1. PERCEPTION AND EXPERIENCE

Should a philosophical account of perceptual knowledge accord a justificatory role to sensory experiences? One might easily suppose that the right answer must be, ‘Yes’. For it might seem (a) that perceptual knowledge is to be explicated in terms of justified belief and (b) that the beliefs implicated by such knowledge derive their justification at least in part from our sensory experiences. Consider, for instance, believing of some surface you see that it is red, or believing of a sound you hear that it is loud. It might seem plausible (i) that such beliefs derive their justification from that on which they are based, (ii) that they are based, at least in part, on sensory experiences, and, accordingly, (iii) that they are justified, at least in part, by those experiences. There are complications.

The very idea of experience is contested. Perception is an indisputably relational notion. To see the sunset is to stand in a relation to something that is mind-independent – the sunset. Experience is not always understood to be a relational notion even in the context of discussions of seeing, hearing, and the like. In (standard) causal theories of perception sensory experiences are taken to be causally dependent on the presence of things in our surroundings but not to be essentially related to such things. Though widely held this conception has in recent decades been challenged by those who hold that the sensory experiences implicated by perceptions of mind-independent objects are essentially relational in that they are essentially episodes of acquaintance with, or awareness of, mind-independent objects. See, for instance, Snowdon (Reference Snowdon1980/81, Reference Snowdon1990), McDowell (Reference McDowell1982), Campbell (Reference Campbell2002), Martin (Reference Martin2002), Brewer (Reference Brewer2011), Logue (Reference Logue2012), and Soteriou (Reference Soteriou2013). The issue here is not merely terminological. It is not just that those who take this view use the term ‘experience’ for what others call perception, for they challenge the view that acquaintance with mind-dependent objects amounts to being in an inner psychological state that is causally related to the presence of such objects. One complication, then, has to do with how experiences are conceived.

Another concerns how experiences are supposed to justify or contribute to the justification of beliefs. We can readily make sense of a belief being justified on the basis of certain considerations, where those considerations entail the thing believed or lend it adequate evidential support, but it is far from clear what it is about the relation between an experience and a belief that is supposed to make it the case that the experience justifies the belief or contributes to its justification. Indeed, Donald Davidson (Reference Davidson and Henrich1983) famously held that while experiences, which he called sensations, cause beliefs they could not justify those beliefs, and more recently, Jack Lyons has argued against the thesis that ‘our perceptual beliefs about external objects are justified, if at all, by appeal to our having certain sensory experiences’ (Lyons Reference Lyons2009: 13).Footnote ¹ This second complication interacts with the first since how one conceives of sensory experiences bears on how one conceives of the supposed justificatory role of those experiences.

The central contention of the present discussion is that to understand what perceptual knowledge is and how it is related to justified belief we need not accord a justificatory role to sensory experiences. Before launching into the main body of the discussion a few preliminaries are in order.

2. PRELIMINARIES

A ‘looks-just-as-if’ specification of a visual experience is any expression of the form ‘… has an experience such that it looks to … just as if …’ where the first gap is filled by an expression referring to a person, the second gap is filled with an expression that picks up the reference of the first expression, and the third is filled by a declarative sentence describing something that could be seen to be so. Expressions of this form are applicable to people in ascriptions of experience such as the following: S has an experience such that it looks to her just as if the sun is setting. To understand which sort of experience is specified by this ascription we have to imagine looking at a setting sun. Of course, setting suns can figure in many diverse scenes. Nonetheless, so long as we visualize some scene or range of scenes satisfying the descriptive phrase in the ascription, we can have a good idea of the sort of experience that the ascription specifies. It will be like experiences that would be gained looking at the visualised scene or scenes. Ascriptions of this sort are non-committal with respect to whether or not the experiences characterized by their means are implicated in actual episodes of perception. The experience implicated in a perfect hallucination could be like one that would be gained if one were looking at a scene containing a spider scurrying across the floor.

When a sensory experience is implicated in an episode of perception of our surroundings – an episode of seeing, hearing, or the like – I shall call it a perceptual experience, but sensory experiences also include experiences that are perfectly hallucinatory and so not perceptual.Footnote ²

As remarked above there is a conception of perceptual experiences according to which such experiences are not essentially related to objects in our surroundings. According to this conception, even though I see a spider the visual experience I have when doing so is such that I could have had just such an experience even if I had seen nothing but had been subject to a perfect hallucination. On this understanding, having the visual experience is merely a component of the essentially relational episode of seeing the spider, but is not essentially a matter of awareness of, or acquaintance with, any mind-independent objects. Those who take this view are non-relationalists by contrast with relationalists (including those cited above) who think that perceptual experiences are essentially relational. On the relationalist view, every perceptual experience is such that, necessarily, it is an episode of acquaintance with, or awareness of, mind-independent objects. So no such experience can be implicated in an episode of perfect hallucination.

It is open to non-relationalists to take the phenomenal character of sensory experiences to be individuated by what is characterized by ‘looks-just-as-if’ specifications.Footnote ³ An experience E and an experience É would have the same phenomenal character if and only if they satisfied exactly the same ‘looks-just-as-if’ specifications. By the non-relationalists’ lights a subject who has an experience implicated in a perception might have had an experience with the same phenomenal character if he or she had been undergoing a hallucinatory counterpart of that perception – a hallucinatory episode implicating an experience that satisfies all and only the ‘looks-just-as-if’ specifications that are satisfied by the experience that figures in the perception.

Consider the case in which I see a spider and, from what I see, believe that a spider is present and the case in which I undergo the hallucinatory counterpart of that perception and on that account believe falsely that a spider is present. Assume that in both cases my visual experiences are such that it looks to me just as if a spider is in clear view. Many endorse what I shall call the Parity Thesis.Footnote ⁴ That is, they hold that cases so related can be on a par with respect to justified belief. That is because the justificatory potential of experiences is held to be fixed by their phenomenal character and the experiences in such pairs of cases would have the same phenomenal character in virtue of the fact that one is the hallucinatory counterpart of the other. On one view of the matter (Pryor Reference Pryor2000, Reference Pryor, Steup, Turri and Sosa2013) there would be parity with respect to justification in the absence of countervailing considerations. On another (Conee and Feldman Reference Conee and Feldman2004, Reference Conee, Feldman and Smith2008) there would be parity provided that a suitable background of beliefs, memories, and so forth is in place in both cases. What matters for my purposes is the idea that cases related as in my example can be on a par with respect to justified belief. On this way of thinking it is not essential to the justificatory status of the belief in the case in which I see a spider, that the case is a case of visual perception. For although in this case the seeing furnishes me with the visual experiences that supposedly have a role in justifying me in believing that a spider is there, the justificatory status of my belief does not depend essentially on the seeing. That is simply because it could have been the same in a hallucinatory counterpart case.

The Parity Thesis is compatible with a relationalist account of perceptual experiences though it does nothing to invite acceptance of such an account. It is compatible with the view that the experiences in perceptual and hallucinatory cases differ in kind even if indistinguishable, but it entails that any such difference in kind is irrelevant to the justificatory status of the beliefs involved in each case. So far as the theory of justified belief is concerned we might as well be non-relationalists.

My own view is that the Parity Thesis is profoundly mistaken. At any rate, it is if justified belief is taken to be well-founded belief. Under that conception, which I take to be the central conception, there is an asymmetry with respect to justified belief between perceptual cases and their hallucinatory counterparts.Footnote ⁵ I shall explain how this can be so in sections 5 and 6.

Nothing I shall say commits me to endorsing a relationalist account of perceptual experience. Relationalists are right in thinking that we need relational notions to account for perceptual knowledge. I believe that we can make do with familiar notions of the modes of perception – seeing, hearing, and the like. More would be needed to establish the relationalist conception of the experiences implicated by perception.Footnote ⁶

Resistance to the asymmetry just mentioned has its roots in the conception of justified belief that underpins it. I shall address this in section 4. In the next section I consider one account of how sensory experiences might be supposed to have a justificatory role. I then identify a problem for any view on which they have such a role.

3. EXPERIENCE, EVIDENCE, AND JUSTIFIED BELIEF

It is not in dispute that from the fact that one has an experience of a certain sort one might, with the help of an assumption about what causes one to have experiences of that sort, form a justified belief about one's immediate surroundings. Acknowledging that this is so is compatible with denying that experiences themselves have a justificatory role. For it is compatible with supposing that only beliefs, or propositions, or truths, or facts have such a role. My present target is any view that accords to sensory experiences themselves a justificatory role that is independent of whether they are perceptual experiences.Footnote ⁷ I shall focus on the view of Earl Conee and Richard Feldman because it has been recently influential and is illustrative of the thinking about justification that I take to be problematic.Footnote ⁸

Conee and Feldman think that experiences can contribute to the justification of beliefs in virtue of being evidence possessed by the subjects of those experiences. Here is a statement of the view about experience.

Conee and Feldman are non-relationalists about experiences. Evidentialism, for them, is a species of epistemological internalism according to which the justification of beliefs supervenes on mental states. Mental states are conceived in keeping with an internalist philosophy of mind. As they put it in another article,

The relationalist conception of experience is not compatible with mind internalism.

Conee and Feldman defend their evidentialist conception of experiential justification with the help of both general considerations and examples. Two general considerations are prominent. The first is that ‘conscious awareness is how we gain whatever evidence we have’. The second is the foundationalist consideration that there must be a kind of justification that does not depend on anything that itself calls for justification. It is suggested that in empirical matters this can be provided only by our experiences.

The first of the passages from Conee and Feldman displayed above nicely illustrates how easy it is to slide from the uncontroversial thesis that it is by conscious awareness that we gain whatever evidence we have to the controversial thesis that our experiences (by implication conceived non-relationalistically) constitute our ultimate evidence for what we believe. Looking at a glass of water I am consciously aware of it. Seeing it is a mode of being consciously aware of it. Because I am in this way consciously aware of it I can have thoughts about it from the way it looks. Included in what I see might be evidence of something, for instance, evidence that someone wearing lipstick has drunk from the glass. Recognizing the evidential significance of the smudge on the glass I might treat it as evidence that someone wearing lipstick has drunk from the glass. All this is common sense about how we come to have and exploit evidence, but there is no obvious move from the role of visual awareness of objects in our environment in enabling us to acquire evidence to the view that visual experiences, under the non-relationalist conception, can themselves routinely constitute evidence we have.Footnote ⁹ Let it be granted that an ingredient of the visual awareness of the glass is the having of a visual experience such that it looks to me just as if a glass of water is before me. Let it also be granted that this experience is not essentially related to any mind-independent object. Still, the experience itself, so conceived, does not fall within the realm of things of which I am bound to be conscious, far less perceptually conscious or conscious in any quasi-perceptual way.Footnote ¹⁰ It is one thing for me to be aware of the glass, quite another to be aware in such an episode that I am having a visual experience such that it is just as if a glass is before me. (I'm likely to be aware that I see the glass, but that again is not the same as being aware of the implicated visual experience.) This is not to deny that I could learn to think of the experience as being such that it looks to me just as if a glass of water is before me, but thinking of one's experience in that way is a fairly sophisticated matter into which we have to be coached.Footnote ¹¹ An explanation is required of how experiences could constitute evidence that one has when they do not routinely figure as evidence in one's thinking.Footnote ¹² Conee and Feldman do not see the need for any such explanation. One might defend such a stance along one of two lines. The first would be to argue that our experiences, under the operative conception, routinely figure as objects of thought. So far as I can see no defence is given of such a view. Alternatively, one might suppose that one can be justified by evidence that one has even if that evidence does not figure in one's thinking. Again, it is unclear how evidence is supposed to justify if not exploited by being taken into account in one's thinking.Footnote ¹³

The second general consideration is foundationalist. It is suggested that there has to be a kind of justification that does not depend on anything that itself calls for justification and that in empirical matters this can be provided only by our experiences. Conee and Feldman (Reference Conee, Feldman and Smith2008: section 2.1) are attracted by a conception of epistemology, inspired by Roderick Chisholm (Reference Chisholm1977), according to which we may uncover the ultimate grounds for our empirical beliefs by reflecting on how we might properly terminate chains of Socratic questioning at each step of which one pushes the question, ‘And what justifies me in believing that?’. No belief about one's current surroundings could bring such questioning to an end since, it is supposed, one could always appropriately ask in a Socratic spirit what justifies one in holding any such belief. The questioning is supposed to be properly brought to an end at points at which one can allude to how it appears to one as if things are, that is, to the character of one's current experience.

A problem arises from the fact that most of us do not engage in the kind of Socratic reflection that takes us to the supposed foundationalist stopping points. So those who take this stance must suppose either that most of us are never justified in our beliefs because we are not aware of what their supposed foundations are or that we are justified by factors that Socratic reflection can expose but which need not ever figure in our thinking – I mean figure as objects of thought. Neither of these options is attractive. If we adopt the first we must suppose that those capable of recognizing cats as cats cannot be justified in believing that a cat is before them when they see one without having followed through appropriate Socratic inquiry. The second option allows that we can be justified without having engaged in Socratic inquiry but at the price of making it obscure how factors about which we have never thought could bear on the justification of our beliefs.Footnote ¹⁴

Looking outside I believe that it's raining. That is part of what it is for me to see that it's raining. If the question were to arise as to why I believe that it's raining I could direct my interlocutor's attention outside or just explain that I see that it's raining. Sensory experiences under the non-relationalist conception are unlikely to figure and if they do it will most likely be by way of offering an explanation of why one believes what one believes rather than evidence for the thing believed.

Perhaps examples make it compelling that experiences can have an evidential role. Conee and Feldman write,

Let's focus on the case of seeing the red car.Footnote ¹⁵ Whether we see things in our surroundings and by that means gain justification for our beliefs is not a seriously contested matter. Whether in seeing things we have visual experiences under the non-relationalist conception is a seriously contested matter. Even granted that we have experiences so conceived, it is simply not evident that they should be regarded as evidence and the claim that they constitute such evidence demands explanation. Again we have to ask, ‘How can something that is rarely an object of thought routinely constitute evidence bearing on the justificatory status of our beliefs?’ The example of seeing the red car gets us no further forward in relation to this question. The epistemic standing of the subject's belief that the car in front is red is explicable in terms of the belief's having been acquired by the exercise of a certain recognitional ability – an ability to tell, and thus come to know, from the look of red things, that they are red. An explanation is needed of how we are supposed to get from this rather obvious claim to what Conee and Feldman need. The Parity Thesis might be invoked with a view to showing that recognition of the car's colour is not essential to the justification of the belief in this case. The trouble is that the present dialectic is such that the truth of the Parity Thesis is not an agreed datum. Whether it is correct is part of what is being contested.

Despite the fact that the treatment of these issues by Conee and Feldman falls within a perspective that they call evidentialism, the notion of evidence actually does very little work for them. They appear to regard anything that they take to contribute to the justification of a subject's belief as evidence that the subject has. If that is what they think then invoking the notion of evidence one has does not explain how experiences can have a justificatory role; it is simply a way of articulating the claim that they have such a role.

For all that I have argued, experiences under the non-relationalist conception could have a justificatory role in relation to beliefs about our surroundings even if they do not, properly speaking, constitute evidence one has for what one believes. Yet, to harp again on what is becoming a theme, experiences so conceived do not routinely figure in our thinking about knowledge or justified belief. We make evaluations of what we, or others, think and, when appropriate, in making these evaluations we take account of what we, or they, perceive. In some contexts, if asked to give our reason for believing something about our surroundings it is appropriate to allude to what we see. If our interlocutors are sceptical about what we claim to see a natural response would be to show them or look again, which is, of course, to direct attention outward, not to ‘look’ within. If our belief relates to a standing circumstance that we take ourselves to have witnessed in the past we could in principle return to the relevant location. If it relates to an event that is past we might try to determine whether others witnessed it. The focus is always on the world and on whether we, or others, have or had perceptual access to the supposed circumstance or event. It is not on our experiences, conceived in non-relationalist terms. All this seems to me to be correct but it might not cut much ice with those who think that experiences have a justificatory role. (I suspect that the theory that they have such a role is widely held to be the only game in town.) We need to tackle head-on the thinking about justified belief that underpins the Parity Thesis. I consider this in the next section.

4. REASONABLENESS AND WELL-FOUNDEDNESS

As I observed in section 2, an implication of views of the type under consideration is that the justificatory status of a belief never depends essentially on perception. For, according to the view, when I see and recognize a colleague in the corridor of our department, the justification for my thinking that this colleague is around does not depend essentially on my having seen her. By the Parity Thesis, I might have had exactly the same justification for believing she is around in the bizarre event of my having had a perfect hallucination of her. So the justification for my belief does not constitutively depend on the perception. I used to find such considerations compelling but it now seems to me that they trade on an epistemologist's conception of justified belief that is out of kilter with our evaluative practices. I want to bring into focus an alternative conception on which justified belief is, in a certain sense, well-founded belief. I shall set out this conception and then, in sections 5 and 6, follow through implications for the kind of justification to which perception gives rise.Footnote ¹⁶

Some of our beliefs are assumption-based – we form them on the basis of certain assumptions that we accept. Suppose that you believe that P on the basis of a number of assumptions but that one assumption on which your belief essentially depends is false. In that case there is a clear sense in which your belief is not well-founded. For it to be well-founded the assumptions in question would need to constitute an adequate reason for you to believe that P and they cannot do so if some among those that are essential to the reason's being an adequate reason are false.Footnote ¹⁷ If you were to realise that there is falsehood in what constitutes your reason for belief you would not be justified in retaining the belief on the basis in question. What about prior to such realization? Might you have been justified then in believing as you did on that basis? Many epistemologists think that whether you were does not depend on whether the assumptions were all true. (This is commonly assumed in discussion of Gettier cases, on which see further below.) In this they are not completely awry since there is a sense in which believing as you did on the basis of those assumptions might have been, as we might quite naturally say, reasonable in the circumstances. A belief is reasonable if it issued from thinking that was apt to yield a well-founded belief. A belief might be reasonable yet not well-founded because the subject was stymied by some quirk of the circumstances that was undetectable from his or her standpoint. For instance, you might have been told a falsehood by someone whose say so on the kind of subject matter in question would normally settle that the thing told is true. Nonetheless, your belief would not have been well-founded and in that sense justified.

Your belief that P is well-founded only if there is an adequate (normative) reason to believe that P, and you believe that P for that reason. In such a case the assumptions that constitute the normative reason also constitute a motivating reason you have for believing that P – a reason for which you believe that P. So in a case in which you believe that P on the basis of the assumptions that Q and R you must believe that P for the reason that Q and R. And bearing in mind that a well-founded belief is a belief for which one has a sound basis, you had better know that Q and R as well. Anything less is liable to make your belief that P unstable because hostage to the possibility of realizing that you don't know that Q or don't know that R.

It is a striking feature of contemporary epistemology that what has been taken to be the central conception of justified belief is not that of well-founded belief as I have just explicated it. On the contrary, it has been widely assumed that the central conception is that which commonly figures in discussion of Gettier cases – cases represented as ones in which a subject has a justified true belief but does not on that account know. Most of the cases that are discussed are assumption-based and involve a false assumption on which the key belief essentially depends. Clearly, if Gettier cases are to be effective in illustrating Gettier's (Reference Gettier1963) claim that knowledge is not justified true belief it should not be that their subjects fail to have knowledge because they lack justification for the belief in question that is adequate for knowledge, that is, adequate to meet the supposed justification condition for knowledge. For the point is to show that even when justification is entirely adequate, and the belief true, the subject need not know. But there is a serious question as to why it should be thought that subjects in Gettier cases have justification adequate for knowledge. In cases in which the subject's belief depends essentially on a false assumption the subject would not know, but it's just as important that it would not be well-founded either and in that sense justified. Indeed, the natural explanation for why the subject did not know, in Gettier cases in which the key belief is assumption-based, is that the belief was not justified in that sense.Footnote ¹⁸

Epistemology appears in a new light if we take seriously the possibility that justified belief, in its most central sense, is well-founded belief and that a merely reasonable belief – belief that is reasonable but not well-founded – is a semblance of well-founded belief and enjoys the positive status that it has in virtue of this relation to well-founded belief. A merely reasonable belief is on the way to being a well-founded belief because the thinking that led to it is of a sort that in favourable circumstances would issue in well-founded belief. That is why one who has a merely reasonable belief escapes criticism on the grounds of carelessness, muddle, and the like, yet falls well short of having a well-founded belief.Footnote ¹⁹

The requirement that assumption-based knowledge implicates well-founded belief respects the principle that our thinking as to what is so is answerable to, and therefore should be rationally constrained by, truths or facts.Footnote ²⁰ When you see that there is milk in your glass, the fact that there is milk in your glass can rationally constrain your thinking. For since you thereby know that there is milk in your glass, the truth that there is could contribute to the well-foundedness of other beliefs, for instance, a belief that there is no need to fetch more milk, or a belief that milk is on hand if you want to drink it. But now what are we to make of the status of the belief that there is milk in your glass gained from looking at the glass? Does the notion of well-founded belief apply to this belief and, if it does, what makes for well-foundedness in this sort of case?

I shall return to this matter in the ensuing two sections. However it is settled it should not be in dispute that seeing that there is milk in your glass is an episode in which your thinking is rationally constrained by how things are. The exercise of the relevant recognitional abilities is an act of the rational intellect – one by which you come to know that there is milk in your glass. So the truth that you thereby know has the potential to serve as a reason or a component of a reason to believe other things. By contrast, on a view that takes justification to derive at least in part from experiences under the non-relationalist conception, or one that endorses relationalism about experiences, but takes the relational character of experiences to be irrelevant to their justificatory status, whether a belief is justified has nothing to do with whether its formation is rationally constrained by how things are. (How could it when it does not constitutively depend on perception?) That raises a question as to whether a belief that is not so constrained could contribute to the well-foundedness of other beliefs. Well-foundedness always depends essentially on how the world is. Beliefs that are not rationally constrained by how the world is cannot contribute to the well-foundedness of other beliefs. We have, then, an incentive to consider whether there is a different, and better, way to think about the justification provided by perception.

5. PERCEPTUAL KNOWLEDGE FIRST

To vary examples, consider believing that a rose is before you on looking at a rose and judging from how it looks that it is a rose. In that situation judging that a rose is before you would be inextricable from making a demonstrative judgement concerning that flower that it is a rose. By contrast, if you had been undergoing a perfect visual hallucination of a rose, you would not be seeing any flower and so would not be making any demonstrative judgement concerning a particular flower to the effect that it is a rose. What should we say about the status of such demonstrative judgements?

At this point it should not be assumed that the first question to ask is, ‘What would it take for your judgement concerning the rose to be justified?’ You are looking at the rose and judge it to be a rose. We may quite naturally, indeed, more naturally, go straight to a question about knowledge. What would it take for this judgement to amount to knowledge?

Obviously, you must know a rose when you see one, and your judgement on this occasion must depend on that knowledge. What we pick out by such language is that you would need to have a certain perceptual-recognitional ability – an ability to recognize roses as roses from the way they look.Footnote ²¹ This ability is an ability to tell, and thus come to know, of roses that they are roses from the way they look. Abilities of this kind are entirely familiar. They include the ability to recognize a person as one's friend so-and-so from the way he looks, the ability to recognize a smell as being that of sizzling bacon from the way it smells, and the ability to recognize from the sound of a piece of music that it is by Mozart. You exercise such an ability only if you do what the ability is an ability to do. So you exercise the ability to recognize roses as roses, only if you recognize some rose to be a rose, thus only if you tell and thus come to know of some rose that it is a rose.Footnote ²² The standing you occupy in recognizing the rose to be a rose from the way it looks counts as knowledge in virtue of being acquired via the exercise of an ability to do such a thing.

Perceptual-recognitional abilities are very specific abilities individuated in part by the concepts employed when they are exercised. They are acquired as we learn to attune our applications of particular concepts to how things appear to some sense modality. Of course, it matters that the things we recognize have a nearly enough distinctive appearance. You can tell from the look of the flower that it's a rose only if your environment is favourable in the sense that in that environment roses have a nearly enough distinctive look, so only if there is a visual appearance such that having that appearance is a very highly reliable indicator of being a rose. Possession of the ability is relative to environments that are favourable in this respect. (Examples of the fake-barn type (Goldman Reference Goldman1976) depict environments that are not favourable to the possession of the relevant ability. In those environments there is some kind such that you cannot tell of instances of that kind that they are of that kind from their appearance because the appearance is not nearly enough distinctive of the kind.)

Perceptual-knowledge is not evidence-based.Footnote ²³ When you tell that the rose is a rose you do so from the way it looks, not by basing your judgement on assumptions as to how it looks. Indeed, you might be perfectly able to recognize roses as roses even if you lacked the resources to frame thoughts about the visible appearance of the rose, whether in terms of visible features, or just in terms of a notion of the look of a rose. What matters is that your judgements should be attuned to how roses look, not that you should rely on some conceptualization of how they look. Thanks to having suitable conceptual resources, most of us would take in at the level of belief or judgement that the rose has such-and-such features or has the look of a rose, but the point here is that there is no necessary connection between being able to recognize roses as roses from the way they look and being able to conceptually represent how roses look.Footnote ²⁴ (Children can adequately recognize species of animal prior to having concepts of visible features.)Footnote ²⁵

From a mainstream standpoint it might seem that recognitional abilities and their exercise cannot be explanatorily basic – that we have to explain how exercises of those abilities can be acquisitions of knowledge in terms that are more basic and do not implicate the concept of knowledge. That is one motivation for supposing that we need to account for perceptual knowledge in terms of evidence or justified belief (or both), but nothing other than entrenched tradition makes this seem so. The phenomena do not require us to think this way. The availability of the recognitional-abilities treatment of perceptual knowledge shows that we are not powerless to understand what it is to see, and in that way know, that something is so other than by the lights of some reductive account of knowledge in terms of justified belief.

None of this is to deny that episodes of seeing yield justified beliefs. Nor is it to deny that such episodes implicate sensory experiences, or that such experiences are important for how we think.Footnote ²⁶ Our ability to think in the ways we do about how things look, sound, feel, or otherwise appear, as when recalling what we have perceived or visualizing how things would look in some situation, depends on our having had suitable sensory experiences. What is being contested is the idea that to make sense of perceptual knowledge we need to explicate it in terms of justified belief and explicate that in turn by according sensory experiences a justificatory role. Perception does yield beliefs that are justified. We can make sense of their being justified in the sense of being well-founded, but it will turn out that their well-foundedness depends on perceptual knowledge and not the other way round. I consider this in the next section.

6. PERCEPTUAL KNOWLEDGE, WELL-FOUNDEDNESS, AND RATIONAL CONSTRAINT

One might be led to think that the notion of well-foundedness has application only to assumption-based beliefs and does not apply to the beliefs implicated in perceptual knowledge. When you see the rose and, by way of exercising an appropriate recognitional ability, judge it to be a rose, you are not basing your belief on any assumption. That is not to say that background beliefs have no role to play here. The point is just that you did not form your belief – come to your belief – prompted by something else that you believe constitutes your reason for believing. Rather, by way of exercising an appropriate recognitional ability you simply saw concerning this rose that it is a rose. Now, I have been working with the idea that your belief that P is well-founded only if there is an adequate reason for believing that P and you believe that P for that reason. Accepting this might lead one to suppose that the notion of well-foundedness does not apply to believing that P in an episode of seeing that P since in such a case you do not come to believe that P prompted by something else you believe. But it would be a mistake to suppose that there is no sense in which you can have a reason for believing that P in such a case and, indeed, believe that P for that reason. There is a reason for you to believe as you do concerning the rose. It is constituted by the truth that you see that that is a rose. It is a reason that is available to you if you have a particular kind of perceptual-recognitional ability – an ability to tell of roses you see that you see them to be, that is, see that they are, roses.

Just as we have concepts like that of a rose that admit of recognitional application to perceived objects, so there are relational concepts that admit of recognitional application to perceived objects. One such concept is that of seeing that … is …, as in the judgement, He sees that the block there [demonstrative] is red or I see that that [demonstrative] is a rose. Assuming that you are armed with such a concept and can apply it in first-person terms recognitionally, in response to what you see, then when you look at the rose, recognizing it to be a rose from the way it looks, you are in a position to recognize, concerning this rose, that you see that it is a rose. It is looking outward that brings the appropriate relational concept into play, and with it a reason to believe concerning the rose you see that it's a rose. Our ability to employ the relational concept in the way described accounts for our ability to have reasons constituted by truths as to what we see, thus know, to be so. Moreover, in the case in hand, the fact that you see concerning the rose that it's a rose can be your reason for believing, concerning that flower, that it's a rose. It is not that you were led to form this belief in view of the fact that you see, concerning the rose, that it's a rose. The belief that it's a rose and the belief that you see that it's a rose come together, yet the latter can sustain the former so that, were you to doubt that you see that it's a rose then, all else equal, you'd be liable to doubt that it's a rose, and so long as you continue to believe that you see that it's a rose you'd resist any doubt that it's a rose.

The upshot is that your belief that the flower is a rose can be well-founded on the truth that you see that it's a rose. This truth, moreover, is itself something you know in virtue of exercising the recognitional ability that implicates the relational concept.

This account would be problematic if we were in the business of giving a reductive account of visual-perceptual knowledge in terms of a justification for belief that has to be explicated without adverting to that knowledge. For then one would be attempting to account for the visual-perceptual knowledge concerning the rose in terms that invoke that very knowledge. The account would be blatantly and viciously circular. But we are not, and from the perspective I am outlining should not be, in that business. We account for the visual-perceptual knowledge in terms of the exercise of a recognitional ability. We account for access to the reason that provides one with justification for believing, concerning the rose, that it's a rose in terms of the exercise of a distinct recognitional ability – one that implicates the relational concept discussed above. At the same time we can explain how the justification in the case can be well-foundedness in the sense explained in section 4, for what constitutes your reason for belief is a truth – a known truth about what you see.Footnote ²⁷

A distinct worry might arise concerning how it can be that our entitlement to believe of the rose that it's a rose consists in a justification for so believing that is provided by a reason constituted by the fact that one sees that it is a rose. For one might suppose that to be in a position to avail oneself of that reason one would need to be independently entitled to believe concerning the rose that it's a rose. The trouble then is that even if the proposed reason is a reason so to believe, it seems to be a reason possession of which presupposes rather than accounts for our entitlement to believe concerning the rose that it's a rose.Footnote ²⁸ On the proposed account, though, we do not explain an entitlement to believe concerning the rose that it's a rose in terms of a justification that presupposes that very entitlement. Rather, we account for the knowledge concerning the rose that it's a rose in terms of the exercise of a recognitional ability linked to one's possession of the concept of a rose, and we explain the justification of the belief concerning the rose that it's a rose in terms of the reason made available by the recognitional ability linked to one's possession of the relational concept of seeing that … is a … .

It is no part of the story that, necessarily, if you have visual-perceptual knowledge concerning the rose that it's is a rose then you will have a belief that is well-founded in the way I have explained. That is because we accommodate the possibility that you lack the conceptual resources to believe and exploit the truth that you see that the rose is a rose, or any truth that is tantamount to that. For most people this possibility is unlikely to be realised. Most of us acquire the conceptual and recognitional resources that enable us to believe and exploit reasons of the sort under consideration. The upshot is that, while one can know that P in virtue of seeing that P, even if one lacks the conceptual and recognitional resources to have a reason to believe that P constituted by the fact that one sees that P, nonetheless, most of us who know that P in virtue of seeing that P, have those resources, and so it can be true of us not only that we know that P in virtue of seeing that P but, in addition, that our belief that P can be sustained by our having a reason to believe that P constituted by the fact that we see that P. We should not multiply necessities without necessity. We can do justice to the facts without assuming that there is a necessary connection between perceptual knowledge and possession of such reasons.Footnote ²⁹

Some might feel a residual unease at the idea that one could have a belief like the envisaged belief about the rose and yet, by my account, not be justified in so believing in the central sense. But note that even if you lacked the resources to have the kind of reason for believing that the rose is a rose that I have been discussing, the standing of your belief would still be very different from the standing of a belief accompanying a perfect hallucination of a rose. For, as previously observed, in the case in which you see the rose your thinking is rationally constrained by the fact that this thing at which you are looking is a rose. Your recognition of the rose as a rose is an act of the rational intellect by which, on seeing the rose, the fact that it's a rose shapes your current thinking. You come to know that this thing is a rose. Yet even though you would acquire knowledge in this way you would lack something important that most of us have when we have perceptual knowledge – a reason to believe the thing we know constituted by a truth as to what we see to be so.Footnote ³⁰