. (I assume that Chudnoff means for the general truths to be necessary and the particular ones contingent, but I won’t try to settle the interpretive question.) According to Chudnoff, when we have low-level intuitions, “The same experience puts you in a position to learn about the general and the particular. And though the general has some epistemic priority, this priority does not take the form of epistemic dependence on background beliefs.” I have no direct argument against low-level intellectual intuitions. But my argument does suggest that there are certain instances of evaluative knowledge in which the necessary does not have epistemic priority over the contingent. And that is all I need, as far as the argument in this section goes. 124 We could formulate a similar proposition that is not in the form of a conditional, e.g., helping agents struggling to achieve their ends is good. I formulate candidate necessities in this section conditionally, but the arguments do not depend on that decision. 73 case, the principle is clearly false, since it is not good to help some agents struggling to achieve their ends (e.g., axe murderers). 125 A defender of the support model might propose a weakening of this principle: if an agent is struggling to achieve her ends and another agent, S, is able to help, then S has a reason to do so. But this principle is also questionable. It does not seem as if we have any reason to help struggling axe murderers to achieve their ends. 126 More importantly, though, it also does not seem as if Sasha needs to know any such principle to know it would be good to help Hans to cross the street. Sasha knows that it would be good to help her struggling neighbor to cross the street, but, intuitively, she need not know any necessary evaluative truths which she might use to support her belief. That is a big problem for the standard and support models. A necessity-requiring perceptualist who defends a standard or support model might try to resist my analysis of the case. One response would be to argue that Sasha must be covertly relying on some necessary truth. But this is an especially bad move for a perceptualist to make, since perceptualism says that when we have basic justification for believing some necessity, the truth must be presented to us in a perceptual-like way (see the Introduction). This means that if an agent forms a belief about an evaluative contingency in part on the basis of her justified belief in a necessity, then she must have (or at least have had) attended to the principle. A second strategy for resisting involves going specific. In my assessment of the case, I imagined Sasha’s being confused about highly general necessary principles. But 125 One might think we can just add that the agent struggling to achieve her ends has to have a good end. But this move does not really help. By putting evaluative content into the antecedent of the principle, we make it too trivial for Sasha’s purposes. For such a principle to be of any use, she would need some evaluative insight into which ends are good. 126 Schroeder [2007a] makes a reasonable case that we need to be suspicious of our intuitions about when there is not a reason. But the point I am making here is about the questionability of the principle. Nothing I say here hinges on whether it is true or not. 74 a necessity-requiring perceptualist who goes specific argues that Sasha might well know some less general necessary truth. For example, perhaps she knows this: if a close friend of agent S is struggling to cross a street and S is easily able to help, then it would to some extent be good for S to help. But even this principle might be false, since a person might have a close friend who aims to cross the street because doing so is part of a plan to poison a friendly but noisy dog. It does not seem as if Sasha must have in mind, much less know, any such specific principle. 127, 128 To insist that she must would be to set the bar on knowledge too high. Visitors at the zoo can know that the animal they are observing is a zebra on the basis of their visual experience as of a zebra; they need not also experience it as not being a painted mule, even though if it were it would not be a zebra. 129 For similar reasons, to know on the basis of her experience that it would be good to help her struggling neighbor to cross the street, Sasha need not experience defeating conditions as not being present. But suppose that Sasha imagines the scenario. Some philosophers assume that knowledge obtained through the imagination in this way is going to be of a necessity. 130 But it is important to notice that the perceptualist can, and should, insist that this need not be so; Sasha could have basic knowledge of an evaluative contingency even if she responds to an imagining. Upon imagining a scenario in which her elderly neighbor, 127 A similar argument can be offered for any candidate principle. One might think, for instance, that the following is more likely true: if an agent is struggling to achieve her ends, the achievement of which would harm no one, and another agent, S, is able to help, then S has a reason to do so. Even if this principle is true (and I doubt it is), Sasha might well be confused about whether it is, and it is highly counterintuitive to insist that agents confused about the truth of such principles cannot know contingent evaluative truths of the sort Sasha seems to know. 128 The supervenience of the evaluative on the non-evaluative ensures that some such complicated principle has to be true. Necessary evaluative truths are also presumably in some sense metaphysically more basic than contingencies. But we need not model the epistemology of value after its metaphysics. 129 This is a variation on an example originally presented by Dretske [1970]. 130 See, for instance, McKeever and Ridge [2006]. 75 Hans, is struggling to cross the street, she comes to believe that if the scenario she imagines were to come about, then it would be good to help him. Whether this counterfactual is true depends on whether in the closest possible world in which the imagined scenario comes about the consequent (that it would be good to help him) is also true. Presumably the closest world won’t be one in which helping him turns out not to be good (e.g., the world in which he is crossing the street to poison a friendly but noisy dog). However, since the actual world could be such that the world closest to the actual world is a world in which it is not good for Sasha to help, the counterfactual is only contingently true. The fact that Sasha is imagining the scenario rather than actually visually experiencing it does not make it more likely that Sasha needs to know any necessary principles. The example I have been discussing is just the tip of an iceberg; it isn’t an exceptional case of evaluative knowledge. If Sasha need not know any necessary evaluative truths in order to know that it would be good to help her elderly neighbor, Hans, to cross the street, it is hard to see why she would need to know any necessary evaluative truths in order to know, say, that it was wrong of Jesse to kill Shannon or that it would be bad for Claude to steal the watch. Sasha might be confused about the conditions under which killing is always wrong or even prima facie wrong. 131 Similarly, Sasha might be confused about the conditions under which stealing is always bad or even prima facie bad. A perceptualist who goes for the standard or support models has a theory which can be expected to predict that people lack evaluative knowledge when we have a 131 As a concession to the necessity-requiring theorists, I allow that for Sasha to know the conditions under which killing is necessarily wrong she need not recognize that those conditions necessitate the wrongness of killings. 76 strong pretheoretical intuition that they have it. If perceptualists allow that there are perceptual-like experiences that immediately justify evaluative beliefs about contingencies, then we open up the possibility of a more extensionally adequate theory that predicts people really do know the contingent evaluative truths we pretheoretically take for granted that they do. In sum, necessity-requiring perceptualists appear forced to be revisionary about when people have knowledge, for there are cases in which it appears people know contingent evaluative truths without knowing any necessary ones. We should try to avoid being revisionary if we can. 4 Toward a Contingency-Allowing Value Epistemology The aim in this section is to illustrate why we should expect that perceptualists can use basic, perceptual-like knowledge of evaluative contingencies to explain our knowledge of evaluative necessities. Perceptualists should thus be contingency-allowing, for on a perceptualist picture, we’d need basic knowledge of contingencies and, moreover, we’d only need to allow basic knowledge of necessities when we stumble into such knowledge (see section 3.1). That is, perceptualist only needs to posit evaluative experiences that, when well-functioning, are sensitive to how things are in this world. 4.1 The picture It is common to use actual or imagined cases to test candidate necessary principles. When we find a case in which a principle seems to get the wrong result, we often give up the principle, unless we have some reason to be skeptical about the response 77 to the case at hand. 132 Here is an illustration of the procedure, using the following principle: necessarily, if an agent, S, intentionally tortures a baby merely for fun, then S acts wrongly. Perhaps we think hard about this principle and come up with no counterexamples. In that case, many ethicists will agree (as seems reasonable) that we get some justification for believing that the principle is true. Now consider a second candidate principle: necessarily, if an agent, S, intentionally kills another agent, then S acts wrongly. We might initially come up with this principle when we notice that in many cases, agents seem to act wrongly by intentionally killing other agents. But then we imagine a case in which an agent intentionally kills a well-known villain because that is the only means to prevent the villain from poisoning a city’s water source. It seems that the killer in this case would be acting rightly. 133 On a standard picture, then, the principle is revealed to be false, unless we have some reason to be skeptical about our response to the imagined case. What about this idea that we have reason to be skeptical of some responses to contingencies? An obvious type of case in which we should be skeptical about our perceptual-like evaluative experiences (familiar from evaluative and non-evaluative inquiry alike) are cases in which we have a personal stake. For example, an extremely wealthy businessman who is considering whether it is permissible for him to hoard 132 This method relies on a common, but controversial, assumption that the imagination can be at least a decent guide to possibility. If the imagination is not such a good guide, then the contingency-allowing perceptualist need not give up her view. She will say that we just do not have much knowledge of (non- analytic) necessary evaluative truths, or at least we cannot know of any truth that it is necessary. Perhaps we just know fairly general ceteris paribus principles. That seems to me not such a bad result. 133 A contingency-allowing perceptualist believes that what we experience as true in imagined cases is a counterfactual conditional. Such counterfactual conditions are (normally) only contingently true. This is because we evaluate the conditional by going to the nearest world in which the antecedent is true and see whether, in that world, the consequent is also true. This allows that there may be more distant worlds in which the antecedent is true and the consequent false. See section 3.2. 78 billions of dollars while his workers scrape by at minimum wage may experience this inequality as permissible. 134 When we are personally interested in the case we are evaluating, or are otherwise skeptical about one of our evaluative seemings, the right move is often to consider other cases. The businessman might try to figure out why his extreme stockpiling of wealth strikes him as permissible. If the reason it strikes him as permissible is because his workers “voluntarily choose” to work for him, then he ought to get clear on the sense in which his workers have a choice and then imagine other cases. He might imagine workers in other countries, see whether the conditions of those workers experience strikes him as acceptable, and then try to figure out whether there is a relevant difference. If the conditions of those workers strikes him as unacceptable, and he can find no relevant difference, then if he is honest he should toss out his judgment about the case in which he has a personal interest. He might also use fictional cases, or shift his perspective, imaginatively occupying his workers’ perspective (to the best of his abilities) to see how things strike him from that angle. Another option, not to be overlooked, would be for him to see how others, especially those who have no personal stake in the situation (but nonetheless understand it well), respond to it. In general, a perceptualist who accepts a contingency-allowing value epistemology seems to be able to make sense of the idea that we can to some extent correct for questionable evaluative experiences by shifting to new cases, shifting perspective, and asking others. There is a reasonable case to be made that this familiar picture of how we justify evaluative necessities is all perceptualists need to explain much of our knowledge of 134 A similar phenomenon arguably occurs in ordinary perceptual experience. A young man might experience a person in the distance as his lover because he deeply longs to see his lover. His interest in finding his lover makes his visual experiences less reliable (with respect to the matter he has an interest in). Similar examples are offered in Siegel [2012]. 79 evaluative necessities; we only need to allow basic knowledge of a necessity when we stumble into such knowledge. The first point to notice is that it is plausible that we come up with general principles, at least in most cases, by first considering cases. Here is Robert Audi, explaining Ross’s view as well as his own: [Ross] holds that we must attend to particular cases in order to determine what generalizations hold, even if it is repeatable features of those cases, such as their being acts of promising, that reveal the general truths we reach through reflection on the cases. This is a point not about what can be known but about the order of knowing; our basic moral knowledge – even of prima facie duties – comes from reflection on particular cases… ([1997: 36 – 7]) Ross and Audi do seem correct that we do not usually concoct evaluative principles in a way detached from our involvement in, or our thinking about, particular cases. It is a good result for the contingency-allowing perceptualist that our thinking about principles is largely shaped by our thinking about cases. The contingency-allowing theorist can say more. She can defend a way of rethinking the common rationalist thought that truly understanding a principle often requires thinking through cases. 135 Here is a principle, which is well-known from Peter Singer’s work: Necessarily, if an agent can prevent suffering at only a trivial cost to herself, then she has good reason to do so. 136 On one perceptualist picture, we can know this principle if our understanding of the principle leads us to experience it as true and, moreover, we believe it on the basis of that experience. However, those who defend this picture normally allow that it is difficult for us to really understand such a principle without thinking through cases, e.g., Singer’s example of an agent who sees a drowning 135 See, for instance, Audi [1997] and Nye [2015]. 136 This principle is entailed by stronger principles Singer defends. See Singer [1972]. 80 child and must decide whether to save the child, which would ruin his new shoes, or let the child die. 137 A contingency-allowing perceptualist says that such cases are not really helping us to understand the principle – at least where understanding is a matter of conceptual grasp – but rather helping us to justify the principle. 138 And that’s not a surprising view. It is standardly the case that for any principle we are highly confident in, we have beliefs about cases that we are at least as (though usually more) confident in that can plausibly be understood as supporting the belief in the principle. 139 Singer’s principle and example illustrate this common state of affairs. My aim in this section so far has been to sketch a picture (not necessarily the only possible such picture) on our knowledge of necessities is downstream from our basic knowledge of contingencies. I have already presented a story about how our responses to cases (actual and imagined) can present contingent evaluative truths (section 3) and also that we need to allow for basic knowledge of contingencies (section 4). Given perceptual-like evaluative experiences that track how things are, we do not need to posit perceptual-like evaluative experiences that track how things must be. I now turn to a worry that proponents of the necessity-requiring view may already have in mind. 137 Nye [2015]. 138 Some principles may include what George Bealer [1998] calls “semantically unstable” terms. The meaning of semantically unstable terms can vary across linguistic groups in qualitatively similar epistemic situations. Natural kind terms, for example, are paradigmatically unstable. Linguistically competent people may fail to understand evaluative principles that include semantically unstable terms. 139 Prichard [1912] remarks, “[S]uppose we come genuinely to doubt whether we ought, for example, to pay our debts owing to a genuine doubt whether our previous conviction that we ought to do so is true, a doubt which can, in fact only arise if we fail to remember the real nature of what we now call our past conviction. The only remedy lies in actual getting into a situation which occasions the obligation, or -- if our imagination be strong enough -- in imagining ourselves in that situation, and then letting our moral capacities of thinking do their work.” 81 4.2 Directly plausible principles As I have already noted, perceptualists who believe in basic knowledge of evaluative necessities hold that our justification for believing those necessities can become independent of our justification for believing anything about evaluative contingencies. Howard Nye has, I think, done a particularly good job of capturing why they think this; it supposed to be because some principles are directly plausible. 140 A principle is directly plausible if it seems true independently of its relationship (logical or otherwise) to anything else; directly plausible principles seem true in themselves. An agent knows a directly plausible principle when the principle strikes an agent as true in itself, the agent believes it on that basis, and the principle is true. Must a perceptualist allow that some evaluative necessities are directly plausible in this way? I doubt it. 141 The most plausible candidates for directly plausible principles are ones that are incredibly obvious. Here is one of Nye’s prime examples. A 19th century white slaveholder has the perceptual-like evaluative experience that it is permissible to force Tom, a black man, into slavery. His intuition here is being influenced by Tom’s race. A natural way the slaveholder could come to see his mistake, Nye argues, is by getting clear about what race per se is, i.e., phenotypic characteristics such as skin color and hair texture which are the result of one’s area of ancestry. He would (hopefully) directly see this to be of no intrinsic ethical significance and would likely search for alternative justifications, e.g., his “greater intelligence,” or so Nye argues. 140 Nye [2015] 141 If perceptualists need to make room for directly plausible principles, then given the arguments in section 3, they may end up needing to defend an awkward combination of intellectual and sentimental perceptualism. 82 There is a competing explanation of what is going on with this example, however, which is friendly to the contingency-allowing perceptualist. Readers of Nye’s work already have a large collection of beliefs about which kinds of things can be of intrinsic ethical significance (e.g., pleasure/pain, certain special relationships, knowledge). And one of the things not included on any of these lists is (hopefully) race. So when we are invited to consider the principle that race per se is ethically irrelevant, we immediately judge that the principle is true. An explanation friendly to the contingency-allowing theorist is that we are immediately inclined to judge the principle true because of our extensive set of background beliefs about which kinds of things can be of intrinsic ethical significance. But what of the slaveholder? It seems that he could agree with us (i.e., that race is of no intrinsic ethical significance) but still continue to experience enslaving Tom as permissible. And so would not contingency-allowing theorists be unable to explain his justification for believing the general principle? No. The slaveholder might agree that he was confused about the nature of race (he didn’t take it to be about phenotypic traits), but he will presumably continue to insist that whatever he thought race was is of intrinsic ethical significance; race is just a “mark,” the slaveholder might insist, of the divinely ordained superiority and inferiority of different people. When we get clear about whatever factor the slaveholder took himself to be responding to, it strikes me as doubtful he will see directly – without the help of any other background evaluative beliefs or reflection on other cases – that it (e.g., the divine rankings of peoples) is of no importance. After all, the principle he is asked to consider is, in the end, not about what he took to be 83 important. And his grounds for believing phenotypic traits are of not intrinsic ethical import could (as easily as ours) depend on his background beliefs about what is valuable. I suspect the contingency-allowing explanation of what is going on in the slaveholder example is at least in one respect better than the explanation in terms of directly plausible principles. Suppose we are thinking about controversial principles, e.g., that we have a fundamental duty to be autonomous, that we ought to act so as to maximize net pleasure, or that torturing innocent human beings is always wrong. Assuming our ordinary practices are not entirely off the mark, it does not appear that we ever really justify a belief in any of these principles merely by experiencing it as true. When we are deciding whether to believe these principles, we should try to imagine whether they have unacceptable implications for particular cases (emphasizing cases that we have no personal stake in). But then if this is how it works for principles we find controversial (i.e., our knowledge of them, if we have it, is epistemically dependent on our thinking about cases), this gives us some reason to prefer the contingency-allowing perceptualist’s account of what is happening when we evaluate uncontroversial principles. The contingency-allowing perceptualist has a simple, unified account of how necessary principles are justified. 5 Conclusion: A Historical Note I close this chapter with a historical observation. The argument I have given bears a structural resemblance to an argument that Francis Hutcheson pressed against Gilbert Burnet. 142 Hutcheson and Burnet agreed that being virtuous requires a commitment to 142 See Burnet and Hutcheson [1971]. 84 promoting the general happiness, but they disagreed over how we learn that this is so. Burnet argued that the proposition is self-evident, knowable on the basis of reason alone. Hutcheson objected that this cannot be so, for there is no logical contradiction in supposing that lesser quantity of happiness is better than a greater quantity. 143 To justify the belief that the greater happiness is better, we need a prior affection for happiness. Hutcheson’s reply to Burnet relies on the assumption that reason alone is only capable of helping us to know propositions guaranteed to be true by their meanings alone (analyticities). Faced with this objection, the strategy for rationalists, or intellectual perceptualists more narrowly, has typically been to defend the synthetic a priori. 144 To this day, the existence of synthetic a priori knowledge is hotly contested. 145 Here is a way of repackaging Hutcheson’s defense (in this instance) of a sentimentalist epistemology. We start out with two conditional theses. First, if our basic evaluative knowledge is only ever analytic, then we should be rationalists. But if our basic evaluative knowledge is sometimes substantive, then we should be sentimentalists. As it turns out, much of our basic evaluative knowledge is substantive. Thus we should be sentimentalists. My strategy in this chapter was structurally similar to Hutcheson’s, although I only aimed to adjudicate between perceptualist varieties of sentimentalism and rationalism. I argued that if perceptualists should be necessity-requiring, then they should be intellectual perceptualists. But if perceptualists should be contingency-allowing, then they should be sentimental perceptualists. I then argued in a preliminary way that a 143 For a more detailed discussion of this dispute between Hutcheson and Burnet, see Gill [2006]. 144 See, for example, Huemer [2005]. 145 For discussion of the synthetic a priori in the context of ethics, see Väyrynen [2008]. 85 contingency-allowing perceptualism is preferable to a necessity-requiring perceptualism. This is because perceptualists need to allow basic knowledge of evaluative contingencies, and moreover, we can use basic knowledge of contingencies to explain our knowledge of necessities. In crucial distinction from Hutcheson, the argument I give for a sentimentalist epistemology is (fortunately) not hostage to well-worn, difficult debates about the synthetic a priori. My view, then, is that perceptualists should be sentimental perceptualists, on the grounds that sentimental perceptualism, but not intellectual perceptualism, pairs with the most attractive view about the modal status of our basic evaluative knowledge. Or, if you are inclined to reject sentimental perceptualism on the basis of some other line of argument (see section 1), then perhaps this chapter gives you a reason to reject perceptualism altogether. 146 But whether one finds the arguments for a contingency- allowing over necessity-requiring perceptualism persuasive, I hope the strategy advanced here at least suggests that significant headway can be made on the question. And, so long as that is true, we should be optimistic about finally resolving the debate between intellectual and sentimental perceptualists. 146 I believe that the familiar arguments against sentimental perceptualism are unpersuasive, but I cannot hope to defend such a claim here. See my [forthcoming b] and [unpublished b] for responses to major lines of objection. 86 Chapter 3: Toward the Best Version of Sentimental Perceptualism, or How Sentimental Perceptualists Can Solve the Content Challenge 0 Introduction ‘Perceptualism’ refers to a broad family of views, and the arguments in Chapters 1 and 2 make the case that sentimental perceptualism is the most promising such view. But sentimental perceptualism is itself a broad church. Over the course of the next three chapters, then, I continue zeroing in on the best version of perceptualism. I do this by considering what kind of commitments a sentimental perceptualist needs to take on board to address one of the major, outstanding challenges for the view: sentimental perceptualists are yet to explain how affective experiences have come, or at least could have come, to have evaluative content. Because it is so controversial whether they have such content in the first place, the failure to come up with an attractive content- determination story easily leads to doubts about whether we should have posited such content in the first place (the content challenge). This chapter explains how best to understand the nature of the challenge, and then explains, given that understanding, the form that an attempt to resolve it should take. In short, the challenge is best understood as a demand to show that it is reasonable to expect that affect gets its evaluative content in the same way that ordinary perception gets its content. And the best way to deal with the challenge, so understood, is to adopt a fairly demanding account how ordinary perception gets its content: if affect stands in the relations to value required by the demanding account, then we have excellent evidence that affect is about value, whatever the right theory of content determination turns out to 87 be. To preview, the relevant relations are causal, covariational, and bio-functional. The claim will be that if affect has the biological function of standing in causal-covariational relations with value, then we have outstanding evidence that affect is about value. 1 The Content Challenge There is no shortage of arguments for the thesis that affect has evaluative content. One popular brand of argument is straightforwardly phenomenological: careful reflection on affective experiences suggests that properly characterizing them requires us to make reference to evaluative properties. 147 Another popular brand of argument is action- theoretic: taking emotions and/or desires to have evaluative content helps us to understand why they can play a rationalizing role in action. 148 There is a cottage industry centered around developments and objections to such arguments. But comparatively little has been said about how affect could have come to have evaluative content in the first place. And this leads to a serious worry about sentimental perceptualism’s psychology. Absent an attractive explanation of how this could work, opponents of sentimental perceptualism may insist that this tips the scales in favor of those who wish to resist the idea that affect has evaluative content. This problem has been explicitly raised in the literature. Karl Schafer [2013: 268 – 69] argues that it is “the most serious problem for any account that accepts [that desires have evaluative content] is the problem of explaining how and why it is that desires involve perceptions with a normative or evaluative content.” 149 Mark Schroeder raises 147 See Roberts [2003] on emotions and Oddie [2005] on desires. 148 See Döring [2003] on emotion and Quinn [1994] on desire. 149 Schafer mentions that Dennis Stampe’s [1987] is one of the only advocates of the view that desires are perceptions of value who actually tries to address the challenge. Stampe’s explanation for why we should 88 the same problem, focusing in particular on Sergio Tenenbaum’s brand of sentimental perceptualism, labeled in what follows as the scholastic view. Schroeder [2008: 127] asks, “[I]f the good is independent of desire, as the scholastic view claims…then how does it get to be about it?” He then helpfully expounds about the nature of the question: Compare the question of how your greenish perceptual experiences get to be about green, rather than about some other thing, such as orange or square. There is a great disagreement in the theory of content determination about just how this happens. But there is widespread agreement that it doesn’t happen simply by magic. Your perceptual state has to somehow latch on to green. What different theories of content determination differ on is exactly how this “latch” is effected. Causal theories require a causal relationship between green and your perceptual state; co-variation theories require that they co-vary in some nomic way; teleological theories require that it be the function—whether supplied by design or by natural selection—of your perceptual state to be tokened in the presence of green; and inferential role theories require that there be some kind of structural match between the inferential relations that your perceptual state actually stands in and the ones that would be licensed if it were about green. [2008: 127 – 28] In his reply, Tenenbaum attempts to sidestep the objection by insisting that he does not really think that desires have evaluative content, even though he accepts that desires are appearances of the good. There is a serious puzzle about whether Tenenbaum can drive a gap between appearances and content in a way that allows him to dodge the content challenge, but let’s set questions about Tenenbaum’s view, including whether it should really count as a version of sentimental perceptualism, aside. Schroeder’s point is that take desires to be value perceptions is that desires co-vary with values (at least in ideal conditions). Schafer [2013: 269] points out, rightly I think, that “it is highly doubtful that a satisfactory account of any form of mental content can be provided solely on the basis of facts about reliable indication.” It will become clearer below why Schafer is almost surely right. 89 anyone who thinks that desires have evaluative content needs to be able to explain how that happens. Sentimental perceptualists need to address the challenge head on. 2 How to Answer the Content Challenge Sentimental perceptualists hold that affect is about value in much the way that visual experiences are about greenness or squareness. The presumption, then, is that such theorists think that affect acquires its evaluative content in structurally the same way that ordinary perceptual experiences do. In any case, arguing as such would be in the spirit of the sentimental perceptualist approach, which is to reduce puzzlement about evaluative domain by comparing it to ordinary empirical domain. In seeking an account modeled on perception, we need to be careful, since different perceptual experiences, even within the same modality, may get their content in different ways. I will illustrate using vision. Everyone agrees that we can visually perceive color, shape, and motion. Many would add to this list agency, prey, and food, among other things. Some of the most highly contested candidates for content, as I have observed in earlier chapters, include artificial and natural kinds. Those who defend such highly controversial content often appeal to other non-perceptual states (normally beliefs) that represent the same properties and then insist that those states alter (penetrate) visual content, generating visual perceptions of such properties. Visual perceptions that get their content in this way are arguably of less epistemological interest – since they seem justification conferring only if the representational states that generate them are – and so I will not be interested in an answer to the content challenge which shows that affect can 90 come to acquire evaluative content in this way. 150 An answer to the content challenge likely to preserve the epistemological import of affect is one that does not rely on appeal to other evaluative representational experiences to explain how affect comes to acquire its evaluative content. Such experiences would arguably be suited for getting us in on “the ground level,” epistemologically speaking, when it comes to value. The sentimental perceptualist, then, should aim to model her content-determination story on that of perceptual states that do not depend for their content on an agent’s being able to represent that content elsewhere. (For simplicity, I will not continually make this qualification throughout.) When it comes to the question of how to address the content challenge, the sentimental perceptualist faces a dialectical puzzle. Owing to the fact that it is controversial how perception gets its content, there are several different options for how to approach the content challenge. One is to list plausible theories of how perception comes to have its content and then to show that a similar story could be told about affect. But this would be a nauseating strategy. It would require a book (maybe more) to show that sentimental perceptualism’s psychological picture is defensible within the framework of each reasonable approach to perceptual content-determination. At the same time, it would be deeply unsatisfying for me to show that sentimental perceptualism is compatible with my own favored approach to perceptual content determination, unless I were prepared to argue that I were really working with the best theory (another book). I adopt a more nuanced, and I think reasonable, approach. Opponents of sentimental perceptualism do not express skepticism about the prospects for offering an 150 I discuss some reasons to doubt the epistemological significance of value perceptions which acquire their content in this way in Chapters 1 and 2. 91 explanation of how affect has evaluative content by putting forward any favored theory of content-determination and then explaining why, given the theory, affect could not have evaluative content. I suspect the worry is not best conceived as particularly theory-laden. But at the same time, it is not simply request to show that the theory can fit with some theory of content-determination or other. The challenge is best thought of as a demand to show why it is reasonable to think affect has evaluative content, whatever the right theory of perceptual content-determination turns out to be. With that in mind, here is the approach I will take. Theories of perceptual content- determination tend to fall into one of the following camps: covariational, causal, and teleological. 151 And these different approaches can be combined in different ways (e.g., a theory that has teleological and causal requirements). What I will do is present the outlines of a demanding account that combines covariational, causal, and teleological commitments; I will then argue that if affect stands in all those relations to value (and does so in the right kind of way), it is reasonable that to hold that affect is about value, even if it turns out that some of the account I give is mistaken. If the account were simply too demanding (say, because it incorrectly requires covariation), then showing that affect stands in the relations would still be sufficient. But if the account is false not merely because it is overly demanding, then showing that affect is about value according to the (supposedly false) account would still provide evidence that affect is about value according to the true account. I will have more to say about this below. 151 There are also inferential role theories of content determination, but these are generally used to explain the content of beliefs rather than perceptual states. 92 3 Perceptual Content Determination As a starting point, many theorists have thought the notion of information important for perceptual content determination. 152 When a mental state carries information about something, this intuitively helps us to understand, at least to an extent, how the mental state could come to be about that something. For example, that perceptual representations of objects as square carry information about squareness seems explanatory of how such representations came to be about squareness. But what exactly is the information relation? A traditional way of cashing out what it is for one thing (mental state or otherwise) to carry information about another is in terms of covariation. 153 One thing, A, carries information about another thing, B, just in case A statistically co- varies, even weakly, with B. More concretely, smoke carries information about fire, because where there is smoke, there is (normally) fire. Sentimental perceptualists would do well to accept an information requirement, so understood, for perceptual content determination. Although not universally accepted, the sentimental perceptualist should not want her response to the content challenge to turn on the hope that information, so understood, is not important for content determination. 154 (In any case, the lack of an informational relation would make sentimental perceptualism a non-starter for explaining evaluative knowledge, never mind the content or armchair challenges.) 152 See, for instance, Dretske [1995], Burge [2010], and Neander [2012] and [2013]. 153 See Burge [2010]. Dretske [1981] defines the information-relation, or what he often calls “indication,” slightly differently. The subtle differences will not matter for our purposes. Neander [2012] defines information in terms of causation rather than covariation. Information in her sense will eventually be part of my account, too. 154 Papineau [1984] and Millikan [1984] deny that information is necessary. I should note, however, that what Papineau and Millikan deny, strictly speaking, is that information figures into the analysis of perceptual representation. I am not concerned here with giving a reductive analysis of representation, which I doubt can even be done; my only concern is with explaining how states with certain representational content came to be. 93 But it is widely agreed that we need more than a (simple) information-relation to explain perceptual content. The reason for this is that perception is about a much narrower range of things than perception carries information about: perceivers do not perceptually represent everything their perceptions covary with. One ingredient that is natural to add here, although it is not enough to complete the picture on its own (or so I would argue), is causation. Causal and covariational requirements can be joined together in an attractive way. Here’s how. Covariational relations often have an underlying explanation. In many cases, the underlying explanation will identify a causal relation between the two things (objects, properties, etc.) that covary. Thus a popular view says that perceptions stand in causal-tracking relations with what they are about. That causal- tracking relations would be important for perception is very intuitive; it helps us to see how a mental state latches on to external objects and properties. Sentimental perceptualists should accept that causation, in addition to mere covariation, is important for perceptual content determination. Adding in this requirement ensures that the account of content determination will have broader appeal. Unfortunately, causation and covariation do not seem sufficient for explaining perceptual representational content. I will (once again) use vision to illustrate. Visual perceptual experiences are about distal (environmental) particulars, properties, and relations; but any such representation is immediately triggered by proximal stimulations, notably light arrays of varying intensity that are registered on the retina. 155 Visual perceptions of a certain type covary with certain proximal stimulations and are also 155 Although retinal stimulation is the primary kind of input in visual experience, it is not the only input. For example, there is also proprioceptive input, which includes information about eye positioning. See Burge [2010: 89]. 94 caused by these proximal stimulations. However, a perceiver does not represent the proximal stimulation; what gets represented (at least in the case of vision) is almost always something about how the world is external to the perceiver. So if we want to understand why an observer represents, say, solidity, rather than the light arrays on the retina that cause the representation of solidity, we need more than causation and covariation. One way to try to solve this problem is to build a teleological component into the account of content determination. Why, we might ask, is a perceptual mental state about, say, solidity, rather than light arrays on the retina? It is because perceptions, or underlying perceptual systems, do not have the biological function of detecting light arrays on the retina. (Or, if we think that perceptions do have the biological function to detect such proximal stimulations, then, following Karen Neander [2012], we can add a qualification: our perceptual systems have the function of producing such perceptions (“solidity perceptions”) in response to certain light arrays because that is a means of producing the perceptions in response to what is solid and not vice-versa.) According to the account I favor, a perceptual state’s content gets fixed by what it has the biological function of standing in causal-covariational relations to. 156, 157 I suggest, then, that sentimental 156 If particular perceptual states do not have functions, then we can instead simply rephrase the point in terms of the underlying perceptual systems that give rise to particular perceptual states. See Neander [2013]. 157 Appealing to biological functions also helps to solve the so-called underdetermination problem for perception. The problem is that the same sorts of proximal stimulation are compatible with a variety of possible distal causes; and so it can at first seem puzzling why given proximal stimulation triggers one representation rather than another. Burge [2010: 344 – 45] has argued that any solution to this problem is going to need to appeal to formation principles, which are (in the case of vision) principles that describe how our visual system takes registrations of light intensities as input and then represents certain distal attributes. These transformations can also go from perceptual representations to other perceptual representations, as when perception of edges and surfaces leads to representations of solidity, even though edges and surfaces do not guarantee solidity. The presence of formation laws helps us to understand our ability to represent objects as the same despite variations in perspective, often referred to 95 perceptualists can give a reasonable answer to their content challenge by arguing that (some sufficient range of) affect has the biological function of standing in causal- covariational relations with value. 158 This account still faces stiff challenges. Are function, covariation, and causation (combined in the way that I just suggested) enough to narrow in on the right content? Consider a potential worry. Might it not turn out, say, that frogs’ visual representations have the function of tracking flies, are caused by flies, and covary with them? Such hypotheses would seem reasonable, given the omnipresence of flies in explanations of frog behavior and well-being. But even if all those hypotheses were true, we might still want to resist the idea that frogs ever represent flies. Burge argues that being able to represent natural kinds requires “an openness to a distinction between surface properties and underlying natures, an openness frogs cannot be expected to have.” 159 The general worry here (which is due to Fodor) for the account of content determination I have been sketching is that function is insufficient for individuating contents. 160 as perceptual constancies. The rim of the jar still appears to us to be circular even as we shift the angle at which we view it, causing dramatic variations in proximal stimulation. We explain the emergence of particular formation laws, and thereby the content of the representations they give rise to, by appeal to biological functions. The biological function of an item (e.g. a mental representation) is what the item was selected for. The idea that perceptual content is constrained by biological function immediately helps with the underdetermination problem. Here is an illustration. Frogs (to take an oft discussed case) have visual perceptions that facilitate preying on flies. What are the contents of the experiences that facilitate this behavior? We want to say that the frog represents bodies of such and such size and shape. But there are other candidates. The proximal stimulation is also compatible with the frog’s representing, say, undetached fly parts or temporal slices of flies. These candidates are ruled out because the frog’s perceptual faculties do not have the function of tracking undetached fly parts or temporal slices of flies. Settling on what exactly the frog’s capacities were selected for is a difficult question, one that requires detailed biological and zoological study; but the point here is just to illustrate how the appeal to biological function brings us closer to the correct content. 158 For views roughly along these lines, see Neander [2013], Prinz [2004], and Burge [2010]. Burge in particular appeals to bio-functions not in order to reduce content, as some others do, but merely to explain how perceptual experiences end up with certain content. Sentimental perceptualists can do something similar. 159 Burge [2010: 323] 160 Fodor [1990] 96 I am neutral about this objection. Sentimental perceptualist do not need to take a stand on such subtle debates from the content-determination literature. What they should be on the lookout for, however, are specific biological or psychological considerations that should lead us to think that even if affect stands in covariational, causal, and functional relationships with value, there is still some reason to suspect it does not figure into the content of affective experience. Burge mentions such a consideration in expressing doubt that frogs represent flies; his thought is that representing a natural kind property requires certain kind of cognitive sophistication we have reason to suspect frogs lack. 4 Closing Thoughts on the Dialectic Let’s return to an issue raised at the beginning of the chapter. Even though my approach to content determination is a demanding one, there are other, reasonable approaches to content determination; and this can lead to worry. Christopher Peacocke, for instance, defends what he calls the action-answerability account of content determination. This approach says that “the correct attribution of representational content to perceptual states is constitutively answerable to the range of properties of actual and counterfactually possible actions of their subject that those perceptual states are capable of explaining (in combination with other states).” 161 On this approach, the explanation for why an animal represents the shape and color of some fruit appeals to the fact that the animal’s behavior is counterfactually sensitive to the presence of such fruit; all else equal, it eats all and only fruit of that shape and color. This is a kind of covariation approach, 161 Peacocke [2014: 477 – 78] 97 but what I say above does not directly speak to it, since I say nothing about action- explanations. That said, Peacocke would agree (or so it seems to me) that if desires or emotions have the biological function of causally-covarying with value, then that would be evidence that the value figures into the content of the relevant experience. 162 My general point, then, is that even if one rejects the approach I have taken to content determination, it does not follow that it fails in its aim, which is to show that it is reasonable to maintain that affect has evaluative content (especially if one is moved by the phenomenological and action-theoretic arguments for such a view). I conclude, then, that sentimental perceptualists can answer the content challenge by showing that affect stands in causal, covariational, and functional relationships with value. More precisely, the claim is that challenge can be answered by showing that affect has the biological function of detecting (standing in causal-tracking relations with) value. It is worth emphasizing, given this dissertation’s broader concern to figure out the best perceptualist theory, that this way of approaching the content objection reveals it to be intimately related to three serious epistemological challenges. First, I take it as obvious that affect cannot be the foundation for our knowledge of value unless it to some extent co-varies with it. Second, as David Enoch [2011] has pointed out, if our evaluative beliefs are not caused by value, there is a serious puzzle about how we can explain the way in which our evaluative beliefs align with the evaluative truths. 163 And, third, evolutionary debunking arguments aim to appeal to the evolutionary origins of our evaluative beliefs 162 Peacocke [2014: 477] asks, “Is biological function not merely evidentially or epistemically relevant to the determination of content, but constitutively involved as a matter of the very nature of perceptual representation itself?” The assumption here is that biological function is at least evidential. 163 If it can be argued that values cause affective experiences, then it is an easy step to the conclusion that values cause evaluative beliefs, since affective experiences are certainly sometimes causes of evaluative beliefs and causation is a transitive relation. 98 to explain why we should expect such beliefs to be unreliable (at least given realist assumptions in evaluative metaphysics). 164 But if affective experiences get their content in the way I have argued that sentimental perceptualist should argue that they do, then there is a simple response to this argument: we can expect our evaluative beliefs to track value because affect plays a perceptual-like role in grounding those beliefs and, moreover, affective experiences have the function of tracking value. It is very important to notice, then, that the dialectic surrounding the content challenge, at least as I am conceiving it, brings us to the heart of things for sentimental perceptualism, both psychologically and (indirectly) epistemologically. 164 On evolutionary debunking arguments, see particularly Joyce [2006] and Street [2006]. 99 Chapter 4: Answering the Content Challenge, Part 1 0 Introduction Sentimental perceptualists can answer the content challenge by showing that affect has the biological function of standing in causal-covariational relations with value. We can helpfully divide this challenge into three tasks. The sentimental perceptualist needs to argue that affect (i) co-varies with value, (ii) can be caused by value, and (iii) has the function of standing in those relations with value. This chapter addresses (i) and (iii), in particular. 165 To keep things manageable, I focus principally on the case of desire. Furthermore, I hold fixed (at least at the start) that desires, at least a certain range of desires, have the biological function of covarying with some kind of value. The question will be what kind of value that most plausibly is. 166 (‘Value’, remember, is used broadly to refer not only to narrowly evaluative notions such as goodness but also normative properties/relations which concern action.) In the next chapter, I consider whether it is plausible that desires really do have the function of detecting the sort of value that they are most plausibly about. I also save for the next chapter the question of whether desires are ever caused by such values. Among those who believe that desires have evaluative content, the most common view is that desires involve representations of goodness. For example, Stampe [1987], Quinn [1993], Oddie [2005], and Brewer [2009] all defend some version of this view. I argue, however, that on the most natural ways of making these views precise, we will not 165 This chapter includes all that I will have to say about (i), but I will also have important things to say about (ii) in the next chapter. 166 It could turn out (though it in fact does not) that there is no kind of value which desires plausibly have the function of detecting. 100 be able to resolve the content challenge. This is because our desires fail to covary with goodness in certain important ways which make it difficult to maintain that they have the biological function of covarying with goodness. But sentimental perceptualists have a way forward. They will have a much easier time with the content challenge if they shift to the view that desires involve representations of the desirer’s practical reasons. 167 This view fits much better with how our desires pattern. 2 What Sort of Goodness? There is an old view that we desire “under the guise of the good.” Kant, in the Critique of Practical Reason, identified this position with Scholastic philosophy, in particular, labeling it “the formula of the schools.” 168 In recent decades, the guise of the good has enjoyed a revival, with philosophers defending it on a number of independent grounds. 169 Here is Graham Oddie: When I desire that P, P has a certain magnetic appeal for me. It presents itself to me as something needing to be pursued, or promoted, or embraced. Now the good just is that which needs to be pursued, or promoted, or embraced. So my desire that P involves P’s seeming good (seeming to be worth pursuing). So the desire that P look as though it just is the experience of P as good. [2005: 55] 167 Alex Gregory [2013] has argued that desires more plausibly involve representations of reasons than they do representations of goodness. The argument I offer here provides independent support for that general position. However, I should flag that his view is more precisely that desires involve judgments about reasons. This changes the dialectic surrounding the content challenge quite a bit, since content determination for judgment may work very differently than for perception. I do not try to answer whether arguments similar to the ones offered in this chapter carry over to a judgmentalist view about the evaluative content of desire. 168 I owe this reference to Tenenbaum [2007]. 169 For example, Brewer [2009] is motivated by action-theoretic considerations while Oddie [2005] is motivated by phenomenological and epistemological considerations. 101 Oddie’s view is that in desiring that P we represent P as good simpliciter, or (what I take to amount to the same thing) impersonally good. This is not the only possible version of the guise of the good, however. Consider two other ways in which we use ‘good’. 170 We also speak of what is good for something or other. And there are important differences between what is good for (say) a person and what is impersonally good. For example, it may be impersonally good that the war criminal be severely punished but not good for the war criminal that he be so punished. We also use ‘good’ in an attributive sense, e.g., when we say ‘John is a good thief’ or ‘that is a good knife’. When we use ‘good’ in this sense, the term functions as a predicate modifier rather than a predicate. Given the variety of ways in which we use ‘good’, the guise of the good theorist needs to tell us in what sense desires involve experiences of their objects, or something systematically related to their object, as good. 171, 172 170 On the different ways we use ‘good’, see Schroeder [2012]. There is an important question of how all of the different senses of ‘good’ are related to one another. For the purposes of this chapter, I set such complexities aside. 171 We should not assume that desires can only involve evaluative perceptions if the object of the desire is what is perceived as valuable. As we shall see below, the best versions of the view that desires represent value do not have this structure. 172 In contemporary philosophy, it is common to talk about agent-relative goodness. This notion of goodness arguably does not correspond to any ordinary notions of goodness. (See Schroeder [2007b]). But if agent-relative goodness is a philosopher’s invention, then we can set it aside here. It is not going to be attractive for sentimental perceptualists to say that the foundational evaluative knowledge desires help us to achieve is knowledge of agent-relative goodness, for this would mean that ordinary folk do not even have a concept which corresponds to the kind of value about which they supposedly have perceptual-like knowledge. One way to try to make sense of agent-relative value is in terms of desirability (or what we have reason to desire). (Cf. Portmore [2011: 125]). But if this is the right way to think about agent-relative value, then, once again, sentimentalists should not think that desires are about such value. There are at least two major reasons why. First, the character of desiderative experience does not seem to be such as to by its very nature make salient thoughts about itself; but that is what we would be forced to say if we hold that desires represent desirability in a perceptual-like way. For example, if I want to drink a coke, my desire makes salient thoughts about the sweet, carbonated taste, the fact that there are cokes in the fridge, and so on; but it does not seem to make salient the thought that it is appropriate to desire a coke. Secondly, if desires are about desirability, then it is hard to see how we could ever understand desires as having the function of detecting the value that they are about. This is because there would not have been anything for desires to evolve to detect prior to the existence of desires. 102 We can straightway set aside the view that desires represent the attributive good. To see why, consider the shape that such a view would have to take. As indicated above, the attributive good is applied to things. But desires are not thing-oriented attitudes, even though we occasionally talk as if this were so. For example, we might say, “Helmut wants the cake,” as if Helmut’s desire took a cake as its object. But the object of the desire is more precisely characterized as either a property or proposition: Helmut wants to eat cake (a property), or wants that he eats a cake (a proposition). Taking desire to be at root a proposition-oriented or property-oriented attitude can help us to understand why Helmut’s desire leads him to eat chocolate rather than, say, wear it on his head for decoration. 173 Despite what some theorists have recently argued 174 , then, desires are not best understood as attitudes toward things. 175 Thus if desires involve representations of the attributive good, it will not be the object of desire (which is a property or proposition) which is represented as good. Of course, a guise of the good theory does not require that the object of the desire be represented as good (even if this is what proponents of the view have almost invariably assumed), for it could be that desires involve perceptions of something systematically related to the object of the desire as good. So there is the possibility that desires involve an evaluative perception of something as attributively good, even though it is not the object of desire which appears attributively good. But this view looks hopeless, for something which is good in the attributive sense is a good member of its kind, and it is 173 See Sinhababu [2015: 7] 174 See Brewer [2006] and Thagard [2006]. 175 Finlay [2014] and Sinhababu [2015] offer examples along these lines. Sinhababu explains in great detail why proponents of the unorthodox view that desires are thing-oriented have gone wrong. I will not engage in a lengthy recounting of that debate here, however. 103 hard to imagine what kind of thing desires could systematically represent as good. So in what follows, I set aside attributive goodness as a candidate for developing the guise of the good theory. Showing why the guise of the good theorist should not appeal to impersonal or personal goodness is a more delicate task. The first thing that we need to do is to get a better understanding of how each view is supposed to work. And in order to understand the nuts and bolts of the two views, we first need to figure out whether desires are best conceived as a proposition-oriented or property-oriented attitude. 3 Desires as Property-Oriented We talk about desires as if they were attitudes capable of taking three different kinds of object: things, propositions, and properties. For example, we can say any of the following: (i) Brent desires an ice-cream, (ii) Brent desires that he (Brent) eats an ice- cream, or (ii) Brent desires to eat an ice-cream. But however we may talk, it is not psychologically attractive to think that there are really three different types of desire. We have already seen, for example, how desires we initially characterize as thing-oriented are at the psychological level best thought of as oriented to properties or propositions. But is there any way to get leverage between the property and propositional views of desire? It might initially seem as if there is no way to get leverage, since we can seemingly always translate propositional desire talk into propertitional desire talk and vice-versa, as a couple of examples will help to illustrate. Consider a case in which a group of Olympic runners each desires to win the race. On the property-oriented view, the desires are 104 ultimately desires to be the winner of the race; and on the proposition-oriented alternative, each runner’s desire is ultimately a desire that she (the desirer) wins the race. But what of a desire, say, for world peace, which might initially seem to be a desire for the world to be some way rather than to possess some property? Here again there is a simple translation. The desire for world peace can be characterized as a desire to be such that world peace obtains, which is a property had by the desirer only if everyone else has the property, too. 176 Graham Oddie [forthcoming] argues that despite the possibility of these translations, we can show in the following way that the property-oriented view of desire is superior. Return to the case of the runners who desire to win the race. In such cases, we often say that the runners desire the same thing. But on the proposition-oriented view, this is not true, since the objects of their desires are inconsistent propositions. On the property-oriented view, however, we can truly say that the runners desire the same thing; and so the property-oriented view better captures how we talk about desires. This is a point in favor of the property-oriented view, or so Oddie maintains. 177 We should not be persuaded by this argument. The linguistic data also permits us to say that the runners do not desire the same thing. Remember, we can say of any of the runners, R, either of the following: (i) R desires to win the race, or (ii) R desires that R 176 Oddie [forthcoming] uses the label “state properties” for properties that are had by one person only if they are had by all. 177 Here is how Oddie puts the argument: “There is, however, a rather natural rival to the propositional view. Suppose Orlando and Oliver are competing for Olympic gold in cross country skiing. Oliver says I really want to win the gold and Orlando chips in So do I! The common object of their two desires is to win the gold. On the propositional view, they can only want the same thing if they want the very same state of affairs to obtain: but there is no state of affairs that is their common desire. But there is a single state of being rather than state of affairs that both want. They both want to have a certain property, the property of winning the gold. More generally what one desires is typically to possess or to have some property, what one prefers is the having of some property to the having some other property. The property view can make literal sense of the claim that Oliver and Orlando want the very same thing.” 105 wins the race. If we are thinking in terms of (i) rather than (ii), then we will be inclined to say that the different runners desire the same thing. But if we are thinking in terms of (ii), which is from a linguistic perspective perfectly fine (though a bit stilted), then we will not be inclined to say that each of the runners literally desires the very same thing. So I do not think Oddie’s argument gets any real leverage one way or the other. A better kind of argument for the property-oriented view of desire will mirror Lewis’s [1979] argument for the property-oriented view of belief. Lewis argues that some beliefs have objects which cannot be understood propositionally. Suppose, to borrow Lewis’s example, that Rudolf Lingens is lost in the Stanford Library. The more books that Lingens reads, the more he will know about the world that he is in. But he might learn every proposition which is true at his world without knowing where he is located within that world. Lingens will not know where he is until he self-ascribes the property of “being in aisle five, floor six, of Main Library, Stanford.” 178 The object of this belief cannot be understood propositionally, since it is a property that some can have in a world while others do not. (If a proposition is true at a world, then it corresponds to a property had by everyone.) Lewis argues that an analogous argument can be made to show that the objects of some desires cannot be captured propositionally. Return to Lingens, the poor soul who is lost in the library. Lingens wants to escape from the library. And the best understanding of this desire is as property-oriented, for the following reasons. There are a number of worlds in which people are lost in libraries. In some such worlds, the individuals who are lost have all the properties Lingens takes himself to have, but only 178 Lewis [1979: 520] 106 some of these people end up escaping the library. Lingens wants one of these worlds in which some of the people who are lost escape, but he does not want just any of those worlds to obtain. He wants one to obtain in which he is among those who escape. Here is another case Lewis offers to push the idea that desires are property- oriented. Suppose that God 1 and God 2 inhabit world W. For every proposition that is true at W, they desire that it be true. Insofar as we are concerned with which propositions the Gods 1 and 2 desire, they are, as Lewis puts it, “as choosy as can be.” However, one of the Gods lives on the tallest mountain. God 1, as it turns out, forms a desire to be the God who lives on the tallest mountain; and this desire cannot be captured propositionally. God 1 desires not just that some world obtain but that she occupy a particular location within that world. Some entities have the property of living on the tallest mountain while others do not, and she wants to be among those who do. Lewis’s argument, then, makes a compelling case that we cannot always translate property-oriented desires/beliefs into proposition-oriented desires/beliefs, after all. But because we can always do the translations in the other direction, the property-oriented view of desire/belief is to be preferred. Taking for granted now that desires are property-oriented, we can give a more a more precise characterization of the impersonal and personal guise of the good theories, given orthodox views about impersonal and personal good. The impersonal good seems to be picked out when predicate ‘good’ of some complementizer phrase, as indicated by the following: (i) it would be good if he comes, (ii) it is good that we are together, (iii) it would be good for it to rain. Complementizer phrases in such sentences pick out propositions, ascribing a monadic property to them: being good. (This seems to be the 107 kind of goodness consequentialists are talking about when they say we ought to do the actions which bring about the best outcomes.) This suggests a very natural way of thinking about the impersonal guise of the good theory. The idea is that when an agent desires O, she perceives it as good that she is O. To illustrate, if Cassandra desires to eat some ice-cream, what appears good to her is that she eats some ice-cream. Now turn to the personal guise of the good theory. We often use ‘good for’ to indicate personal good, as when we say ‘it is good for her that she won’. 179 This sentence is not referring to a monadic property of a proposition but rather a kind of goodness which is a relation between an individual and a proposition. On the personal guise of the good theory, then, when an agent desires O, it appears good for “someone” that she is O. But who is the “someone” in this schema? The most natural candidate seems to be the desirer herself. When Cassandra desires to eat some ice-cream, she perceives it as good for her that she eat some ice-cream. Over the course of the next two sections, I will explain why both versions of the guise of the good theory, so understood, fail. As we shall see, however, one way to try to fix the problems for the impersonal guise of the good view is to take up the (unorthodox) view that goodness is really a second-order property rather than a property of proposition, as I have suggested. Part of the interest in this alternative is that Graham Oddie – one of the most prominent proponents of the guise of the good theory – holds that goodness is a second-order property. I argue that ultimately the revised guise of the good theory falls victim to the same basic problem. 179 One potential source of confusion is that the ‘for’ in ‘good for’ can sometimes function as a complementizer, in which case it will refer to impersonal goodness. 108 In the face of these difficulties, a committed proponent of the guise of the good may return to the drawing board and attempt to construct a new version of the theory. After all, I did not claim to offer the only possible versions of the personal and impersonal guise of the good theories. 180 The trick would be to construct a version of the theory which avoids the objections in a way that is not ad hoc and that retains the initial motivations for the theory (including in action theory and value epistemology). I invite others to carry out this project, if they think it a promising one. 181 What I will do, instead, is point out how elegantly we can solve the troubles that the guise of the good theory runs into by shifting to a guise of reasons view of desires. 182 3 The Accuracy Constraint I argued in the previous chapter that covariation is important for perceptual content determination. In this section, I formulate a specific covariational constraint – the accuracy constraint – which we can use to get evidence for or against various hypotheses about the contents of perceptual states. The intuitive idea behind the accuracy constraint is that the presence (or absence) of a systematic covariation relation between an experience and what it is hypothesized to be about at least provides evidence for (or against) the hypothesis. 183 To give an 180 For example, an alternative way to characterize the personal guise of the good theory is that desires involve perceptions not of what is good for me but what is good for someone. 181 It is worth emphasizing that if it turns out, contrary to what I argue here, that there are plausible versions of the guise of the good, then this just provides the sentimental perceptualist with more options for how to develop her theory. 182 Mark Schroeder and I [unpublished] go into more detail on the attractions of the guise of reasons. 183 I take this constraint (which will be refined below) to be amenable to explanations of perceptual content offered by Millikan [1989], Dretske [1995], Tye [2000], Prinz [2004], Burge [2010], and Neander [2013], among others. Not all of these theorists hold that covariation is part of the fundamental explanation of fundamental explanation of content. Neander, for instance, has a causal-teleological theory of content, but nevertheless she accepts that covariation is evidential. For example, Neander 109 illustration, consider the hypothesis that blueish perceptual experiences (i.e., perceptual experiences with blueish phenomenology) are about greenness. I take it that this is an implausible view. On the assumption of this hypothesis, the blueish perceptual experiences of human agents with normal, well-functioning psychologies would be systematically in error, since as a matter of fact such agents tend to have such experiences in the presence of blue objects, not green ones. That there would be such systematic inaccuracy tells against the initial hypothesis, and it tells against it because they would be inaccurate in this systematic way. Of course, sometimes there are systematic sources of error in our perceptual experiences. Perceptual illusions, including famous cases such as the Müller-Lyer illusion, occur systematically. So we should not say that any systematic source of error tells against the hypothesis that the perceptual state has a content that makes it inaccurate. But we can get some insight into when systematic error is to be expected by considering such ordinary cases of perceptual error or illusion. It turns out that we actually know quite a lot about how perceptual illusion works in ordinary cases. In the Müller-Lyer illusion, for example, most humans, who have been raised in an ordinary built environment, perceive one of two lines as longer than another, despite the fact that they are exactly the same length. This turns out to be easily explained by the fact that in a normal built environment, the line segment that visually appears to be longer actually is longer. Even though the images projected on the retina are identical [2013: 29] says, “If our ancestors experienced color contrast illusions, or the light wasn’t always white light, with the result that [reddish visual experiences] were sometimes produced in the absence of red, certain pathways in the human visual system could nonetheless have been adaptive and could have been selected because often enough the [reddish visual experiences] they produced carried information about red.” 110 in size, the line flanked by outward facing arrowheads corresponds in normal settings to something farther away; and if two objects project the same size retinal image, but one is perceived as farther away, we perceive the one that is apparently at a great distance as larger. 184 Other illusions cut across the senses in interesting ways. To take a familiar example, ventriloquists make a living creating auditory illusions which are triggered by visual cues. In particular, a ventriloquist moves a puppet in a way that that corresponds to how we would expect someone to move who is speaking in the way that the ventriloquist is speaking, but the ventriloquist herself does not move her mouth at all. This helps to create the auditory illusion that the sounds are coming from a different direction than they in fact are. 185 In both cases, then, the dispositions which lead to the illusory perceptions are such as to produce accurate perceptions in typical environments. Similar explanations go for other visual and auditory illusions. In fact, it should not be a surprise that this is so, because one of the standard methods of research into perceptual cognition presupposes that it is so. According to this methodology, one of the 184 See Gregory [1997]. This explanation has not gone uncontested (e.g., Rock [1995]), but the important thing is not the details of the explanation but rather the general (uncontested) idea that the dispositions which lead to the illusions are such as to be accurate in more typical cases. 185 In general, when visual experiences influence the direction from which we perceive as sound as coming from, this is known as ‘the ventriloquism effect’ (Howard and Templeton [1966]). 111 principle sources of insight into how perceptual cognition works is to pay special attention to perceptual illusions. We can learn a lot about how a perceptual process actually works by assuming that the process in an illusion is an extension of a process that is accurate in its core cases. 186 Here, then, is the constraint that we are looking for, formulated in a general way: Accuracy Constraint A hypothesis about the content of a perception is plausible only if perceptions of that sort arising out of normal, well-functioning psychologies are accurate, or else can plausibly be understood on the model of an illusion. In the next section, I use this accuracy constraint to argue that the guise of the good theory is not very plausible. This is because the desires of agents with normal, well-functioning psychologies fail in typical contexts to covary with what is good, which indicates that desires do not involve perceptions of anything as good, after all. 4 The Dilemma I am now in a position to precisely formulate the central problem for the guise of the good theory. The crux of the problem is that neither the impersonal nor the personal guise of the good theories can satisfy the accuracy constraint, because in the normal case, well-functioning, non-illusory desires deviate from both personal and impersonal value. 187 186 Textbooks on perception make liberal use of this strategy. See, for instance, Gregory [1997], Snowden et al [2012], Goldstein [2013], and Wolfe et al [2014]. 187 Although for simplicity I speak of the impersonal guise of the good theory and the personal guise of the good theory, I do not mean to suggest that I offer the only possible characterizations of these theories. This is a point I emphasized above (see section 3). 112 The easiest way to begin to see the trouble is by thinking about the two “flavors” that normal, well-functioning desires come in: egoistic and altruistic. Consider Larry. Larry wants to help the homeless in his neighborhood. He does not want to do this merely because doing so is a way of, say, impressing his peers or creating pleasure for himself. Of course, he is not a complete altruist. He really does have those desires to impress his peers, get pleasure for himself, and so on; and he does not have those desires merely because he sees satisfying them as a way of helping others. Most of us are like Larry. We have a mixture of genuinely egoistic and other-regarding desires. A genuinely other- regarding, or altruistic, desire is simply one that is not had simply as a means of satisfying some egoistic desire, and vice-versa for genuinely egoistic desires. 188 The difficulty for the guise of the good is that its personal form runs into trouble with altruistic desires, but its impersonal form runs into trouble with egoistic desires. 189 It is easy to see how altruistic desires make trouble for the personal form of the guise of the good. According to the theory, when an agent desires O, she perceives it as good for her that she is O. So if Kunal learns that a friend of a friend has put in a bid on a house, and he desires for that person’s offer to be accepted, then, according to the personal guise of the good theory, he has an associated perception as of its being good for him that her offer is accepted. But unless there is some coincidence, in general it is not good for Kunal that friends of his friends win bids on houses. So the vast majority of such cases turn out to be inaccurate, on this view. 188 Some have challenged the claim that there are genuinely altruistic desires. The view that all desires are ultimately egoistic, or instrumental to the satisfaction of egoistic desires, is called psychological egoism. But I side with those who take psychological egoism to be misguided (see Sober and Wilson [1998] for forceful objections to psychological egoism). 189 Actually, as we will see shortly, the impersonal form of the guise of the good theory runs into trouble with both egoistic and at least some altruistic desires. 113 This is not yet inconsistent with the accuracy constraint. To settle the question of whether the personal view runs afoul of the constraint, we need to answer: (i) Are desires like Kunal’s normal and well-functioning? and (ii) If so, can explain away inaccuracy in those cases on the model of an illusion? To begin, such altruistic desires seem to arise out of normal, well-functioning psychologies. There is nothing defective about the desiderative system of a person who wants friends of friends to win bids on houses; we expect normal human agents to have such altruistic desires. But now consider the possibility of an illusion. If under normal evolutionary conditions, our ancestors would have seen a friend of a friend making a bid on a house only under conditions under which it would have been accurate to conclude that it was better for the desirer for the bid to be successful, then that would be the right kind of thing to explain why it is that we experience an illusion. Or, more seriously, if under normal evolutionary conditions what is good for the friends of friends were also good for oneself, then this would help us to understand Kunal’s desire as an illusion, However, under normal evolutionary conditions not everything that is good for the friends of friends one encounters is good for oneself. In general, from the perspective of the personal guise of the good theory, the trouble with altruistic desires is, as the name suggests, that they are more other-regarding than self- regarding, which makes a stretch to maintain that they evolved to track the personal value of the desirer. 190 This brings us to the impersonal guise of the good theory, which is the historically most prominent version. On this front, it is equally straightforward why egoistic desires make trouble. Recall that, according to the theory, when an agent desires O, she perceives 190 For an additional worry about the personal guise of the good theory, see n. 196 below. 114 it as good that she is O. To see how egoistic desires make trouble for the impersonal form of the guise of the good, consider them as they arise in competitive contexts: Race An Olympic runner, Shelly, is about to compete for the 400 meter gold. She desires to win the race and is averse to any of the other runners winning. The other runners, of course, also want to win the race and their opponents not to win. 191 As is often the case in such competitions, it makes no difference (or no significant difference) to the total goodness of different possible outcomes who wins the race. So if proponents of the guise of the good want it to turn out that Shelly’s desires are associated with accurate perceptions, they will be forced to say that the other runners’ desires are associated with inaccurate perceptions. This is for the simple reason that, according to the theory, Shelly represents it as good that she wins the race, and the other runners represent it as not good that Shelly wins. But it cannot be both impersonally good and not good that Shelly wins the race. Cases like Race are quite typical and could be multiplied. Now let’s think about such competitive desires in light of the accuracy constraint. Similar to before, we need to answer: (i) Are egoistic desires in competitive contexts normal and well-functioning? and (ii) If so, can explain away inaccuracy in those cases on the model of an illusion? To begin, egoistic desires in competitive contexts seem to arise out of normal, well-functioning psychologies. There is nothing defective about the desiderative system of a runner who wants to win a race and is averse to other runners winning, for example; we expect normal human agents to have egoistic desires when in such contexts. Now, to clarify, one might think it would be better all things considered 191 This example is adapted from Oddie [forthcoming]. 115 if we just never had egoistic desires; but we should not confuse what would be best all things considered with the notions of normal and well-functioning relevant to the accuracy constraint. For example, it might be better all things considered if our visual dispositions were slightly different so that the world looked a bit prettier than it in fact does, but even if this were so, there is nothing in the relevant sense defective about our visual faculties when we (say) represent dandelions as plain yellow rather than rainbow colored. If the agents’ desires in Race, and in similar cases, are well-functioning, then proponents of the impersonal guise of the good theory will need to insist that there is an illusion. That is, they will need to say that the runners’ desires are well-functioning but nonetheless misrepresent the good. The trouble for such a move is that in competitive contexts, we do not seem to be responding to considerations which covary closely with the impersonal good. A competitor’s desires will normally be highly and systematically sensitive to thoughts of her glory, her status, her future endorsements and so on; and these egoistic considerations just are not relevant to the impersonal good. At this point, proponents of the guise of the good may make a last ditch play to salvage the view. The move is to claim that there are actually two types of desire, egoistic desires which are associated with perceptions as of personal goodness and altruistic desires which are associated with perceptions as of impersonal goodness. Shifting to a disunified picture of desire can seem like an ad hoc move in the face of the guise of the good’s trouble with the accuracy constraint, but even if we look past this, there is still a problem. 116 The problem is that our paradigmatic altruistic desires are typically centered on the desirer in a way that makes it implausible that they track the impersonal good. 192 Consider the following, more harrowing kind of competition: Hungry Jasmine and Jacqueline are members of the same tribal community, but they’re not close family members. During one especially harsh winter, food is limited and survival is in question. On occasions when there is not enough food for everyone, Jasmine desires that her family get the food and that Jacqueline’s not get it. Jacqueline’s desires are similar. In Hungry, as in Race, it makes no difference to the impersonal good who “wins” the competition, and so the guise of the good theorist, even on the disunified version of the theory, cannot have it that both Jasmine’s and Jacqueline’s desires are accurate. But at the same time, both of their desiderative systems seem to be functioning well. And, moreover, it is hard to see why we should expect illusions. Jacqueline’s is sensitive to the fact that a certain family is hers, and similarly for Jasmine. But these considerations are irrelevant to the impersonal good, and yet their desires are highly and systematically sensitive to such considerations. So once we have the accuracy constraint on board, it becomes very implausible not only that desires in general are systematically associated with perceptions of the impersonal good but even that altruistic desires are. The trouble is, we might say, with the desirer-centeredness of desire; even our altruistic desires are often centered on the desirer in a way that is difficult even for the disunified theory of desire to accommodate. 192 The discussion of Kunal’s altruistic desire discussed above hinted at the desirer-centeredness of most altruistic desires, but I draw the point out more clearly here. 117 5 Revising the Impersonal Guise of the Good I now want to consider a strategy for salvaging the impersonal guise of the good theory. At first pass, it seems (for reasons I will point out momentarily) as if a proponent of the theory can avoid the problems with the accuracy constraint by shifting to the view that ‘good’ in the impersonal sense actually denotes a second-order property, rather than a property of a proposition. Given this second-order property view of goodness, the natural way to state the view is as follows: when an agent desires O, she represents O (which is a property) as good. Graham Oddie [forthcoming] has recently defended the guise of the good theory understood in this way. This kind of view generates the result that in Race, each of the runners’ well- functioning desires to win the race involves an accurate perception of the good. The object of the desires is a property - to win the race; and the theory says that in desiring they each represent this property as good. And since it is true that winning the race is good, their desires are all accurate. Similar points go for Hungry. Jasmine and Jacqueline desire for their children to survive, even if that means that other people’s children do not. On the present theory, their desires do not involve inaccurate perceptions of value here, for it is better to be such that your children survive (if only some children will survive). 6 Trouble for the Revised Impersonal Guise of the Good We might not want to grant that ‘good’ in the impersonal sense denotes a second- order property. 193 But for my purposes, I grant that the second-order view of goodness is correct, since the view still has problems with the accuracy constraint. 193 See Milona and Schroeder [unpublished] for some reasons why. 118 Trouble starts with the following key assumption: the guise of the good theory says not only that desires are associated with perceptions of something as good but as good to some degree. Unless desires represent degrees of value, their epistemological import will be severely limited. Remember, the sentimental perceptualist needs to explain not only our knowledge of what is good but also our knowledge of how good things are. Otherwise her value epistemology will be crucially incomplete. Assuming, then, that desires represent degrees of goodness, we can now ask about the feature of desiderative experience which represents degrees of goodness. Oddie proposes what I think is the natural view here, namely that the strength, or intensity, of a desire corresponds to the degree of goodness which the desire represents. 194 Going forward, I take for granted Oddie’s model. 195,196 The trouble with this picture is that the intensity of our desires are systematically sensitive to factors which have nothing to do with degrees of goodness. To illustrate, take the following case: Future Race Shelly is about to compete for the 400-meter gold in the 2016 Olympics. She desires to win the race and is averse to any of the other runners winning. As it turns out, the runner also plans to compete at the 2020 Olympics. She wants to win the gold in 400 meters at the 2020 Olympics, too, but her desire is much weaker than her desire to win the gold at the 2016 Olympics. 194 See Oddie [2005]. 195 Why not talk about preferences (as Oddie himself occasionally does) rather than strengths of desires? I will address this matter below. The claim will be that it does not really make a difference. 196 This point about the intensity of desires also deepens the trouble that the personal guise of the good theory has with respect to the accuracy constraint. To illustrate, suppose that Kunal sees a stranger who is suffering and forms an intense desire for the stranger not to be in pain. Even if it is good for Kunal for the stranger’s pain to stop (e.g., because it makes him feel bad to see others in pain), it is hard to imagine that the intensity of his altruistic desire corresponds to the degree to which it would be good for him for the stranger’s pain to stop. Cases like this can be multiplied in ways that add additional weight to the case that desires have not evolved to track what is good for the desirer. 119 Being the 2016 Olympics 400-meter champion is no better or worse than being the 2020 Olympics 400-meter champion. Nevertheless, her desire to win the 2016 Olympics is more intense. In general, the intensity of our desires is sensitive to our temporal location, even though the goodness of properties is typically not. 197 Given the assumption that intensity of desire corresponds to the degree of goodness the object is represented as having, there will be predictable, systematic mismatch between perceptions of degrees of goodness and actual degree of goodness. If desires had the function of covarying with goodness, we would expect them to be much less sensitive to time than they in fact are. The guise of the good theorist thus needs to insist on an illusion. Oddie makes just this move when he argues that (nearly all) desires distort degrees of value to some extent, because desires are like perceptions in being influenced by perspective. Here is what he has to say: Suppose two objects A and B are exactly the same size. Dom is closer to A than to B while Eric is closer to B than to A. To Dom, A will appear larger than B, while to Eric, B will appear larger than A. But the differences in how these two things appear to them does not signal any defect in the perceptual apparatus of either perceiver. It is completely normal for objects to appear differently to the two observers, given the different relations the observers bear to the perceived objects. Perception is always perception of objects as they stand in relation to the perceiver. Visual appearance is always an appearance from a certain point of view, and as such the appearances may legitimately incorporate perspectival differences. [forthcoming] Perhaps, then, time affects desiderative appearances of value in a way analogous to how physical distance (inter alia) affects visual experience. What is temporally distant seems 197 The objection I push in this section focuses on the sensitivity of desires to time. But an analogous objection could be developed focusing on the sensitivity of our desires to our physical location with respect to the object of our desire. 120 less good (or bad) just as what is physically distant seems smaller. Here, again, is Oddie: Time can be thought of as a dimension in value space, one which should affect the perception of value states. The further away a state of affairs is in time, the less it looms in our experience of its value. [forthcoming] What are we to make of this analogy with vision? The analogy ultimately rests on a mistake. To illustrate, suppose that Wanda walks into an auditorium and observes the array of seats. The seats are uniform in size, but if we are following Oddie’s picture, the closer chairs should appear to Wanda to be larger. But this claim is either false or very misleading. To be sure, the closer chairs take up more space in Wanda’s visual field, and so perhaps in some sense we can say that they “appear” larger to her (although this seems a stretch), but if her visual faculties are well- functioning, she will represent them as the same size. It is not as if the chairs strike her as being of all different sizes, and then she engages in some post-perceptual correction of her experience. It is a standard view in philosophy of perception and visual science that human visual faculties have evolved in such a way that the visual system itself often represents objects as being the same despite shifts in perspective (often referred to as perceptual constancies). 198 The nearby dogwood tree does not appear to us to be larger than the distant redwood, even though the former takes up more space in our visual field. Our visual faculties are amazingly adept at correcting for our shifting perspectives, at least in worlds like our own. By contrast, our desires do not seem to be such as to correct for temporal distance. The intensities of our well-functioning desires are, apparently, 198 For two philosophers (among many) who defend this picture, see Burge [2010] and Orlandi [2014]. Burge and Orlandi also make numerous helpful references to work in empirical psychology and cognitive science. 121 sensitive to time; something would be going wrong if our desires for temporally distant properties were typically just as intense as our desires for temporally near properties (of equal value). Because the strength of our desires is sensitive to temporal distances in a way that impersonal goodness is not, we should be skeptical that desires have the function of detecting goodness (to some degree). We want a theory of the kind of value that desires are about which fits better with the kinds of considerations desires have apparently evolved to be sensitive to. And it is not just temporal distance that makes trouble. For example, our desires are also sensitive to physical distance. Our desire, say, to save a nearby drowning child will typically be more intense than a desire to save a more distant child (one we cannot see). Why would physical distance from the desirer matter? It is not typically better for a nearby child to be saved than a distant one. In sum, even once we shift to the view that goodness is a second-order property, the impersonal guise of the good theory runs still runs into serious trouble in light of the accuracy constraint. Some readers may wonder whether we might be able to recover an attractive version of the guise of the good by appealing to preferences rather than strengths of desires. Let me explain. Historically, most philosophers have taken preferences to be analyzed in terms of the strengths of our desires: to prefer x to y is to have a stronger desire for x than for y. Decision theorists, however, often take desires to be analyzable in terms of preferences. 199 On this picture, the fundamental state is that of preferring one thing to another. 200 The thought, then, is that the guise of the good theorist can argue that 199 See von Neumann and Morgenstern [1994], referenced in Timothy Schroeder [2015]. 200 I am very skeptical about the view that desires should be analyzed in terms of preferences rather than vice versa, for reasons pointed out by John Pollock [2006] and Timothy Schroeder [2010]. 122 preferences cue us in directly to what is better than what, no need to worry about the intensity of desires. But this is not quite correct. Whether or not desires or preferences are basic cannot really help the guise of the good theorist, so long as we agree to the following: (i) desires have evolved so as to be sensitive to factors which typically have nothing to do with how good something is (e.g., temporal and physical locations), and (ii) if desires reduce to preferences, then preferences will be sensitive to such good-irrelevant considerations, too. 201 7 The Guise of Reasons I argue in this section that sentimental perceptualists can elegantly resolve the trouble we have been having with the accuracy constraint by shifting to a guise of reasons theory. 202 In order to see why this is so, we first need get clear about what the view is. According to the guise of reasons, desires are systematically associated with perceptions as of reasons. In particular, these reasons will be considerations which favor actions which are means to the satisfaction of the desire. For example, if you want to drink a coke and there are cokes in the fridge, then you will be disposed to have a perception as of a reason to go to the fridge. Of course, if you do not believe that there are cokes in the fridge, then it won’t seem to you that you have a reason to go to the fridge. So we need to be careful in formulating the theory. The guise of reasons says when you desire to drink a coke, what appears to you to be a reason will depend on your beliefs; if given 201 I think that we often use ‘preference’ to refer to our judgments about what is better than what, but of course preferences in this sense are of no help to a perceptualist. 202 Scanlon [1998] is a prominent proponent of this theory. 123 your beliefs, some action would help bring it about that you drink a coke, then that will appear to you to be a reason. Put generally, Guise of Reasons When a person desires O she is disposed to represent herself as having reasons to do actions which, given her beliefs, would help to bring it about the object of her desire. 203 Now let’s work through how the guise of reasons helps with the problem cases for the impersonal guise of the good theory, focusing first on the cases which made trouble for the version of the guise of the good theory which takes ‘good’ to ascribe a property of proposition. The primary example which made trouble for this view, Race, involved a collection of competitors who desire to win the 400-meter Olympic gold. According to the guise of reasons, each of the runners will see herself as having reasons do actions which, given her beliefs, would help to bring it about that she wins the race. For example, if the race is two weeks away, each runner will see herself as having reasons to eat well, train hard, and avoid distractions, among other things. And these perceptions are entirely accurate; the runners really do have reasons to take such means to help bring about the object of their desire. By contrast, the impersonal guise of the good theory – the version of the theory which takes ‘good’ to ascribe a property to a proposition – ran into trouble with Race, since (ex hypothesi) it makes no significant difference with respect to impersonal goodness who ultimately wins. So even though the runners’ desires are normal and well-functioning, it cannot be that they are each associated with accurate normative perceptions (as the guise of reasons theorist can say). And, moreover, it is hard 203 This is how the guise of the reasons is characterized in Milona and Schroeder [unpublished]. There may be other subtly different ways in which we can characterize the view. 124 to see how the desires (in this case and all similar ones) can be understood on the model of an illusion. The guise of reasons also does well when competitive desires are altruistic rather than egoistic. In Hungry, for example, there is not enough food for all the children and Jacqueline wants her children to get the food. She will thus see herself as having reasons to take steps which help ensure that her children, rather than the other children, get the food. And to get her perceptions of reasons to come out as accurate, we are not forced to say that there is anything objectionably special about Jacqueline’s children, for other parents in the tribe have analogous desiderative profiles to Jacqueline and accurately perceive themselves as having reasons to bring it about that their children get the food. (The reasonable assumption at work here is that, when survival is at stake, we normally have reason to favor those with whom we have the closest of bonds.) 204 The guise of the good theory also does well with desires which are wholly altruistic and non-competitive. For example, Kunal sees someone in pain and desires for her pain to stop. According to the theory, Kunal will see himself as having reasons to bring about the cessation of her pain (e.g., a reason to call an ambulance). And, of course, 204 One can imagine a philosopher – perhaps a utilitarian – who insists that Jasmine and Jacqueline should be impartial: all else being equal, they have no more reason to see to it that their family gets the food than that some other family does. A couple of points about this. There is space even for a utilitarian to say that people typically have reason to be partial in cases like ones Jasmine and Jacqueline are in, since aiming to be impartial in such cases would not ultimately be the best way to promote the good. But if one insists that they really have most reason to be impartial, then I think this is actually just ground for doubting sentimental perceptualism. The reason is simple. I have already argued that desires do not have the function of tracking goodness. And if Jasmine and Jacqueline have most reason to be impartial, then similar objections will apply to the guise of reasons alternative. We would thereby have ruled out the two most common, and I think prima facie plausible, views about the sort of value desires are about, which should lead us to question the idea that desires are really about value. Fortunately for the guise of reasons theorist, however, it does not seem as if Jasmine and Jacqueline (who are in a life or death situation) should be impartial. 125 Kunal does have such reasons. 205 The guise of reasons, then, has a much easier time with the accuracy constraint than its guise of the good cousin, at least so long as we have the standard (and I think correct) view that the impersonal ‘good’ ascribes a property of a proposition and the view that the personal ‘good’ ascribes a relation between an individual and a proposition. Of course, as we saw, the (impersonal) guise of the good theorist may revise her theory to say that ‘good’ really ascribes a second-order property, and this helps with the problem cases considered above. However, the accuracy constraint still makes trouble when we consider how desires could help us to learn about degrees of value. The most promising way to do this is to take the intensity of desires to correspond to the degree to which an object is represented as good. But because the intensity of our desires exhibits systematic sensitivity to time (among other things, e.g., physical distance), which is irrelevant to how good something is, the view that desires involve representations of degrees of goodness is not very plausible. But the guise of reasons theory does well in explaining the sensitivity of our desires to time. In criticizing the revised guise of the good theory, I focused on Future Race, a case in which an Olympic runner’s desire to win the (temporally close) 2016 Olympics is much greater than her desire to win the (temporally distant) 2020 Olympics. And so on the guise of reasons theory, she sees the reasons to do what promotes winning the 2016 Olympics as much weightier than her reasons to do what promotes winning the 205 In some cases, an agent may not see anything as a reason. For example, if Kunal desires for a friend of a friend to win a bid on a house, then he may not see anything as a reason. But this is only because he does not recognize any way in which he can help to bring about the satisfaction of his desire. However, he is disposed to perceive reasons should they arise. For example, if he comes to realize that he knows the sellers of the house, he may see himself as having a reason to urge them accept the offer. 126 2020 Olympics. And this seems easy enough to understand. Temporal distance can make a major difference to the likelihood of being able to satisfy a desire. All else equal, we are more likely to be able to satisfy a desire for what is temporally near than what is temporally distant. And, crucially, the likelihood of being able to achieve what we have reason to promote is relevant to the weight of those reasons to promote it. The point that temporal distance is relevant to strengths of reasons is easier to illustrate if we take a more extreme case than the Olympic runner. Consider Sandy, a high-school runner who has never come close to winning any race despite her best efforts. Even if winning an Olympic gold would bring Sandy tremendous joy and fulfillment, she has little reason to pursue that end since she is so incredibly unlikely to achieve it. However, Sandy might have weighty reasons to take the means toward the end of winning local races, which would bring her less joy and fulfillment, but are much more within her grasp. One might point out that our desires are still too sensitive to temporal distance such that even on the guise of reasons theory we get fairly widespread misrepresentation of the weights of reasons, even when our desires are normal and well-functioning. For example, people in early adulthood have weighty reasons to prepare for retirement, if they are able, but their desires to do so are not typically very intense. But here is a case in which it really is reasonable to appeal to affective illusions. Let me suggest one plausible (if a bit speculative) line of thought. Our desiderative systems evolved in contexts in which life tended to be much shorter and less predictable, among other things. For most of our evolutionary past, we could not accumulate money, store it in banks, and then be sure that it would be available decades later for us to spend on whatever goods we would 127 need to continue living. Preparing for the future in any major way has only become possible in the very recent past. Because life today tends to be lengthier, more predictable, and so on, our reasons to focus on the distant future are weightier than they once were. So it is no wonder why our desires would lag behind, making it seem to us as if our reasons to prepare for the future are weaker than they actually are. Similar points can, I think, be made for the sensitivity of our desires to physical distance. We exhibit greater compassion for those nearby, even if we bear no special relationship with them, than those who are far away. For example, a desire to save a suffering children that we know to be nearby will typically be stronger than our desire to save those who are far away. In many cases, we will be more likely to be able to help those who are closer, which makes it plausible that we have greater reason to help them. That said, in the modern world, many of us have the opportunity to take actions which will reliably help those who are very far away (e.g., by donating to charity), and in these cases, we arguably have weighty reasons to do so. Of course, our desires may not involve accurate representations of the weights of reasons when we have weighty reason to help those who are far away, but this is not difficult to explain. In our evolutionary past, physical distance patterned in more systematic ways with the weights of reasons than it does today. Finals Thoughts The aim of this chapter has been to argue that sentimental perceptualists need to be very careful about the sort of evaluative content that they take affective experiences to have. Otherwise the content challenge may be insurmountable. I focused in this chapter 128 on the case of desire. Traditionally, philosophers who have taken desires to have evaluative content have been proponents of the guise of the good. I argued that in order to see whether this theory is plausible, we should see how it fairs in light of the accuracy constraint. This constraint says that a hypothesis about the content of a perception is plausible only if perceptions of that sort arising out of normal, well-functioning psychologies are accurate, or else can plausibly be understood on the model of an illusion. But there are a number of cases in which our desires fail to covary with the good in ways which cannot easily understood on the model of an illusion, making it doubtful that desires are about the good. 206 But sentimental perceptualists are not without recourse. As it happens, the guise of reasons is an alternative which does much better in light of the accuracy constraint. The cases which made trouble for various guise of the good theories fit well with the guise of reasons view. What we have learned, then, is that the content challenge is a serious challenge which rules out prominent psychological frameworks for developing sentimental perceptualism, but fortunately there is a way forward for sentimental perceptualism who wish to take desires as foundational in their value epistemology: the guise of reasons. I close with a final thought about emotions. Even though I have not said anything about emotions here, I have offered a schema for evaluating various hypotheses about the evaluative content of emotions. The accuracy constraint is no less relevant to perceptual models of emotions than it is to perceptual models of desires. For any hypothesis about 206 Again, I do not take myself to have shown that there is no possibility of a guise of the good theory which avoids trouble with the accuracy constraint while maintaining independent plausibility, and I invite those who are optimistic to develop such a theory. For my part, I am highly skeptical. No version of the guise of the good theory which has ever been defended comes close to doing the trick. 129 the evaluative content of an emotion, we can explore the hypothesis in light of the constraint. This means looking at the kinds of considerations which agents with normal, well-functioning emotional responses are sensitive to. Should we carry out such investigations, I suspect we will find it most plausible that our emotions, at least many of them, are about reasons, if they are about any sort of value at all. This is because our emotions also tend to be sensitive to time, physical distance, personal attachments, and so on; and it was a sensitivity to these kinds of considerations which led us to think that desires were about reasons rather than (some sort of) goodness. 130 Chapter 5: Answering the Content Challenge, Part 2 0 Introduction Sentimental perceptualists, as we know from Chapter 3, can answer the content challenge by showing that affect has the biological function of standing in causal- covariational relations with value. In the previous chapter, we learned that sentimental perceptualists need to be very careful about the kind of value that they take affect to be about. In particular, we learned that the historically prominent view that desires are about goodness makes it difficult to defend the claim that desires have the biological function of covarying with the value that they are supposed to be about. A more promising view, I argued, says that desires are about the desirer’s reasons. A similar point may hold for a wide range of our emotions, as indicated at the close of the previous chapter, but I left detailed explorations of particular emotions for a later date. The aim of this chapter is to complete the answer to the content challenge. I address the following two questions: (i) are desires plausibly caused by the values that they are hypothesized to be about (namely, reasons)? (ii) Is it really plausible, at the end of the day, to say that desires (and affective experiences more generally) have the biological function of tracking reasons (or whatever sort of value that they are hypothesized to be about)? I will argue that the answer to both of these questions is yes. 131 1 Are Desires Ever Caused by Reasons? 1.1 Commensurateness as a guide to causation In attempting to determine whether desires are ever caused by the values that they are hypothesized to be about (namely, reasons), the best case scenario would be to have in hand the true analysis of the causal relation and then see whether that relation ever obtains between desiderative experiences and the relevant values. But the question of how to analyze causation is one of the most vexed questions in philosophy. 207 The next best thing, then, would be to have a reliable test for determining whether a causal relation obtains. Oddie offers an intuitive counterfactual test for causation. Oddie argues that commensurateness is guide to causation. Commensurateness is a matter of adequacy and dependency. 208 Let’s start with the former. The notion of adequacy captures the idea that a cause is sufficient in the circumstances for its effect. Suppose a father walks into a clothing store with the intention to buy his son a shirt. He ultimately opts for the magenta-colored shirt. The magenta color of the shirt is, let’s suppose, adequate for his deciding to buy it. Being magenta is a determinate of the determinable, being colored; but the shirt’s being colored is not a cause of his buying it because it is not in the relevant sense adequate. This is because his father would not buy his son a green, yellow, or blue shirt, for he knows his son does not like those colors. So being colored is not enough, but being magenta is. Now turn to dependency. Dependency captures the idea that causes are necessary in the circumstances for their effects. It turns 207 Some philosophers have given up hope for anything approaching a classical analysis of causation. See Godfrey-Smith [2011] and Hitchcock [2003]. 208 See Oddie [2005: 191 – 95]. Oddie is adapting some ideas from Yablo [1992] on mental causation. Oddie and Yablo actually uses ‘contingent’ rather than ‘dependent’. I tweak the language in order to avoid confusion, since the earlier way in which I use ‘contingent’ (in Chapter 2) is different from the way in which Oddie and Yablo use it. 132 out that the shirt’s magenta color does not cause the father to buy it because he knows his son’s favorite color is purple and would thus would have bought him any purple colored shirt. (Build into the case that there were numerous non-magenta purple shirts in the store.) Oddie insists that adequacy and dependency together make up the idea that causes are commensurate with their effects, i.e., that they are in the circumstances necessary and sufficient. In this case, it is the purple color of the shirt, rather than, say, the fact that the shirt is colored (not adequate) or magenta (not dependent). 1.2 Reasons commensurate with desire Reasons are often commensurate with desires, or so it seems. Consider the following example: Bullying: A father is going to school to pick up his daughter. He arrives a bit early, and upon arriving, observes a group of her classmates physically abusing her. He immediately forms a desire to help his daughter. Could it be that the desire is caused by his reason(s) to help her, or at least a subset of those reasons? We need to be very careful about how we understand this question. There is a sense in which reasons are just non-evaluative facts. But certain facts count as reasons because they favor certain actions. More precisely, then, what we want to know is whether the father’s desire to help his daughter is sensitive to the presence of the favoring relation which obtains between the non-evaluative features of the situation and the act of helping his daughter. Very plausibly, the father will be sensitive to considerations which favor helping. To see why, consider what goes wrong with the following story: the father’s perception 133 of his daughter being physically abused in the particular way that she is causes the father to form the desire to help her. On this picture, a highly determinate non-evaluative fact – that his daughter is being abused in the particular way that she is – causes the desire (by way of first causing the very determinate perceptual experience). But if we accept commensurateness as a guide to causation, this cannot be quite right. We can agree, of course, that the non-evaluative fact will be adequate to produce the desire. But the circumstances that he finds his daughter in could have been different in all sorts of ways and yet the father’s desire to help would still be triggered. Here are some examples. The other children may have been physically abusing his daughter in some slightly different way. Or maybe they were just taunting his daughter. It could even be that there were no other children involved at all; his daughter could just have fallen off the jungle gym. The desire to help would plausibly have arisen in each of these contexts, which makes it attractive to think that father’s desire is dependent on the presence of considerations which favor helping his daughter. 209 Sentimental perceptualists need not think that desires are always caused by reasons. The point is simply that if commensurateness is a good guide to causation, as it indeed seems to be, then there is no major obstacle to viewing reasons (or more precisely, favoring relations) as causes in a great many cases. This removes a major obstacle to answering the content challenge: desires can be caused by reasons and so it is not on that 209 One may complain that the favoring relations typically will not be adequate to desires. For example, in the example at hand, the situation could be such that the father has reasons to help his daughter but he fails to desire to do so. The sentimental perceptualist can allow this result. Strictly speaking, we can speak reasons to help within a certain range as the cause of the desire. But this is not a surprising qualification, for a similar qualification goes for ordinary perception. For example, an object’s squareness can cause a visual representation as of its being square, but of course, strictly speaking, what we normally mean is squareness within a certain range. (The square object needs to be within a certain size range, for instance.) 134 account implausible to hold that they have the biological function of standing in causal- covariational relations with reasons. 2 Wedgwood’s Objection 2.1 The objection Ralph Wedgwood argues that we do not need to appeal to particular values to causally explain affective experiences, even when those experiences reliably track value. In this way, Wedgwood argues, affect is very much unlike perception. 210 The asymmetry stems from the fact that affective experiences, in contrast with perceptions, are responses to other mental states. To explain the objection, I stick with the bullying case, offering a Wedgwood- style explanation of why the father’s desires to help his daughter reliably indicate the presence of such reasons. We start with the idea that affective responses are always responses to other mental states, paradigmatically non-evaluative beliefs and perceptions. 211 In the case at hand, the desire is going to be a response to the father’s non-evaluative perceptual experiences and beliefs about the kids’ treatment of his daughter. If the desire is reliable, it would also need to be that these perceptual experiences and beliefs are reliable about their non-evaluative content. Second, we need to appeal to the father’s dispositions to experience a desire to help in response to certain perceptual experiences and beliefs. If the desire is a reliable indicator of reasons to help, then the disposition would need to be such as to respond to all and only sets of perceptual 210 Wedgwood’s specific target is Johnston [2001]. 211 I follow Wedgwood in simplifying the discussion by speaking simply of perceptions and beliefs, but of course desires can also be a response to other mental states, too, e.g., memories. 135 experiences and beliefs that are indicative of reasons to help. A complete explanation of the desire’s reliability would appeal to a principle explaining why such perceptual experiences and beliefs are reasons-to-help-indicating. What is important here, though, is just the idea that we can causally explain the occurrence of desires (and emotions), even reliable ones, by reference to other mental states (paradigmatically, perceptual experiences and beliefs) and to dispositions; there is no need to appeal to particular values/reasons. Wedgwood insists that the disanalogies with perception run even deeper than it might first appear. Because affective experiences are not caused by particular value facts, we should not think of them as providing us with direct epistemic access to particular value facts. Here is how he puts it: [A]n affective state does not have immediate access to any particular evaluative fact. Its access to such facts is mediated by our beliefs or experiences about the particular case, and by a rough correspondence between our affective dispositions and certain general evaluative truths or principles. In short, the evaluative truths to which affective states give us most direct epistemic access are not particular, but general. [2001: 223] One may wonder why Wedgwood slides from the thought that because particular values do not cause affective experiences to the conclusion that they are not directly about particular values. But a natural way of understanding the transition is to see Wedgwood as having in mind a picture on which perception gets to have particular content by standing in causal relations to the property instances that they are about; and so if affect fails to stand in such relations, it cannot similarly be about particulars. 136 2.2 How sentimental perceptualists should respond Although Wedgwood raises two worries for sentimental perceptualism, the source of the second worry – that affective experiences do not give direct epistemic access to particulars – is a function of the first about causation. 212 In what follows, I explain how the sentimental perceptualist can maintain the thesis that affective experiences are caused by values, or more precisely, reasons. To begin, let’s step back for a moment and consider some features of ordinary perception. Imagine an observer looking from directly above at the circular opening of a jar. There is a visual experience as of the jar’s having a circular opening. Now imagine the agent looking down at the jar from an angle. There is still a visual experience as of the jar’s having a circular opening, but the proximal stimulation is different. The visual experience occurs even though the angle of observation, and thus proximal stimulation, varies; an attractive explanation of this is that the observer’s visual system operates according to certain formation principles. 213 Formation principles describe certain dispositions, or formation laws, which ensure that certain types of proximal stimulation lead to the distal environment being represented in the same way. Formation principles thus also help to explain how we can be subject to illusions: proximal stimulations of the right sort will cause the representation of certain environmental attributes even when those attributes are not present. 212 It is worth noting that although Wedgwood is not a sentimental perceptualist, his own view assigns an important epistemological role to affect and underlying affective dispositions. But the explanation he offers of how affect justifies is importantly different from the sentimental perceptualist’s. See Wedgwood [2007]. 213 I borrow talk of formation principles (and the formation laws those principles describe) from Burge [2010] and Orlandi [2014]. 137 Now let’s ask the following: on the basis of the above, is it plausible to conclude that the distal attributes visual experiences are about are never caused by those attributes? Transposing what Wedgwood says about affect, the idea would be that visual representations are explained entirely by proximal stimulation and formation laws/dispositions; there is no need to invoke distal attributes. But of course this reasoning is mistaken. We often do need to invoke the distal attributes to explain the type of proximal stimulation (e.g., the type that generates visual representations of circularity) and the visual experiences that such stimulation causes. Return to the observer looking at the circular opening of a jar from directly above. First, that the jar is circular is adequate to explain the type of proximal stimulation and thus the visual experience as of circularity. (Of course, there are some qualifications we need to make here in order to be precise. If the circle’s diameter were too large or small, for instance, then the visual experience of circularity would not occur; but once such stipulations are made, then the fact that the jar’s opening is a circle of a certain sort will be adequate. For simplicity, I will continue to speak simply of circularity.) It also seems as if visual experience is dependent on the circularity of the jar. We can reasonably expect the following counterfactual to be true: if jar had not been circular, the observer would not have experienced it as being circular. To be sure, there are ways the jar could have failed to be circular yet the agent still experiences it as such; but such possibilities only threaten the counterfactual if the possibilities obtain in nearby worlds. If in the nearest possible world in which the jar is not circular the visual experience of circularity does not occur, then, given a standard understanding of the truth conditions of counterfactuals, the relevant counterfactual comes out true. 138 These remarks about ordinary perceptual experience help us to see why Wedgwood’s explanations only exclude causation of affect by particular values by halting the explanation at what appears to be an arbitrary point. The crucial observation above was this: the proximal stimulation is a cause of the visual experience of circularity insofar as it is a proximal stimulation of a certain type, namely the type that triggers visual representations of circularity. If we type the proximal stimulation too narrowly, then the visual representation will not be dependent on it. Now return to the bullying scenario. Wedgwood’s picture has it correct that the father’s desire to help is a response to certain other mental states, but we must take care not to identify too narrow an aspect of those mental states, or else the purported cause will not satisfy the dependency requirement. Take the perception of the kids shoving his daughter into the mud. The desire to help will not depend on the perception, so characterized, because the desire to help would have arisen had the perception been of the kids taunting her for being unusually tall. This is the crucial point that Wedgwood’s account misses and that leads him to leave particular values (reasons) out of the explanation. The perceptions and beliefs on which the desiderative experience is dependent are given by the formation principles (descriptions of formation laws/dispositions) for that type of experience. And it may well be that a set of perceptions and beliefs within the scope of the formation laws for that kind of experience would not have arisen had the reasons to help not been present. 214 214 For this response to work, we need to assume that causation is transitive, or at least that it is in these kinds of cases. 139 6 A Return to Function I am nearly to the finish line with the answer to the content challenge, but a bit more ought to be said about the idea that at least some types of affect have the biological function of detecting value. (I am broadening the focus here from desires to affect more generally.) We should begin by reminding ourselves that it is a widely accepted view that some types of affective experiences – joy, parental concern, fear, lust, disgust, anger, etc. – were selected for. Psychologists standardly refer to this class of affective experiences as basic. Basic types of affect “have a fixed set of neural and bodily expressed components, and a fixed feeling or motivational component that has been selected for through longstanding interactions with ecologically valid stimuli.” 215 The key question, though, is why they were selected for. Were they selected for (inter alia) because they help us to detect value of some sort? 216 Many philosophers balk at the idea that affect (or any mental state) could have such a function. 217 A principal worry expressed by, among others, Blackburn [1988] and Street [2006], is that there is an apparently competing hypothesis about the function of affect which is supposedly more scientifically respectable. The basic thought is that it is far more plausible that affect was selected for because of how it caused us to behave rather than because it allowed us to track, or detect, some evaluative property or relation. 215 Jessica Tracy and Daniel Randles [2011: 398]. In a special volume on basic emotions in Emotion Review, each contributor accepts this way of thinking about basic affect and also agrees that such affect exists. See Ekman and Cordaro [2011], Izard [2011], Levenson [2011], and Panksepp and Watt [2011]. For some philosophers who accept a similar thesis, see D’arms and Jacobson [2003], Prinz [2007], and Brady [2013]. 216 We also know from the last chapter that if desires have the biological function of detecting any sort of value, then reasons, in particular, are the best candidate for the sort of value they have the function to detect. I suggested – but did not argue in detail – that similar conditional thesis holds for emotions, too. 217 See particularly Blackburn [1988], Joyce [2006], and Street [2006]. That said, both James [2009] and Artega [2015] argue that certain of our mental states have evolved to track value, or at least that existing arguments against that thesis are unpersuasive. 140 Blackburn is the first philosopher that I am aware of to formulate an objection roughly along these lines 218 : It is noteworthy that the account will insist upon the nonrepresentative, conative function for the stance. The evolutionary success that attends some stances and not others is a matter of the behavior to which they lead. In other words, it is the direct consequences of the pressure on action that matter. Evolutionary success may attend the animal that helps those that have helped it, but it would not attend an allegedly possible animal that thinks it ought to help but does not. In the competition for survival, it is what the animal does that matters. [1988: 363] Blackburn seems to think that no mental state could have the function of detecting or representing value. The objection that I wish to focus on says more precisely that no affective experiences have the function of detecting an evaluative property or relation; such experiences instead have the function of getting us to behave in ways which typically promote our reproductive success. The trouble with this line of objection is that it takes the hypothesis that affect has the function of detecting value to be in competition with the hypothesis that it has the function of getting us to behave in certain ways; but the sentimental perceptualist can insist that affect is best understood as having both sorts of functions. The easiest way to see this is just to consider an example from ordinary perception. 219 The ability of many animals to detect predators is almost surely an adaption, but this ability would not have 218 Street formulates a similar objection, and does so in more detail. Her formulation is less than ideal for our purposes, however. She argues that certain patterns of evaluative judgment were selected for “not because they constituted perceptions of independent evaluative truths, but rather because they forged adaptive links between our ancestors’ circumstances and their responses to those circumstances, getting them to act, feel, and believe in ways that turned out to be reproductively advantageous” [2006: 127]. What makes this version of the objection non-ideal is not simply that she puts it in terms of judgment rather than affect but that Street actually grants her opponent that we were selected to have certain representations of value (albeit not ones we can assume are accurate). But my whole overarching concern is whether affect represents value. 219 Artega [2015] makes this point in a recent reply to Street. The reply is really a very simple one, as I will illustrate. The core of the response is simply to point out the typical kinds of descriptions which are offered by psychologists, cognitive neuroscientists, and biologists for how perceptual capacities evolve. See, for instance, Comer and Leung’s [2004] description of how cockroaches came to represent predators. 141 conferred any advantage whatsoever were it not intimately linked to action. 220 That is, the ability to perceptually detect predators would not have been selected for unless it tended to trigger fight or flight responses (or at least not unless it were linked up to action in some evolutionary advantageous way). So there are at least two functions of a predator- response system: (i) to produce a perceptual state, S, in response to inputs which indicate the presence of a predator, and (ii) to engage in fight or flight when S occurs. The sentimental perceptualist insists that matters are much the same with affect and value. We have value-response systems with at least two important kinds of functions: (i) to produce an affective state, A, in response to inputs which indicate the presence of some value, V, and (ii) to engage in action of the relevant sort when A occurs. 221 Even if the perceptual case provides a promising model for sentimental perceptualists, more needs to be said in order to motivate that this model really extends to the case of affect and value. Street presses those who defend the view that mental states of any sort have evolved to track evaluative properties or relations to explain how it could ever make a difference for an organism’s reproductive success to reliably track evaluative facts. What advantages could be conferred by perceiving, for instance, what really is good or bad, right or wrong? In pressing this challenge, Street sets up a contrast with certain kinds of empirical properties that we presumably have evolved to accurately detect through ordinary perception: Surely, one might think, an organism who is aware of the truth in a given area, whether evaluative or otherwise, will do better than one who isn't. But this line of thought falls apart upon closer examination...[C]onsider truths about a creature's manifest surroundings - 220 On the claim that the ability to perceive predators is a common kind of adaption, see the evidence compiled by Burge [2010] and Artega [2015]. Both Burge and Artega also emphasize the important connections between perception and action. 221 The appropriate sort of action will vary depending on the nature of V. 142 for example, that there is a fire raging in front of it, or a predator rushing toward it: the fire might burn it to a crisp; the predator might eat it up. But there are many other kinds of truths such that it will confer either no advantage or even a disadvantage for a given kind of creature to be able to grasp them. [2006: 130] Street goes on to illustrate the sort of truth it would make no difference to grasp: Take, for instance, truths about the presence or absence of electromagnetic wavelengths of the lowest frequencies. For most organisms, such truths are irrelevant to the undertakings of survival and reproduction; hence having an ability to grasp them would confer no benefit. [2006: 130] Street thinks that affect-independent evaluative truths would be among the truths it would confer no advantage to grasp. There are two important kinds of thoughts which might drive one to agree. First, as Street points out, one could think that values are importantly different from, say, predators in that the latter can cause things to happen whereas the former cannot. 222 Second, even if we allow that values can enter into causal relations, we might think that detecting them would be of no more evolutionary significance than, to borrow Street’s example, detecting low frequency electromagnetic wavelengths. Such low frequency wavelengths cause things to happen, to be sure, but it would confer us no advantage (and may even confer a disadvantage) to have the ability to perceptually track them. But sentimental perceptualists have a very reasonable response to this line of objection. I have already argued above that values can enter into causal relations. So taking this for granted, the key question is whether values are similar to predators and fires in that they predictably bear causal relations to outcomes which reliably correlate with what would promote or diminish our opportunities for reproductive success, or 222 Street [2006: 130 – 1] argues that those who think values are causally inert should agree with her objection. 143 whether they are more like low frequency electromagnetic wavelengths which (we can assume) do not predictably bear causal relations to outcomes which so correlate. 223 It seems to me that if value properties and relations stand in causal relations to anything at all, then much of what they so relate to will be significant from an evolutionary perspective. Just consider some schematic examples which illustrate the kinds of things we typically appeal to values in order to explain: (i) the unreasonable verdict led to the riot, (ii) the good deed brought the community together, and (iii) bad parenting is at the root of their depression. Such examples could easily be multiplied and fleshed out. The key, however, is just that things like riots, social cohesion, and depression are obviously significant from an evolutionary perspective and those are the sorts of things we often appeal to values to explain. 224 Sentimental perceptualist can thus lean on their perceptual analogy: just as the (empirical) properties and relations we evolved to perceptually detect are causally networked and evolutionarily significant, so too are the evaluative properties and relations we evolved to affectively detect. The claim that affect has the function of detecting value is ultimately an empirical rather than philosophical thesis. So the ideal situation for sentimental perceptualists would be if the standard view among psychologists and cognitive neuroscientists were that we must appeal to value in describing the function of affect. Unfortunately, there is no standard view about the matter, perhaps because scientists are not terribly worried 223 That said, we can easily imagine how it might be advantageous to grasp low frequency electromagnetic wavelengths. Fraser-Smith et al. [1990] argue that spikes in low frequency electromagnetic wavelengths may precede the occurrence of large earthquakes, which if that were true, is the right sort of thing to make the capacity to perceive such wavelengths more evolutionarily advantageous than we might initially have thought. 224 The basic point that value properties and relations seem to be significant from an evolutionary perspective has been observed by many philosophers. See, for instance, James [2009], Cowan [2015b], and Artega [2015]. 144 about the philosophical implications of the question and perhaps also because it can be difficult to adjudicate between competing hypotheses about what a state has the function of detecting. 225 That said, many psychologists and cognitive neuroscientists have used broadly evaluative language in describing the function of affect. In particular, many think that the way in which at least some basic affect promotes reproductive success is by helping us to detect so-called core-relational themes, e.g., personal loss in the case of sadness, danger in the case of fear, wrongings in the case of anger, and so on; and these themes are plausibly understood in evaluative terms. 226 To illustrate, consider the following description from Richard Lazarus: I have proposed that each emotion involved a special, and different relational meaning, which I referred to as core relational themes. For example, anger is the result of a demeaning offense against me and mine, anxiety is facing uncertain, existential threat, sadness is having experienced an irrevocable loss, pride is enhancement of one’s ego- identity by taking credit for a valued object or achievement, and relief is a change of a negative condition for the better. [2003: 126] 227 More empirical work remains to be done before we can confidently conclude that evaluative notions will appear in the most accurate descriptions of the biological function of emotions. And since different forms of affect have different functions 228 , this work will be messy and piecemeal. I cannot hope to settle the issue here. 225 One famous illustration of the difficulty here is the question of whether a frog’s visual system has the function of detecting flies or small, black moving objects. For discussion, see Neander [2012]. 226 Lazarus [2003] describes the idea of a core-relational theme in detail. See Prinz [2007] for a defense of the claim that emotions have the function of detecting such themes. 227 Similar language can be found, for example, in Arnold [1960] and Scherer [2001]. Many psychologists believe that emotions involve appraisals, which are defined as representations of how something or other bears on well-being, paradigmatically the well-being of the appraiser. In the case of basic emotions, the appraisals have the function of responding to some class of considerations which bears on well-being. See Prinz [2007, Ch. 2] for detailed discussion. 228 See Nesse [1990] and Tracy [2014]. 145 In sum, the hypothesis that affect has the function of detecting value is a reasonable one worth taking seriously. I have tried to speak to the unease or resistance that many philosophers have to this proposal by making a series of observations about affect and value. I first observed that many psychologists and cognitive neuroscientists agree that we have basic affective experiences which were selected for. I then briefly defended the following two theses: (i) the hypothesis that affect has the function of detecting value is compatible with the hypothesis that affect has the function of getting us to behave in ways which promote our reproductive success, and (ii) values seem to reliably stand in causal relations with much that is systematically relevant to our reproductive success. If we combine these points about affect and value with the old philosophical idea (defended by philosophers as diverse as Aristotle, Shaftesbury, and Graham Oddie) that affective experiences play a crucial role in helping us to detect value, and are the basis for many of our evaluative beliefs, then the hypothesis that affect has been selected to play this role becomes an eminently reasonable empirical hypothesis. 7 A Final Objection An opponent of sentimental perceptualism’s psychological framework may try to exploit the fact that perceptual representation does not reduce to causal, covariational, and functional relations. Focusing on the case of desires, the objector concedes that evolved desires stand in causal, covariational, and functional relations with value, but then he says that, for all we have said, affective experiences are responses to non-affective perceptions of value but are not themselves representations of value. (Strictly speaking, the objector may prefer to say that affect has the biological function of detecting non- 146 affective perceptions of value, rather than conceding that they have the function of detecting value.) Responding to this objection will help us better understand my response to the content challenge on behalf of sentimental perceptualism. The opponent we are considering here concedes that perceptions of value are systematically associated with affect. This goes a long way to giving the sentimental perceptualist everything she wants. In particular, it is easy to think that once you have gone this far along the sentimental perceptualist path, there is not much epistemological interest in the distinction between sentimental perceptualism’s psychology and the objector’s alternative. After all, both pictures can be the basis of a value epistemology according to which our evaluative knowledge is grounded in experiences systematically associated with affect. While the above observation seems to me entirely correct, it is also helpful to recognize that this objection is not really an objection to my response to content challenge. We can illustrate by analogy with vision. According to the account of content determination I have been working with, “solidish” visual experiences (i.e., visual experiences we know to be about solidity) come to be about solidity by having the biological function of standing in causal-covariational relations with the property of being solid. But because visual representations of solidity do not reduce to such a bio- functional relation (or so I assume), one could in principle try to argue that it is not the solidish visual experience themselves which are about solidity; it is rather some other, non-visual state which actually does the representing. 229 But this picture seems clearly 229 Or, as before, the better way to put the alternative may be that solidish visual experiences have the biological function of standing in causal-covariational relations with non-visual perceptions of solidity, rather than that they have the biological function of detecting solidity itself. 147 false, since there is no other experience to appeal to other than the visual experience. Moreover, if we give an accurate description of the phenomenal character of solidish visual experiences, we end up describing some distal object as being solid. And the sentimental perceptualist holds that a similar point goes for affect. When we accurately describe an affective experience, we typically end up describing (often among other things) the world in some value-laden way. 230 And, the story continues, there is no alternative experience around which is a rival candidate for the state which represents value. Of course, opponents of sentimental perceptualism may disagree, but it is important to see this disagreement as one which is independent of the content challenge. A good answer to the content challenge does not guarantee that affect is about value any more than good answers to analogous content challenges for hypotheses about the contents of perceptual experiences guarantee that the perceptual experiences have the hypothesized content. In sum, the reply I have offered to the content challenge is structurally very simple. I began by arguing that if affect has the biological function of standing in causal- covariational relations with value, then we have excellent evidence that affect is about value. I then offered some reasons to think that the antecedent of this conditional is true, at least so long as we are careful about the kind of value that we take affect to be about. And this solution to the content challenge has epistemological significance. For as I 230 Consider how Robert Roberts [2013: 72] tries to make the case that anger and joy go beyond bodily sensation: “The person in an emotional state, like the person describing what he sees, largely describes the part of the world that, in her emotion, she is attending to. She says, “The dirty bastard took me for all I had” or “This is a wonderful day!” We can, of course, imagine someone saying, “My blood pressure is up” or “My body feels light as a feather!”…But such ways of talking are both less common and less informative. Just as the person who is reporting on his visual impressions most often tells you what he sees, the person reporting on his emotions most often tells you how the world appears to him.” 148 pointed out in Chapter 4, the relations that help to explain content – bio-functional, covariational, and causal relations – are each of epistemological significance. If one of these relations fails to obtain, we might worry not only whether affect is about value but also whether it can be a good source of evidence about the presence of value. But there are additional epistemological challenges that sentimental perceptualists need to address, as we shall see over the next two chapters. 149 Chapter 6: Perceptualism and The Puzzle of Armchair Evaluative Knowledge 0 Introduction We seem to be able to acquire evaluative knowledge by mere reflection, or, as we might say, “from the armchair.” But how is such knowledge possible? This question is especially pressing for proponents of perceptualism. However useful and illuminating the perceptual analogy initially appears to be, it apparently breaks down in important ways when it comes to evaluative reflection. For one, we do not seem to obtain empirical knowledge by mere reflection. And, furthermore, the empirical knowledge that we do have is plausibly explained in part by causal relations between perceptions and properties in the world. But in evaluative reflection, the relevant evaluative properties are not present and so cannot cause anything. Thus it comes as no surprise that perceptualists who take seriously the idea that we can get evaluative knowledge from the armchair typically drift somewhat from the perceptual analogy. In particular, they take evaluative knowledge from the armchair to be of necessary evaluative truths and offer non-causal explanations of how we acquire such knowledge. 231 But such solutions are only available to intellectual perceptualists, not sentimental perceptualists. As I argued in Chapter 2, intellectual perceptualism fits best with a necessity-requiring value epistemology while sentimental perceptualism fits best with a contingency-allowing approach. In this chapter, I show how sentimental perceptualists can allow for knowledge of contingent evaluative truths by mere reflection and preserve 231 Prominent views here include understanding-based reliabilism and Chudnoff’s direct realism. For discussion of such views, see Chapter 2. 150 a causal model of evaluative knowledge. The theory that I present leans heavily on the solution to the content challenge defended over the course of the last three challenges. In fact, my claim is that the solution to the content challenge in effect already is a solution to the armchair challenge; we simply have to see why. The chapter proceeds as follows. I begin with a more detailed description of armchair challenge. Following that, I outline some of the historically prominent perceptualist solutions to the challenge. Most of the solutions discussed will be familiar from Chapter 2, but it is worth revisiting them in the context of the armchair challenge. The discussion will highlight that nearly all perceptualists who have taken the armchair challenge seriously have ended up defending a necessity-requiring value epistemology, and this is of concern for sentimental perceptualists, since their theory does not pair well with the necessity-requiring approach. I then consider whether sentimental perceptualists might be able to treat armchair evaluative knowledge as a kind of contingent a priori knowledge. Familiar models of the contingent a priori will not do. But as it happens, the solution to the content challenge reveals how sentimental perceptualists can make sense of armchair evaluative knowledge as a kind of contingent a priori knowledge, even if it bears little resemblance to more familiar putative instances of contingent a priori knowledge. 1 Armchair Evaluative Reflection Reflective, or “armchair,” evaluative inquiry can be made familiar with some examples. Almost anyone who has taken an introductory course in ethics has thought about whether it would be right to flip a switch to divert a runaway trolley onto a separate 151 track, thereby saving the lives of five strangers but also killing a single stranger who would not otherwise have been killed. This is a paradigmatic instance of armchair evaluative reflection; there is no actual interaction with a runaway trolley. It is difficult to overstate just how central reflective evaluative inquiry is, for we engage in it whenever we try to decide whether some possible course of action would be good or bad, right or wrong. The perceptualist view is that we can learn evaluative truths through reflection by having evaluative experiences in response to our suppositions and imaginings. The easiest way to see how the armchair challenge arises for perceptualism is to notice an apparent asymmetry between armchair evaluative and armchair empirical reflection. 232 Although we can think about empirical questions from the armchair (e.g., what would happen if I pushed my laptop off the table?) and such reflection can help us to learn empirical propositions, such armchair inquiry is typically understood to be epistemically subordinate to actual empirical inquiry. The ultimate arbiter for empirical questions is what we observe in the world. But there does not always seem to be the same kind of asymmetry in the evaluative case. The evaluative experiences that we have in response to our perceptions and beliefs about the world – our online evaluative experiences – do not seem to be privileged (or at least not always privileged) over the evaluative experiences that we have in response to suppositions and imaginings – our offline evaluative experiences. This is just the perceptualist’s way of framing the traditional idea that we can obtain evaluative knowledge by mere reflection. 233 One 232 I do not insist that there is always such an asymmetry between empirical and evaluative reflection. My aim is simply to identify a familiar assumption about empirical inquiry which can help us to see the force of the armchair challenge for perceptualists; the asymmetry is in no way essential to the challenge. 233 Richard Swinburne [2015: 620] articulates the core idea as follows: “When examples of particular situations (e.g., the trolley problem) are adduced in order to persuade us that some general moral principle is or is not true, it is quite irrelevant whether the examples are examples of an actual event or of an 152 intuitive way to push this thought, as Sarah McGrath [2010] observes, is that while we rely on actual experiments to figure out the nature of the empirical world, evaluative inquiry only seems to require thought experiments. The task, then, is to explore whether perceptualists can develop their theory in a way that respects the intuition that online evaluative experiences are not always privileged over offline experiences when it comes to grounding evaluative knowledge. Or, more simply, perceptualists need a theory which respects the idea that we can get evaluative knowledge simply by thinking. 234 Crucially, the challenge here is about knowledge rather than mere justification. Some perceptualists (e.g., Michael Huemer [2005]) focus primarily on justification, but it is arguably much easier to make sense of evaluative justification from the armchair. Suppose, following Huemer [2005], we accept the principle that if it seems to you that P then you are defeasibly justified in believing that P. On this kind of picture, there is not going to be a deep puzzle about reflective evaluative justification, so long as we have evaluative experiences from the armchair. But forming a belief on the basis of an accurate evaluative (or perceptual) experience is certainly not enough for knowledge, since the experience may be only accidentally correct. For example, if due to a brain lesion a person hallucinates a red apple and believes (justifiably) on the basis of that experience that there really is a red apple, then it would not follow that the person knows that there is a red apple, for it is an accident that they got it right. 235 What is required for a perceptual experience to be non- imagined event. What matters is what it would be right to conclude about which actions in that situation would be good or bad; whether or not the situation actually occurred is irrelevant.” 234 I do not take myself to have said anything definitive against an empiricist model of evaluative knowledge, according to which our online evaluative experiences are always privileged over our offline experiences. But this should be a fallback position. If they can, perceptualists should respect the intuition that we can get evaluative knowledge by mere reflection. 235 Armstrong [1973] 153 accidentally correct? The most common idea is that perceptual knowledge requires (perhaps among other things) a causal connection between the experience and what the experience is about. But perceptualists in evaluative epistemology seemingly cannot take advantage of anything like this idea, for when we reflect, we are not in causal contact with any evaluative properties. 2 Why Sentimental Perceptualists Cannot Borrow Intellectual Perceptualist Solutions Historically, intellectual perceptualists have been the ones to take the armchair challenge the most seriously. (I do not know any contemporary sentimental perceptualists who even take up the challenge.) The question for us is whether sentimental perceptualists can “borrow” any of these solutions from their intellectualist cousins. As we shall see, it turns out that they cannot, and the reason why might at first be thought to show that sentimental perceptualists are in no position to solve the challenge. The starting point for each of the intellectual perceptualist solutions is a second analogy, added to the first with perception, this time between value and mathematics: just as our mathematical knowledge is grounded in intellectual presentations of mathematical truths, so too is our evaluative knowledge grounded in intellectual presentations of evaluative truths. There is no difference in the type of experience which does the justifying, only in the objects of the experience: in the one case a mathematical object and in the other an evaluative one. I am aware of three ways in which this picture can be developed in detail. The most popular elaboration says that an understanding of mathematical propositions often gives rise to intellectual experiences which can – in virtue of arising 154 out of understanding alone – be non-accidentally correct. 236 Typically, understanding is taken to consist in a grasp of the concepts involved in the target proposition. So long as such an experience in fact arises on the basis of understanding alone and is accurate, it can generate knowledge. 237 Some who have this picture of mathematical knowledge argue that such experiences can also arise when thinking about propositions in other domains, e.g., the evaluative. And the explanation for why such evaluative experiences can often be non-accidentally correct is the same as in the mathematical case, namely that they arise merely out of an understanding of the target proposition. Some perceptualists resist the understanding-based model. Elijah Chudnoff has recently developed an alternative view (which he traces back to Plato and Husserl, among others) that he applies directly to mathematics but believes could be extended to other domains, notably the evaluative. 238 According to this alternative, we are capable of intellectual experiences that are based not in understanding but rather on direct awareness of abstracta, i.e., objects which are non-spatiotemporal and causally inert. (This model, as the reader might expect, takes its cue from direct realist views about perception.) On this view, an intellectual experience, say, that circles are symmetrical about their diameters is an experience whereby circularity figures in the explanation of what the experience is. That is, circularity – an abstract object – figures into the mereological explanation of the experience. This awareness can ground knowledge that circles are 236 Contemporary proponents of this picture include Huemer [2005] and Stratton-Lake [2015]. Bealer [1999] has done much to develop this view, although he is not primarily concerned with evaluative knowledge. 237 Some (e.g., Robert Audi [1997]) argue that there is no need to appeal to intellectual experiences to explain how understanding generates knowledge. Understanding is sufficient on its own. For discussion of this non-perceptualist view, and how it compares to the perceptualist alternative, see Bedke [2010] and Tropman [2014]. 238 See Chudnoff [2013]. 155 symmetrical about their diameter, since in having the experience, we are directly aware of the truth-maker for the proposition. Finally, there is a third view (not explicit in the literature but for dialectical reasons worth mentioning) which assumes that the mathematical or evaluative propositions we learn from the armchair are necessary. And if an experience accurately represents a necessary truth, then, according to this view, that experience is thereby non- accidentally correct. The idea is that this will be so for the same reason that (as some maintain) beliefs in necessary truths are always non-accidentally correct. Here is how David Lewis puts it: [I]f it is a necessary truth that so-and-so, then believing that so-and-so is an infallible method of being right. If what I believe is a necessary truth, then there is no possibility of being wrong. That is so whatever the subject matter of the necessary truth, and no matter how it came to be believed. [1986: 114 – 15] There are reasons to worry about each of these proposals. The understanding- based view, for instance, may only be a good model for explaining necessary truths which are analytic. 239 To illustrate, perhaps I can know that all bachelors are unmarried is true on the basis of my understanding of the proposition, but it is not obvious I can know that torturing babies just for fun is wrong on the basis of my understanding of the proposition. Chudnoff’s view avoids this objection, but does so at a cost, for it presumably inherits all of the challenges that direct realist views about perception face. And the Lewis-inspired view, as John Bengson [2015b] points out, seems misguided. Bengson imagines that due 239 This is a very old complaint, one that Hutcheson leveled at Gilbert Burnet’s brand of moral rationalism (see Burnet and Hutcheson [1971]). For a contemporary statement of the objection, see Väyrynen [2008]. For different objections to the understanding-based view, see Chudnoff [2014] and Bengson [2015a]. 156 to a brain lesion, he experiences forty mathematical propositions as being correct; and as it happens, one of them really is correct. Does he know? Bengson thinks not, because intuitively he is only accidentally correct, despite what Lewis suggests. Even if all these concerns could be overcome, it would be little comfort to sentimental perceptualists. They just are not in a position to borrow any of these solutions, since each is only good for explaining basic knowledge of necessities. Recall that I have already argued at length (in Chapter 2) that affect can give us basic knowledge of contingent evaluative truths if it can give us knowledge of any evaluative truths. 240 That the intellectualist views are limited to explaining knowledge of necessities is most obvious in the case of the third, Lewis-inspired view, which builds in that assumption from the start. But it is also true of the other views. The understanding-based strategy of illuminating evaluative knowledge only works for direct knowledge of necessary truths, for a fairly simple reason. A contingently true proposition is not true in all worlds, so it seems as if we would need to know something about what the actual world is like before we could know the proposition. But merely grasping a proposition does not involve knowledge of the world, and so the understanding-based view is only going to help us to know necessary truths. 241 240 I argued in Chapter 2 that this point holds even if we limit ourselves to affective experiences which arise in evaluative reflection. There is no good reason (which is not ad hoc) to hold that while online affective experiences give us knowledge of contingencies, offline affective experiences can only give us knowledge of necessities. The offline experiences can be expected to give us knowledge of conditional evaluative propositions which in typical cases will be contingently true if they are true at all. (I did allow for the possibility of stumbling into knowledge of a necessary evaluative truth through reflection.) 241 There are special cases in which this may not be true. To illustrate, Kripke [1980] argues that we can know a priori the standard meter stick is one meter long. We know this a priori because the referent of ‘meter’ is just whatever length the standard meter stick happens to be. But the length of the stick could have been other than it is and so our a priori knowledge is of a contingency. Even if we grant that this is an example of contingent a priori knowledge grounded in understanding, it does not look like a good model for evaluative knowledge. It is not as if there is a standard action – say, whatever type of action Adam does after time t – which fixes the reference of an evaluative term – say, ‘good’. For more on this point, see McKeever and Ridge [2006] and Chapter 2 of this dissertation. 157 Chudnoff’s direct realist has the same limitation. His requirement that we be aware of a truth-maker pushes us to say that intellectual experiences, so understood, can only give us knowledge of necessities. 242 Even though a contingently true proposition may itself be an abstract object, the truth-makers for a contingent proposition will not themselves be abstracta (i.e., non-spatiotemporal and causally inert). By observing that these views are only good for explaining knowledge of necessities, I do not take myself to be uncovering anything surprising. But given that affect can give us knowledge of contingencies if it can give us any knowledge at all, we need to look for an alternative solution. This is arguably bad news for the sentimental perceptualists. Perhaps the intellectual perceptualist rival will emerge from his crypt, claiming to have seen this problem coming ever since we learned in Chapter 2 that affect is suited for giving knowledge contingencies. The sentimental perceptualist who takes the armchair challenge seriously has to argue that we can get knowledge of evaluative contingencies by mere reflection, which looks like an incredibly difficult kind of thing for which to argue. That said, I will make the case that the sentimental perceptualist actually has a highly attractive solution to the armchair challenge, one suggested by the solution to the content challenge and which preserves a causal story of armchair evaluative knowledge. 3 Answering the Challenge The structure of my solution to the armchair challenge on behalf of sentimental perceptualists is simple. The idea is that if sentimental perceptualists can answer their 242 We could give up this requirement, but then it would be puzzling how these experiences could generate knowledge. For discussion, see Chudnoff [2013]. 158 content challenge, then they have a solution to the armchair challenge, given only very minimal assumptions. And I have already argued (Chapters 3, 4, and 5) that we should be optimistic that the content challenge can be solved. Let me explain. Recall that the content challenge is a demand for an explanation of how affect could have come to be about value. And we can ask similar questions about perception. Philosophers of perception inquire about how perceptual experiences of various sorts (e.g., solidish perceptual experiences) came to have certain contents (e.g., being solid). I argued that sentimental perceptualists should stick close to their perceptual analogy and model the content-determination story for affect on that of ordinary perception. I proposed that perceptual experiences get their content by virtue of having the biological function of standing in causal-covariational relations with what they are about, or at least that the presence of such relations provides very good evidence of content. So turning to affect, if affect has the biological function of standing in causal-covariational relations with value, then we at least have very good evidence that affect is about value. I made the case (in Chapters 4 and 5) for optimism about the antecedent of this conditional. Crucially, this answer to the content challenge does not extend to our offline affective responses, unless we are willing to say that offline affective experiences never fulfill their function, since they cannot be caused by the values that they represent. But we can extend the story to the offline case given very minimal assumptions, and this doubles as an attractive explanation of evaluative knowledge from the armchair. Offline affect is incredibly similar to online affect, and the former plausibly gets its content by virtue of being a simple extension of the latter. As I have already pointed out, online and offline affect respond to the same contents, but they respond to contents 159 which are represented in different ways. To illustrate, an affective experience which is a response to content represented in perception counts as online; but if the experience were a response to sensory imaginings with the same content, then it would be offline. Here, then, is the key thought about how offline affect gets its content: offline affective experiences have the content that they do because they have the function of tracking what would be the case evaluatively (which will often only be contingently true); and they can have this function without standing in causal relations to value because they are developmentally downstream from our online affective experiences, which do stand in causal relations with value. According to this picture, there are not really two affective systems, the online and the offline; there are just single sets of dispositions which trigger in response to contents which are represented both online and offline. It is worth making an observation in favor of the thesis that offline affect is developmentally downstream in the right way, at least once we grant some background observations (which I have already discussed in previous chapters). The first observation builds on the thought that online affective experiences have the biological function of standing in causal-covariational relations with value. If that is right, then for our ancestors it promoted reproductive success to have such experiences. And if this observation about reproductive success is true, then it would presumably also promote reproductive success to have offline experiences which track what would be the case evaluatively. This would, for instance, allow us to effectively plan for the future. We would be able to learn what would be best for us in advance, rather than having to wait for the future to come to us. Of course, this point about promoting reproductive success does not entail that offline experiences have the function of tracking value. But if we agree that online experiences 160 do, it is natural to think that offline experiences also have that function (presumably among other functions). By now I suspect it is becoming clear how this solution to the content challenge can help perceptualists to accommodate pretheoretical assumptions about the possibility of evaluative knowledge by mere reflection. Online affective systems came to be about value by having the biological function of standing in causal-covariational relations with the target values, but those same dispositions evolved so as to respond in kind offline to the same inputs. These offline responses are developmentally downstream from the online responses, since only the latter hook causally to the values that they are about and those connections were needed for offline tracking to be possible; but once the offline tracking abilities come about, their nature is such as to track value – at least in worlds like ours – without further reliance on the online responses. On this picture, even though offline affect is developmentally dependent on online affect, they are not epistemically dependent in a way that blocks us from getting evaluative knowledge by mere reflection. This is because a core set of the dispositions to affectively represent values in the imagination are native rather than learned. A standard way of thinking about innateness, and the one I have in mind, is that innate characteristics are organized in advance of experience. 243 Innate characteristics need not be universal or unmalleable; the idea is that genetic structure supplies, as Gary Marcus puts it, a first draft. We are prewired in certain ways, though it may be that experience edits, or alters, that prewiring. It is often taken for granted that some “valuations” or 243 See Marcus [2004]. Marcus’s understanding of innateness is accepted by Haidt and Joseph [2004], and something like it seems to be assumed by other psychologists, or so Haidt and Joseph think. 161 “proto-evaluative judgments” must be innate. Here is how Cosmides, Tooby, and Barrett put it: [T]here must be an irreducible core set of initial, evolved, architecture- derived content-specific valuation assignment procedures, or the system could not get started. The debate cannot sensibly be over the necessary existence of this core set. The real debate is over how large the core set must be, and what the proper computational description of these valuation procedures and their associated motivational circuitry is. [2005: 317] Psychologists as diverse as Laurence Kohlberg and Jonathan Haidt have taken this innate structure to be wholly, or at least in large part, affective. 244 By making the reasonable bet that these psychologists are right, the sentimental perceptualist is in a position to explain how we get basic knowledge from the armchair. Because these capacities are native, there is no pressure to say that knowledge grounded in the offline experiences is always somehow epistemically subordinate to knowledge obtained via online experiences. After all, we do not want to say that the evaluative beliefs we form on the basis of offline affective experiences are epistemically dependent on the online affective responses of our phylogenetically distant ancestors. 244 Haidt and Joseph [2007: 374] succinctly describe Kohlberg’s (and Piaget’s) picture: For Piaget and Kohlberg, there was just one word written in the first draft of the moral mind: empathy (or perhaps a compound word: “perspective-taking”). There were also some words written in other (non-moral) chapters of the first draft: like and dislike. As long as the child liked some things (such as pleasure, candy, or friendship) and disliked others (such as pain, frustration, or rejection), then as she became increasingly good at taking the perspective of others during the concrete operational stage, the child could feel for herself (empathetically) that actions that hurt others were bad, while actions that made others happy were good. In this way children come to understand the value of different kinds of rules, and to appreciate that rules and social practices have to be justified by reference to something else. For Kohlberg, that something else was tradition, authority, and society for “conventional” moral reasoners, but it was justice for the most advanced moral reasoners. The knowledge of justice was not innate; it was the crowning achievement of the editing process, and the editing process was constructivism driven by the experience of role-taking. This approach is elegant in explaining how so much can be derived from so little innate knowledge, and it is reasonable if you believe that the moral domain is restricted to matters of harm, rights, and justice. 162 As we now see, this model is very different from the intellectual models considered in section 2. To reiterate an earlier observation, existing intellectualist models are only good for explaining direct knowledge of necessities, since they rule out any information from the world having an influence on the intellectual experiences. But the model I have sketched does allow worldly information to get in: causal interaction with the world shapes the affective dispositions. And when we shift to offline inquiry, the influence of that interaction is still, we might say, felt. A major payoff of the strategy is that it allows the sentimental perceptualist to adopt a causal explanation of how we get basic evaluative knowledge from the armchair, which even some rationalist-minded philosophers concede is the most philosophically- secure model available. 245 Even though there are no causal relations between offline affective experiences and the values they represent, causal relations between affect and value still play a role in explaining how the offline experiences track value. The drawback of this strategy, however, is that it depends on empirical assumptions about the nature and origins of affect; and all I can do in this dissertation is motivate cautious optimism that the assumptions are true. In the end, though, this not too big a drawback relative to intellectualist alternatives considered above, since those views also depend on as yet unsettled empirical assumptions. For example, intellectualist views which hope to explain the evaluative knowledge that we pretheoretically take ourselves to have depend on empirical assumptions about how good evaluative reasoners form evaluative 245 Enoch [2010] is candid in explaining why realists are likely to struggle to explain how so many of our evaluative beliefs are true if there is not a causal relation between those beliefs and values. Enoch himself tries to make do in his value epistemology without such a relation. 163 judgments (e.g., by making them on the basis of experiences which arise out of understanding alone). 246 6 Summing Things Up Sentimental perceptualists propose that we can demystify evaluative knowledge by analogizing it with perception. But, as I pointed out in section 1, the analogy seems to come up short when we turn to evaluative knowledge acquired by reflection. 247 Sentimental perceptualists thus face the armchair challenge. Familiar perceptualists strategies for addressing the challenge carry a commitment to a necessity-requiring value epistemology, which we know from Chapter 2 fits much better with intellectual perceptualism than sentimental perceptualism. But we also know from Chapter 2 that a necessity-requiring value epistemology is difficult to motivate on independent grounds, and so it would be pretheoretically desirable to have an attractive solution to the armchair challenge which does not have the necessity-requiring limitation. Nevertheless, with respect to the armchair challenge, in particular, giving up the necessity-requiring limitation might initially seem to make the armchair challenge all the more difficult, since it requires an explanation how we could learn evaluative contingencies by reflection. However, if this stiffer version of the armchair challenge can be met, then this would remove much, if not all, of the remaining attraction of intellectual perceptualist approaches. 246 Additionally, they make the empirical assumption that there even are intellectual experiences of the relevant sort. See Williamson [2007]. 247 Perhaps you think that we can get (some) basic empirical knowledge from the armchair. I am sympathetic to this possibility, although I do not explore it here. 164 How can affect ground knowledge of contingencies from the armchair? I argued in this chapter that sentimental perceptualists can actually reduce the armchair challenge to an independent, further challenge – the content challenge. If they solve the latter, then they also receive a solution to the former. And as we know from Chapters 3, 4, and 5, there is evidence which motivates cautious optimism that the content challenge can be solved. Ultimately, the answer to the armchair challenge which emerges from the solution to the armchair challenge is highly philosophically attractive. For one, it avoids the necessity-requiring limitation of intellectual perceptualist alternatives. Furthermore, it preserves a causal story of evaluative knowledge without giving up the possibility of acquiring evaluative knowledge by mere reflection. 165 Chapter 7: Taking the Perceptual Analogy Seriously 0 Introduction In this final chapter, I continue defending and developing sentimental perceptualism in the face of important objections. In particular, András Szigeti and Michael Brady have independently developed a new and important set of objections. (They object to versions of sentimental perceptualism which have emotions, rather than emotions and desires, playing the foundational role. But I take it that they would push similar objections against a more inclusive version of sentimental perceptualism.) The objections have a common structure: they begin by conceding that emotions have some important epistemic role to play, but then go on to argue that understanding how emotions play that role means that there must be some alternative, emotion-independent route to obtaining knowledge of value. 248 If there has to be such an emotion-independent route, then the perceptual analogy breaks down in a significant way; emotions would not be the final arbiter in the evaluative domain in the way that perceptual experiences are in the empirical. In this chapter, I argue that the right ways for sentimental perceptualists to respond to each of these objections are revealed by thinking through how analogous objections applied to perception and the empirical domain would be answered. Thus Szigeti’s and Brady’s objections should not persuade sentimental perceptualists to give up their view. However, my response shows that these objections put important constraints on what a form of the view has to be like in order to do exciting metaethical work. Throughout this dissertation I have argued that sentimental 248 Szigeti [2013] develops three such arguments and Brady [2013] adds two more. 166 perceptualists must go beyond a minimal analogy with perception. This chapter continues that same theme; for their view to be plausible, sentimental perceptualists must, we might say, take seriously the analogy with perception. 1 Trouble for Sentimental Perceptualism? Sentimental perceptualists insist on a central epistemological role for emotions in value epistemology, but historically many philosophers have argued that emotions are of no epistemic value. Here, to illustrate, is a memorable passage from Seneca’s 85 th Epistle: [I]t makes no difference how great the passion is; no matter what its size may be, it knows no obedience, and does not welcome advice. Just as no animal, whether wild or tamed and gentle, obeys reason, since nature made it deaf to advice; so the passions do not follow or listen, however slight they are. Tigers and lions never put off their wildness; they sometimes moderate it, and then, when you are least prepared, their softened fierceness is roused to madness. Vices are never genuinely tamed. Again, if reason prevails, the passions will not even get a start; but if they get under way against the will of reason, they will maintain themselves against the will of reason. For it is easier to stop them in the beginning than to control them when they gather force. This half-way ground is accordingly misleading and useless; it is to be regarded just as the declaration that we ought to be “moderately” insane, or “moderately” ill. Virtue alone possesses moderation; the evils that afflict the mind do not admit of moderation. You can more easily remove them than control them. [2011] For Seneca, and many other Stoics, we should strive to rid ourselves of all ordinary emotions, for once such emotions get a start, reason cannot stand against them. Our ordinary emotions are always a source of evaluative confusion. 249 249 The claim that Stoics were against emotion requires qualification (Sorabji [2000]). Seneca, for instance, thought that the wise could cultivate special emotions (or something like emotions) to replace the emotions of unenlightened folk. 167 In contrast, contemporary rationalists typically concede that emotions play at least some important epistemic role(s). 250 One reason for the shift – though not the only one – is that the work of many psychologists and neuroscientists makes it apparent that we would struggle to detect much of what we think of as valuable without emotion (see Damasio [1994]; Ellsworth [1994]; LeDoux [1998]; Frijda [2007]). To note one well- known example, Damasio [1994] has found that frontal lobe damage impairing emotional processing leads to poor practical decision making, even while scores on intelligence tests remain the same. One might think that it would be at least a small victory for sentimental perceptualists if everyone must concede that emotions are in some way epistemically important. But András Szigeti and Michael Brady have independently developed provocative arguments that aim to show otherwise. Whereas the Stoics have an uncompromising attitude toward the emotions – insisting that they have no value – Szigeti and Brady argue at length that emotions do have value; but they then try to show – in different ways – that the value emotions have helps us to understand why there has to be an alternative, emotion-independent route to evaluative knowledge. I call these shrewd rationalist objections, since they begin by making a sort of concession to sentimental perceptualism. Szigeti develops three shrewd rationalist objections and Brady adds two more. 251 In what follows, I begin by explaining and critiquing the three from Szigeti. I then turn to the two from Brady. 252 The method 250 In fact, I am not aware of any rationalist who still clings to the old Stoic idea that our ordinary emotions have no helpful role to play. For some examples of rationalists who describe in detail the roles that they believe emotions to play, see Peacock [2004] and Audi [2013]. 251 Szigeti’s target is the view that emotions are the foundation of our evaluative knowledge and he does not assume proponents of that view make any analogy with perceptual experience. But my argument is that Szigeti’s opponents can respond by making the analogy and by developing it in the ways that I outline. 252 Brady [2013] develops other objections to sentimental perceptualism. I am only concerned here with so-called shrewd rationalist objections. 168 of defending sentimental perceptualism will be the same in each case. I explain why analogous arguments can seem to apply to perception, but of course we do not want to deny that our empirical knowledge is ultimately dependent on perception. Understanding why the arguments do not work for perception illuminates the theses we would expect sentimental perceptualists to accept in order to avoid the apparent problems. The lesson, once again, is that sentimental perceptualism is plausible only if its proponents take the perceptual analogy seriously. 253 2 Shrewd Rationalist Objections: Part 1 2.1 The argument from recalcitrance Szigeti argues that emotions are heuristics for value. To say that emotions are heuristics is to say that they are mental shortcuts or rules of thumb. When we rely on a heuristic we substitute the target attribute for the heuristic attribute. 254 This generally allows us to reason more quickly than we would otherwise be able; heuristics are frequently said to be “fast and frugal” (Gigerenzer et al. [1999]). Relying on heuristics is often superior to non-heuristic based reasoning, though not always. For example, when we rely on the “availability heuristic,” we estimate the probability of something based on the examples that come readily to mind. Since plane crashes are commonly reported in the media, relying on the availability heuristic, rather than the statistics themselves, can lead us to a misguided picture of how dangerous air travel is (Szigeti [2013: 847]) 253 I do not claim that sentimental perceptualists need to say that emotions are literally perceptual. There may still be important disanalogies. To get a sense for some of the ways in which emotions might be thought analogous to perceptions, see Salmela [2011]. 254 Szigeti borrows talk of substitution from Kahneman and Frederick [2002]. 169 Szigeti argues that there is a great deal of empirical evidence supporting the conclusion that emotions are a biological system subserving heuristic functions [2013: 848 – 53]. I will review just a bit of that evidence here. Some supposedly comes from experimental psychology. Recent psychological research suggests that people rely on emotions directly in forming a variety of different kinds of judgments and decisions. For instance, if information does not trigger an emotional response, it is often treated as less significant in one’s deliberation. Recognizing that something raises the probability of getting cancer causes a significant emotional reaction for most people, whereas recognizing that something raises the probability of a car accident has a much smaller effect. This difference seems to impact our judgments about which hazards are riskiest and most in need of regulation (Slovic et al. [2002: 410]). Another example is the “outrage heuristic” (Sunstein [2005: 538]). When deciding how severe a wrongdoer’s punishment should be, we tend to use the degree of our outrage as the measure. Szigeti hypothesizes that psychological evidence for other emotional heuristics (e.g., a shame heuristic) could be discovered. There is also supposed to be evolutionary evidence for the heuristics model. Emotions such as sadness, anger, disgust, and fear seem to have evolved to help with “fundamental life tasks” (Ekman [1992]). Fear, for example, has evolved to help us respond quickly and effectively to danger. We often (though not always) take something to be dangerous, and behave accordingly because we are afraid of it, which is often better for responding effectively than if we stopped to engage in rational reflection about whether the object of our fear really is dangerous. (By the time we answer that question, we may already be dead.) More generally, these emotions – often called “basic 170 emotions” – are phylogenetically hard-wired response-mechanisms which respond quickly to environmental cues and, at least in familiar environments, obviate the need for slower, cognitive reflection. There is considerable controversy about which emotions are hard-wired in this way; but there is a good deal of agreement that some are. 255 Szigeti argues that there are several different ways to see why the fact that emotions are heuristics makes trouble for sentimental perceptualism. The first is the argument from recalcitrance (Szigeti [2013: 855 – 56]). A recalcitrant emotion, as Szigeti understands it, is an emotion that persists even though it is judged unwarranted. For example, an agent may fear falling even though she acknowledges that there is no real chance she will fall, e.g., because she’s securely fastened to a safety harness. The phenomenon of recalcitrance is adequately accounted for by the heuristics model of emotions. Affect heuristics have evolved to respond to certain environmental cues; and they often do not cease responding to those cues even when we recognize the cue to be misleading. But then this looks like a problem for sentimental perceptualism. To use Szigeti’s words, “If emotions can persist despite our best judgment that they are wholly unjustified in some cases, then it’s hard to see on what grounds emotions should be stipulated to be the highest court of appeal in matters axiological” [2013: 856]. Thus recalcitrance is supposed to be a problem for sentimental perceptualism, because in cases of recalcitrance we are apparently not relying on our emotions and, moreover, we seem (at least in many instances) justified in not doing so. 255 I take it Szigeti assumes that if the basic emotions are heuristics, then this somehow guarantees, or makes it probable, that non-basic emotions are, too. There is no need to speculate about the matter here. 171 2.2 Responding to the argument from recalcitrance The argument from recalcitrance relies on two premises. The first is that there can be cases of recalcitrance. The second is that if there are cases of recalcitrance, then all evaluative beliefs are not epistemically dependent on emotion. But there is something odd about this conditional premise. Here, again, is how Szigeti puts it: If emotions can persist despite our best judgment that they are wholly unjustified in some cases, then it is hard to see on what grounds emotions should be stipulated to be the highest court of appeal in matters axiological. [2013: 856] A sentimental perceptualist ought to say that in whatever sense emotions can be unjustified, so too can perceptions. But notice how unpromising the conditional premise looks if we substitute “perceptions” for “emotions” and “empirical” for “axiological”: If perceptions can persist despite our best judgment that they are wholly unjustified in some cases, then it is hard to see on what grounds perceptions should be stipulated to be the highest court of appeal in matters empirical. But perceptions are the highest court of appeal in matters empirical. Perceptions seem to be the only basic source of knowledge about how the world contingently is (or nearly the only source, if we allow for special cases of the contingent a priori). Any judgment that a given perception is unjustified depends in some way on other perceptions. One may complain that there is no sense in which perceptual experiences are ever unjustified. But it seems to me that sentimental perceptualists can reasonably maintain that perceptual experiences can be unjustified in the same ways as emotions. Szigeti seems to have two sorts of cases in mind. First, there are cases in which our emotional 172 systems are working as they should, but they nonetheless misrepresent the world evaluatively. 256 A person’s fear of falling while riding a roller coaster is in some sense incorrect (there is a negligible chance of crashing or the harness failing), but the fear system is still working as it should. This is arguably like the Müller-Lyer illusion. Our visual system is working as it should in representing one line as longer, but the two lines are in fact the same length. There are other cases, though, in which our emotional or perceptual systems are not working as they should. This could be a result of our own irrational choices or beliefs. For example, a person who unwisely cultivates a strong habit of watching movies or playing video games for lengthy periods may be sad when she cannot indulge in the activity for significant stretches of time; but, in this case, the emotion is not merely incorrect but also, we might say, indirectly unjustified due to the irrationality of her choices to cultivate her emotional tendencies in that way. An analogous case involving perception would be someone who has a variety of irrational beliefs about the paranormal and so develops both auditory tendencies to hear creaking noises as voices and visual tendencies to see distant planes as flying saucers. There are also cases in which our emotional or perceptual faculties are not well-functioning yet we’re not responsible for the defect. In sum, the sentimental perceptualist’s analogy with perception reveals how Szigeti’s conditional premise – if emotions persist despite judgments that they are unjustified, then emotions must not be the highest court of appeal in matters axiological – can be resisted in a way that is natural for the theory. The very same phenomenon occurs with perception and empirical judgments (or so the sentimental perceptualist can 256 We might prefer to speak of correctness/incorrectness here. 173 reasonably maintain) and yet this is not grounds for denying that perception is the final court of appeal in the empirical domain. One may worry that although I have responded to the letter of the argument, the intuitions behind it remain. The thought, perhaps, is that it seems as if we are relying on reason alone in cases of emotional recalcitrance. There are at least three theses that sentimental perceptualists will naturally accept – theses suggested to us by reflecting on perception – that help dissolve the intuition. Here is the first: Background Principle Override: we often have emotion-backed background beliefs about what is valuable that we use to override how things appear evaluatively at a given time. Imagine a person unwittingly strolling down a bike path. A biker who passes shouts at him to move to the walk path. The person becomes angry, but then immediately judges the anger to be incorrect on the basis of the following principle which he accepts: looking out for the safety of others is good. The belief in this principle is emotion-backed, even though it is not backed by any occurrent emotions. The agent’s credence in the principle is quite high (e.g., because it is based on his beliefs about a wide range of cases) and so he judges his occurrent emotional experience to be misleading. In similar fashion, an agent who visually experiences a strawberry as purple may judge her perceptual experience somehow mistaken (it’s not a real strawberry or it’s not purple), since she has the background perception-backed belief that strawberries are red. 257 257 What does it mean for a principle to be supported by emotion or perception? The idea is that the principle is a generalization from past emotional/perceptual experiences: when we consistently experience some object (e.g., an act-type, a natural kind) as having some property, we often form the justified, general belief that objects of that sort possess the property in question (e.g., looking out for the safety of others is good, strawberries are red); and we often use those general beliefs to correct future experiences which suggest otherwise. 174 The walker who experiences the misleading anger may also have a belief about the conditions under which his emotions are reliable. Here is a second thesis that is natural for sentimental perceptualists to accept: Background Conditions Override: we often have background beliefs about when our emotions are likely to be reliable that we use to override how things appear evaluatively at a given time. Just as we have beliefs about when our perceptual experiences are reliable, so too do we have beliefs about when our emotions are. When lighting conditions are poor, for instance, we know to be highly skeptical of our visual experiences. Or, if a person has just brushed her teeth, she knows to be skeptical of her ability to detect the flavors of food. Similarly, we know that there are conditions when our emotions are likely to mislead, e.g., when we’re taking criticism or when we’re intoxicated. There is an important question of how we know when our emotions are reliable. But since there is a similar question for perceptual experience, the sentimental perceptualist hypothesizes that we know in an analogous way. 258 The sentimental perceptualist should emphasize that our emotion-backed beliefs in background principles or ideal conditions will often not be foregrounded in our thinking about whether an occurrent emotion is correct; and if we are not careful, this can mislead us into thinking that we’re not relying on emotions to correct an emotion. To illustrate, take a scenario in which a person is afraid of a snake slithering near her feet. Her fear (let’s suppose) represents the snake’s potential bite as a threat to her welfare. 258 One may wonder whether background conditions override is really distinct from background principle override, since presumably the sentimental perceptualist thinks our beliefs about when emotions are reliable is emotion-backed (just as it is with perception). What makes them distinct is that principles about ideal conditions make explicit reference to emotions and not (just) to values. 175 However, she figures out that the emotion is misrepresenting once she realizes it’s only a harmless corn snake. Foregrounded in her thinking about whether her fear is correct are certain non-evaluative facts, in particular that the snake has the visible marks of a corn snake and that corn snakes have a non-venomous, non-aggressive nature. She need not rely on emotion to learn such information. That said, to rationally conclude that the snake is no threat to her welfare she also needs to have emotion-backed beliefs about the kinds of things that can detract from her welfare (e.g., pain), or so the sentimental perceptualist can for all we have said maintain. Finally, there is a third thesis that illustrates how we might rationally set aside an occurrent emotional experience without appealing to pure evaluative reasoning: Testimonial Override: we often trust others evaluatively more than ourselves, at least in certain kinds of situations. Consider an example of a youth who has all too often come to recognize that his mother’s judgments about his welfare are more accurate than his own. And so when he is deciding whether to do an act, or series of acts (e.g., drinking multiple beers), that his mother has advised against, he judges that he should not do it. He does so on the basis of her testimony and despite the fact that the act affectively appears good. Examples of testimonial override are especially common in religious contexts. A person might trust a spiritual advisor or the authors of a religious text to know the evaluative truth. The existence of evaluative testimony is no more a threat to sentimental perceptualism than is empirical testimony to the idea that our empirical beliefs are epistemically dependent on perception. Just as empirical testimony is a route to empirical knowledge only if 176 perception is, the sentimental perceptualist says that evaluative testimony is a route to evaluative knowledge only if emotion is. So long as sentimental perceptualists take the perceptual analogy seriously, in the ways I have been suggesting, they can reasonably contend that recalcitrance is no more a problem for their theory than it is for the theory that empirical judgments ultimately depend on perception. Failing to keep the override theses in mind can mislead us into thinking that in cases of recalcitrance we rely on emotion-independent reason, or so the sentimental perceptualist maintains. 2.3 The argument from collective action Next we have the argument from collective action (Szigeti [2013: 856 – 57]). Szigeti uses guilt as his primary example. He says, “in many cases which involve harm due to the aggregation of actions by several people—think of environmental pollution, multinational corporations or intergovernmental organizations—contributors tend to experience no guilt feelings” [2013: 856]. The heuristics model can explain this: guilt has evolved to help one respond to one’s own actions in situations familiar to our ancestors. Paradigmatically, guilt responds to “a single action of visible impact and short duration by one agent causing harm to one or few victims” [2013: 857]. As we diverge from paradigmatic cases, guilt responses tend to weaken, or even disappear. But behavior in the non-paradigmatic cases, e.g., the actions comprising the collective harms mentioned above, are often no less wrong. We cannot always rely on our guilt responses as a measure of the degree of wrongness of an action. We have to, and indeed often do, step back from those responses to consider matters rationally. 177 2.4 Responding to the argument from collective action The first premise of the argument from collective action is that we often make reasonable judgments about the evaluative status of different actions despite never emotionally representing them as being that way. This often happens with collective action cases. Sentimental perceptualists should grant this premise. The trouble arises with the second, which says that if evaluative judgment comes apart from emotional evaluative representation in the way just mentioned, then we are not epistemically relying on emotions to make those judgments. In responding, the sentimental perceptualist should lean on the analogy. It is not the case that we can only know about an empirical property if there is a perceptual experience that accurately represents it. Many philosophers deny that we can perceive, say, the property of being a table, but no one should deny that we need perception to learn about the presence of tables. Or, to take some less controversial examples, we do not perceive protons or planet Earth (at least not until space travel), but we can learn about them using perception. There is a major outstanding question for philosophers of perception and sentimental perceptualists alike: what is the range of properties that can be experientially represented and how exactly do we know about the properties – empirical or evaluative – that are not? Although we cannot take up such questions in any detail here, it will be useful to take the abstract point made in this paragraph – namely, that it is natural for the sentimental perceptualist to hold that the class of evaluative properties we emotionally represent is smaller than the class of evaluative properties we know about on the basis of emotion – and point to one concrete path forward. 178 Many philosophers and psychologists believe that emotions are principally about factors significant for the agent who experiences the emotion, often labeled core- relational themes. 259 (I’ll assume these theorists are right, but other sentimental perceptualist might have different views about which properties get represented and so the story they tell will be slightly different.) Here is Richard Lazarus’s eloquent description of the kind of value that emotions tend to be about: An emotion is always about certain substantive features of the relationship between a person and an environment. Although this relationship can occur with the physical world, most emotions involve two people who are experiencing either a transient or stable interpersonal relationship of significance… [R]elational meaning refers to the juxtaposition of two conditions, the goal that is at stake in the encounter and the action or inaction of the other person that bears on the fate of that goal. Both events must be united in terms of a meaning, which must be appraised by the person as either a relational harm, potential or actual, or a benefit to produce an emotion. This is what relational meaning is all about… I have proposed that each emotion involved a special, and different relational meaning, which I referred to as core relational themes. For example, anger is the result of a demeaning offense against me and mine, anxiety is facing uncertain, existential threat, sadness is having experienced an irrevocable loss, pride is enhancement of one’s ego- identity by taking credit for a valued object or achievement, and relief is a change of a negative condition for the better. [2003: 126] On this picture, emotions are for the most part about the welfare of the agent experiencing the emotion. But we can allow that some emotions are altruistic in the sense that they target the welfare of others. Our ability to engage in mind-reading and/or perspective 259 See, for example, Ben-Ze’ev [2000]; Martha Nussbaum [2001]; Lazarus [2003]; Prinz [2004]. Nussbaum, however, allows that certain atypical emotions (e.g., wonder) represent impersonal value. 179 taking may be what makes such altruistic responses possible. 260 It is natural for sentimental perceptualists to accept the following: Content Limits: Emotions do not represent all types of values and are for the most part, albeit not exclusively, about what is significant for the agent experiencing the emotion. 261 Once this thesis is on board, not only do sentimental perceptualists have grounds for resisting a key premise in the collective action argument, but also have a way of speaking to the specific cases involving collective moral wrongs that Szigeti has in mind. Emotions are not directly about properties like moral wrongness, even though they play a role in helping us learn about such properties. Still, one may worry about how this learning takes place. Sentimental perceptualism, recall, only says that substantive knowledge of value is grounded in emotion. But just as we can have conceptual knowledge of tables, protons, mountains, etc. not grounded in perception, so too can we have conceptual evaluative knowledge. We can learn about the existence of evaluative properties not represented in emotional experience by combining our knowledge of values directly represented, 260 For discussion of various hypotheses about how altruism might emerge, see Nichols (2004). 261 There is an alternative way in which we might understand the idea that emotions are about what is significant for the agent experiencing the emotion, which I will briefly flag here. Readers of this dissertation will recall that in Chapter 4 I suggested that emotional experiences (may) involve representations of the experiencer’s reasons. (I did not explore the matter in detail, however.) If emotions are about reasons, then it may be that in the cases Szigeti has in mind, emotions will typically misrepresent the values that they are about (although this is open to doubt). So, for example, we may have weighty reasons to avoid driving a typical gas-powered vehicle, even though we do not feel terribly guilty when we do. But this is not a problem for the sentimental perceptualist, for such misrepresentations are very plausibly understood on the model of a perceptual illusion. Humans have only recently developed the capacity to collectively influence the climate in dramatic ways, and so it comes as little surprise that we would struggle to recognize the force of the reasons which this new ability gives rise to. But it would be a mistake to think that the presence of perceptual illusions means that we need a perception-independent route to knowledge. 180 conceptual evaluative knowledge, and non-evaluative knowledge. Take the collective action case. Our emotions are arguably not suited for tracking an impersonal evaluative notion such as moral wrongness. However, suppose we accept Philippa Foot’s [1958] idea that it is a conceptual truth that we have strong moral reasons to unselfishly promote human flourishing or welfare. 262 If we accept this, then we can see the rudiments of a story for how we could know that polluting is very wrong, even though it fails to generate a strong emotional response: we are in some way combining our emotion-based knowledge of welfare with our knowledge that morality demands unselfish promotion of welfare. It is important for sentimental perceptualists to eventually give a detailed story for how this combining works (it probably will not work by an inference transparent to the reasoner, for example), but those concerned with empirical knowledge will arguably face a similar question. For my purposes, it’s enough to note that sentimental perceptualists naturally take the class of evaluative properties we emotionally represent to be smaller than the class we know about on the basis of emotion; and moreover, it’s reasonable to expect they will be able to tell a story about how we come to know about the properties not directly represented. 2.5 The argument from difficult cases Szigeti’s third argument is the argument from difficult cases [2013: 858 – 59]. Here he uses Sophie’s Choice as his example. In this case, Sophie’s Nazi captors force her to choose one of her two children to be saved, or else both will be killed. Sophie ends up selecting one of her children. After making the choice, Sophie may experience guilt. 262 Foot’s idea is also endorsed by Michael Smith [1994] and Terence Cuneo [2007]. 181 After all, she has made a decision that she knew would lead to the death of one of her children. On an intellectual level, she may judge that she acted rightly and is not to blame, but her emotions are not meant to deal with such “extreme and difficult cases” though the “affective heuristic nevertheless continues to function as an information channel” [2013: 858]. The sentimental perceptualist may suggest that Sophie’s emotion is more likely to be regret. But then this leads straight to a supposedly serious problem for sentimental perceptualism. It is often very hard for agents to tell which emotions they are feeling in difficult cases. But if we do not know what we are feeling, then “the emotional response cannot be used as evidence in the evaluation of the relevant situation because one could only rely on the emotional response if one knew how to classify the emotion” [2013: 858]. 2.6 Responding to the argument from difficult cases The argument from difficult cases has the by now familiar two premise structure. The first says that it is often difficult to tell which emotions we are experiencing when we face puzzling evaluative situations, yet in those cases we often do come to determinate conclusions. I have no qualms here. The second is that if it is difficult to figure out our emotions in puzzling cases and yet we come to a determinate evaluative conclusion, then those judgments must not be supported by emotions. 263 There are a couple ways to resist this premise. First, it’s worth noting that the three “override” theses mentioned above can help the sentimental perceptualist. Take Background Principle Override and 263 Szigeti in fact only says that we would not be able to rely on the occurrent emotional response. But I take him to mean something stronger than this, namely that we would not be able to rely on emotions. Only the stronger claim makes trouble for sentimental perceptualism. 182 Background Conditions Override. If we imagine Sophie trying to figure out whether she has done anything bad, she may not be able to make sense of which emotions she’s experiencing – e.g., guilt, regret and/or sadness – but she may recognize that she’s in less than an ideal situation for having accurate emotional responses and also that according to principles she accepts (e.g., that people are not to blame for the bad situation they are in unless they knowingly behave in ways that are likely to cause the situation) she has not done anything to be ashamed of. The second, and key, problem with the argument is this: it is actually a good result for sentimental perceptualism that difficult evaluative questions are more-or-less co- extensive with the cases in which it is difficult to figure out exactly how our emotions are presenting things. Suppose, to return to the perceptual analogy, you are looking through the fog at a large object in the distance. It may be difficult to sort out precisely what you are seeing. What color and shape is the object, exactly? Is that the fog moving or the object? And so on. With respect to some of these questions, your perceptual experiences may be presenting to you a certain answer (e.g., your experience is only presenting the fog as moving, not the object), though it difficult for you to interpret them correctly. In other cases, the experience may not present a determinate answer. Any judgments we make will be fallible, although they will nonetheless be grounded in perceptual experience. The defender of sentimental perceptualism should insist that matters are much the same in difficult ethical situations. We occasionally have trouble recognizing how our emotions present things or, as a matter of fact, our emotions do not present things determinately. (Nevertheless, the evaluative belief we form in such situations may be very determinate; and it may enjoy a high-degree of justification, e.g., if we rely on a 183 well-supported background principle.) Difficult cases, then, are not obviously a problem for sentimental perceptualism, and they may even help the sentimental perceptualist’s cause, given that the view seems to make accurate predictions about which cases will prove difficult. 2.7 Final remarks on the idea that emotions are heuristics Szigeti’s official objections do not slide directly from the purported fact that emotions are heuristics to the conclusion that at least some substantive evaluative knowledge is independent of emotion. But what should the sentimental perceptualist think of the claim that emotions are heuristics? Sentimental perceptualists probably ought to reject the idea that emotions are by their nature heuristics, or allow that they are heuristics only in a highly qualified sense (see below), even if there are certain cases in which we might use them as heuristics. According to Szigeti’s understanding of a heuristic, we rely on a heuristic when we substitute the target attribute for the heuristic attribute. Szigeti’s central example of an emotional heuristic is the outrage heuristic: we often answer the question of how severe punishment ought to be by answering the question of how outraged we are. But a sentimental perceptualist will resist the idea that, at least in normal cases, that there is any substitution going on here. Recall Döring’s description of indignation (which is arguably the same emotion that Szigeti has in mind): In experiencing indignation at the harsh punishment of the toddler, it seems to you that the punishment is in fact unjust: your occurrent emotional state puts forward your indignation’s content as correct. This is in analogy to the content of a sense perception. In perceiving that the cat is on the mat, it seems to you that the cat is actually there. [2007: 377] 184 For Döring, indignation, or outrage, is the way in which we become aware of the target attribute; it represents it directly. 264 If we want to call outrage, at least in normal cases, a heuristic, then the sentimental perceptualist should insist that we ought to say the same about perception. We could treat perception as a heuristic – we answer the question of whether the cat is on the mat by answering the question of whether we see a cat on the mat – but this runs roughshod over an apparently important distinction between perception and typical heuristics, namely that perception, in contrast with the typical heuristic, is by its nature about the target attribute. But even if we want to think of perceptions and emotions as heuristics, there is no trouble, so long as we recognize that this fact alone has little epistemological upshot for how knowledge in some domain can be achieved. 265 In arguing that it is misleading to say that emotions are heuristics, sentimental perceptualists are not forced to deny any of Szigeti’s scientific data points. The reasons are simple. None of the data points, as Szigeti himself notes, show that emotions are not perceptual-like experiences of value [2013: 850]. But then if that is right, we have granted (or at least allowed) sentimental perceptualism’s favored psychological framework, and so the only way that Szigeti can generate trouble for sentimental perceptualism is by making evaluative assumptions and/or assumptions about how sentimental perceptualism is to be developed. As a case in point, Szigeti has the following to say in his discussion of the evolutionary reasons to treat emotions as heuristics: 264 The point I make here is independent of whether we think Döring has gotten outrage’s evaluative content correct. 265 Szigeti [2013: 851] seems to treat perception as a heuristic, too. 185 But also these evolutionary accounts appear to converge with the heuristics-model as to what emotions cannot do for us. As phylogenetically hard-wired response-mechanisms, emotions are too rigid to track fine-grained evaluative features of situations which require us to make normative judgments or decisions. [2013: 851] It seems to me that there are a lot of crucial assumptions here. Most obviously, there is an evaluative assumption about the fine-grainedness of the evaluative landscape. More subtly, though, there appears to be a background assumption about what a sentimental perceptualist theory has to be like, namely that it must say that the only values we can know about are ones directly tracked by emotion. But, as we have already seen, that is a commitment we would expect those sentimental perceptualists who take the perceptual analogy seriously to reject. In sum, Szigeti’s shrewd rationalist objections to sentimental perceptualism do not leave the sentimental perceptualist without recourse. By hewing close to the perceptual analogy, sentimental perceptualists can have a theory that is not only compatible with the cases Szigeti has in mind but also predicts them. 3 Shrewd Rationalist Objections: Part II 3.1 The argument from indiscriminacy Using an alarm-bell metaphor to characterize at least some of our emotions is fairly common (see, for instance, Ben Ze’ev [2000] and Greene [2013]). Typically, the metaphor is applied to so-called “affect program” emotions. This group of emotions includes anger, fear, sadness, disgust, joy, and surprise. Here is Brady’s description of affect programs and affect program responses: 186 Affect programs are “relatively discrete special-purpose mechanisms that are sensitive to some important aspect of human life”, and which evolved because they were of adaptive value with respect to various recurring and universal human situations. Affect program responses are short-term, reflexive, and phylogenetically ancient reactions to a limited class of perceptual inputs. On this line, fear is an automatic, reflexive response to potential danger, which results from an affect program which has evolved to deal with threats. [2013: 20 – 1] 266 Affect-program responses are indiscriminate. That is, they tend to trigger on the basis of a narrow range of inputs, operating on a principle of “better safe than sorry.” Emotions such as fear often emerge before there is good evidence that the emotion would be accurate to the situation; and this is by-and-large a good thing. Having certain emotions be highly sensitive, or indiscriminate, helps us to survive; the indiscriminacy of fear, say, can help us to avoid a predator we might not otherwise of avoided if our fear only triggered upon excellent evidence of a predator (Ellsworth [1994]). But Brady insists that it is not just affect program responses that are indiscriminate. Emotions such as guilt, resentment, pride, and shame also tend to be responses to a narrow range of environmental inputs, or so it would appear. (In other words, the sets of inputs sufficient to generate the emotion, at least in many cases, do not make it highly likely that evaluative content of the emotion is accurate.) Like affect programs responses, this second group of emotions has a pervasive influence on our behavior; but unlike affect program responses, they are not so much hardwired as they are the product of our social and cultural surroundings. Now we are in a position to identify a fourth shrewd rationalist objection to sentimental perceptualism, the argument from indiscriminacy. If our emotions are 266 Brady quotes in the passage above from D’arms and Jacobson [2003: 138]. 187 indiscriminate, then we need a more discriminating evaluative system to determine whether our indiscriminate emotional responses are getting things right. Brady puts the objection in characteristically rationalist language: At this point it is common for philosophers and psychologists to appeal to our capacity for reason as a solution to the problem of the lack of discrimination in our emotional responses. For there is a long-standing and venerable tradition according to which emotion needs to be monitored and controlled by reason: it is our capacity for rational reflection on our emotion situation that is essential as a check-and-balance on our emotional reactions. [2013: 100] The indiscriminate nature of emotions explains why emotions often distort and why we need emotion-independent reason as a check on emotions. 267 3.2 Responding to the argument from indiscriminacy I suspect Brady is right that emotions are indiscriminate (although see below for a qualification about the alarm-bell metaphor); and in any case, sentimental perceptualists can accept this. What they should challenge is the following conditional premise: if our emotions are indiscriminate, then we need a more discriminating evaluative system to determine whether our emotional responses are getting things right. To see how they should challenge it, let’s return to the perceptual analogy. Take a case in which a hiker has a visual experience as of a shallow stream flowing over a bunch of rocks. Because the rocks are so far away, the shapes and colors she sees that trigger her visual experience of a stream are very much compatible with the absence 267 Ben-Ze’ev [2000: 170 – 71] develops a similar line of thought. 188 of a stream. 268 The experience is thus indiscriminate in that it triggers on the basis of limited input (though it may not be indiscriminate in virtue of functioning according to a “better safe than sorry” principle). As she gets closer, she realizes that what she initially experienced as water is really just sunlight shimmering on a bed of rocks. She corrects her initial judgment by getting a better perspective on the matter. This example helps us to notice something that can be misleading about the alarm- bell metaphor. Most alarms trigger on the basis of certain kinds of input, but are not so sophisticated that they cease firing when they get new input revealing the initial input to be misleading. For example, standard smoke alarms aren’t capable of registering that even though there is smoke around, there’s no need to sound the alarm, e.g., because the cook has everything under control. In contrast, when rustling sounds coming from the nearby foliage cause an agent to be afraid, the fear will (normally) dissipate if she learns that the rustling was caused by the wind. Similarly, when A steps on B’s toe, and B becomes angry, we can expect B’s anger to dissipate if he realizes that A unintentionally stumbled. Here, then, is the worry about the indiscriminacy argument: our emotions do tend to emerge on the basis of limited information about their object, but they also have a tendency to become more accurate as we learn more non-evaluative information. It is thus natural for sentimental perceptualist to accept the following: Increasing Accuracy: As we become aware of more non-evaluative information about a situation, our emotional responses tend to become more reliable. 268 As a matter of fact, the proximal stimulations, or lower-level perceptual representations, that trigger a perceptual experience are almost always compatible with the representation’s being inaccurate. This is the so-called underdetermination problem for perception (see Burge [2010]). 189 In this way, emotions are, we might say, more akin to a guard-dog than an alarm-bell. 269 3.3 The argument from stimulation Now for the final shrewd rationalist objection. Brady argues that emotions have the function of raising evaluative questions; and he contends that this is ultimately a problem for those who believe that emotions are a basic source of justification similar to perceptual experience. When Brady says emotions have the function of raising evaluative questions, he only means that it is something emotions do (emotions do not have the biological function of calling their accuracy into question, for example). The argument for this functional claim is simple. There is a great deal of evidence that emotion and attention are closely linked. An agent who experiences fear, for example, will generally have her attention drawn to the object of her fear. The fear will also generally compel her (all else equal) to continue focusing on the emotion’s target. And this seems to be true of emotions generally. Although it may be obvious to many from experience, Brady notes that there is overwhelming scientific (neurophysiological and psychological, specifically) evidence for this connection. Given the connection between emotion and attention, and given the view that emotions present their objects as valuable in some way, it is a short step to the following: I propose that one of the important things that attentional persistence can do is to enhance our representation of potentially significant objects and events, precisely by enabling us to discover reasons which bear on the accuracy of our initial emotional appraisals….[T]he persistence of attention in emotional experience can facilitate, by motivating the search for and discovery of reasons, a judgment as to whether emotional appearance in this instance really does match evaluative reality. [2013: 93] 269 What if the emotion does not dissipate? Here I refer the reader back to my response to the argument from recalcitrance. 190 In Brady’s view, emotions play a crucial role in directing our attention to objects of potential importance which then often prompts reflection on whether the object really is important in the way the emotion presents it as being. The well-supported thesis that emotions are prompters of reflection purportedly makes trouble for sentimental perceptualism: [T]he proposal that emotions involve persistent attentional focus, and that this motivates the search for and discovery of reasons that are relevant to our emotional situation, suggests, interestingly enough, that it is emotions themselves that rule out our taking our initial emotional appraisals at face value. In other words, it is our emotions themselves that raise, rather than silence, the justificatory question… [2013: 96 – 7] Although Brady allows that emotions present value in a perceptual-like way, he thinks it is a mistake to go from that psychological thesis to the normative thesis that emotions are the foundation of our evaluative beliefs in the way that perceptual experiences are the foundation of our empirical beliefs. Perception silences questions; emotion raises them [2013: 81 – 90]. 3.4 Responding to the argument from stimulation In what sense is it true that emotions prompt evaluative reflection? The remark that it is “emotions themselves that rule out our taking our initial emotional appraisals at face value” suggests that emotions by their very nature prompt evaluative reflection. But I doubt that is true. What emotions do by their nature is focus attention. That focusing of attention will only prompt evaluative reflection if certain other conditions are met, e.g., 191 if we are careful and conscientious people. 270 All else equal, our emotions seem to cause us to act in ways stereotypical for whatever emotion is in question (Frijda [2007]). For instance, anger naturally leads to retaliation, fear to fight or flight, and so on. Emotions themselves have a tendency to lead us to take the emotional presentation for granted and to act on that basis. To drive home the point that it is not in the nature of emotions to prompt evaluative reflection, consider that for many emotions, humans and non-human animals are capable of experiencing those emotions, but animals do not engage in evaluative reflection on the accuracy of their emotions. If there is only a highly contingent relationship between emotion and reflection, is this a problem for sentimental perceptualism? Sentimental perceptualists can reasonably argue that it is not. Let’s return, one last time, to the analogy. Our visual system, for instance, is bound up in intimate ways with visual forms of attention. Here is a helpful summary remark from a recent overview of the vast literature on visual attention: [A]ttention allows us to optimize performance in visual tasks while overcoming the visual system’s limited capacity. Attention optimizes the use of the system’s limited resources by enhancing the representations of the relevant, while diminishing the representations of the less relevant, locations or features of our visual environment. Selective attention thus enables us to gather relevant information and guides our behavior – key factors in for the evolutionary success of an organism. (Carrasco [2011: 1486 – 7]) 270 Brady [2013: 158 – 191] has much to say about the “virtuous regulation of attention” that indicates he would ultimately agree. 192 Many psychologists believe that visual experience is not best thought of as a purely sensory phenomenon but rather as one that involves both sensation and attention. 271 And if so, we might say, reasonably enough, that our visual systems promote empirical reflection on certain phenomena by focusing visual attention on certain spatial locations, properties, or objects. That the visual system might prompt reflection in this way, however, is no cause for us to worry that empirical knowledge must be obtained without reliance on perception. And as far as I can tell, matters are much the same with emotions. There is a way that emotional experience can lead to reflection on the accuracy of emotional experience, but only for the epistemologically benign reason that emotions make salient what they are about. 4 Conclusion Sentimental perceptualists argue that emotions (and/or desires) are the foundation of our evaluative knowledge just as perceptions are the foundation of our empirical knowledge. Historically, many rationalists have argued that not only are emotions not the ultimate source of evaluative knowledge but are actually of no epistemic use at all. Contemporary rationalists, however, concede that emotions have an important role to play in everyday learning about value. Some of these rationalists go on to propose what I have called shrewd rationalist objections: if we think more broadly about the uncontroversial epistemic roles emotions play (e.g., as heuristics, alarm-bells, or prompters of reflection), we will learn that they actually cannot also play a perceptual-like foundational role. But 271 Visual attentional phenomena occur at a very basic level. Carrasco notes, “Initially, there was a great deal of interest in categorizing mechanisms of vision as pre-attentive or attentive. The interest in that distinction has waned as many studies have shown that attention actually affects tasks that were once considered pre-attentive, such as contrast discrimination, texture segmentation and acuity” [1485]. 193 I have argued that whatever we think about the ultimate attractions and prospects of sentimental perceptualism, the shrewd rationalist objections should not convince us to reject it. Throughout the chapter I spelled out some auxiliary commitments that we should expect sentimental perceptualists who take their perceptual analogy seriously to accept. Here are some of those commitments (the ones that were labeled): Background Principle Override: we often have emotion-backed background beliefs about what is valuable that we use to override how things appear evaluatively at a given time. Background Conditions Override: we often have background beliefs about when our emotions are likely to be reliable that we use to override how things appear evaluatively at a given time. Testimonial Override: we often trust others evaluatively more than ourselves, at least in certain kinds of situations. Content Limits: Emotions do not represent all types of values and are for the most part, albeit not exclusively, about what is significant for the agent experiencing the emotion. Increasing Accuracy: As we become aware of more non-evaluative information about a situation, our emotional responses tend to become more reliable. For any case in which rationalists insist that we do not need emotion to know a substantive evaluative proposition, the sentimental perceptualist, I predict, is going to have a reasonable way of recasting the phenomena, using ordinary perception as her model. 194 References Anscombe, Elizabeth. Intention. Cambridge, MA: Harvard, 1957/2000. Aristotle. Rhetoric. in The Complete Works of Aristotle: Volume 2. ed. Jonathan Barnes. Princeton, NJ: Princeton University Press, 1984. 2152 – 2269. Armstrong, David. Belief, truth, and knowledge. New York: Cambridge, 1973. Arnold, Magda. Emotion and Personality. New York: Columbia, 1960. Artega, Marc. “Rescuing Tracking Theories of Morality.” Philosophical Studies 172.12 (2015): 3357 – 3374. Audi, Robert. Moral Knowledge and Ethical Character. New York: Oxford, 1997. Audi, Robert. “Moral Perception and Moral Knowledge.” Proceedings of the Aristotelian Society Supplementary Volume 84 (2010): 79 – 97. Audi, Robert. Moral Perception. Princeton, NJ: Princeton, 2013. Balguy, John. A Collection of Tracts Moral and Theological (selections). in British Moralists 1650 – 1800: Hobbes – Gay. ed. D.D. Raphael. Indianapolis: Hackett, 1991. 389 – 408. Bealer, George. “Intuition and the Autonomy of Philosophy.” in Rethinking Intuition: The Psychology of Intuition and Its Role in Philosophical Inquiry. eds. Michael DePaul and William Ramsey. Lantham, MD: Rowman and Littlefield, 1998. 201 – 240. Bealer, George. “A Theory of the A Priori.” Philosophical Perspectives, 13, Epistemology (1999): 29 – 55. Bedke, Matthew. “Intuitional Epistemology in Ethics.” Philosophy Compass. 5.1 (2010): 195 1069 – 1083. Bengson, John. “The Intellectual Given.” Mind 124.495 (2015a): 707 – 760. Bengson, John. “Grasping the Third Realm.” in Oxford Studies in Metaethics. eds. John Hawthorne and Tamar Gendler. New York: Oxford, 2015b. Ben-Ze’ev, Aaron. The Subtlety of Emotions. Cambridge, MA: MIT Press, 2000. Blackburn, Simon. Spreading the Word: Groundings in the Philosophy of Language. New York: Oxford, 1984. Blackburn, Simon. “How to Be an Ethical Antirealist.” Midwest Studies in Philosophy 12 (1988): 361 – 375. Boghossian, Paul. “Analyticity Reconsidered.” Nous 30 (1996): 360 – 391 Bonjour, Lawrence. In Defense of Pure Reason. New York: Cambridge, 1998. Boyd, Richard. “How to Be a Moral Realist.” in Essays on Moral Realism. ed. Geoffrey Sayre-McCord. Ithaca, NY: Cornell University Press, 1988. 181 – 228. Brady, Michael. Emotional Insight: The Epistemic Role of Emotional Experience. New York: Oxford University Press, 2013. Brewer, Talbot. “Three Dogmas of Desire.” in Values and Virtues: Aristotelianism in Contemporary Ethics. ed. Timothy Chappell. Oxford: Oxford University Press, 2006. 253 – 285. Brewer, Talbot. The Retrieval of Ethics. New York: Oxford, 2009. Brogaard, Berit. “Do We Perceive Natural Kind Properties?” Philosophical Studies 162 (2013): 35 – 42. Burge, Tyler. The Origins of Objectivity. New York: Oxford, 2010. Burnet, Gilbert and Francis Hutcheson. “Letters between the Late Mr. Gilbert Burnet, and 196 Mr. Hutchinson, Concerning the True Foundation of Virtue or Moral Goodness.” Illustrations on the Moral Sense. ed. Bernard Peach. Cambridge, MA: The Belknap Press, 1971. Carrasco, Marisa. “Visual Attention: The Past 25 Years.” Vision Research 51: 1484 – 1525. Chappell, Timothy. “Moral Perception.” Philosophy 83 (2008): 421 – 437. Chudnoff, Elijah. Intuition. New York, Oxford: 2013. Chudnoff, Elijah. “Is Intuition Based on Understanding?” Philosophy and Phenomenological Research 89.1 (2014): 42 – 67. Chudnoff, Elijah. “Moral Perception: High Level Perception or Low Level Intuition?” in Phenomenology of Thinking. eds. Thiemo Breyer and Christopher Gutland. Routledge, forthcoming. Church, Jennifer. “Seeing Reasons.” Philosophy and Phenomenological Research 85.3 (2010): 638 – 670. Church, Jennifer. Possibilities of Perception. New York: Oxford, 2013. Clarke, Austin. A Theory of Sentience. New York: Oxford, 2000. Cohen, Stewart. “Bootstrapping, Defeasible Reasoning, and A Priori Justification.” Philosophical Perspectives 24 (2010): 141 – 159. Cohon, Rachel. Hume’s Morality: Feeling and Fabrication. New York: Oxford, 2008. Comer, Chris and Vicky Leung. “The Vigilance of the Hunted: Mechanosensoryvisual Integration in Insect Prey.” in Complex Worlds from Simpler Nervous Systems. ed. F.R. Prete. Cambrige, MA: MIT Press. 313 – 335. Cosmides, Lisa, John Tooby, and H.C. Barrett. “Resolving the debate on innate ideas: 197 Learnability Constraints and the Evolved Interpenetration of Motivational and Conceptual Functions.” in The Innate Mind: Structure and Contents. eds. Peter Carruthers, Stephen Laurence, and Stephen Stich. New York: Oxford, 2005. 305 – 337. Cowan, Robert. “Perceptual Intuitionism.” Philosophy and Phenomenological Research 90.1 (2015a): 164 - 193. Cowan, Robert. “Cognitive Penetrability and Ethical Perception.” The Review of Philosophy and Psychology 6.4 (2015b): 665 – 682. Cowan, Robert. Review of Robert Audi, Moral Perception. Mind 123.492 (2014): 1167 – 1171. Cudworth, Ralph. A Treatise Concerning Eternal and Immutable Morality (selections). in British Moralists 1650 – 1800: Hobbes – Gay. ed. D.D. Raphael. Indianapolis: Hackett, 1991. 105 - 119. Cullison, Andrew. “Moral Perception.” European Journal of Philosophy 18.2 (2009): 159 – 175. Cuneo, Terence. The Normative Web: An Argument for Moral Realism. New York: Oxford, 2007. Cuneo, Terence and Russ Schafer-Landau. “The Moral Fixed Points: New Directions for Moral Nonnaturalism.” Philosophical Studies 171.3 (2014): 399 – 443. D’Arms, Justin and Daniel Jacobson. “The Significance of Recalcitrant Emotion (or, anti- quasijudgmentalism).” in Philosophy and the Emotions. ed. Anthony Hatzimoysis. Cambridge: Cambridge, 2003. 127 – 146. Damasio, Antonio. Descartes’ Error. London: Picador, 1994. 198 Dancy, Jonathan. Ethics without Principles. New York: Oxford, 2004. Dancy, Jonathan. “Moral Perception.” Proceedings of the Aristotelian Society Supplementary Volume 84 (2010): 99 – 117. Doggett, Tyler and Andy Egan. “How We Feel About Terrible, Non-Existent Mafiosi.” Philosophy and Phenomenological Research 84.2 (2012): 277 – 306. Dӧring, Sabine. “Explaining Action by Emotion.” The Philosophical Quarterly 53.211 (2003): 214 – 230. Dӧring, Sabine. “Seeing What to Do: Affective Perception and Rational Motivation.” Dialectica 61.3 (2007): 363 – 394. Dretske, Fred. “Epistemic Operators.” Journal of Philosophy 67 (1970): 1007 – 1023. Dretske, Fred. Knowledge and the Flow of Information. Cambridge, MA: MIT Press, 1981. Dretske, Fred. Naturalizing the Mind. New York: Cambridge, 1995. Dwyer, Susanne. “How Not to Argue that Morality Isn’t Innate Comments on Prinz.” Moral Psychology, Vol. 1: The Evolution of Morality: Innateness and Adaptation. ed. Walter Sinnott-Armstrong. Cambridge, MA: MIT Press, 2008. 407 - 418. Dwyer, Susanne. “Moral Dumbfounding and the Linguistic Analogy: Implications for the Study of Moral Judgment.” Mind and Language 24 (2009): 274 – 296. Ekman, Paul. “An Argument for Basic Emotions.” Cognition and Emotion 6 (1992): 169 – 200. Ekman, Paul and Cordaro, Daniel. “What Is Meant by Calling Emotions Basic.” Emotion Review 3 (2011): 364–370. Ellsworth, Phoebe. “Levels of Thought and Levels of Emotion.” in The Nature of 199 Emotion. eds. Paul Ekman and Richard Davidson. New York: Oxford, 1994. 192 – 196. Enoch. David. Taking Morality Seriously: A Defense of Robust Realism. New York: Oxford, 2011. Finlay, Stephen. Confusion of Tongues: A Theory of Normative Language. New York: Oxford, 2014. Fiocco, M. Oreste. “Conceivability, Imagination, and Modal Knowledge.” Philosophy and Phenomenological Research 74.2 (2007): 364 – 380. Fodor, Jerry. A Theory of Content and Other Essays. Cambridge, MA: MIT Press, 1990. Foot, Philippa. “Moral Arguments.” Mind 58.268 (1958): 502 – 513. Fraser-Smith, A.C., A. Bernardi, P.R. McGill, M.E. Ladd, R.A. Helliwell, O.G. Villard, Jr. “Low-Frequency Magnetic Field Measurements Near the Epicenter of the Ms 7.1 Loma Prieta Earthquake.” Geophysical Research Letters 17.9 (1990): 1465 – 1468. Frijda, Nico. The Laws of Emotion. Mahwah: Lawrence Erlbaum Associates, 2007. Gettier, Edmund. “Is Justified True Belief Knowledge?” Analysis 23 (1963): 121 – 123. Gigerenzer, Gerd, Peter Todd, and the ABC Research Group. Simple Heuristics that Make Us Smart. New York: Oxford, 1999. Gill, Michael. The British Moralists on Human Nature and the Birth of Secular Ethics. New York: Cambridge, 2006. Gill, Michael. “Moral Rationalism vs. Moral Sentimentalism: Is Morality More Like Math or Beauty?” Philosophy Compass 2.1 (2007): 16 – 30. Gill, Michael. “Moral Phenomenology in Hutcheson and Hume.” Journal of the History 200 of Philosophy 47.4 (2009): 569 – 594. Godfrey-Smith, Peter. “Causal Pluralism.” in Oxford Handbook of Causation. eds. Helen Beebee, Christopher Hitchcock, and Peter Menzies. New York: Oxford, 2011. 326 – 337. Goldie, Peter. The Emotions: A Philosophical Exploration. New York: Oxford, 2000. Goldie, Peter. “Misleading Emotions.” in Epistemology and Emotions. eds. Georg Brun, Ulvi Doguoglu, and Dominique Kuenzle. Burlington: Ashgate. 149 – 165. Goldstein, Bruce. Sensation and Perception, 9 th edition. Belmont, CA: Wadsworth Cengage Learning, 2013. Greco, John. Putting Skeptics in Their Place: The Nature of Skeptical Arguments and Their Role in Philosophical Inquiry. New York: Cambridge, 2000. Greene, Joshua. Moral Tribes: Emotion, Reason, and the Gap between Us and Them.” New York: Penguin, 2013. Gregory, Alex. “The Guise of Reasons.” American Philosophical Quarterly 51.1 (2013): 63 – 72. Gregory, Dominic. Imagery, the Imagination, and Experience. The Philosophical Quarterly 60.241 (2010): 735 – 753. Gregory, Richard. Eye and Brain, 5 th edition. Princeton: Princeton University Press, 1997. Haidt, Jonathan and Craig Joseph. “Intuitive Ethics: How Innately Prepared Intuitions Generate Culturally Variable Virtues.” Daedalus: On Human Nature 133 (2004): 55 – 66. Haidt, Jonathan and Craig Joseph. “The Moral Mind: How 5 Sets of Innate Moral 201 Intuitions Guide the Development of many Culture-specific Virtues, and perhaps even Modules.” in The Innate Mind, Vol 3. eds. Peter Carruthers, Stephen Laurence, and Stephen Stich. New York: Oxford, 2007. 367 – 391. Haidt, Jonathan. The Righteous Mind: Why Good People Are Divided by Politics and Religion. New York: Vintage, 2012. Harman, Gilbert. “Using a Linguistic Analogy to Study Morality.” Moral Psychology, Vol. 1: The Evolution of Morality: Innateness and Adaptation. ed. Walter Sinnott- Armstrong. Cambridge, MA: MIT Press, 2008. 345 – 351. Hauser, Mark, Liane Young, and Fiery Cushman. “Reviving Rawls’s Linguistic Analogy: Operative Principles and the Causal Structure of Moral Action.” in Moral Psychology, Volume 2: The Cognitive Science of Morality: Intuition and Diversity. ed. Walter Sinnott-Armstrong. Cambridge, MA: MIT Press, 2007. 107 – 144. Hitchcock, Christopher. “Of Humean Bondage.” British Journal of the Philosophy of Science 54.1 (2003): 1 – 25. Howard, Ian and William Templeton. Human Spatial Orientation. New York: John Wiley and Sons, 1996. Huemer, Michael. Ethical Intuitionism. New York: Palgrave MacMillian, 2005. Hume, David. Enquiry Concerning Human Understanding. Indianapolis: Hackett, 1993. Hutcheson, Francis. An Inquiry Concerning the Original of Our Ideas of Virtue or Moral Good (selections). in British Moralists 1650 – 1800: Hobbes – Gay. ed. D.D. Raphael. Indianapolis: Hackett, 1738/1991. 261 – 299. Irwin, Terence. The Development of Ethics: A Historical and Critical Study: Volume II: 202 From Suarez to Rousseau. New York: Oxford, 2008. Izard, C. E. “Forms and Functions of Emotions: Matters of Emotion–Cognition Interactions.” Emotion Review 3 (2011): 371–378. Jackson, Frank. From Metaphysis to Ethics: A Defense of Conceptual Analysis. New York: Oxford, 1998. James, Scott. “The Caveman’s Conscience: Evolution and Moral Realism.” Australasian Journal of Philosophy 87.2 (2009): 215 – 233. Jérôme Dokic and Stéphane Lemaire. “Are Emotions Perceptions of Value?” Canadian Journal of Philosophy 43.2 (2013): 227 – 247. Johnston, Mark. “The Authority of Affect,” Philosophy and Phenomenological Research 63.1 (2001): 181 – 214. Joyce, Richard. The Evolution of Morality. Cambridge, MA: MIT Press, 2006. Kahneman, Daniel and Shane Frederick. “Representativeness Revisited: Attribute Substitution in Intuitive Judgment.” in Heuristics of Intuitive Judgment: Extensions and Applications. eds. Thomas Gilvovich, Dale Griffin, and Daniel Kahneman. New York: Cambridge, 2002. 49 – 81. Kant, Immanuel. Critique of Pure Reason. trans. Werner Pluhar. Indianapolis: Hackett, 1996. Kauppinen, Antti. “A Humean Theory of Moral Intuition.” Canadian Journal of Philosophy 43.3 (2013): 360 – 381. Kripke, Saul. Naming and Necessity. Cambridge, MA: Harvard University Press. 1980. Lazarus, Richard. “Appraisal: The Minimal Cognitive Prerequisites of Emotion.” in What 203 Is an Emotion? Classic and Contemporary Readings (2 nd edition). ed. Robert Solomon. New York: Oxford, 2003. LeDoux, Joseph. The Emotional Brain: The Mysterious Underpinnings of Emotional Life. London: Wiedenfeld and Nicolson, London. Levenson, Robert. “Basic Emotion Questions.” Emotion Review 3 (2011): 379 – 86. Lewis, David. “Attitudes De Dicto and De Se.” The Philosophical Review 88.4 (1979): 513 – 543. Lewis, David. On the Plurality of Worlds. Oxford: Basil Blackwell, 1986. Macpherson, Fiona. “Phenomenal ‘Presence as Absence’ in Visual Experience.” in Phenomenal Presence. eds. F. Dorsch, F. Macpherson, ad M. Nida-Rumelin. Oxford: Oxford, forthcoming. Mackie, J.L. Ethics: Inventing Right and Wrong. London: Penguin Books, 1977. Marcus, Gary. The Birth of The Mind: How a Tiny Number of Genes Creates the Complexities of Human Thought. New York: Basic Books, 2004. McBrayer, Justin. “A Limited Defense of Moral Perception.” Philosophical Studies 149 (2010a): 305 – 320. McBrayer, Justin. “Moral Perception and the Causal Objection.” Ratio 23.3 (2010b): 291 – 307. McDowell, John. Mind, Value, and Reality. Cambridge, MA: Harvard, 1998. McKeever, Sean and Michael Ridge. Principled Ethics: Generalism as a Regulative Ideal. New York: Oxford, 2006. McGrath, Sarah. “Moral Knowledge by Perception.” Philosophical Perspectives 18 (2004): 209 – 228. 204 McGrath, Sarah. “Moral Knowledge and Experience.” in Oxford Studies in Metatehics: Volume 6. ed. Russ Schafer-Landau. New York: Oxford, 2010. 107 – 127. Mikhail, John. “Universal Moral Grammar: Theory, Evidence and the Future,” Trends in Cognitive Science 11 (2007): 143–52. Millikan, Ruth. Language, Thought and Other Biological Categories. Cambridge, MA: MIT Press, 1984. Millikan, Ruth. “Biosemantics.” Journal of Philosophy 86 (1989): 281-287. Milona, Michael. “On the Epistemological Significance of Value Perception.” in Evaluative Perception: Aesthetic, Ethical, and Normative. eds. Anna Bergqvist and Robert Cowan. Oxford: Oxford, forthcoming a. Milona, Michael. “Taking the Perceptual Analogy Seriously.” Ethical Theory and Moral Practice, forthcoming b. Milona, Michael. “Intellect vs. Affect: Finding Leverage in an Old Debate.” unpublished a. Milona, Michael. “Armchair Moral Knowledge and the Perceptual Analogy.” unpublished b. Milona, Michael and Mark Schroeder. “Desiring Under the Proper Guise.” unpublished. Moss, Jessica. Aristotle on the Apparent Good: Perception, Phantasia, Thought, and Desire. New York: Oxford, 2013. Neander, Karen, “Teleological Theories of Mental Content.” The Stanford Encyclopedia of Philosophy (Spring 2012 Edition), Edward N. Zalta (ed.), URL = <http://plato.stanford.edu/archives/spr2012/entries/content-teleological/>. Neander, Karen. “Toward an Informational Teleosemantics.” in Millikan and Her Critics. 205 eds. Justin Kingsbury, Dan Ryder and Kenneth Williford. Malden, MA: Blackwell, 2013. 21 – 40. Nesse, Harold. “Evolutionary Explanations of Emotions.” Human Nature 1 (1990): 261 – 289. Nichols, Shaun. Sentimental Rules: On the Natural Foundations of Moral Judgment. New York: Oxford, 2004. Noë, Alva. “Conscious Reference.” The Philosophical Quarterly 59.236 (2009): 470 – 82. Nussbaum, Martha. Upheavals of Thought: The Intelligence of the Emotions. Cambridge: Cambridge, 2001. Nye, Howard. “Directly Plausible Principles” in The Palgrave Handbook of Philosophical Methods. ed. Christopher Daley. New York: Palgrave MacMillan, 2015. 610 – 636. Oddie, Graham. Value, Reality, and Desire. New York: Oxford, 2005. Oddie, Graham. “Value Perception, Properties and the Primary Bearers of Value.” in Evaluative Perception: Aesthetic, Ethical, and Normative. eds. Anna Bergqvist and Robert Cowan. Oxford: Oxford, forthcoming. Orlando, Nico. The Innocent Eye: why Vision is not a Cognitive Process. New York: Oxford, 2014. Panksepp, Jaak and David Watt. “What Is Basic about Basic Emotions? Lasting Lessons from Affective Neuroscience.” Emotion Review 3 (2011): 387–396. Papineau, David. “Representation and Explanation.” in Philosophy of Science 51 (1984): 550–72. 206 Peacocke, Christopher. The Realm of Reason. New York: Oxford, 2004. Peacocke, Christopher. “Perception, Biology, Action, and Knowledge.” Philosophy and Phenomenological Research 88.2 (2014): 477 – 484. Pollock, Jackson. Thinking about Acting: Logical Foundations for Rational Decision Making. New York: Oxford, 2006. Portmore, Douglas. “The Teleological Conception of Practical Reasons.” Mind 120.477 (2011): 118 – 153. Price, Richard. A Review of the Principle Questions in Morals (selections). in British Moralists 1650 – 1800: Hume – Bentham. ed. D.D. Raphael. Indianapolis: Hackett, 1991. 132 – 198. Prichard, H.A. “Does Moral Philosophy Rest on a Mistake?” Mind 21 (1912): 21 – 37. Prinz, Jesse. Gut Reactions: A Perceptual Theory of Emotions. New York: Oxford, 2004. Prinz, Jesse. The Emotional Construction of Morals. New York: Oxford, 2007. Pryor, Jim. “What’s Wrong with Moore’s Argument?” Philosophical Issues 14 (2004): 349-78. Quinn, Warren. “Putting Rationality in Its Place.” in Morality and Action. New York: Cambridge, 1993. 228 – 255. Roberts, Robert. Emotions: An Essay in Aid of Moral Psychology. New York: Cambridge, 2003. Roberts, Robert. Emotions in the Moral Life. New York: Cambridge, 2013. Rock, Irvin. Perception. New York: Scientific American Library, 1995. Ross, W.D. The Right and the Good. New York, Oxford: 2002. Salmela, Mikko. “Can Emotion be Modeled on Perception.” Dialectica 65.1 (2011): 1 – 207 29. Sayre-McCord, Geoffrey. “Mill’s Proof of the Principle of Utility: A More than Half- Hearted Defense.” Social Philosophy and Policy 18.2 (2001): 330 – 360. Scanlon, T.M. What We Owe to Each Other. Cambridge, MA: Harvard. Schafer, Karl. “Perception and the Rational Force of Desire.” Journal of Philosophy 110.5 (2013): 258 – 281. Scheler, Max. Formalism in Ethics and Non-Formal Ethics of Values. trans. Manfred Frings and Roger Frunk. Evanston: Northwestern University Press, 1973. Scherer, Klaus. “Appraisal Considered as a Process of Multilevel Sequential Checking.” in Appraisal Processes in Emotion: Theory, Methods, and Research. eds. Klaus Scherer, Angela Schorr, and Tom Johnstone. New York: Oxford, 2001. 92 – 120. Schroeder, Mark. “Cudworth and Normative Explanations.” Journal of Ethics and Social Philosophy 1.3 (2005): 1 – 27. Schroeder, Mark. Slaves of the Passions. New York: Oxford, 2007a. Schroeder, Mark. “Teleology, Agent-Relative Value, and ‘Good’. Ethics 117 (2007b): 265 – 295. Schroeder, Mark. “How Does the Good Appear to Us?” Social Theory and Practice 34.1 (2008): 119 – 130. Schroeder, Mark. “Value Theory.” Stanford Encyclopedia of Philosophy (Summer 2012 Edition) ed. Edward Zalta. URL = <http://plato.stanford.edu/archives/sum2012/entries/value-theory/>. Schroeder, Mark. “The Price of Supervenience.” in Explaining the Reasons We Share: 208 Explanation and Expression in Ethics, Volume 1. New York: Oxford, 2014. 124 – 144. Schroeder, Timothy. “Desire and Pleasure in John Pollock’s Thinking about Acting.” Philosophical Studies 148 (2010): 447 – 454. Schroeder, Timothy. “Desire.” Stanford Encyclopedia of Philosophy (Summer 2015 Edition) ed. Edward Zalta. URL = <http://plato.stanford.edu/archives/sum2015/entries/desire/>. Seneca. 85 th Epistle. in Seneca’s Complete Epistles: Volume 2. eds. Neill Carson and Christian Selbrede. trans. Richard Gummere. Carson and Selbrede Inc.: Amazon Digital. Shaftsbury, the Third Earl of. Anthony Ashley Cooper. An Inquiry Concerning Virtue, or Merit (selections). in British Moralists 1650 – 1800: Hobbes – Gay. ed. D.D. Raphael. Indianapolis: Hackett, 1991. 169 – 188. Siegel, Susanna. “Which Properties Are Represented in Experience?” in Perceptual Experience. eds. Tamar Gendler Szabo and Jonathan Hawthorne. New York: Oxford, 2006: 481 – 503. Siegel, Susanna. The Contents of Visual Experience. New York, Oxford: 2011. Siegel, Susanna. “The Contents of Perception.” Stanford Encyclopedia of Philosophy (Spring 2013 Edition) ed. Edward Zalta. URL = <http://plato.stanford.edu/archives/spr2013/entries/perception-contents/> Siegel, Susanna. “Cognitive Penetrability and Perceptual Justification.” Nous 46.2 (2012): 201 – 22. Silins, Nicholas. “The Significance of High-Level Content.” Philosophical Studies 161.1 209 (2013): 13 – 33. Silins, Nicholas. “Perceptual Experience and Perceptual Justification,” The Stanford Encyclopedia of Philosophy (Summer 2015 Edition), Edward N. Zalta (ed.), URL= <http://plato.stanford.edu/archives/sum2015/entries/perception-justification/> Singer, Peter. “Famine, Affluence, and Morality.” Philosophy and Public Affairs 1 (1972): 229 – 43. Sinhababu, Neil. “The Advantages of Propositionalism.” Pacific Philosophical Quarterly 96.2 (2015): 165-180. Slovic, Paul, Melissa Finucane, Ellen Peters, and Donald MacGregor. “The Affect Heuristic.” in Heuristics and Biases: The Psychology of Intuitive Judgment. eds. Thomas Gilvovich, Dale Griffin, and Daniel Kahneman. New York: Cambridge, 2002. 397 – 420. Smith, Michael. The Moral Problem. Malden, MA: Blackwell, 1994. Snowden, Peter, Peter Thomson, and Tom Troscianko. Basic Vision: An Introduction to Visual Perception, 2 nd Revised Edition. Oxford: Oxford University Press, 2012. Sober, Elliot and Robert Sloan Wilson. Unto Others: The Evolution and Psychology of Unselfish Behavior. Cambridge, MA: Harvard University Press, 1998. Sosa, Ernest. “Intuitions: Their Nature and Epistemic Efficacy.” Grazer Philosophische Studien: Internationale Zeitschrift für Analytische Philosophie 74 (2007): 51 – 67. Sorabji, Richard. Emotion and Peace of Mind. New York: Oxford, 2000. Sripada, Chandra. “Nativism and Moral Psychology: Three Models of the Innate 210 Structure that Shapes the Contents of Moral Norms.” Moral Psychology, Vol. 1: The Evolution of Morality: Innateness and Adaptation. ed. Walter Sinnott- Armstrong. Cambridge, MA: MIT Press, 2008. 319 – 343. Stratton-Lake, Philip. “The Basis of Objective Judgments in Ethics.” Ethics 125.2 (2015): 521 – 524. Stampe, Dennis. “The Authority of Desire.” The Philosophical Review 96.3 (1987): 335 – 381. Street, Sharon. “A Darwinian Dilemma for Realist Theories of Value.” Philosophical Studies 127 (2006): 109 – 166. Sunstein, Cass. “Moral Heuristics.” Behavioral and Brain Sciences 28.531 (2005): 531 – 573. Swinburne, Richard. “Necessary Moral Principles.” Journal of the American Philosophical Association 1.4 (2015): 617 – 634. Szigeti, Andres. “No Need to Get Emotional? Emotions and Heuristics.” Ethical Theory and Moral Practice 16 (2013): 845 – 862. Tenenbaum, Sergio. Appearances of the Good: An Essay on the Nature of Practical Reason. New York: Cambridge, 2007. Tenenbaum, Sergio. “Appearing Good: A Reply to Schroeder.” Social Theory and Practice 34.1 (2008): 131 – 138. Thagard, Paul. “Desires are Not Propositional Attitudes.” Dialogue 45.1 (2006): 151 – 156. Tracy, Jessica and Daniel Randles. “Four Models of Basic Emotions: A Review of Ekman 211 and Cordaro, Izard, Levenson, and Panksepp and Watt.” Emotion Review 3.4 (2011): 397 – 405. Tracy, Jessica. “An Evolutionary Approach to Understanding Distinct Emotions.” Emotion Review 6.4 (2014): 308 – 312. Tropman, Elizabeth. “Varieties of Moral Intuitionism.” Journal of Value Inquiry 48 (2014): 177 – 194. Tye, Michael. Consciousness, Color, and Content. Cambridge: MIT Press, 2000. Väyrynen, Pekka. “Some Good and Bad News for Ethical Intuitionism.” The Philosophical Quarterly 58.232 (2008): 489 – 511. Väyrynen, Pekka. “Doubts about Moral Perception.” in Evaluative Perception: Aesthetic, Ethical, and Normative. eds. Anna Bergqvist and Robert Cowan. Oxford: Oxford, forthcoming Von Neumann, John and Oskar Morgenstern. The Theory of Games and Economic Behavior. Princeton: Princeton University Press, 1944. Walton, Kendall. “Fearing Fictions.” The Journal of Philosophy 75 (1978): 5 – 27. Wedgwood, Ralph. “Sensing Values.” Philosophy and Phenomenological Research 63.1 (2001): 215 – 223. Wedgwood, Ralph. The Nature of Normativity. New York, Oxford: 2007. Werner, Preston. “Moral Perception and the Contents of Experience.” Journal of Moral Philosophy (2014): 1 – 24. Whiting, Demian. “Are Emotions Perceptual Experiences of Value?” Ratio 25.1 (2012): 93 – 107. Wiggins, David. Needs, Values, and Truth: Essays in the Philosophy of Value. New York: 212 Oxford, 1998. Williamson, Timothy. The Philosophy of Philosophy. New York, Oxford, 2007. Williamson, Timothy. “Knowing by Imagining.” in Knowing through Imagination.” eds. Amy Kind and Peter Kung. New York: Oxford, forthcoming. Wolfe, Jeremy, Keith Kluender, and Dennis Levi. Sensation and Perception, 4 th edition. Sunderland, MA: Sinauer, 2014. Wollheim, Richard. Painting as an Art. Princeton, NJ: Princeton, 1987. Yablo, Stephen. “Mental Causation.” Philosophical Review 101 (1992): 245 – 80. Yablo, Stephen. “Is Conceivability a Guide to Possibility?” Philosophy and Phenomenological Research 53 (1993): 1 – 42.

Abstract (if available)

Abstract We know many things about what is right or wrong, good or bad. But how? According to the view that I call ‘perceptualism’, such knowledge (evaluative knowledge) is not so different from ordinary empirical knowledge: just as we have perceptual experiences which justify our empirical beliefs, so too do we have evaluative experiences which justify our evaluative beliefs. Perceptualism has proven a seductive theory throughout the history of moral philosophy in large part because it seeks to demystify an initially very puzzling species of knowledge, evaluative knowledge, by showing that it is really not so different from a comparatively much less puzzling species of knowledge, ordinary empirical knowledge. But perceptualist theories can take a number of different forms depending on how one characterizes the relevant evaluative experiences which are supposed to do the justificatory work. The tasks for this dissertation are, first, to determine the best form of a perceptualist theory and, second, to address the most important challenges which arise for that view. On the initial question, I argue that the most promising sort of perceptualist theory takes evaluative experiences to be affective in nature (sentimental perceptualism). The key advantage to the sentimentalist approach over alternatives (e.g., those which take evaluative experiences to be intellectual) is that the sentimentalist theory is the only one capable of explaining the full range of our evaluative knowledge. Other models invariably leave some of our knowledge unexplained. But a sentimental perceptualist theory faces two major challenges. The first is to explain how affective experiences ever came to have evaluative content, and the second is to explain how affective experiences ground moral knowledge by mere reflection. The latter half of the dissertation explains how to develop the sentimental perceptualist theory in the face of these important challenges.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Positivist realism

PDF

Feeling good and living well: on the nature of pleasure and its role in well-being

PDF

A deontological explanation of accessibilism

PDF

Rationality and the primacy of the occurrent

PDF

Rethinking reductive realism in ethics

PDF

Cognitive boundaries for rational coherence requirements

PDF

Process-oriented rationality

PDF

Beliefs that wrong

PDF

Belief as credal plan

PDF

The case for moral skepticism

PDF

Capturing moods: a philosophical exploration

PDF

Morality with Humean foundations: a proposal for a solution to the accommodation problem

PDF

The virtue of reasonableness: on the normative epistemic implications of public reason liberalism

PDF

Minimal sensory modalities and spatial perception

PDF

Molyneux's question answered!

PDF

Mistaken defense and normative conventions

PDF

Reasons, obligations, and the structure of good reasoning

PDF

Contrastive reasons

PDF

Contract, from promise to commodity

PDF

Units of agency in ethics

Asset Metadata

Creator Milona, Michael (author)

Core Title A perceptual model of evaluative knowledge

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Philosophy

Publication Date 06/26/2016

Defense Date 03/28/2016

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag moral epistemology,OAI-PMH Harvest,perception,rationalism,sentimentalism

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Schroeder, Mark (committee chair), Finlay, Stephen (committee member), John, Richard (committee member), Levin, Janet (committee member), Wedgwood, Ralph (committee member)

Creator Email milonamc@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-258321

Unique identifier UC11281301

Identifier etd-MilonaMich-4475.pdf (filename),usctheses-c40-258321 (legacy record id)

Legacy Identifier etd-MilonaMich-4475.pdf

Dmrecord 258321

Document Type Dissertation

Format application/pdf (imt)

Rights Milona, Michael

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA