Is Perception Cognitive?

As noted here, one of Alston’s four concepts of perception is the doxastic one. He glosses it as x looks in the way that would ordinarily dispose the perceiver to take x to have a certain feature.

I’ve been obsessing about noncognitivism in epistemology lately, and this doxastic looks concept raises the issue again for me. In particular, suppose that there is a complex perceptual state to which Alston’s looks concepts answer. Such a supposition requires that the perceptual state is intrinsically motivational: it involves a look which disposes the perceiver to hold a particular belief. In this way it is analogous to the kinds of states that dispose toward action in the theory of practical rationality, where the motivational feature is cited as grounds for taking the states in question to be noncognitive. So, for example, if the existence of a moral belief is intrinsically motivational, then that is supposed to be grounds for taking such beliefs to be noncognitive.

Applying this reasoning to the case of perceptual states on the supposition above, it looks like we should conclude that perceptual states are noncognitive. What would it mean to say that perceptual states are noncognitive? That they are not representational, e.g., that they have no propositional or conceptual content? Frankly, that makes noncognitivism look silly: it infers that perceptual states lack propositional content precisely because they motivate belief. Why would anyone think that? Maybe perceptual states lack propositional content, but that’s not a good reason for thinking that they do. If noncognitivism is to be adopted on the grounds above, we’ll need some other way of characterizing what it is for perceptual states to be noncognitive. Moreover, desire states are intrinsically motivational with respect to action, and they are usually assumed to have propositional content. So maybe we should think about perceptual states being noncognitive in the way that desire states are? The prominent aspect that makes desire states noncognitive is that they are mind-to-world states, rather than world-to-mind states such as we find with ordinary belief states. But surely we don’t want to think of perceptual states on the mind-to-world model, either. Perhaps that’s the basis for treating perceptual states as cognitive, but then we need a reason for treating the grounds usually cited for noncognitivism to be inapplicable to perceptual states.

By now, it should be apparent that I suspect that thinking carefully enough about the connections between epistemology and ethics will give us grounds for doubting noncognitivist views in ethics. We’re not to that point yet, however, in large part for me at least in virtue of my failure to quite grasp exactly what noncognitivism is in ethics (though I can do what Socrates ridicules when asking about some concept: I can cite examples of the view and characterize some of them).

Taking a cue from Jamie’s comment on internalism, let’s suppose that judgment internalism threatens noncognitivism but not existence internalism. Then there is no threat that perception is noncognitive from the fact that it motivates belief. One version of the judgment view is still troubling, however. If I judge that I am nonfactively perceiving it to be the case that p, that (ceteris paribus) motivates me to believe that p is the case. So even if we distinguish judgment internalism from existence internalism, maintaining that only the former threatens noncognitivism, it’s not clear that perception is out of the woods, cognitively speaking.


Is Perception Cognitive? — 11 Comments

  1. It seems to me that it just can’t be a sufficient condition on noncognitivity that the state in question possess so minimal a motivational power. After all, beliefs possess the power to dispose those who have them to form particular other beliefs. E.g., my belief that Fa disposes me to believe (Ex)Fx. And is anyone a noncognitivist about beliefs? (I’m not counting the likes of the Churchlands here!)

  2. Jonathan, you’re obviously right, and leaves the question I’m worried about concerning how to characterize noncognitivism. I take it the motivational feature is supposed to be part of the grounds of noncognitivism: the “magnetism” of the good, etc., a la Stevenson. As noted in other comments here, the Stevenson grounds carry over fairly straightforwardly into epistemology, so something more is needed here if noncognitivism is to be avoided in epistemology. Especially, I would think it ought to be avoided about perceptual states.

  3. I’m a bit out of my area here, but I thought the motivation provided by judgments of good and bad was supposed to be analogous to the motivation provided by desires — so it wasn’t necessary to have an additional desire to motivate action, once the value judgment was in place. But beliefs aren’t explained by desires (except in abnormal and disreputable cases), so it seems the perceptual experience isn’t like the value judgment in having a motivational role.

    A little off topic, but in keeping with my practice of saying not p whenever Jon says p, I was rather distressed to realize that I had apparently talked myself into something like non-cognitivism about epistemology in my recent Phil Studies paper, where I argue that we should regard ‘know’ as functioning as praise for testimony that meets certain standards, where the standards may evolve over time in something like the way the prescriptivist suggests for ethical standards.

  4. Steven, I understand your second point but not the first. On the second point, it is clear that attributions of knowledge have a functional role beyond mere description. The usual function is some sort of honorific one, and your suggestion is a version of that. To be a noncognitivist, however, more is needed, since an ascription could be both honorific and descriptive. So I don’t see your suggestion as requiring non-cognitivism about knowledge ascriptions.

    On your first point, is this the idea? Value judgements, as well as desire, typically motivate action, but neither motivates belief. Perceptual states motivate belief, so are disanalogous to value judgements.

    If that’s the idea, I still think there’s a worry here. Besides the motivational question, the basis for taking value judgments to be noncognitive appeals to moral disagreement as well. I’m not sure that’s a good reason for noncognitivism, but if we suppose the combination of the two is, then perceptual states are perhaps noncognitive as well. It won’t matter that they motivate different entities, it’s the fact that they are motivational that is central. And as to the disagreement issue, I doubt there is much mileage there. People’s perceptions differ in the same objective circumstances, and if we are allowed to rely on radically different cultures as has to be done in the discussion of ethics, I see no reason to suppose that all such differences are resolvable, either empirically or otherwise.

  5. Yes, that does sound a lot like Gibbard. Whether it is noncognitivist to the core depends on what else one says, I think. I’ve been hoping someone would trot out the functionalist line that Gibbard likes. There is something deeply troubling to me about these lines, and it is easier to say why in the epistemological domain than in the moral domain, though I think much of it will carry over.

    So suppose we have a functionalist account of knowledge ascriptions, and we use it to explain why people use the language of knowledge to the extent they do and for the purposes they do. I think that is an interesting project and that much can be learned from it. The question is, how does it connect with the project of understanding what knowledge is? Here a deflationary attitude can creep in: after we give the functionalist account of knowledge ascriptions, there is nothing left to explain. That seems wrong to me. Assuming that knowledge has the kind of value I pre-theoretically thought it had, I want to know. Knowledge is valuable not only because I need to be able to rely on the word of others, but because I view it as an important achievement in the first-person case; and this sense remains even if I come to accept the view that a proper account of the presence of our talk of knowledge is explained in terms of the need to trust people as sources of information. To carry through the deflationist attitude, one has to treat such first-person valuing as a kind of societally inculcated false consciousness, where what really has only third-person significance becomes internalized so that it comes to be taken to have first-person significance. I am enough of a social animal that I also value being a useful contributor to society, but that aspect does not exhaust the way in which knowledge is significant for me.

    Of course, I don’t quite believe the above, since my preferred account of intellectual value proceeds in terms of understanding rather than knowledge, but the point transposes well into either context. At bottom, functionalist explanations of certain linguistic expressions–how we came to talk the way we do, and use the particular concepts we do–will lead us to focus on third-person value questions in epistemology. If one then combines such a functionalist approach with a deflationary attitude that after the functionalist account is given, there is nothing left to explain, one will have a distorted answer to the first-person issue of why one wants to have knowledge (or understanding) rather than some other cognitive state. One can try to avoid the ascription of false consciousness here by noting that, as enduring beings, memory functions to tell our later selves what our earlier selves know, thereby finding further ways for the third-person answer to help with the first-person question. I don’t think the analogy at the root of the extension will work, but even if it does, I want more: I don’t want knowledge or understanding just because my present state of information may need to testify to a later stage of myself.

    If one refuses the deflationary invitation, one can find much that is useful and insightful in the functionalist approach to knowledge ascriptions. But it’s the deflationary attitude that is needed to get from such an approach to noncognitivism, and I don’t see the attraction of that attitude in epistemology.

  6. “To carry through the deflationist attitude, one has to treat such first-person valuing as a kind of societally inculcated false consciousness, where what really has only third-person significance becomes internalized so that it comes to be taken to have first-person significance.”

    I don’t see why this has to involve false consciousness, or a lack of genuine first person value. It seems to me just a fact of human psychology that lots of what we value, for ourselves, we would not have come to value if others had not praised it. Think of sports and the arts. My brother is proud of his golf swing, and is so because others valued that skill first, but it would be foolishly reductive to say that he values it now only because he thinks others admire it. A functional account of ‘knows’ of the sort I give doesn’t mean that we now value knowledge only for the sake of the praise from, or its value to, others — that’s an essential part of the history of how we came to value it, but not an account that makes it only of instrumental value — instrumental to those other goals — now.

    I also readily concede that there are other reasons to value knowledge for oneself. Seeking knowledge may be the best way to secure the truth for ourselves, or it may fit an ideal of human living that we value now for its own sake. My idea is that the social goal of improving testimony explains some features of our use of ‘know’ — why it requires truth, belief, and justification for starters (see the paper), but not that individuals have that goal of improving testimony as their personal goal, anymore than people who seek money for their personal reasons value it as a convenient medium of exchange. (Most of them probably never think of it that way and don’t realize that that’s why we have the institution of money.)

    It seems to me that where the non-cognitivism comes in, if it does, is in the openness to revision in the states or things we apply it to (without it being a change in the meaning of the term) that I have suggested ‘know’ exhibits. There is clearly a factual component to our ascriptions of knowledge — truth and belief and the miscellaneous matters that (currently) figure into justification and Gettier considerations are facts on my view. If those requirements were the end of the the meaning of ‘know’, a correct and complete account of how we use it now (or what it is true of), then the history of our coming to use the term ‘know’, starting with it as a term of praise for acceptable testimony, wouldn’t make it non-cognitive. Also, any word can become the locus of a linguistic tug of war over usage — that doesn’t make it an anti-realist or non-cognitive term in its prior usage. Where it seems to me that non-cognitivism has something right about it is if the term is treated in its prior, normal, usage as open to revision in that way — as maybe at one time proof wasn’t required for knowledge of mathematical generalizations, in the way the word was used, but it came to be so required. If that could happen without a change in the meaning of the word (or of the equivalent terms in Greek or Egyptian), then we shouldn’t think of the terms as purely factual. Maybe even then we could see the revisions as focusing in on something that was implicitly there all along, a sort of Platonic essence that is seen with increasing clarity over time. Maybe the idea of the circle was like that — the idea of the really round as exhibited in visual examples gradually gives way to a Euclidean definition of the circle as a set of points equidistant from a given point. But my impression is that the way we use ‘knowledge’ is too messy and provisional — too open to negotiation — to expect that.

  7. Steven, I think the first two paragraphs in your response, plus the beginning of the third, show that you are not a deflationist in the sense of the term I was using. The functionalist account is supposed to be something like an archeology of a the term in English, without any particular limitations on the nature of the semantic value of the term as it is presently used.

    I don’t think I quite understand the remainder of the last paragraph, though. Here’s the source of my confusion, I think: every term in natural language is open to revision and subject to usage-tugs-of-war. The latest entry in the dictionary for ‘literally’ is “figuratively”! (It’s a southern California phenomenon…) I’m sure there’s a way to make your points that doesn’t commit you to anything outlandish about usage, but I don’t see what it is. It must be something about openness of meaning to revision, but I don’t understand what that could be.

  8. Sorry about the confusion. I meant to be allowing that every term is subject to tugs of war over usage, and to describe a way in which variations of what ‘know’ is true of could be negotiated without changing the meaning of the term. Maybe the way to put it is to say that the function of the term, as indicating acceptable testimony, is primary, so that as long as it serves that function it hasn’t changed its meaning. That allows for changes in what is currently regarded as acceptable testimony. So currently what we call knowledge would be true and believed and have certain kinds of histories involving perception, testimony, memory, and/or reasoning (etc) in particular ways. But what is accepted as the appropriate history might change, in response to assertions or denials of knowledge in new sorts of cases, made in the hope of encouraging the production of better testimony than the old rules allowed (more likely to be true, but also easier to recognize as such, or allowing usually helpful testimony on important topics we currently don’t accept testimony on, etc.). Some assertions that so and so has knowledge should be treated rather like the first attempts to get others to accept banknotes in exchange for goods and services, instead of minted coins — retrospectively it looks like money, but at the time there doesn’t yet seem to have been a fact of the matter about it being money or not. It had something of the character of a proposal — not just ‘here’s money’ but ‘let’s count this stuff as money too’. Similarly the people who first proposed not to count the old surveyor as knowing the Pythagorean generalization about relations the squares on the sides of right triangles, in the absence of his mastering a proof of it, were not simply asserting a fact (that he didn’t know it) — they were doing something more like saying “let’s not allow people to testify to mathematical generalizations anymore unless they have followed a proof.” Although couched as a denial of knowledge, it was in effect a bit of proposed legislation — but not for that reason a change in the meaning of ‘knows’ (or of the Greek equivalent) anymore than the offering of banknotes was a change in the nature of money.

    Does that make sense?

  9. Steven, the money example especially helps, so let me try the following. Let’s say that x is money iff it plays a certain functional role in an economic exchange, and let’s say that this formula is, in some sense, an account of the meaning of the term ‘money’. Then what is money at one time may not be money at another time; what counts as money is conventional in a certain respect, since the functional role depends on the conventions in place at a time. But all of this can be true and yet the meaning of the term be invariant across time.

    If we want to do the same with the term ‘knowledge’, we’ll say something like: x is knowledge iff it plays a certain functional role that is social in nature… Then what counts as knowledge will display some of the same conventionality that we see with other functional terms that are conventional in nature, even though the meaning of the term is constant.

    The money example is interesting here in another respect. As social systems become more complex, the concept of money can become rigidified so that it is no longer open-ended in the way functional concepts are. For example, if I use my carpentry skills in a well-structured barter system, the use of those skills won’t count as money even though they function in the same way that money functions. If this rigidification process goes far enough, the term ‘money’ will lose its functional origins, even though a functional archeology of the concept will still be accurate.

    This possibility raises the issue of how to tell whether the functional origins of a terms are still in place, and I don’t know how to begin thinking about that question. But in any case, the worry about noncognitivism is still pretty distant, since a functional account of money certainly doesn’t threaten that…

  10. The idea is not merely that what counts as money changes over time, but that some statements apparently about money aren’t simply true or simply false — they count rather as proposals to change the conventions about what counts as money. If a member of a society that has hitherto had only coined money offers a banknote with the comment “here’s the money I owe you”, what he’s saying isn’t simply false, even if the offer is rejected as “not really money”. Nor is it simply true even if it leads directly to a widespread practice of exchanging bank notes for goods and services, so that we can see it in retrospect as the first use of a new kind of money.

    My suggestion about some attributions of knowledge is that they play a similar role — they are not simply false, or simply true, but rather attempts to change what will be counted as knowledge. Ordinarily attempts to change what counts as an F would be regarded as attempts to change the meaning of the term, but the constancy in the functional role — for ‘knows’, on my view, indicating acceptable testimony, for ‘money’ indicating a conventional medium of exchange for goods and services — makes it continue to have the same intuitive meaning.

    Maybe a realist’s way of describing the same facts about money would be to say that the assertion about the banknotes is true if exchanging such notes for goods and services becomes an accepted practice, false if it doesn’t, and that the role of the assertion as an attempt to establish a change in the convention is independent of its truth value. Whether it is money depends upon its relation to future social practices, not on the intentions of the speaker. That wouldn’t deny that there is an attempt to change the conventions there, but just insist that it has nothing to do with whether the assertion is true or false: it could function, in effect, as an urging “Let’s do this…” even though it is also a true (or false) statement. Or maybe the realist could just treat the assertion in those cases as abnormal, and showing nothing about whether there is anything prescriptive about normal descriptions of these bits of paper, or records in banks, etc., as money? Likewise we don’t need to treat ascriptions of knowledge in the overwhelming majority of cases as anything other than factual, in order to treat a few unusual cases as attempts to change the conventions about what counts as knowledge. (Even if that is conventional in something like the way money is, which of course many would deny.) My own feeling is that ordinary use of ‘know’ does have a kind of prescriptive force (because it functions to encourage certain kinds of behavior in testifying) which is maybe lacking in our talk of money, so that talk of knowledge is more likely to need an anti-realist account.

  11. I guess I find the money example too underdescribed to think that we have to give up excluded middle here. If some authoritative body such as a central bank offers a new currency, one may feel it risky to accept the new currency, but I don’t see any reason to think it’s not money. If its something more akin to an IOU, it’s just false that it’s money. At least in our own system, the introduction of what we now call paper money falls into the latter category.

    I’m also dissatisfied with thinking that licensing testimony is the function of the concept of knowledge. The concept plays many roles besides this. It is an honorific notion; it allows one to keep score on what needn’t be investigated further; and it applies to nonlinguistic animals as much as linguistic.

    But what I ought to do is read your article!

Leave a Reply

Your email address will not be published. Required fields are marked *