Kelly and McDowell on Perceptual Content

Fred Ablondi

Department of Philosophy

Hendrix College


[0] In a recent issue of EJAP, Sean Kelly [1998] defended the position that perceptual content is non-conceptual. More specifically, he claimed that John McDowell's view that concepts involved in perception can be understood as expressible through the use of demonstratives is ultimately untenable. In what follows, I want to look more closely at Kelly's position, as well as suggest possible responses one could make on McDowell's behalf.

[1] Kelly begins by considering Gareth Evans' view that (in Kelly's words) “whereas the content of our beliefs, thoughts, and judgments necessarily involves 'conceptualization' or 'concept application', the content of our perceptual experiences is...'non-conceptual'” [1]. [1] Kelly examines four distinct arguments which Evans made for this claim: [2] (1) that perceptual content is the same for humans and animals; (2) that perceptual content is belief-independent; (3) that perceptual content is (at least at times) tied to action; and (4) that perceptual content is more finely grained than are our concepts. It is the fourth to which Kelly devotes the majority of his attention, and it is the one on which I too will concentrate.

[2] Evans' own reason for holding (4) is stated in Varieties of Reference in the form of a question: he asks, “do we really understand the proposal that we have as many colour concepts as there are shades of colour that we can sensibly discriminate?” [1982, 229]. Evans assumes the answer to this is a clear 'No'. McDowell, however, disagrees; in Chapter 3 of his Mind and World he argues that it is (at least sometimes) demonstrative concepts which are active in our more fine-grained perceptual experiences. He writes: “[O]ne can give linguistic expression to a concept that is exactly as fine-grained as the experience, by uttering a phrase like 'that shade', in which the demonstrative exploits the presence of the sample” [1994, 57]. In other words, while Evans may be right that we do not possess as many general concepts as there are discernable experiences, this “ignores the fact that we can credit even ordinary subjects of experience with conceptual capacities that are exactly as fine-grained as necessary, because they are expressible with the help of demonstratives” [1998, 414].

[3] McDowell's position on perceptual content is succinctly stated in his remark that “according to the position I am recommending, conceptual capacities are already operative in experience itself… Having things appear to one a certain way is already itself a mode of actual operation of conceptual capacities” [1994, 62]. On this view, to hold that there are such things as experiential “bare presences” would be to fall prey to the Myth of the Given. Again, according to McDowell, our perceptual experiences, while the result of impressions made on our senses by the world, already have content which is conceptual. He writes:

In experience one finds oneself saddled with content. One's conceptual capacities have already been brought into play, in the content's being available to one, before one has any choice in the matter. The content is not something one has to put together oneself, as when one decides what to say about something. [1994, 10]

And later:

[W]e must insist that the understanding is already inextricably implicated in the deliverances of sensibility themselves. Experiences are impressions made by the world on our senses, products of receptivity; but those impressions themselves already have conceptual content. [1994, 46]

As discussed above, McDowell takes it that the relevant concepts are expressible via demonstrative expressions. Further (and I will return to this below), he is insistent that experiences must have conceptual content because it is only in this way that “we can coherently credit experiences with rational relations to judgement and beliefs” [1994, 162].

[4] Against McDowell, and in agreement with Evans' (and Kelly's) view that perceptual content is non-conceptual, Christopher Peacocke argues [1992, 1998] that demonstrative concepts are too fine-grained to capture the contents of perceptual experiences. He writes: “Consider 'that shade', 'that red', and 'that scarlet'. These are all different conceptual contents. It seems to me quite implausible that just one of these, and not the others, features in the representational content of the experience of a shade of red” [1998, 382]. I (as well as Kelly) take his point here to be that if we follow McDowell, we must hold that only the most finely grained of the concepts in the perceiver's repertoire of concepts would be brought to bear in the experience, and that this intuitively seems wrong.

[5] Yet even if we grant Peacocke's argument—and Kelly thinks it has problems—McDowell still has a couple of responses. One which Kelly offers is to

accept the option Peacocke offers him, of taking the most specific concept in the repertoire of the perceiver to capture the content of the experience. To do this successfully he would just have to deny Peacocke's intuition that my experience of the color of the scarf is exactly the same whether my conceptual repertoire includes 'scarlet' or stops with 'red'. The denial of this intuition is at least plausible on the face of it, since it seems reasonable to think that the painter or the interior decorator, with her mastery of the various color minutiae, just sees things differently than I do with my limited array of color concepts. Part of what she sees, it might be argued, is that this scarlet wall looks like color chip r-235, but not like r-110… [20, my emphasis]

In other words, we can take up Peacocke's suggestion that we “just take the most specific concept in the repertoire of the perceiver to capture the fine-grained content” [1998, 382], but not follow his subsequent intuition to reject this suggestion on the grounds that two subjects perceiving the same color patch will have the same experiential content, regardless of the level of sophistication of the concepts in their possession.

[6] While he says that at first glance denying Peacocke's intuition might be plausible, Kelly goes on to argue that it is in fact not a move open to us. It is here that he makes the main claim of his article:

[T]he important point about the perception of properties is two-fold: first, that properties are not, as presented in experience, independent of the context in which they are perceived, and second, that they are not, as presented in experience, independent of the object that they are perceived to be a property of. [22]

Kelly asks us to consider Merleau-Ponty's remark that “the blue of the carpet would not be the same blue were it not a wooly blue” [26] in support of the contention that properties are not independent of the object of which they are perceived to be a property. And he thinks that it is this lack of independence which shows that we are no longer justified in denying Peacocke's intuition that my experience of the color of the scarf is the same regardless of the sophistication one's conceptual repertoire.

[7] Just why are we no longer so justified? According to Kelly, “[I]f a perceived color isn't describable independently of its object, then it must be false that the painter's perception of color is explicable in terms of resemblance to an objective measure” [27]. So, for example, “the color chip r-235 presents an independently specifiable property that any object could have, while the scarlet of the scarf is not presented in perception as a color identifiable independently of the scarf” [ibid.]. [3] How does this impact McDowell's position? In the Afterword to Chapter 3 of Mind and World he writes that

we can express all the concepts we need, in order to capture the finest detail of all our colour experience, by utterances of 'that shade'. We do not have all these concepts in advance, but we do have whichever we need, exactly when we need them.

An utterance of 'that shade' depends for its meaning on identity of a sample shade. We might lay down the rule that something counts as having that shade just in case it is indiscriminable in colour from the indicated sample. [1994, 170]

Recall that according to Kelly, the prima facie reason for rejecting Peacocke's intuition was that is seems reasonable that whereas I see scarlet when I look at the scarf, the interior decorator sees r-235. But Kelly maintains that the color expert does not perceive the red of the scarf and the red of the chip as the same, and that appealing to demonstrative concepts (i.e., 'that color' is the same as 'that color') does not help us. The reason is that the shade of the scarf is 'more particular', one could say, than 'just' the shade of r-235, considered independently of any object. The shade of the scarf is not merely that shade, but that shade as possessed by that object. While measurement of wavelength might reveal that the colors are 'the same', the original perceptual experience cannot so inform us.

[8] What can we say at this point on McDowell's behalf? First, it can be noted that Kelly does not provide us with an argument for Merleau-Ponty's remark, but offers it only as a suggestion. So we can ask if it's a suggestion we should accept. Is it true that the decorator's perceptual experiences of the sample and the scarf would not be the same? If not, and if Kelly is right for the reason why it would not, the same would apply, it would seem, to those subjects who have less training in discriminating perceptual experiences. That is, it would seem to follow, e.g., that a child might not be able to perceive the color of the fire engine as that same color as the red crayon in her Crayola 8 pack of crayons, for the latter could serve as much of an 'objective standard' for the child as the chip r-235 does for the interior decorator. If the child's deepest level of discernment is between red and orange (i.e., brick red, scarlet, crimson, etc., are all perceived by her at the most fine-grained level as 'red'), then Kelly's position seems to require he maintain that the child could not perceive the fire engine and the crayon as the same color, or, more specifically, perceive them as both colored thusly . And here McDowell's point comes in: if they are not perceived as both colored thusly, how could they be judged as colored thusly?

[9] Consider another example: a child hears a bouzouki for the first time one day while at the market with her mother (thus making it her 'sample') and the next week hears another bouzouki at school; would she be unable to perceive the second sound as that sound, the one she had heard the week before in a different context and on a different occasion (and thus be unable to make the judgment that a bouzouki is playing)? Kelly claims that “concepts, even demonstrative ones, pick out situation independent features, but the perceptual experience of a property is always dependent upon…context and object” [28]. And herein lies (at least part of) the mistake, I think: while the properties may indeed be dependent (in some sense) on the objects of which they are properties, this does not imply that the concepts—especially if they are demonstrative concepts—are likewise dependent. Now I don't have a developed theory of properties, and a survey of the literature offers an array of views on the topic, [4] but it seems to me that whatever concepts are—and in my view they are best understood functionally—they are distinct from properties. [5] I take a concept to be that C, the possession of which enables me to make judgements about, form beliefs concerning, and pick out those things of which C is a concept. Properties, whether understood as universals, particulars, classes of possibilia , causal powers, or what have you, are, as I see it, something else. As the examples we've looked at show, there can be more than one concept of the same property—the scarf has the property of being that color, but is perceived as 'red', 'scarlet', and 'r-235', depending on the conceptual repertoire of the perceiver.

[10] Now I agree with Kelly that a necessary condition of concept possession is the ability of the subject to consistently and correctly classify objects falling under that concept. For example, if I truly possess the concept 'book', I will on most if not all occasions on which a book is presented to me be able to tell you that the object before me is a book. And I also agree with him that it is possible that “even if I see the color of the carpet to be the same as the color of some other object—a shiny steel ball, for instance—I can always rationally wonder whether they are in fact the same color” [2001, 607]. But if in fact I can see them as the same color—or, more specifically, see they are both colored thusly —it doesn't matter if I am certain, or even correct every time, that they are the same color in order to be in possession of the demonstrative concept in question.

[11] Finally, even if we agree with Kelly that the interior decorator wouldn't perceive the red of the scarf as being the same color as whatever it is that served as her sample of r-235, it does not follow that perceptual content in non-conceptual. We can suppose that she, in this perceptual experience, acquires a new concept, that of being colored thusly, where 'thusly' refers to the ad hoc concept of something like 'wooly scarlet'. [6] If she had previously perceived 'wooly scarlet' (perhaps she bought a scarlet scarf at the same store Kelly got his), she would perceive the colors as the same; if she has not seen such a scarf, this perception now serves as her sample for the concept.

[12] The particulars of his argument aside, no doubt the chief complaint with Kelly's position which McDowell would have is that it is, like Evans' own view, a version of the Myth of the Given. To hold that initially there is the experiential input, and then concepts are applied to it, is, in the first place, either to cut us off from the world, or commit us to the skepticism which motivated Berkeley's idealism. (Of course, some philosophers might not mind having to go one of these routes.) But McDowell's concern goes further. To make perceptual content non-conceptual “offers us at best exculpations where we wanted justifications” [1994, 13]; as he explains:

The idea of the Given is the idea that the space of reasons, the space of justifications and warrants, extends more widely than the conceptual sphere. The extra extent of the space of reasons is supposed to allow it to incorporate non-conceptual impacts from outside the realm of thought. But we cannot really understand the relations in virtue of which a judgement is warranted except as relations within the space of concepts: relations such as implication or probabilification, which hold between potential exercises of conceptual capacities. The attempt to extend the scope of justificatory relations outside the conceptual sphere cannot do what it is supposed to do. [1994, 7]

The point is that if we accept the notion of a non-conceptual 'given' in our perceptual experiences, we cannot rationally explicate how that given justifies our beliefs about the world; a merely causal explanation of the impressions on our sensory surfaces cannot provide the needed justificatory relation—we cannot give reasons for why we believe as we believe. [7] In the words of another epistemologist, Lawrence Bonjour, “if [direct awarenesses or immediate apprehensions] are construed as noncognitive, nonjudgmental, then while they will not themselves need justification, they will also be incapable of giving it” [1986, 69]. Or, to put it as McDowell might, there is indeed a constraint placed on our thinking by the receptivity of sensibility, but that constraint must be rational.


Bonjour, Lawrence (1986). The Structure of Empirical Knowledge . Cambridge, MA: Harvard University Press.

Evans, Gareth (1982) The Varieties of Reference . John McDowell (ed.). Oxford: Oxford University Press.

Horwich, Paul (1998) Meaning . New York: Oxford University Press.

Kelly, Sean D. (1998) “What Makes Perceptual Content Non-Conceptual?”. Electronic Journal of Analytic Philosophy , Issue 6.

-------- (2001) “The Non-conceptual Content of Perceptual Experience: Situation Dependence and Fineness of Grain”. Philosophy and Phenomenological Research Vol. LXII, No. 3, May: 601-608.

McDowell, John (1994) Mind and World . Cambridge, MA: Harvard University Press.

-------- (1998) “Reply to Commentators”. Philosophy and Phenomenological Research Vol. LVIII, No. 2, June: 403-431.

Peacocke, Christopher (1992) A Study of Concepts . Cambridge, MA: MIT Press.

-------- (1998) “Nonconceptual Content Defended”. Philosophy and Phenomenological Research Vol. LVIII, No. 2, June: 381-388.

Putnam, Hilary (1967) “Psychological Predicates”. Art, Mind, and Religion . W.H. Capitan andD.D. Merrill (eds.). Pittsburgh: University of Pittsburgh Press.


©9 Fred Ablondi

EJAP is a non-profit entity. EJAP may not be printed, forwarded, or otherwise distributed for any reasons other than personal use. Do not post EJAP on public bulletin boards of any kind; use pointers and links instead. EJAP retains the right to continuously print all the articles it accepts, but all other rights return to the author upon publication.

EJAP is hosted at The University of Louisiana at Lafayette . Please send comments and questions to .


[Note 1] Notes References are to section numbers in Kelly [1998].

[Note 2] As Kelly notes, all four arguments can be found in Evans [1982].

[Note 3] Kelly does say that we could measure the wavelength of the light reflected by the chip and the scarf and find them to be the same, but, he notes, “this doesn't tell me anything about the content of the original perception, since it's on the basis of the new, measuring experience that I come to believe in the equivalence” [26].

[Note 4] For example, compare Peacocke's claim that “to mistake the requirements on a theory of a concept for the requirements on a theory of a property, and vice versa, can lead to errors and spurious problems” [1992, 7] with Paul Horwich, who says that he sees “no good reason not to identify properties with concepts” [1998, 21].

[Note 5] Of course, there are those who will want to hold that concepts and properties are individuated in exactly the same way, and on this basis argue that we should identify the two. For example, such a person would argue that while the concepts 'red', 'scarlet', and 'r-235' are three in number, so too are the properties 'redness', 'scarletness', and 'r-235ness'. Putnam offers an example which strongly argues against such an identification: “'temperature is mean molecular kinetic energy' appears to be a perfectly good example of a true statement of identity of properties, whereas 'the concept of temperature is the same as the concept of mean molecular energy' is simply false” [1967, 38-9].

[Note 6] This suggestion was made to me by John Haugeland.

[Note 7] I think both Sellars and McDowell are right on this point; the question of “How did I come to believe that x is the case?” calls for a different sort of answer than does a question such as “How did it come to be that the last domino in the series fell?” In this regard McDowell writes that “[Peacocke] has to sever the tie between reasons for which a subject thinks as she does and reasons she can give for thinking that way. Reasons that the subject can give, in so far as they are articulable, must be within the space of concepts” [1994, 165].