International Phenomenological Society
Philosophy and Phenomenological Research
The Non-Conceptual Content of Perceptual Experience: Situation Dependence and Fineness of
Grain
Author(s): Sean D. Kelly
Source: Philosophy and Phenomenological Research, Vol. 62, No. 3 (May, 2001), pp. 601-608
Published by: International Phenomenological Society
Stable URL: [Link]
Accessed: 08-11-2015 00:30 UTC
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at [Link]
info/about/policies/[Link]
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@[Link].
Wiley, International Phenomenological Society and Philosophy and Phenomenological Research are collaborating with
JSTOR to digitize, preserve and extend access to Philosophy and Phenomenological Research.
[Link]
This content downloaded from [Link] on Sun, 08 Nov 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions
Philosophy and Phenomenological Research
Vol. LXII, No. 3, May 2001
The Non-conceptualContentof
PerceptualExperience:Situation
Dependenceand Fineness of Grain
SEAN D. KELLY
Princeton University
I begin by examining a recent debate between John McDowell and Christopher
Peacocke over whether the content of perceptual experience is non-conceptual.
Although I am sympathetic to Peacocke's claim that perceptualcontent is non-concep-
tual, I suggest a number of ways in which his arguments fail to make that case. This
failure stems from an over-emphasis on the "fine-grainedness"of perceptual content -
a feature that is relatively unimportantto its non-conceptual structure. I go on to de-
scribe two other features of perceptualexperience that are more likely to be relevant to
the claim that perceptual content is non-conceptual. These features are 1) the depen-
dence of a perceived object on the perceptual context in which it is perceived and 2)
the dependence of a perceived propertyon the object it is perceived to be a propertyof.
1. Recently in this journal, ChristopherPeacocke published an article
defending the idea that the content of perceptualexperience is non-concep-
tual.1 Peacocke defends this claim against John McDowell's attack on non-
conceptual content in chapter 3 of Mind and World.2 The debate between
Peacocke and McDowell focuses on one of the ways (there are others) in
which perceptualcontent is often said to be differentin kind from the content
of our (conceptual) linguistic utterances or thoughts: namely, perceptual
content is often said to be "finergrained"than the concepts in terms of which
we reportour perceptualexperiences.I am myself sympatheticto the idea that
perceptual content is non-conceptual,and I am therefore, in a general way,
sympatheticto the projectin which Peacocke is engaged. I believe, however,
that the defense he offers of non-conceptualcontent is weaker than it should
be. This weakness stems, in my view, from the fact that the fine-grainedness
of perceptualexperience is not as relevantto its non-conceptualstructureas a
certain kind of situation dependence is. In this paper I will first suggest a
numberof ways in which Peacocke's argumentsfail to make the case for the
1
Peacocke, Christopher, "Nonconceptual content defended," in Philosophy and
Phenomenological Research, Vol. LVIII, No. 2, June 1998, pp. 381-88.
2
McDowell, John, Mind and World(Cambridge:HarvardUniversity Press, 1994).
THE NON-CONCEPTUALCONTENTOF PERCEPTUALEXPERIENCE 601
This content downloaded from [Link] on Sun, 08 Nov 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions
non-conceptualstructureof perceptualcontent. My criticism of Peacocke's
argumentis different, I believe, from the criticism that McDowell proposes
in his response to Peacocke's paper, though I will not attempt to compare
our views here.3Having shown that Peacocke's argumentsdo not succeed, I
will go on in ?3 to describetwo featuresof perceptualexperience-the depen-
dence of a perceivedobject on the perceptualcontext in which it is perceived
andthe dependenceof a perceivedpropertyon the objectit is perceivedto be a
propertyof-which, I believe, are more likely than fineness of grain to be
relevantto the claim thatperceptualcontentis non-conceptual.
Before I get startedon my criticism of Peacocke's argument,let me first
give a generalidea of the contoursof the debate between him and McDowell
(as I understandit). The question at issue, to begin with, is whether percep-
tual experience is more finely grainedthan the concepts we have to describe
it. For the standardexample of this, think of the various shades of color we
can discriminateperceptuallybut for which we have no generalcolor concept.
Insofar as our perceptual discriminations are more finely grained than our
generalcolor concepts-we can discriminatemore colors than we have color
concepts for-one might be tempted to claim that perceptualexperience is
non-conceptual:the general color concepts seem to be insufficientto charac-
terizeadequatelythe contentof the experience.
Once we understandthis basic idea, then roughly speaking, the debate
between Peacocke and McDowell goes like this: McDowell claims that even
if perceptual content is not articulable in terms of general concepts, it is
articulablein terms of demonstrativeconcepts, concepts like <that shade>;
Gareth Evans, he notes, the original proponent of non-conceptual content,
failed to consider this possibility. Once we allow for the possibility of
demonstrativeconcepts, accordingto McDowell, the need for a non-concep-
tual perceptualcontent is no longer pressing:demonstrativeconcepts can do
thejob of characterizingthe contentof our perceptualstates.
Peacocke counters by claiming that demonstrativeconcepts aren't up to
the task. I'll explore this claim more carefully in ?2, but the general idea is
thatperceptualcontent, accordingto Peacocke, is inadequatelycharacterized
both by generalconcepts, like the color concept <mauve>, and by demonstra-
tive concepts, like <that shade>: it cuts more finely than the one (general
concepts), but less finely than the other (demonstrativeconcepts). The result,
Peacocke argues,is thatperceptualcontentis not conceptualeven if we allow
for McDowell's trick of using demonstrativeconcepts.
The details of the debate will become clearer, I hope, as this paper
proceeds. For the moment what is importantto rememberis that McDowell
introducesthe following idea into the debate:a demonstrativeconcept may be
See McDowell, John, "Reply to Commentators,"in Philosophy and Phenomenological
Research, Vol. LVIII, No. 2, June 1998, esp. pp. 414-19.
602 SEAN D. KELLY
This content downloaded from [Link] on Sun, 08 Nov 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions
able to be used to characterizethe content of a given perceptualexperience.
Now, I'll turnto Peacocke's treatmentof the situation.
2. Peacocke distinguishes between three levels of description that are
applicable to a subject perceiving the shape of an object (or generally, some
visible propertyF of an object). Roughly, these levels are:
(i) the shape itself.
(ii) the shape as perceived in experience (or, as we might say, the
"perceivedshape").
(iii) the shape as demonstrativelyconceptualized(as, for instance, in the
utterance"thatshape"or "thatsquare").
The goal of the non-conceptualist, according to Peacocke, is to show that
levels (ii) and (iii) come apart.(We can agree that this is at least a necessary
condition for the success of the non-conceptualistposition.) McDowell, on
the other hand, as a conceptualistabout perceptualcontent, wants to explain
level (ii) in terms of level (iii). The argumentcomes over whether this is
possible.
In orderto show that it is not possible to explain perceived propertiesin
terms of demonstrativeconcepts, Peacocke first tries to show that demonstra-
tive conceptsare too fine-grained:
I think McDowell is right when he complains that Evans, for all his importantcontributions,
overlooked demonstrativeconcepts. But it seems to me that these demonstrativeconcepts slice
too finely to capture the ways of level (ii). Consider "that shade", "that red", "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 representationalcontent of the experience of a shade
of red.4
I don't think this is a very strong [Link] crux of the argument
seems to reside in the claim thatmany differentdemonstrativeconcepts must
feature in the representationalcontent of a single experience. It is "quite
implausible,"as Peacocke says, that this should fail to be the case. But I'm
not convinced. Here's why.
It is true, of course, that "that shade", "thatred", and "that scarlet" are
different concepts, and it is also true that they could all pick out the same
color swatch. If that color swatch is both colored and red and scarlet, then
naturally all those demonstrative concepts will pick it out. But what the
demonstrative concepts pick out in this case is the property itself, as
described at level (i). To claim further that they all pick out the same
Ibid., p. 382.
THE NON-CONCEPTUALCONTENTOF PERCEPTUALEXPERIENCE 603
This content downloaded from [Link] on Sun, 08 Nov 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions
perceived property,as describedin level (ii), seems to me to requireindepen-
[Link] all, at least on the face of it, the fact that the color of
my scarf is accuratelydescribableas a shade and as a shade of red and as a
shade of scarlet doesn't indicate that my experience of it as a shade is the
same as my experience of it as a shade of red and my experience of it as a
shade of scarlet. Indeed, it seems plausible to think that if I'm grouping it
with a variety of different red things I may experience its color differently
than if I'm groupingit with a variety of different scarlet things. And in fact,
if I'm grouping it with a variety of different red things then it's at least
conceivable that the demonstrativeconcept "thatred"will get the experience
right, but the demonstrativeconcept "thatscarlet"won't. If that's right, then
demonstrativeconcepts don't slice too finely at all. They slice just about
right.
Now, I'm not sure how seriously to take this criticism of Peacocke. At
least it seems to require a response if the non-conceptualist wants to stick
with his strategyof showing that demonstrativeconcepts are too fine grained
to account for perceptualexperience, and I don't see that Peacocke has one.
Since I don't think this strategy does much to help the non-conceptualist
anyway, I'm temptedto leave the issue to one side. But Peacocke pursuesthe
problemone stage further,so let me just say this extra bit.
Suppose it could be more convincingly argued,pace the criticism above,
that demonstrativeconcepts are too fine grainedto account for the content of
perceptualexperience. Still, it seems that the conceptualist has available to
him either of two moves. In the first place, he could accept the option
Peacocke offers him of taking "themost specific concept in the repertoireof
the perceiverto capturethe fine-grainedcontent"5of the [Link]
thinks this won't work because of the intuitionhe has that the experience of
the color of, for instance, a scarlet scarf is exactly the same for the person
whose conceptual repertoireincludes "scarlet"as it is for the person whose
repertoirestops with "red".As Peacocke says, "[T]hereis a single shade ...
that they both experience, and in the same ways".' But is this intuition
enough?
Peacocke leaves the intuitionundefended,and it seems at least primafacie
plausible to deny it. For instance, we might reasonably imagine that the
painter or the interior decorator, with her mastery of the various color
minutiae,just sees things differentlythan I do with my limited arrayof color
concepts. To defend this idea we might say that part of what she sees is that
this scarlet scarf looks like color chip r-235, but not like r-110; and this
could mean, if we chose the color chips properly,that she saw it as scarlet,
but not as a more canonical shade of red. If for me there is no distinction
Ibid.
6
Ibid.
604 SEAN D. KELLY
This content downloaded from [Link] on Sun, 08 Nov 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions
between the experienceof these two shades, while for her thereis, then surely
the content of my experience is differentfrom the content of hers. Therefore,
if this account of color perception is right, then although there is a single
shade that we both experience, we neverthelessexperience that single shade
differently. And furthermorethis difference seems to be attributableto our
differentconceptualrepertoires.
Now, I don't think this account of color perceptionis ultimately defensi-
ble, for reasons I'll explain in ?3. But the reasons I bring to bear against it
there have nothing to do with the intuitions that motivate Peacocke's view.
So I don't see thatPeacocke has a decent responseto this line of argument,at
least not on the basis of the account he develops in the article I'm consider-
ing. So much for the first conceptualistline of response.
The other option for the conceptualist is also suggested by something
Peacocke says. The conceptualist could simply accept Peacocke's intuition
that the experience stays the same no matter what fine-grained concepts I
have, and explain the content of the perception in terms of the "medium-
grained"demonstrativeconcept "thatshade".If it is true, as Peacocke says,
that "thereis a single shade.. that [the variously adeptobservers]experience,
and in the same ways,"7then the demonstrativeconcept "thatshade" should
properlypick out the right perceived shade, and hence get the content of the
experienceright. Peacocke's descriptionof the claim makes this unavoidable,
since once he identifies the content of the experience in terms of the single
perceived shade, pointing to it with the phrase"thatshade"seems a perfectly
reasonableway to pick it out. The way Peacocke gets aroundthis explanation
in his example is by stipulatingthat neitherof the observers has the general
concept "shade".In the example in questionPeacocke wants to focus on:
the fine-grainedrepresentationalcontent of experience of two people, neither of whom has thle
general concept "shade", but one of whom has the concept "scarlet",and the other of whom
has only "red"but not "scarlet"...8
But this seems to me [Link] you have the specific concept "red",then you
must know that it refers to some featureof the object, and what is that feature
if not its color or shade? So it seems to me that if McDowell is going to
accept Peacocke's intuition that the possession of fine-grained concepts
doesn't change experience, then it is still open to him to explain perceptual
content in terms of demonstrative concepts of the medium-grained sort-
concepts like "thatshade,""thatshape,"and so on. Again, I will argue in ?3
that this rejoinderdoes not work, but I remain unconvinced that Peacocke's
stance against it is satisfying.
7 Ibid.
8 Ibid., my italics.
THE NON-CONCEPTUALCONTENTOF PERCEPTUALEXPERIENCE 605
This content downloaded from [Link] on Sun, 08 Nov 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions
3. It looks to me, then, that the argumentthat demonstrativeconcepts
are too fine-grainedto account for experience is not a very convincing one.
But in a way that's all by the by, since I don't think that it was getting at the
importantphenomenon anyway. I think that the importantpoint about the
perceptionof propertiesis twofold: first, that propertiesare not, as presented
in experience, independentof the context in which they are perceived, and
second, thatthey are not, as presentedin experience,independentof the object
they are perceivedto be a propertyof. I suspect thatPeacocke has believed in
these two types of dependencies at various points in his career, though I'm
not sure he's ever advocated both simultaneously. I think he should, and I
also think that if he does, he will have the resources necessary to block the
possible responses I consideredjust now on McDowell's behalf. So let me
say a bit aboutthe dependencies.
The first kind of dependency-the dependencyof a perceived propertyon
the context in which it is perceived-is admirably illustrated by the
phenomenon of perceptual constancy. In the case of color this is the
phenomenonwhereby I experience an object to be the same color in various
lighting conditionseven thoughthese conditionschange the way I experience
the color. For instance, I see the color of my entire office wall to be white,
and indeed the same shade of white, even when some parts of the wall are
better lit than others. At the same time, however, my experience of the
poorly lit section is not the same as my experience of the well-lit section:
one looks better lit than the other. Peacocke uses this phenomenon to great
effect in chapter 1 of Sense and Content9in order to argue that perceptual
experiencehas an essential sensory component,the componentthatcharacter-
izes "what-it's-like-to-have-the-experience".The French phenomenologist
MauriceMerleau-Ponty,on the other hand, uses this phenomenonto argue,
among otherthings, that perceptualexperiencehas an essential informational
component, the component that tells something about the features of the
objectbeing experienced.'
Properly understood, both of these are important aspects of the
phenomenon of perception. It may be, however, that they are not properly
understoodas sensory and informationalcomponentsof experience. Indeed, I
suspect that Peacocke now thinks the conclusion he then reachedin consider-
ing these phenomena-the conclusion that there is an ineliminable sensory
9 See Peacocke, Christopher,Sense and Content: experience, thought, and their relations
(New York: Oxford University Press, 1983). I don't think Peacocke actually groups his
examples under the heading of perceptualconstancy, but the examples he considers-the
color of a wall in different lighting contexts, the size of a tree at different distances, and
the loudness of a car engine when far away or close-are obvious examples of that
phenomenon.
( See the chapterentitled "The Thing and the NaturalWorld,"in Merleau-Ponty,Maurice,
Phenomenology of Perception, tr. Colin Smith (London: Routledge and Kegan Paul,
1962).
606 SEAN Do KELLY
This content downloaded from [Link] on Sun, 08 Nov 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions
component to perceptualexperience-is a faulty one. I suspect this because
the argumentfor that claim dependedupon the premise that representational
content is always conceptual,and of course he doesn't believe that anymore.
But even if the conclusion is faulty, the phenomenonit was meant to explain
is still an importantone to consider. And it's especially importantin this
context because I thinkthat,rightlyconsidered,the phenomenonof constancy
shows why medium-graineddemonstrativeconcepts can't completely capture
the content of [Link] me try to say why.
On my view the phenomenon of perceptual constancy shows us some-
thing crucialaboutthe context dependenceof [Link] partic-
ular, it shows us that the complete and accurate account of my perceptual
experience of the color of an object must contain some referenceto the light-
ing context in which that color is perceived. Without a reference to the
context we won't have the resourcesnecessaryto explain the change in expe-
rience that occurs when the lighting context is varied. If it is right, as all
perceptual psychologists agree, that this change is not a change in color
(hence the name "color constancy"), then no color concept, not even a
demonstrativeone, could completely describe the content of a color experi-
ence. So even if McDowell were to try to explain perceptionin terms of the
medium-grainedconcepts mentioned above, such an explanation would be
inadequatebecause the phrase "thatcolor" is unable to distinguish between
that color as presented in the sun and that same color as presented in the
shade. Because the relevant difference is not a difference in color, no color
term could make such a distinction. Since such a distinction is clearly made
in experience-the color looks different in the sun than in the shade-the
demonstrativeconceptis inadequateto accountfor the experience.
The second kindof dependency-the dependencyof a perceivedpropertyon
the object it's perceived to be a propertyof-is shown by Peacocke's exam-
ple of the height of the window and the height of the arch,11and also by
Merleau-Ponty'sequivalent claim' that "theblue of the carpetwould not be
the same blue were it not a woolly blue." The basic idea is that when I
perceive a propertylike height or color, what I see is not some independently
determinablepropertythat any otherobject could share;ratherwhat I see is a
dependent aspect of the object I'm seeing now. The dependency of the
perceived propertyon the object is so complete 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 rationallywonderwhetherthey are in fact the
same color. I can, of course, satisfy myself that they are the same color by
See Peacocke, Christopher,"PerceptualContent," in Themesfrom Kaplan, ed. Almog,
Perry, and Wettstein (New York: Oxford University Press, 1989).
12 I see these examples as equivalent,both pointing to the second dependency. But it may be
that Peacocke is making a different point when he uses the example in "Perceptual
Content".
THE NON-CONCEPTUALCONTENTOF PERCEPTUALEXPERIENCE 607
This content downloaded from [Link] on Sun, 08 Nov 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions
measuring the wavelength of the light they reflect, just as I can satisfy
myself that the window and the arch are the same height by measuringthem
with a tape measure. But this doesn't tell me anything about the content of
the originalperceptualexperience, since it's on the basis of the new, measur-
ing experiencethatI come to believe in the equivalence.
Now, if it's really truethat this second type of dependencyobtains,then it
seems to block the possibility of the initial conceptualist line of thought.
Rememberthat this is the line of thought according to which we accept the
claim thatthe most specific concept in the repertoireof the perceivercaptures
her perceptualexperience, while denying Peacocke's intuitionthat perceptual
experience doesn't vary with conceptual [Link] justification for
this denial is found in the primafacie plausibleclaim thatthe paintersees the
scarlet scarf in terms of its resemblance to a certain color chip, not some
other. But if a perceived color isn't describable independentlyof its object,
then it must be false that the painter's perception of color is explicable in
terms of resemblanceto an objective [Link] all, 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. The point is much like that made
above concerning context. A demonstrativeconcept like "that scarlet" can
only pick out one scarlet among others. But the difference between the
experience of the scarletscarf and the experienceof the scarletsteel ball is ex
hypothesi not due to a difference of color (this shade of scarlet versus that
shade of scarlet), but rather is due to a difference in the object seen to
manifest that color. No color term alone could make that distinction.
If these two observations about perception are right, then demonstrative
concepts are too coarse-grained,not too fine-grained,to capture perceptual
content. Concepts, even demonstrativeones, pick out situation independent
features,but the perceptualexperienceof a propertyis always dependentupon
the two aspects of the situationI mentionedabove-context and object. It is
still open to the conceptualistto arguethatperceptualcontent is explicable in
terms of the conjunction of a variety of demonstrativeconcepts-one that
picks out the property,one that picks out the object that manifests that prop-
erty, and then a large set of demonstrativeconcepts that picks out the relevant
featuresof the context in which the propertyis being perceived. But it seems
as though this last set will present a sticking point, since there could be an
indefinitely large numberof relevant contextualfeatures, and which features
of the context are relevant will change from situation to situation. This
seems to me a more likely reason thatperceptualcontent is non-conceptual-
because it's situation dependent,and situations aren't specifiable in concep-
tual terms.
608 SEAN D. KELLY
This content downloaded from [Link] on Sun, 08 Nov 2015 [Link] UTC
All use subject to JSTOR Terms and Conditions
