Looks and Perceptual Justification*

Matthew McGrath

University of Missouri

Introduction

Imagine I hold up a Granny Smith apple for all to see. You would see that it was an apple, indeed that it was a Granny Smith apple. You’d also see that it was roughly round, green, and many other things. These seeings-that would be instances of perceptual knowledge. But you wouldn’t only have knowledge. You would be quite reasonable – “justified” – to believe these propositions to be true. 1 A familiar view in epistemology, which we may call liberal experiential foundationalism about perceptual justification, holds that in such cases, your experience provides you immediate justification to believe propositions like these about things in the world around you.2 Of course, in some cases of seeing-that-P, your experience doesn’t provide immediate justification to believe that P. If you know I’m planning to take the 9 a.m. train, and after seeing the 9 a.m. train depart a few moments ago, you see me arriving at the station rushing for my train, you can be said to see– and be justified in believing – that I missed my train. The justification you have in this case is one that you have only in virtue of having justifications to believe these things about my plans, the train’s departure, etc., and so is mediate rather than immediate. Nonetheless, on liberal experiential foundationalism, or “liberal foundationalism” for short, there is a large class of beliefs about our local environment that our experience immediate justifies us in having – including beliefs about what are traditionally classified as sensible qualities (colors, shapes, sizes, etc.) as well beliefs about their kinds (apple, Granny Smith, oboe, etc). 3

** For helpful discussion, I thank an audience at the 2013 Central APA meetings as well as Frances Dickey, Jeremy Fantl, David Henderson and Jack Spencer. For extensive written comments which changed the paper substantially, I thank Nico Silins and Susanna Siegel.

1 By ‘justified’ I mean only reasonable. I use ‘justification’ because of its convenient grammatical features only, in particular because of the existence of the count noun ‘justification’ and the verb ‘justifies’.

2 Following James Pryor’s (2000, 532) definition, let us say that your justification to believe P is mediate just if you have this justification in virtue of having justifications to believe propositions other than P; otherwise your justification is immediate. I will assume that, in cases in which a person has one justification j only in virtue of having another justification j*, the justification j is at least partly constituted by the justification j*, i.e., j* helps to make up, is a component part of, j. Thus, if you have a certain justification j to believe the neighbors are away only in virtue of having a justification j* to believe their grass hasn’t been moved as per usual then the justification j is partly constituted by the justification j*.

1

Here I focus on perceptual beliefs that are prima facie good candidates for immediate justification under liberal foundationalism. The case of the apple is our paradigm. More generally, the class of beliefs we will focus on – call these simple perceptual beliefs -- consists of beliefs in which the subject is merely using her perceptual F-categorization skills (such as they are) to categorize a perceived object as F, where F is a sensible quality or kind. I also stipulate that in simple perceptual beliefs the subject is not concluding the object is F based on inference from a belief that it has certain (other) sensible qualities/relations or kind properties. For instance, a belief that this is a blue-footed booby based on inference from this has blue feet and it is a bird will not count as a simple perceptual belief. In the case of the apple, you don’t base your belief that it is an apple on it’s round and it’s green and …. You can – as you say – “just tell” it is an apple. The parenthetical “such as they are” qualifying “categorization skills” is added so as not to exclude cases in which the skills involved aren’t good ones for successful categorization. If I’ve learned, incorrectly, to categorize clarinets by sight as oboes, this is still a categorization skill, in the sense in which I’m using the term. 4

Liberal foundationalists do not take all simple perceptual beliefs to be justified, immediately or otherwise. Some such beliefs are held despite the subject’s possessing strong counterevidence. Others involve the faulty use of categorization procedures, in a way that prevents the resulting beliefs from being justified (e.g., the gold-digger categorizes a pebble as a gold nugget in part because he wants to (Markie 2006)). What the liberal foundationalist thinks is that perceptual experience can and often does provide subjects immediate justification to believe propositions categorizing perceived objects as F, where F is very wide-ranging indeed, including any kinds or sensible qualities that are capable of being recognized perceptually. Subjects who appropriately base their belief on such immediate propositional justification will have immediately justified simple perceptual beliefs.

In this paper, I argue for a claim about simple visual beliefs, namely that they are justified mediately, in virtue of our having justification to believe propositions about how things look. I call this the looks view of justified simple visual beliefs. To see our simple visual beliefs as only immediately is to miss out on the public reasons we have for them and which are essential to

3 I use ‘sensible quality’ stipulatively to pick out the properties traditionally thought to be directly perceived, the “proper sensibles.” Examples are color, shape, size, pitch, loudness, timbre, roughness, smoothness, sourness, sweetness. I use ‘kind’ very broadly and loosely to include qualitative properties beyond sensible qualities, including artifactual kinds at all levels of genus and species (e.g., musical instrument, oboe, Fox Renard oboe), and natural kinds at all levels (e.g., animal, dog, Scottish Terrier), as well as just about any sortals we ordinarily apply to categorize objects, again at all levels (e.g., rock formation, mountain, Mt. Moran), etc.

4 My interest, in this paper, is with the “property” side of perceptual beliefs rather than the “object” side, and even among properties my concern is with qualitative ones, not properties such as being Obama. I also leave aside, as problems for another occasion, cases for which there is no perceived object, as in cases of hallucination.

2

their justification when they are justified. 5 I present and defend the looks view in second part of the paper (sections 3-7). The first part of the paper (sections 1 and 2) focuses on liberal foundationalism. It argues that if we accept experiential foundationalism about some simple perceptual beliefs – such as ones about the sensible properties of things – it is difficult to hold off on accepting full-blown liberal foundationalism. If you let some simple perceptual beliefs count as immediately justified, in other words, you’re opening the floodgates. This is not meant to be a problem for the view. To the contrary, I take it to show its appeal.

Finally, although I focus my critical attention on liberal foundationalism, similar criticisms apply to other epistemological views about perception. Consider other experientialist views, i.e., views which take experience to play an essential role in perceptual justification, views such as the conservatism of Crispin Wright (2004) as well as classical foundationalism. Proponents of these views, like the liberal foundationalist, typically think that a large class of simple perceptual beliefs are justified only by broadly experiential factors, factors such as one’s experience, one’s justified beliefs about one’s experience, and one’s justification to take one’s experience to be reliable. Omitted, again, are the public reasons provided by considerations about looks. The same omission is found in standard nonexperientialist views, such as Jack Lyons’ (2008) reliabilism, according to which it’s enough for perceptual justification that one’s belief be produced by a reliable subpersonal perceptual system. Such a view, also, fails to give looks their due.

1. The Path to Liberal Foundationalism

Suppose you start off granting that experience can justify some propositions about the world around you, at least propositions such as that’s green concerning the apple. You thus think some simple perceptual beliefs are immediately justified by experience. This can seem very plausible indeed. You look at the apple. It might seem that you can be justified – from your experience alone – in believing that it is green. (In the second half of the paper, I challenge this view. But I grant its initial plausibility.) Granting this, you still might want to resist liberal foundationalism and its claim that simple perceptual beliefs about a thing being a Granny Smith, or even an Ipod Touch 5th Generation can be immediately justified. You might maintain, on principled grounds, that there are severe limits to what experience can immediately justify. I’ll begin this section by exploring why one might think there are such limits.

A perceptual experience of a table as being red provides prima facie justification to believe it is red, but an imaginative experience does not (suppose you’ve just seen a brown table and now, eyes closed, you imagine it as being red). Why the difference? Both cases involve

5 I hope to extend the account beyond looks to appearances more generally, i.e., to sounds, feels, smells, etc. However, here I limit myself to looks and so to claims about simple visual beliefs.

3

conscious experience of a table as red. The difference, plausibly, is that the perceptual experience can justify this belief because it presents as true the proposition that the object is red, whereas the imaginative experience fails to do so. If this is how perceptual experience provides immediate justification to believe propositions about the world around us, we can ask whether there are limits on just which propositions can be presented as true in experience. If there are in-principle reasons for thinking the only propositions about sensible qualities can be so presented, then there are in-principle reasons for thinking that simple visual beliefs about things being apples, Ipods, etc. cannot be immediately justified by perceptual experience. It isn’t hard to see what the in-principle reason would be: experiences can only present as true propositions which are among their representational contents6; and the representational content of experience extends only to objects and sensible qualities, not to kinds. Thus, an epistemological debate between illiberal (hereafter moderate) and liberal foundationalists appears to turn on with a debate within the philosophy of mind about how rich the content of perceptual experience is.

The liberal foundationalist has a good reply to the moderate: the envisaged account of perceptual justification is false. The envisaged account -- call it the content-based account -- holds that one’s experience provides prima facie justification to believe P if and only if P is among its contents. The ‘only if’ conditional, conjoined with the claim that the content of experience excludes kinds, rules out liberal foundationalism. But, independently of this, the conditional in the ‘if’ direction is too strong, and when we see what is needed to fix it, we find ourselves moving toward a view friendly to liberal foundationalism. I’ll take these things step by step, beginning with the ‘if’ conditional.

Why is the conditional in the ‘if’ direction too strong? The problem of the speckled hen shows us why.7 Here, in brief, is the problem. The moderate foundationalist’s content-based account implies:

If one has a perceptual experience representing x as F, then one is immediately perceptually justified prima facie in believing of x that it is F.

6 If one accepted some other view of the metaphysics of experience, this in-principle point could be recast accordingly. For any plausible view of experience, I take it, there will be a difference between properties “found in” or “present in” experience and those that are not. Experiences will count as presenting as true only propositions that ascribe properties “found in” experience. (The sense datum theorist will think redness is found in experience, insofar as sense data are sometimes red; by contrast, being an apple isn’t found in experience, since sense data cannot be apples. The adverbialist will think redness is found in experience insofar as one can sense red-ly.)

7 Ernest Sosa (2003) raised the problem as a problem for classical foundationalist’s claims about which introspective beliefs about experience are immediately justified. However, as Michael Pace (2008) has noted, the problem generalizes to other forms of experientialism about perceptual justification.

4

‘F’ here is a schematic predicate-letter, not a singular term for properties. An instance of this schema is:

If one has an experience of a hen representing it as 48-speckled, then one is prima facie justified in believing it is 48-speckled.

When one looks at the hen with exactly 48 clearly visible speckles, one’s experience represents it as 48-speckled even though one isn’t justified in believing it is 48-speckled. Nor does one have prima facie justification to believe this which is somehow defeated. Thus, one isn’t prima facie justified in believing it is 48-speckled but one’s experience represents it as 48-speckled, contra the content-based account.

Broadly, there are two ways for the moderate who cleaves to the content-based account to respond to apparent counterexamples like this. One is to deny that the experience attributes the relevant feature F-ness (48-speckledness in this example), and the other is to claim that the subject has prima facie justification to believe the object is F but that this justification is defeated. Let’s consider these in turn.

The moderate grants that the contents of experience include propositions attributing sensible qualities to objects. Qualities such as numerosity of speckles are sensible qualities. So, why is 48-speckledness not part of the content of one’s experience when one sees the hen? Here is one answer the moderate might give. Visual acuity drops off outside of the fovea. Outside of this area, one can fail to notice changes in perceived objects when they are properly masked. One might think that if we fail to notice them, this is reason to think that there are no changes in what the experience is like, and if there are no changes in what the experience is like, then the content of the experience must stay the same the same. But if the content stays the same across changes in speckles’ numerosity, presumably the content is more determinable than 48-speckledness, e.g., many-speckledness. The claim would then be that one is prima facie justified in believing the hen possesses the more determinable feature, e.g., that it is many-speckled.

We could argue about the hen case, but there are other examples which might better serve the same purpose. Sticking with numerosity, consider a figure like this:

I can foveate the dots if I hold this piece of paper far enough from my eyes. I can’t make out how many there are, although I can see each and every one of them without changing my point of fixation. I might even be able to tell, if you present other dot circles, whether they involve the same or different numbers of dots. Suppose I can. Then the case for representation of 13-dottedness seems strong. Still, I am not justified in believing there are 13 dots until I count.

5

Wayne Davis (2005) and Michael Pace (2008) discuss even more compelling counterexamples. Suppose someone lacking absolute pitch hears a middle C played on the piano. The experience represents it as having a certain pitch, indeed as a C. It certainly doesn’t sound the way an A sounds. Play the two serially and they clearly sound different to the person. So, here is a case in which one has an experience as of a sound being a C and yet one isn’t prima facie justified in believing it is a C. Or take the case of unusual colors. Here one might be able, unlike the pitch case, to recognize the color as the same again more easily. Still, one might see a color that is in fact puce, and one’s experience represents it as puce, and yet one is not prima facie justified in believing it is puce but at best that it is a shade of red (compare Putnam’s elm/beech case). So, I take it that it is hopeless to argue that in all the relevant proposed counterexamples, the relevant property F-ness isn’t attributed by the experience.

The other way to resist the counterexamples is to claim that they are cases of defeated prima facie justification. But the defeat doesn’t take place through any standard sort of defeater – such as “the lighting conditions are abnormal” and the like. One might suggest that the defeater is one’s knowledge that one can’t recognize the relevant feature to be present. Let me mention two worries about this suggestion. First, it doesn’t feel like there are competing prima facie justifications here, one of which defeats the other. As Mark Schroeder (2007) points out, sometimes there are competing prima facie reasons even when it doesn’t feel there is competition because one is extremely weak and the other very strong. But here the prima facie justification provided by experience must be very strong, because it is the same sort found in cases of visual knowledge, say, that a thing is green.8 Second, the moderate who accepts the content-based view herself must beware of appealing to second-order defeaters. She thinks experiences can immediately justify beliefs in their contents, and so presumably can do so independently of whether the subject is justified in thinking she can recognize a feature. Thus, there ought to be cases in which one has the same experience without having the extra I can’t recognize it defeater and in these cases it ought to justify. But intuitively it won’t.

I conclude that the content-based account is refuted. However, the leading idea motivating it, that experience immediately justifies when it presents a certain proposition as true – i.e., when it offers up a proposition as “face value” – might be salvaged. If you lack absolute pitch, then when you hear the C played, although you have an experience as of a C pitch, your experience does not present as true the proposition this tone is a C as true. Thus, not every proposition that is a content of an experience is a face value of the experience.

The moderate might now concede: fine, not every content of experience is among its face values; still, only contents are among face values. However, the liberal foundationalist has made dialectical progress. If some but not all contents are among an experience’s face values, this

8 Might it not feel this way because one doesn’t know that one has the experiential prima facie justification even though one has it? If so, we might expect that in describing the justificatory facts from a third-person perspective, the mention of competing strong justifications would feel correct. But it doesn’t.

6

must be because the subject is disposed to believe those contents noninferentially in response to an experience with that content – or, alternatively disposed to have appropriate noninferential seemings with those contents in response to the experience, if seemings are sui generis states distinct from beliefs (Huemer 2001, Tucker 2010). That is, what qualifies a content P as a face value is the subject’s having the dispositions to form beliefs/seemings with the content P noninferentially in response to the experience. So, if the liberal can dislodge the idea that the dispositions linking the experience with a belief/seeming cannot introduce “new content” – that is, that the belief’s or seeming’s content must be conservative with respect to the experience’s content – then she can motivate the liberal view.

Consider limited cases of absolute pitch. I, for example, can certainly recognize many notes when struck on the piano. However, for others, it can be difficult. When Ab or Eb is played, especially beyond the third octave above middle C, I can be confident it is one of those two pitches, but I can’t be confident of which, except by an effort of mentally lowering the sound a half-step. Suppose an Ab is played. Now if my experience ever represents sounds as having pitches – which the moderate allows -- this would seem to be a case in which my experience represents a tone as an Ab, despite the fact that I’m inclined to believe, upon hearing it, at most that it is either an Ab or Eb. It’s the disjunctive proposition that is a face value of the experience, not the proposition that it is an Ab. One might suggest that my auditory experience, in addition to having the Ab content has a disjunctive Ab or Eb content. However, it’s one thing to think experiential contents can be determinable; it is quite another to think that a disjunction of widely separated pitches a fifth apart could be part of the content. This would be a surprising result. I take it to be implausible. Consider that if I hear the notes Ab and Eb serially, not knowing which is which, and then a few moments later I hear the Ab and an F serially (where I can tell that the F is an F), there won’t seem to be any more auditory phenomenal similarity between the Ab experience and the Eb experience than between the Ab and the F experiences, contrary to the disjunctive content hypothesis.

Disjunction isn’t the only problem, so is negation. Many musicians can tell whether a tone within the octave above middle C is an A, but they can’t tell much more than this. Play an Eb and they will not be justified in believing it is an Eb, but they will be justified in believing it is not an A. Are we to think it is part of the content of their experience that the tone is not an A? The same concerns arise for colors. I might know that a chip, which is indigo, is clearly not turquoise. Must this be inference from this is indigo? No, I might be able to recognize indigo as indigo. (I “forgot” which shade is indigo.) Or take a strange shape for which I lack a name. Can’t I just tell it’s not a square, not a triangle, etc.? Is not being a square, etc. part of the content of the experience? The worry is that to retain the “contents only” part of the moderate’s content-based view, one is going to have to attribute rather gerrymandered contents to experience – contents which don’t seem to have a phenomenological basis, and which are posited only

7

because they have to be to make the epistemology come out right.9 I conclude that the moderate ought to drop the “contents-only” constraint on face values.

The moderate should concede, then, that whether a proposition P is a face value is a matter of whether one has a disposition linking the experience-type to noninferential beliefs/seemings that P, and these dispositions can introduce new content beyond that of the experience. What is the barrier, then, to allowing the dispositions to introduce contents concerning kinds? The moderate, I think, should become a liberal: so long as the dispositions are noninferential transitions from experiences to beliefs/seemings, there is no restriction on the sorts of contents the beliefs can have. This is not to say that any old disposition to move from an experience with content P to a belief/seeming with content Q is enough to make that Q a face value of the experience. To go down that path would be to deprive experience of its epistemic role. (Down this path lies epistemological conservatism. More on this below.) Restrictions should be imposed on the relation between P and Q. Perhaps the restriction should be P reliably indicates Q or P reliably indicates Q in normal environments, etc. But, in any case, it is not clear why Q cannot concern kinds, even kinds like Granny Smith and Ipod Touch Generation 5.

If I am right, there is pressure on a moderate foundationalist to give in to liberal foundationalism, and allow that simple perceptual beliefs, even about very specific kinds, can and often are immediately justified by experience, regardless of the outcomes of debates in philosophy of mind over rich content.

What is not to like about liberal foundationalism? It promises broad epistemological rewards. One’s foundationally justified beliefs could include beliefs about things being Ipods!

2. Immediate justification and reasons

In this section, I argue that, on a standard and plausible conception of reasons, immediate justification is never justification by reasons. If this is right, the liberal foundationalist account of the justification of our simple perceptual beliefs does not impute to us reasons, at least on this standard conception. This is a problem, of course, only if we do have such reasons for our simple perceptual beliefs. The section following this one one takes up that question.

9 Silins (2011) gives other examples, which involve degrees of justification. If, looking at an array of dots, I’m immediately justified to some degree in believing there are 10 and to some degree in believing that there are 11. Should we think both of these are parts of the content?

8

I begin by giving an account and defense of the standard conception of reasons. We are concerned throughout with reasons to believe, act in a certain way, etc., rather than reasons why one believed, acted etc.

Although we can speak of objects, events, and other non-propositional entities as reasons

to , they are so only derivatively. The dirty cup in the dishwasher is a reason to think the dishwasher wasn’t just run. This is surely only so insofar as something like the fact or proposition that the particular cup in the dishwasher is dirty is a reason to think this. The same goes for internal states: my twinge of guilt is a reason for me to believe I’ve done wrong. It is so only in virtue of the fact that I have this twinge being such a reason.10

This take on non-propositional reasons explains a number of otherwise puzzling facts. Surely, it is not a basic fact that about the dishes that they are a reason for me to believe that P. It is surely something about the dishes which helps explain why the dishes are a reason for me to believe P. But this something about them, too, is a reason for me to believe P, and it is not a reason for me to believe this because the dishes are. Nor is it merely a coincidence that whenever a thing is a reason to believe P there are facts/propositions which are reasons to believe P. We understand why a thing is a reason to believe P – and why, more generally, it is a reason to believe the things it is a reason to believe – by taking its being a reason to derive from certain facts/propositions about it being reasons.

So far so good: reasons are fundamentally facts/propositions. Further reflection on the concept of reason makes clear that there is a difference between a proposition (or fact) being a

reason there is for you to and a proposition or fact being a reason you have to .11 Ordinary talk is loose on how this distinction is expressed. It’s perfectly good English to say that if the building you’re in is burning down you “have a reason to leave it,” even if you have no inkling of that reason. But this use of ‘have’ concerns only the subject whose reason it is and does not imply anything about epistemic possession of the reason. But ‘have’ can be used to imply

epistemic possession. Here we regiment English a bit, stipulating that ‘has’ in ‘has a reason to ’ is to be understood to imply epistemic possession.

The important point about reasons-there-are vs. reasons-one-has is that when one is

justified in -ing by virtue of a reason, it is always by virtue of having a reason. If the building is burning down and you have no idea of this, this does not make you justified in leaving. But

10 Cf. Williamson (2000), Byrne (2005). Williamson’s focus is the concept of evidence, but it is natural to think that epistemic reasons would be evidence and vice versa.

11 See Schroeder (2008) and Byrne (2005).

9

when you have that reason, it can. Thus, where P is a reason for you to , it’s only when you

have the reason, P, that it can justify you in -ing.

Next, what sort of epistemic condition must one meet with respect to P to “have” it, to epistemically possess it? Looking at examples, e.g., “he had a reason to leave the building, viz. that it was on fire,” it seems one must at least be being justified in believing that P.12 If you have only a little evidence that the building is on fire, you might have a reason to leave, but this reason wouldn’t be that it is on fire, but rather something like it might be on fire or possibly it somewhat likely it is on fire. Thus, we can say: where P is a reason for one to believe Q, one’s having the reason P implies that one is justified in believing that P. 13

Finally, suppose one becomes justified in believing Q in virtue of one’s having a reason, P, to believe Q. By the above, one must be justified in believing P in order to have P as a reason to believe other things, including Q. But this is not merely a necessary condition. When one is justified in believing Q by virtue of having a reason P to believe Q, this justification is partly constituted by one’s justification to believe P. Suppose that the building is burning down is a reason I have to believe that I’ll be burnt if I don’t leave right away. Suppose it’s a very good reason and so by virtue of having it I am justified in believing that I’ll be burnt if I don’t leave right away. My justification is hardly immediate. It comes from my justification to believe I’ll be burnt.14

Thus, we have argued for three premises:

1. If one has a reason-based justification for believing a proposition (say, Q), there is a proposition (P) which is a reason for one to believe Q, and which one “has”, i.e., which one epistemically possesses.

2. Epistemically possessing a reason P requires being justified in believing P.

12 Some (e.g., Hawthorne and Stanley 2008) might think more is required: e.g., knowledge.

13 Might having an experience with content or face value P be enough to epistemically possess P. No. At best, this will be enough only when one lacks defeaters. If one has an experience as of red but knows the lighting is such that things that look red might well not be red, then this is red is not among the reasons one has. But what if it isn’t defeated? Then – assuming liberal foundationalism – one is justified in believing P. At best, then, we can say that experience with content/face value P is enough for epistemic possession only when it justifies belief that P. And this is compatible with our claim that epistemic possession of P requires that one be justified in believing that P. Thanks to Susanna Siegel for discussion here.

14 For this reason, even if P can be a reason to believe P, one cannot become justified in believing P because one has a reason, P, to believe P.

10

3. If (1) and (2) are true, then when one has a reason-based justification for believing a proposition (Q), this justification is not immediate.

These three premises entail that

4. When one has a reason-based justification for believing a proposition, this justification is not immediate.

Contraposing, (4) tells us that then when one has immediate perceptual justification to believe something, this justification isn’t reasons-based. Hence, from (4), we can conclude that if liberal foundationalism is true, then, in the class of cases in which experience gives us a face value – including the case of the apple and your beliefs that it is an apple and that it is green – one has a justification that isn’t reasons-based.

The standard reply to arguments like this isn’t to concede that under liberal foundationalism what provides immediate justification isn’t a reason, but rather to insist that premise (1) is false.15 Experiences, when they have face values, can be reasons and not only because facts or propositions about them are reasons. Moreover, experiential reasons don’t need to be epistemically possessed to justify; they must merely be states of the subject. Thus, justification from experience can be both reasons-based and immediate.

I don’t want to argue that we cannot extend the term ‘reason’ to cover such cases. I merely note this: reasons on this standard and plausible conception – I’ll call them henceforth “Reasons” with a capital ‘R’– cannot provide immediate justification. Thus, if liberal foundationalism is right, in a great many cases of justified simple perceptual beliefs, any Reasons we have are epistemically extraneous. This is because, if liberal foundationalism is true, then in these cases whether we have a Reasons-based justification or not, we have a justification that isn’t Reasons-based but which is sufficient all by itself to justify us in the belief.

3. The Looks View

Do we have Reasons for our simple perceptual beliefs? Consider the apple and your belief that it is an apple. Intuitively, you do have a reason – a Reason – to believe it’s an apple, one which your friend has as well when she looks at the apple. Your Reason is not about you or anyone else. It is about the apple itself – its looks, the way it looks. The reason is simply this: it looks like an apple. There are other variations: its look is distinctive of apples; it has a look which it wouldn’t have unless it was an apple. We have such Reasons in cases in which we visually categorize things as apples, oboes, Ipod Touches, green, vermillion, etc.

15 See Pollock (1975) and more recently Conee and Feldman (2004, 289ff) for an account in which experiences are reasons/evidence.

11

One might think that if we want to acknowledge the place of Reasons in perceptual justification, we should drop liberal foundationalism in favor of the competing experientialist views such as classical foundationalism or Wright-style conservatism. On these views, perceptual justification is always mediate, and we can see such justification as coming from Reasons. But note the character of the Reasons on these views: they are propositions about one’s experience (perhaps that it has a certain face-value) and propositions about the reliability of one’s experiences. Thus, these Reasons are propositions about oneself and one’s mental life. When someone else looks at the same apple you do, they have similar but distinct Reasons – propositions about themselves and their mental life. Under experientialism in general, then either Reasons play no essential part in perceptual justification in simple perceptual cases or if they do play an essential part, the Reasons involved are Reasons about one’s own experience. But this neglects the sort of Reasons you have in the apple case – to say the thing looks like an apple isn’t to say anything about one’s experience.

In the remainder of this paper, I defend the following view of the justification of simple visual beliefs:

The looks view:

In cases in which one is justified in having a simple visual belief, one has a Reasons-based justification for one’s belief, where the Reasons concern propositions about the thing’s looks.

From the start, I want to note three points about the looks view.

First, the looks view is a thesis about propositional justification (about justification in having or to have a belief). However, as I will note in section 5.3, it is also plausible as a theory about the doxastic justification of our justified simple perceptual beliefs.

Second, the smooth fit with the standard conception of reasons comes with theoretical debts that need paying. If a perceptual belief is justified by virtue of a Reason about looks, how is the Reason itself justified? It seems the answer is that this proposition itself must be perceptually justified. If you tell me that your house – which I’ve never seen before – looks Victorian, and I believe it looks Victorian based on this testimony alone, then if I go on to conclude that your house is Victorian, I don’t have a perceptual belief that it is Victorian. So, on the view I’m recommending, simple visual beliefs don’t really get perceptual justification going. We are perceptually justified to believe that this is an apple mediately based on our perceptual justification to believe some proposition about its looks, e.g., that it looks like an apple. In section 7 of the paper, we will consider what to say about the perceptual justification of propositions about looks.

12

Third, the looks view is not incompatible with liberal foundationalism.16 The liberal might wish to say when one’s simple visual beliefs are justified one has two justifications – one immediate, from the experience, and the other mediate, via the looks-based Reason. So, while liberal foundationalism omits Reasons, perhaps it merely needs to be supplemented with the looks view. We will examine this issue in section 6.

4. Interlude on the metaphysics of looks

What are looks? A thing’s having a certain look amounts to there being a certain way it looks. A look of a thing, then, is a way it looks. If this is what looks are, looks are not adverbial on visual experience or sensing, contra Chisholm (1957).17 In general, a way a thing looks is a way that a non-mental thing like a chair or an apple could in principle be, e.g., red, round.18 By contrast, modes of sensing – e.g., visually, auditorally, etc. – are not ways non-mental things could be. Looks are therefore not modes of sensing.

Looks can be more or less determinate and more or less comprehensive. A fairly determinate way the apple looks is yellowish green. A less determinate but more comprehensive way the apple looks might be round, green, and shiny. It is difficult to characterize a thing’s more comprehensive and determinate looks except by relating them to the looks certain sorts of objects have (in certain conditions). We say it looks the way Granny Smith apples look, etc. However, just because a thing looks like a Granny Smith does not mean that like a Granny Smith is one of its looks. Rather, a particular thing looks like a Granny Smith insofar as its looks are similar to those of Granny Smiths. The similarity is not part of the look, but holds in virtue of the looks of the particular thing and the looks of Granny Smiths or some appropriate class of Granny Smiths. This leaves us free to think of looks as complexes of sensible qualities.19

16 Thanks to Susanna Siegel for this observation.

17 Contra Breckenridge (forthcoming) as well, who holds that ways things look are ways looking events occur, where a looking event is an event of a subject’s being sensorily affected in a certain way.

18 This is not to reject out of hand error theories about color. It is only to note that the way things look, including the colors they look, are the sorts of features about which it makes prima facie good sense to think objects like chairs can have.

19 Once we consider looks over time, we might want to include dynamic properties such as motion among the properties that make up a thing’s looks. I leave these issues aside.

13

Thinking of looks this way, we need to distinguish two things we might mean by saying that a thing “has” a look. On the one hand, looks can be had by objects in the sense that the object looks that way. On the other, looks can be had in the sense that the object instantiates the complex of properties that is the look. As we see from cases of illusion, these are not the same. To avoid this sort of confusion about ‘has’, I will use ‘instantiate’ rather than ‘has’ for property exemplification.

Next, we need to distinguish between the look(s) a thing has simpliciter vs. the look(s) a thing has from a viewing perspective and/or in certain viewing conditions vs. the looks it has to a perceiver. Let’s discuss each of these.

In some cases we simply speak of “the look” of an object. Consider objects whose looks we take in and appreciate, say a statue such as Michelangelo’s David. We say it has a certain look. But it of course also has different looks, from different spatial perspectives, in different viewing conditions. To say David has a certain look isn’t to say that it has the same look from all these perspectives in all these conditions. It obviously doesn’t. If I want to appreciate the look of the statue, I need to move with respect to it, taking in its looks from the different perspectives (for outdoor statues, perhaps I need to take in its looks at different times of day). If we want to find an entity to serve as the denotation of “the look of the statue,” that entity must be something appreciated, better taken in, from a number of privileged perspectives and from a number of privileged viewing conditions. The natural candidate for such entity is a complex of invariant sensible properties, including shape, size, and color properties, which the object in fact instantiates.20 But even though a thing’s looks simpliciter should be understood in such a way,21 objects have looks from particular perspectives in particular conditions, 22 including ones which are not good ones for taking in its look simpliciter. (Think of the look of David from six inches directly above the statue, or the look of David when a red spotlight is directed on his head only.) These looks include more than just determinates of color, shape and size. They include features such as orientation (e.g., one way a coin might look from a certain perspective is titled away) and

20 I therefore agree with Michael Martin’s (2010) “parsimonious view,” when it comes to a thing’s looks simpliciter. These are properties it has anyway, its observable features such as color, shape, size. (However, see the next footnote for possible doubts.) I depart from Martin in claiming that a thing has looks with respect to a perspective or viewing condition which can be different from how the thing is.

21 However, what should we say about the following cases? Consider a structural pole half-immersed in clear water in a public aquarium. It might be natural to speak of “the look” – the look simpliciter – of the pole as bent, even though it isn’t bent. Or consider a sample of the Cornsweet illusion. Is “its look” described by “uniformly colored except for the middle” or by “two squares, the left of which is darker than the right”?

22 Viewing conditions include the likes of illumination, background, foreground, and context. I do not include facts about adaptation, transcranial magnetic stimulation (TMS), etc. as part of viewing conditions.

14

luminousness (e.g., the shadowed parts of David have a less luminous look in the prevailing lights; another example is specular highlights such as the gleam on the apple.)23 We, finally, can speak of the way a thing looks to a person, either the way it looks to her right now or generically. All these looks can vary in determinacy and comprehensiveness.

A natural question to ask is about the objectivity of looks. There are many things one might mean by talk of objectivity here. I will merely have in mind objectivity in the sense of modal independence from particular perceivers and/or token perceptions. I will apply the notion of objectivity not to the looks themselves but to facts about their being had by objects. Start with things’ looks simpliciter. That David has the look simpliciter it does is independent of whether anyone is perceiving it. Indeed, if we all went out of existence but David was unchanged, it would retain its look simpliciter.24 The same is true of looks with respect to perspective/conditions combinations. Consider the look of Grand Teton Mountain from a perspective facing it across from Amphitheater Lake on a clear afternoon in late July. Grand Teton has this look from this perspective and in these conditions, whether anyone looks at it, and would have this look if all perceivers went out of existence but this perspective and the conditions were still in place. Looks simpliciter and looks with respect to perspectives/viewing conditions are thus objective. By contrast, the fact that the apple looks green to me right now is a fact about me and the present moment. It wouldn’t obtain if I didn’t exist and if I didn’t have a certain perception. I will call such looks subjective looks.25 26

23 There are, so to speak, intermediate looks-properties other than “the look” simpliciter and looks in particular perspectives and conditions. Consider the look of a statue when bathed in red light. This is not the same as the look of the status when bathed in red light and viewed from such and such spatial perspective.

24 “I am not disclosing a fact about myself, but about petrol, when I say that petrol looks like water.” (Austin 1962, 43).

25 One complication. A thing could look generically red to me, I take it, whether or not I’m looking at it or even have ever looked at it or even whether anyone had ever looked at it. Still, that it looks this way to me is a fact about me and so obtains only if I exist. Thus, we might distinguish two sorts of objectivity with respects to the having of looks. First, there is existence-objectivity, i.e., modal independence from the existence of perceivers. Second, there is token-perception-objectivity kind is modal independence from the any token perception of the object. The former kind implies the latter but not vice versa. For the most part, I won’t bother to distinguish these two sorts of objectivity in what follows. When I talk of “objective looks” in what follows I’ll allow both sorts of objectivity to suffice.

26 Do we need to claim that a looks are had relative to particular kinds of perceiver? When we consider the possibility of very different kinds of perceivers, this relativization becomes tempting. There are a number of ways to make the relativization. We needn’t make the parameter part of the content of claims about looks. Perhaps having looks is in this way like having a certain weight. When in ordinary life we say thinks have a certain weight we are not saying anything about Earth. Thanks to Jack Spencer for discussion on this.

15

So far, I’ve spoken of the looks of particular objects: their looks simpliciter, their looks from perspectives and in viewing conditions, and their looks to subjects. We can also speak of looks of certain kinds of objects. Granny Smith apples can be said have a certain sort of look. This is a generic: not every Granny Smith will have this sort of look. Moreover, other things, not being Granny Smiths, will have the look of a Granny Smith, say a plastic “apple” or an unripe apple of another variety. And we can speak of the look simpliciter of Granny Smiths or looks of Granny Smiths from certain perspectives and in certain conditions, and of course the look of a Granny Smith to a particular person.

Finally, note that a kind of subjectivity can creep into our comparative judgments about things’ looks, even when these looks are had mind-independently and so objectively. Does young Tom look more like his mother or his father? It’s a matter of judgment, like many similarity judgments (cf. Martin 2010). But that he doesn’t look as much like, say, his friend, or the toaster, as he does his mother, is just fact.

5. In favor of the looks view

Defending the looks view requires showing that when our simple visual beliefs are justified, we have Reasons concerning looks, Reasons which make us justified in believing the target proposition. Earlier, I suggested that, in effect, that the looks view is intuitive. It does seem that, in the case of the apple, that I do have a Reason for my belief that it is an apple, and that my Reason is one you have as well (and could have even if I wasn’t here), and this Reason is about the apple and not about my or anyone else’s experiences. That I have such public Reasons for believing it is an apple, I think, is highly intuitive. And the looks view gives us exactly such Reasons.

Below I supplement this intuitive case by arguing that on the hypothesis that we had justification from looks-Reasons for our simple perceptual beliefs, we would expect to find certain evidence – evidence which I will try to show that we do find.

5.1. Patterns of defeat

Let’s begin with predictions about defeaters. If justified simple visual beliefs that a thing is a K are justified by Reasons concerning looks, we should expect that strong evidence against these looks-Reasons or against the link between them and the proposition that the thing is a K should defeat one’s justification for the belief that the thing is a K. We should expect this because these are two ways of undermining the line of support from Reasons: one is to attack the Reason, the other to attack it connection to the target proposition.

16

We first consider cases of defeaters that are evidence against the relevant looks proposition. Suppose I love the sound of an oboe, and can recognize it well from recordings and in concerts (though from my cheap seats I can’t usually see the oboes at all well.) I have the concept of an oboe. But suppose I’ve gotten them mixed up, by sight, with clarinets. Show me a clarinet and I’ll think it is an oboe. My mistake isn’t verbal. I do not use ‘oboe’ to mean clarinet. I use it to mean oboe. Now, suppose, you’ve in effect set up an identification task for me. You’ve showed me an instrument, about which I had no previous information, and asked me what it is. I declare it is an oboe, expressing my visual belief that it is an oboe. You, who know the looks of oboes, tell me: “you’ve mixed up clarinets and oboes; this is not what oboes look like.” By saying this, you defeat my justification, just as we would expect if my justification was based on the Reason that it looked like an oboe. One might worry that my justification is defeated here because the information about looks acts as a rebutting defeater. If so, we can add that that you preface your remark with “This just so happens to be an oboe disguised to look like a clarinet, but....” 27 This gives defeats my original justification, intuitively, while also giving me a new independent one (your testimony).

This prediction of the looks view is borne out in a great many cases of simple visual beliefs classifying things as falling under kinds. One can certainly be shown to be wrong about the looks of things – both “their look simpliciter” as well as their looks in certain circumstances, from certain perspectives, etc.28

You might worry about the following kind of case.29 The leaves of willow oaks don’t have the characteristic “oak tree leaf” look. Still, their leaves do have a characteristic sort of look. You might develop a categorization skill with respect to such leaves, so that you can have a justified belief this is an oak leaf when you see such a leaf. In such a case the evidence this is not what oak leaves look like will obviously not defeat your justification to believe it is an oak leaf. Doesn’t this show that the defeater predictions for the looks view aren’t borne out in some cases? No. It is not part of the looks view that one’s Reason for this is a K must always be this

27 For contrast, compare a case in which someone ignorant of perceptual psychology believes an object he sees is moving. You point out to him that no efference copy of a muscular sensation was transmitted to the brain. He will just stare at you. His justification has not been undermined. But if his justification came from his having a Reason an efference copy of an ocular muscular sensation was transmitted to my brain, his justification would be undermined.

28 Could one be shown wrong about a simple visual belief that what you see is car in anything like this way? Yes. Here the undermining is probably not going to be achieved by giving the person evidence that will completely overthrow their beliefs about what cars look like. Rather, it will merely show that this particular look is not one any sort of car has, contrary to what the person thought.

29 Thanks to Nico Silins for raising a worry much like the one to be discussed.

17

has the look characteristic of Ks. In this case, plausibly your reason is something like this is a look of a certain kind of oak leaf. Evidence against the latter will defeat your justification. In my discussions, I usually focus on the common has the K look or looks like a K reason, but this is an oversimplification. The appropriate looks-Reason will vary from case to case. The character of the categorization skill employed is the key to determining the appropriate looks-reason.

Second, consider the prediction that evidence weakening the support relation between the looks-reason and the target proposition should defeat one’s justification for believing the target proposition. Suppose in an ordinary identification case, my justification for thinking this is an oboe comes from the Reason this looks like an oboe. We’d expect that evidence that the looks of things in the particular situation are misleading as to their true identity would defeat my justification for this is an oboe. And this is what we find. There are many examples of this. Suppose you tell me that a lot of clarinets in this shop have been made to look like oboes. Again, this defeats my justification for this is an oboe, just as it would if that justification was based on the Reason that this looks like an oboe.

What we have argued here about simple visual kind beliefs applies mutatis mutandis to simple visual beliefs attributing sensible qualities. Our justification to believe this is red isn’t often defeated by evidence that a thing doesn’t look red. But it can happen. Suppose, I learn that I’m wearing a contact lens with a small red spot and the projection of a white teacup across the room perfectly aligns with that small red spot. This evidence undercuts my grounds for thinking the thing is red, and on the looks view there is an explanation why: it defeats my justification to believe that the object looks red, which is an essential element in the line of justification for my belief that it is red. Other sorts of defeaters are more common, and predicted by the looks account. For instance, our justification for that’s gray is sometimes defeated because of information that the way things look in this situation isn’t a good indication of their true colors (this can obviously happen for red as well). This needn’t always involve unusual lighting conditions. A blue square might look gray in certain backgrounds, or when juxtaposed with certain other objects, even viewed in normal lighting. Thus:

18

If a person knows about these contrast effects, they can provide defeaters for the belief that the square is gray.

So, the looks view’s predictions about patterns of defeat are borne out. Still, the liberal foundationalist might not be impressed. She might claim that her view, too, predicts the same patterns. This is because gaining counterevidence about looks or about the relation between looks and reality can defeat one’s immediate justification from experience. Such evidence is evidence that one could easily have had one’s experience despite the fact that the thing seen is not a K, which is a classic form of undermining evidence for justification from experience.

5.2. Epistemic dependence and a straightforward argument for the looks view

The above objection is well-taken. However, it leaves open the possibility that there is an important sort of epistemic dependence relation between the relevant looks propositions and proposition that the thing is a K. This dependence relation will be a key premise in a better argument for the looks view.

Consider again the oboe identification case. Like ordinary adults in this sort of case you will certainly understand the proposition that the thing looks like an oboe. Now, as we saw above, if you become justified in believing that it doesn’t look like an oboe, you’ll cease to be justified in believing it is an oboe. Suppose, though, that you’re neither justified in believing nor in disbelieving the proposition that it looks like an oboe; rather, you’re justified in suspending judgment on the matter. Could you still be justified in believing it is an oboe? This is hard to accept. Certainly if you announced aloud, “Granted, I can’t be confident either that it looks like an oboe or that it doesn’t, but still it’s an oboe,” someone overhearing you would think that surely you had some grounds – independent of your use of your visual categorization skills – for thinking the thing was an oboe. Now, if you explained that, “well, it has the look of a very special kind of oboe that doesn’t really like the way most oboes do,” this would make perfect sense. But this is simply to appeal to a different looks proposition. What seems to make little sense is to form a simple visual belief that it is an oboe while suspending judgment on all the eligible looks propositions.30 31 Why would this be if not because one must be justified in believing the relevant looks proposition – which, in the oboe identification case for the normal

30 The evening before writing this footnote my wife and I took a walk on a country trail. We saw something in the distance. She said, “could it be a building? or is it a big pile of sand?” I said, “it’s a building.” She asked, “how can you tell? It looks like a pile of sand.” My answer was that I knew there was a plant there, because I saw a sign a ways back. Imagine, counterfactually, we both knew I had no such independent evidence about what the thing was. Then if I had said, “from its looks, I have no idea of what it is, but it is a building,” my wife would have been perplexed indeed.

31 I realize I leave some externalists behind here. But again, I emphasize that my topic is reasonableness, not some notion of justification on which being justified amounts to being well along the way to knowledge.

19

perceiver, would be the proposition it looks like an oboe – in order to be justified in having the simple visual belief?

So, I conclude that in this oboe identification case, to be justified in having the simple visual belief that it is an oboe, you must be justified in believing it looks like an oboe. But it is not as if one’s justification for believing it looks like an oboe comes from one’s justification for believing it is an oboe in these cases. Rather, in order to be justified in believing it is an oboe, in such cases, one must also be independently justified in believing it looks like an oboe. That is to say: one’s justification for the looks-belief must not be grounded in one’s justification for the oboe-belief.

So, in order for your oboe belief to be justified in this case, you have to be independently justified in believing it looks like an oboe. Moreover, as we saw in the section on defeaters, in order to be justified in the oboe belief, you must not have defeaters for the reason-link between it looks like an oboe and it is an oboe. Putting these together, it follows that in order for your oboe belief to be justified, you have to have a justification to believe it is an oboe from the Reason it looks like an oboe. Similar arguments establish this for other justified simple visual beliefs. We need only ferret out the appropriate looks-proposition from the nature of the categorization skill used and rerun the argument.32

All this gives us a straightforward argument for the looks view. Take a case of a justified simple visual belief that a thing is a K. There will be an appropriate looks proposition that one must be independently justified in believing in order to be justified in believing it is a K. Moreover, this looks proposition will support the belief that the object is a K (this support relation will not be defeated). But if one is independently justified in believing such a looks proposition, which supports the target proposition that the thing is a K, and this support is undefeated, then one will have a mediate justification from a looks-based Reason to believe the thing is a K. Extending the same argument to beliefs attributing sensible qualities, we arrive at the looks view.

5.3. Doxastic justification?

Suppose the looks view is true, i.e., that we are justified to have simple visual beliefs, when we are, because we have a looks-based Reason. Is there any reason to think that we also base our beliefs in these cases on these Reasons, in whatever way is necessary for these Reasons to make our simple visual beliefs doxastically justified? If not, we might think there surely must be another line of justification on which we do base our beliefs, presumably an immediate one from experience.

32 Indeed, a good account of what a visual categorization skill is, I think, is a skill (perhaps a bad one) at categorizing seen objects merely on the basis of their looks.

20

One obstacle to a looks-based view of doxastic justification is a worry about whether we ordinarily have beliefs about looks in the relevant cases. Let’s consider how this might go in our apple case. I look at the apple and I believe it is an apple. Now, do I form beliefs about its looks? Sure, I could, but I do I? We’re invited to think: no. So, I don’t believe it is an apple based on beliefs about its looks, and therefore the looks view extended to doxastic justification is wrong.

This style of argument has its uses (e.g., consider my example of efference copies in note 24), but it should be used with care. Consider the apple again. In ordinary cases of seeing an apple, one doesn’t form the belief the thing looks like an apple in the sense of making a conscious judgment to this effect. Still, one obviously knows and so believes it looks like an apple. When I ask you, “does it look like an apple?” you answer “of course it does.” It doesn’t seem to you that you are forming a belief; you already had the belief.33 Whether we want to say the belief/knowledge is implicit, tacit, or whatever, it is there. And it can do epistemic work. Thomas Senor (2008) gives a nice example of the epistemic work of a belief not currently manifested in judgment: I look at a sunset, I judge and say it’s a beautiful sunset. I don’t consciously judge that it’s evening. But I know it’s evening, and this is surely part of my justification for believing it is a beautiful sunset. Numerous other examples demonstrate the same thing.34

So, I take it that we normal adults (and surely children at early ages), have looks-beliefs35, and our having them, in the way we do makes them available to play roles in doxastic justification of further beliefs. However, even if they are available to play such roles, we need reason to think they do in fact often play such roles. I try to provide this in the remainder of this section.

33 Contrast this sort of case with one in which a person does really form a belief. I have a friend who is a spitting image of Samuel Adams. I might ask you, about this friend, doesn’t he look like Sam Adams? Your response: “you know, he does! I never would have thought of that.”

34 Silins (2013) gives the example of seeing a DVD still in its wrapping near a friend’s TV. You believe that she hasn’t watched the movie yet. This belief is justified, and presumably this justification – this doxastic justification -- is mediate, despite the fact that one goes through no conscious reasoning involving all the relevant pieces of information. One believes all these pieces of information, but doesn’t consciously affirm them. The same is true of the case in which you believe I just missed my train.

35 I leave it to the experts to determine if animals do. The folk certainly ascribe such beliefs. From Google: “my cat thinks [pink slime] looks and smells like cat food,” “my cat thinks it [an Afrikanische Rote bean] looks and sounds like a piece of dried cat food,” and of course “my cat thinks they [hamsters] look delicious.” In my house, we put up a realistic cardboard cut-out of a cat on a landing in our stairway. Our cat at first hissed but soon ignored the cut-out. The explanation around the house was that the cat “soon figured out that, although it looked like a cat, it wasn’t a cat.”

21

If the looks view were true of doxastic justification, then in ordinary life we would presumably debate, explain, and generally think and talk in ways that would be appropriate if it were true. (The idea is that if it were true, we would implicitly be on to its truth in our ordinary thinking and talking.) Thus, we can check whether we debate, explain, etc. in the relevant ways. Such evidence is of course defeasible – e.g., it could be defeated if our best psychological theories entail that we do not rely on beliefs about looks. But it is evidence nevertheless. I’ll focus on simple visual beliefs about kinds.

5.3.1 Evidence from dialectics:

If we relied on looks-Reasons for our simple visual kind beliefs, we’d expect that debate over the truth of the belief that this is a K between two people looking at the same object and both relying solely on their visual K-categorization skills would regularly lead to discussion of the relevant looks-propositions – either whether these propositions are true or whether, even supposing they are, they are strong enough evidence for the belief that this is a K. This is what we find. For example, suppose you and a friend look at an animal in the distance about which you have no prior information. “See the elk,” you say. Your friend says, ‘It’s not an elk. It’s just a deer.” You say: “but it looks like an elk – see its big rack!” Partner: “not really – do you know what an elk looks like?” Such examples are easy to multiply. I’m not claiming one must turn to propositions about looks when doubts about a simple visual belief are raised, but only that is entirely natural. It does not feel like the introduction of new line of evidence, but rather like one subject matter – looks, objective looks – you’re using to get at another – the kind of thing it is.36

5.3.2. Explaining oneself in cases of false judgments:

If we relied on looks-Reasons for our simple visual kind beliefs, we’d expect that in a case in which the object seen didn’t have the property one attributed to it, one could and would often explain why one believed what one did by referring to its looks. Doing so would be to show one’s belief to have been reasonable in light of the evidence. So, you see the lawn ornament deer and shout to your son, “ooh, see the deer there.” It then becomes apparent it isn’t a deer. “Dad, why did you think it was a deer?” Answer: “well, it did look like a deer from back there, didn’t it?” Compare this to: “I had an experience of it being a deer” or even “it looked to me like a deer from back there.” Similar considerations apply to explaining others’ false judgments: “Dad thought it looked like a deer.”

5.3.3. Unclear cases:

36 I thought of the looks view after months of watching my twin sons arguing about what sorts of birds were at the birdfeeder. The arguments always turned on the looks of the bird. Bird books were hauled out, as was Google Image and the Audubon Society website.

22

If we relied on looks-Reasons in our visual kind beliefs, we’d expect that in cases in which one is trying to figure out whether to believe a thing one sees is a K, and in which the answer wasn’t obvious, one would consciously consider whether various looks-propositions are true. In reasoning out-loud with someone else trying to discover the same thing, we’d expect this as well. This is what we find. You look at the beautiful bird at the feeder. “What sort of woodpecker is it?” “It looks like a Downy, but like a Hairy, too; hmm, more like a Hairy – see how big it is.” 37

All this is evidence in favor of a looks view applied to the doxastic justification of simple visual beliefs.

6. Two Justifications?

As mentioned earlier, the looks view is not incompatible with liberal foundationalism. It could be that in many cases in which one has a looks-based mediate justification for P one also has an immediate justification from experience to believe P. If this is right, one’s justification for many simple visual beliefs is overdetermined.

Consider again the case in which I’ve visually mixed up clarinets and oboes. Suppose a clarinet is before me. Now, I am reasonable to believe it looks like an oboe. To insist otherwise – to think that I can’t be reasonable in thinking this unless it did look like an oboe – is to be too demanding for justification as reasonableness. I might have learned this “skill” through repeated “instruction,” say, from musicians independently playing the same joke on me, or from a book in which the instruments are mislabeled. Hopefully, such acquisitions of “skills” don’t preclude us from having justified beliefs about looks. But if I am perceptually justified in believing that the thing looks look an oboe, and I have no defeaters for this reason, it seems I am justified in believing it is an oboe. That is: I have the looks-based justification to believe it is an oboe.

37 Are the dialectical, explanation, and unclear-cases predictions borne out in the case of simple visual beliefs attributing sensible qualities? Since the connection between looking a color and being that color is tighter than that between looking like an oboe and being one, we should expect there to be fewer cases in which the topic whether it looks a certain color should arise dialectically. Still, it can arise. If I say, “it’s gray” and you disagree, “I don’t think it’s gray” and the perceptual conditions aren’t obviously idea, I might want to check that we agree at least that it looks gray – whether it is gray is something we can work out by moving the chip onto a white background.

As far as explanation is concerned, we can indeed and do refer to things looking a certain color when we explain false beliefs that a thing is that color (say in a color contrast case). “Well, I see it isn’t gray, it’s blue; but it sure looked gray.” For some of the nonbasic chromatic colors, we can easily make mistakes in judging whether a thing looked the color. Imagine a color contrast illusion in which a square looks puce but isn’t. The square is not vermillion and doesn’t look vermillion. I say “it’s vermillion.” (I do have a concept of vermillion, and know it is roughly reddish, but I mix it up with puce.) You move the square to a white background, and I say, “ah, I guess it isn’t vermillion.” You ask me, “why did you think it was vermillion?” My answer: “it looked vermillion.” You then show me a vermillion square on Google Image. You ask, “why did you think it was vermillion?” My answer: “well, I thought the thing you showed me looked vermillion, but I was wrong about what vermillion looks like.” In the case of basic colors, like red, this typically happens only when one comes to know that one’s perceptual apparatus was nonstandard, as in the case of the contact lens with the red spot.

23

Next, let’s ask: do I have immediate justification from experience to believe that it is an oboe? If not, then we have a case of the looks-based justification without an immediate justification. I will argue that even the liberal foundationalist should concede that I lack such immediate justification in this case.

My argument is as follows, in outline. Either the experience includes kinds in its content or it doesn’t. Either way, the only way to see how the experience could provide immediate justification for the oboe-belief is by its disposing me to have a noninferential oboe-belief/seeming. But such dispositions can explain immediate justification only if some strong form of epistemological conservatism is true, one which makes experience irrelevant to immediate justification, contrary to liberal foundationalism. Thus, I conclude the liberal foundationalist ought to deny that there is immediate justification in the oboe/clarinet mix-up case. Let’s go through this in detail.

Suppose in the mix-up case my experience includes kinds in its content. If it includes being a clarinet, our argument is simple: an experience attributing this property presumably cannot immediately justify a belief that the thing is an oboe. Certainly there is no reliable connection, in actual fact or in normal conditions, between these incompatible kinds. Could the content include instead being an oboe? How could this get to be part of the content of my experience? Not through the experience’s being of a type that is tokened in the presence of oboes – the experience-type I enjoy in the example is caused by clarinets and not oboes. If my experience has being an oboe as part of its content, this would have to be because the experience disposes me to form the oboe-belief/seeming noninferentially in response to it. (This itself is a very surprising thesis. Could an experience get to have the content that is a ghost if I am disposed to respond to it by believing that is a ghost?) But if this is what makes it the case that the experience has an oboe-content, it presumably must also be what is making it the case that it immediately justifies me in having the oboe belief.38

Suppose instead that the content of my experience doesn’t include kinds. Assume it merely attributes various sensible qualities and relations to objects. Could it justify me immediately in the oboe belief? There certainly isn’t a reliable connection in the actual world or normal worlds between the content of the experience I have when looking at clarinets – contents attributing clarinet-ish sensible qualities – and oboes. Here again, it seems that it must be the fact that the experience disposes me to form the noninferential oboe-belief/seeming which explains how I could have the immediate justification.

38 One highly metaphorical but intuitive idea, motivating the idea that taking an experience at face value is a source of justification, is that an experience’s having a face value is a matter of experience “speaking” to one – of one’s receiving the “testimony of the senses.” But if this speech is merely a matter of one’s being disposed to believe the relevant proposition, it is not testimony from experience but from oneself. It is a case of what Charles Travis (2004) calls autorepresentation rather than allorepresentation.

24

So, in the mix-up case, whether the content of the experience includes kind properties or not, we have the same upshot: it’s the disposition to have oboe-beliefs/seemings that accounts for the immediate justification. But how would such a disposition account for immediate justification unless a strong form of epistemological conservatism is true – a form such as this: being noninferentially disposed to believe P (or to have it seem to one that P) prima facie justifies one in believing that P? We can then ask why we shouldn’t just apply such a conservative principle to explain perceptual justification in general. The liberal, we saw, already had to appeal to dispositions to have beliefs/seemings. Her hope was to keep experience relevant by imposing constraints on the relation between this disposition and the experience’s content (the content it had that didn’t depend purely on the disposition to form beliefs/seemings) – constraints such as reliable indication. These constraints provided “friction” on spontaneity, to put it in McDowell’s language. In the absence of these constraints, one would have to concede that experience itself is not epistemically significant, except as a trigger for the dispositions to have beliefs/seemings; anything else could be such a trigger (e.g., a subpersonal state). To demote the role of experience in this way is to give up liberal foundationalism. Liberal foundationalism is an experientialist view.

The liberal, then, I conclude, ought to regard the mix-up case as one in which the subject has a looks-based justification for a simple perceptual belief but lacks an immediate justification for it from experience.

Might there nevertheless be some cases in which one has both justifications? To answer yes, we need reason to think that postulating immediate justification in addition to the looks-based justification is necessary to account for the epistemic facts. If the postulation of such justification wouldn’t account for anything which is not already accounted for by looks-based justifications, its postulation is unmotivated and unwarranted. Where can the liberal foundationalist look to show how immediate justifications from experience earn their keep? As far as I can see, the most promising place to look is to the epistemic facts about doxastic justification. If there are cases that show that relying on looks-based justifications wouldn’t -- without further immediate justifications – explain the full extent of our doxastic justification, this might be a good reason to postulate immediate justifications.

Consider the following pair of cases. Case 1: you have a normal simple visual belief that a thing is an oboe, while enjoying standard oboe-ish experiences – that is, experiences whose contents involve appropriate oboe-ish shape-color-gestalts. Case 2: you are in the oboe/clarinet mix-up case and are looking at a clarinet but believe it is an oboe. If your belief is more justified in Case 1 than in Case 2, this might seem best explained by claiming that your experience provides immediate justification in Case 1 but not in Case 2.

What to say about this pair of cases? There surely are epistemic differences between the two cases. In one case, presumably, one’s belief that it looks like an oboe is true and even knowledge in the other case it clearly isn’t. But is the belief more reasonable in one case than

25

another? I cannot see that it is. And even if you think the belief is more reasonable – more doxastically justified – in Case 1 than in Case 2, before you conclude that we must posit immediate justification to explain the difference, you would need to rule out the possibility that the belief that this looks like an oboe is more doxastically justified in Case 1 than in Case 2. This difference could explain a difference in the doxastic justification of the target beliefs in the cases.

In the absence of an account of how postulating immediate justification is needed to explain the epistemic facts, we should assume that the one justification (from looks) is our only one.

7. The justification of beliefs about objective looks

One might feel that merely defending the looks view doesn’t take us to the heart of the epistemological issues about perception. Yes, when simple visual beliefs are justified they are justified by virtue of our possession of Reasons concerning the objective looks of things. But this only pushes back the problem. One set of visual beliefs is justified by another. Now we have to ask: how do we get to be justified in believing propositions about things’ objective looks?

One way of understanding how Reasons concerning objective looks enter the picture is for them to be justified by Reasons concerning subjective looks, i.e., Reasons concerning about how things look to one at the present moment. Reasons concerning objective looks, on this view, would merely be epistemological go-betweens. In the end, the looks view would be embedded within a kind of classical foundationalism.

Consider these propositions, about the apple:

(1) it looks green

(2) it looks green to me right now.

Could the same sort of argument I’ve given in previous sections be used to show in cases like the apple case that we have a justification to believe (1) based on having a justification to believe (2)? 39

39 One common use of sentences like (2) is to indicate something about how one does or is inclined to classify the thing looks with respect to color. Used this way, one could express one’s thought more clearly as follows: “in my assessment, the way the apple looks is green.” If one could be shown that one’s classification is incorrect (e.g., say, because a moment before one had been staring at a bright red expanse), what one indicates with (2) would be true, even though one’s classification of the way the apple looked was wrong. This is more common in the case of non-basic colors. A remark “the square looks vermillion to me” might be one that the person wouldn’t retract even after being shown that she was visually mixing up vermillion with puce. She might reaffirm her statement, explaining that she was only saying that, in her assessment at the time, the look of the thing was vermillion.

26

I have no strong argument that our justification for (1), say in the apple case, does not derive entirely from a justification for (2). But I find the view counterintuitive. Consider my belief about the apple that it is green. I feel my Reasons, the ones I rely on, are about the apple and not about me, even at one further inferential remove. My Reason is (1), and there is no further Reason about myself and my experience – such as (2) – backing up (1). 40

This is just an intuition, though. To supplement it, we would need a plausible account of how one could be justified in (1) but not in virtue of having a Reason of the form (2). I can only offer the barest sketches of how such an account might ago. I sketch below two such proposals.

The first proposal takes a nonstandard view of the content of visual experience. Consider the stick half-submerged in water. It looks bent. This is not merely to say that it looks similar to bent sticks not half-submerged in water. Bent is a noncomparative look it has. Of course, it isn’t bent. Now, if someone comes along and for whatever reason – say TMS stimulation or adaptation – it looks straight to them, then something is wrong with their experience; the experience gets the situation wrong. By contrast, if the stick looks bent to a person, her experience gets the situation right, to this extent. Similar things go for other objective illusions, i.e., illusions in which how things look objectively is not how they are. This suggests the possibility that visual experience represents objects as having objective looks. On this view, an experience gets an object right, with respect to a component F of a look, just if the object does look F from the relevant perspective and in the prevailing conditions of observation.

If visual experiences represent objects has having objective looks, then what would the epistemology look like? Here is one suggestion. Suppose x looks a certain way W to you, i.e., your experience represents it as looking W. Then you will be prima facie immediately justified in believing that the thing looks that way, where this visual perceptual demonstrative picks out

When (2) is used to state “in my assessment, the way it looks is green,” I think it is fairly clear that (2) is not being used to state something that would be a person’s reason to believe a thing looks green (i.e., 1).

40 What about the considerations from dialectics, explanations of errors, and unclear cases? Do these show that we do rely subjective-looks propositions as Reasons for objective-looks propositions? I find these considerations less clear-cut than was the case for objective looks. When a belief about a color is disputed, do we naturally turn to the examination of considerations about how it looks to each of us? When we explain errors, do we do this? Perhaps. But often I think our turning to subjective looks is part of an effort to explain our judgment rather than to cite a reason we had. Consider a variant of the contact lens case. Suppose one mistakes puce for vermillion and has a puce spot on one’s contact lens. I believe the object seen is vermillion. I think the best explanation of the false belief that the object was vermillion is this: I thought it looked vermillion; it in fact looked puce to me, but I was wrong about what puce looks like, mistaking it for vermillion. We don’t need to amplify this as follows: I thought it looked vermillion because I thought it looked vermillion to me, even though it really looked puce to me, because I was mistaken about what puce and vermillion look like. Finally, in unclear cases, we don’t usually turn to an object’s subjective looks to settle disputes about its objective looks. We do say “it looks to me like a hairy woodpecker,” but this is the “in my assessment” use (see the previous footnote).

27

W; and if you are in turn justified in believing that that way is F, then you are justified in believing of x that it looks F. Here, the immediate justification from experience would be simply to believe this looks that way. This immediate justification together with the justification for that is a way of being F would secure the justification for this looks F.

A second proposal relies on a more standard conception of the content of experience. On this proposal, visual experiences do not represent objects as having objective looks but rather as instantiating the properties that comprise those looks. So, when you look at the stick half-submerged in water, your experience represents it as bent, and therefore is to that extent inaccurate. On such a view, how do we become justified in believing, say, that a thing looks green? Suppose x looks a certain way W to you, i.e., your experience represents x as W. (It represents it as W; it doesn’t represent it as looking W.) Then you will be prima facie immediately justified in believing of x that it looks that way, where the visual perceptual demonstrative picks out W; and if you are in turn justified in believing that that way is F, then you are justified in believing of x that it looks F. The challenge for this proposal would be to explain why, when you have an experience of x as W, you get to be justified in believing of x that it looks that way, where the demonstrative picks out W.

One thing these two proposals have in common is that they appeal to our justification to have beliefs that classify a look or aspect of a look as an F (in the kind case, as the look of a K). Thus, further questions about the justification for these beliefs will arise. That we are sometimes but not always justified to have these beliefs, I think, is the key to solving the problem of speckled hen. I will not try to give an account of the justification of these beliefs here, and not merely for reasons of space. 41

Conclusion

The bulk of the paper has argued for the looks view: when are simple visual beliefs about objects are justified, they are justified in virtue of our possessing reasons in the form of propositions about those objects’ looks, their objective looks. This view allows us to explain how simple visual beliefs, when justified, are justified by the having of reasons which are essentially public – the reasons are not about us about our experiences but about how the object is, and they are reasons that others can and do have just as much as we can and which we can discuss, dispute, verify. We saw that the looks view is compatible with liberal foundationalism. However, I suggested that we lack good reason to postulate a separate stream of immediate justification in addition to mediate looks-based justification. We also saw that the looks view is

41 One thought: although I do feel that, when I look at the apple, I have a clear reason to believe it is red, viz. that it looks red; I do not feel I have a clear reason to believe that is red, where ‘that’ picks out a color aspect. On the other hand, my belief doesn’t come out of the blue. It is something I feel I “just know” or that I “remember.”

28

compatible with classical foundationalism. Indeed, if one thinks our justification for propositions about objective looks is grounded in our justification for propositions about subjective looks – how things look to us right now – then one might think the looks view requires a classical foundationalist framework. In the final section of the paper, I sketched two proposals for how our justifications to believe propositions about objective looks might not derive from justifications to believe propositions about subjective looks, i.e., about how things look to us right now. If either of these proposals is on the right track, the looks view would not need to be embedded within classical foundationalism.

The looks view attempts to steer between the Scylla of liberal foundationalism, with its neglect of reasons (or at least Reasons with a capital ‘R’) in perceptual justification, and the Charybdis of classical foundationalism, which sees us as having reasons, but fundamentally reasons about our experiences. I hope to have given some grounds for thinking that there might be such a safe passageway.

Works Cited

Austin, J.L. (1962). Sense and Sensibilia. Oxford University Press.Breckenridge, Wylie (forthcoming). Visual Experience: A Semantic Approach. Oxford

University Press.Byrne, Alex (2005). "Perception and Conceptual Content" in Contemporary Debates in

Epistemology, eds. E. Sosa and M. Steup (Blackwell).Chisholm, Roderick (1957). Perceiving. Cornell University Press.Conee, Earl and Richard Feldman (2004). Evidentialism. OUP.Davis, Wayne (2005). “Concepts and Epistemic Individuation,” Philosophy and

Phenomenological Research 70 (2): 290-325.Dretske, Fred (1969). Seeing and Knowing. University of Chicago Press.Hawthorne, John and Jason Stanley (2008). “Knowledge and Reasons,” Journal of Philosophy.Huemer, Michael (2001). Skepticism and the Veil of Perception. Rowman and Littlefield. Jackson, Frank (1977). Perception. Cambridge University Press.Lyons, Jack (2008). Perception and Basic Beliefs. OUP.Markie, Peter (2005). “The Mystery of Perceptual Justification,” Philosophical Studies 26(3):

347-373.Martin, Michael (2010). “What’s in a Look?” in Nanay (ed) Perceiving the World. Oxford

University Press.Pace, Michael (2008). “Foundationally Justified Perceptual Beliefs and the Problem of the

Speckled Hen,” Pacific Philosophical Quarterly 91 (3): 401-441.Peacocke, Christopher (1992). A Study of Concepts. MIT Press.Pollock, John (1975). Knowledge and Justification. Princeton.Pryor, James (2005). “There is Immediate Justification” in Contemporary Debates in

Epistemology. Blackwell.Pryor, James (2000). “The Skeptic and the Dogmatist,” Noûs 34:4: 517–549

29

Senor, Thomas (2008). “Epistemological Problems of Memory,” Stanford Encyclopedia of Philosophy.

Siegel, Susanna (forthcoming). “The Epistemic Impact of the Etiology of Experience,” Philosophical Studies.

Silins, Nicholas (forthcoming) “Experience and Defeat,” forthcoming in Philosophy and Phenomenological Research.

Silins, Nicholas (2013) “The Significance of High-Level Content,” Philosophical Studies 162 (1):13-33.

Silins, Nicholas (2011). “Seeing Through the Veil of Perception,” Mind 120 (478): 329-367.Sosa, Ernest (2003). Epistemic Justification, coauthored with Laurence Bonjour. Blackwell.Schroeder, Mark (2007). Slaves of the Passions. OUP.Tucker, Christopher (2010). “Why Open-Minded People Should Endorse Dogmatism,”

Philosophical Perspectives.Travis, Charles (2004). “The Silence of the Senses,” Mind 113 (449): 57-94.Wright, Crispin (2004). “Warrant for Nothing (and Foundations for Free)?” Aristotelian Society Supplementary Volume, 78 (1):167–212.

30