zyxwvutsrqpo
zyxwvutsrqp
zyxwvutsrq
zyxwv
zyxwvutsrqp
Mind & Language ISSN 0268-1064
Vol. 9 No. 4 December 1994
@ Basil Blackwell Ltd. 1994, 108 Cowley Road, Oxford OX4 l / F ,
UK and 238 Main Street, Cambridge, M A 02142, USA.
Article
Peacocke’s Argument Against the
Autonomy of Nonconceptual
Representational Content
zyxwvu
JOSE LUIS BERMUDEZ
The idea that there might be mental states with nonconceptual representational content is receiving an increasing amount of discussion among
philosophers of mind, and it is potentially of considerable theoretical
importance to psychologists. This paper considers an argument that has
been put forward by Christopher Peacocke against the suggestion that
nonconceptual content is autonomous-that is, against the possibility of
a creature being in states with nonconceptual representational content
although not possessing any concepts at all. Peacocke’s argument, if sound,
would circumscribe the theoretical uses to which the notion of nonconceptual representational content could be put.
The paper is divided into five sections. In the first section I outline
some of the motivations that philosophers and psychologists might have
for endorsing a notion of autonomous nonconceptual representational
content. In the second section I give a more specific characterization of
the notion, together with a sketch of the specific form of nonconceptual
content that Peacocke puts forward. In the third Peacocke‘s argument
against the autonomy of nonconceptual content is presented. In the fourth
section it is argued that Peacocke’s argument effectively destroys the
distinction between conceptual and nonconceptual content, while the last
section considers two possible responses that a defender of Peacocke’s
position might make.
I am very grateful to Julie Jack and an anonymous reader for comments on earlier
versions.
Address for correspondence: Faculty of Philosophy, Sidgwick Avenue, Cambridge CB3
9DA, UK
zyxwvu
zy
zy
zyxwv
zyxwv
The Autonomy of Nonconceptual Representational Content
403
1. Theoretical Uses for the Notion of Nonconceptual
Representational Content
The general idea underlying the notion of nonconceptual representational
content is that it is theoretically legitimate to refer to mental states that
represent the world but that do not require the bearer of those mental
states to possess the concepts required to specify the way in which they
represent the world. Alternatively put, a particular content is taken to be
a nonconceptual content iff it can be attributed to a creature without
thereby attributing to that creature mastery of the concepts required to
specify that content.’
One very general motivation for introducing a notion of nonconceptual
content is the thought that in many cases an adequate description of the
content of experience will be more detailed and fine-grained than it could
possibly be if given only in terms of concepts possessed by the experiencing subject (Peacocke, 1992, ch. 3). There seem good grounds, for example,
for describing a perceiver as seeing something as magenta-coloured, without that requiring that he possess the concept of magenta. He can perceive
the particular colour even though he has no conceptual grip on it. The
same applies to shape. Many would hold that it is legitimate to describe
someone as seeing a chiliagon or a rhombus, even though he has no
conceptual understanding of what a chiliagon or a rhombus might be.
The suggestion is even more pertinent in the case of the modalities for
which very few people have a developed descriptive vocabulary, such as
taste and smell. Here it seems clear that restricting the characterization of
experience to concepts actually possessed by a perceiver will often succeed
only in picking out a range of possible perceptual experiences, rather than
fixing uniquely on the particular perceptual experience in question.
Another very general reflection on the nature of perceptual experience
that might make the notion of nonconceptual content seem tempting is
that it offers a way of accommodating the insight that the nature of
perceptual experience is partly determined by the concepts brought to
bear in that experience without falling into the extremes of perceptual
relativism. S o , even if one holds that two observers with different conceptual abilities in the same situation (say, a young child and an ecclesiatic
confronting a religious painting) will have relevantly different perceptual
It should be noted that this paper is concerned only with ascriptions of content at
what might be termed the personal level. The notion of nonconceptual content is
also employed in the context of subpersonal computational states, the idea being,
for example, that early visual processing involves representations of the environment
that do not implicate mastery of the relevant concepts (e.g. the concept of binocular
disparity). Although I do think that the notion of nonconceptual content can be
legitimately employed in the subpersonal context, my interest, and that of Peacocke
in the argument under discussion, is with the content of experience.
0Basil
zyx
zyxwvutsrq
Blackwell Lfd. 1994
404
zyxwvutsrq
zyxwvut
zyxwvu
Mind & Language
experiences, one can do justice to the intuition that at a basic level they
both represent the world in similar ways by placing the similarities at the
level of nonconceptual content and the divergences at the level of conceptual content.2 By the same token, the notion of nonconceptual content also
allows one to capture any continuity one might think there is in the way
the world is represented in experience before and after the acquisition of
a concept.
Perhaps the most obvious advantage of the notion of nonconceptual
content is the promise it offers for explaining what it is to possess concepts
and to be in states with conceptual content. It is important to distinguish,
however, between two broad forms that such an explanation might take.
The first form is what one might term a developmental explanation of
conceptual content. In such an explanation the acquisition of the capacity
for being in states with conceptual content is explained in terms of a
developmental progression over time from being in states possessing
nonconceptual content. On this sort of view, nonconceptual content comes
first and conceptual content emerges from it. Cussins (1990) gives an
account of conceptual content along these lines, suggesting that the gradual
development of a conception of objectivity should be understood in terms
of the progressive construction of concepts from a basis of structured
nonconceptual content-what he describes as ’the transition from a preobjective stage where no concepts are possessed to an objective conceptexercising stage’ (p. 409). The possibility of a developmental explanation of
conceptual content has obvious relevance to developmental psychologists
studying the ontogeny of concept development, as well as to evolutionary
biologists considering the phylogeny of conceptual abilities.
A different way in which conceptual content can be grounded in nonconceptual contents is offered by Christopher Peacocke in various writings
(most comprehensively in Peacocke, 1992). He suggests that what it is to
possess a particular concept (i.e. an explanation of the possession conditions
for that concept) can in certain fundamental cases be explained in terms
of more primitive representational contents, which are egocentric and
nonconceptual. The general idea is that the possession conditions of some
basic concepts involve the subject’s being willing to apply the concept
when in a perceptual state with the appropriate nonconceptual content.
Using the notion of nonconceptual content in this way allows non-circular
characterization of the possession conditions for the relevant concept,
because the subject’s possession of the concept is not mentioned in specifying the content of the appropriate perceptual experience.
These two proposed forms of explanation involve differing degrees of
zyx
zyxwv
zyxwv
zyxwvutsr
zyxwvutsrq
Of course, there are other ways of accommodating this point. One could maintain,
as McGinn does (1989, pp 59ff), that the similarities in perceptual experience are a
function of the basic observational concepts they have in common. Alternatively,
one could hold that the similarities are qualitative features of the experience, rather
than representational features.
0Basil
Blackweil Lfd. 2994
zyxwvu
zyx
zyx
The Autonomy of Nonconceptual Representational Content
405
commitment to what we can, following Peacocke (1992, p. 90), term the
Autonomy Thesis (AT):
The Autonomy Thesis (AT) It is possible for a creature to be
in states with nonconceptual content, even though that creature
possesses no concepts at all.
A philosopher who holds that the notion of nonconceptual content can
provide a developmental explanation of conceptual content is likely to be
sympathetic to AT, to the extent that his project is one of explaining how
a creature possessing only states with nonconceptual content can develop
into a fully-fledged concept-user.3 A philosopher committed only to possession conditions explanation, however, does not have to accept AT
(although there is no requirement to deny it). He can, for example, maintain
that the nonconceptual content of experience will yield a non-circular
account of possession conditions for certain primitive concepts, even if
AT is denied. What makes this possible is, of course, that nonconceptual
content is defined as content that does not require mastery of any of the
concepts required to specify it-it is not defined as content that requires
mastery of no concepts whatsoever. Indeed, one can be in a state with
nonconceptual content despite possessing the relevant concepts. The point
is that being in such a state is not dependent on possession of the concepts
required to specify it.
On the other hand, however, if AT is rejected, then it will be ips0 fucto
impossible to give at least one type of developmental explanation of
conceptual content in terms of nonconceptual content-namely, the type
of developmental explanation that involves explaining how a creature in
states with only nonconceptual content, such as for example a newborn
human infant, can develop into a fully-fledged concept-user. By the same
token, it will also rule out an account of the phylogenetic development of
conceptual content of the sort required by any account of conceptual
content that sees it as appearing at a far later state of evolution than
nonconceptual content.
In addition to this, moreover, it should be clear that rejecting AT rules
out a further potential theoretical use for the notion of nonconceptual
content. Some philosophers and psychologists have been struck by the
thought that a primitive form of intentional explanation is required to
account for the behaviour of creatures one might not want to describe as
concept-using, and obviously any form of intentional explanation requires
attributing to such a creature representations of its environment. One
area in which this emerges is animal learning theory (see, for example,
zyx
There seem, however, to be possible versions of developmental explanation that are
compatible with denying AT. For example, it could be argued that, although in
general nonconceptual and conceptual contents form a holism, there are certain
conceptual contents that develop out of nonconceptual contents.
@ B u d BIackweii Ltd. 1994
406
zyxwvutsrq
zyxwvut
zyxwvu
z
Mind b Language
Dickinson, 1988, where it is argued that certain cases of instrumental
conditioning in rats support an intentional interpretation; see also Premack
1988), but it also seems highly relevant to the study of infant cognition.
The suggestion that there are experiential states that represent the world
but that do not implicate mastery of the concepts required to specify them
is potentially very important here. Clearly, however, it can only be of
theoretical use if AT is accepted.
zyxwv
zyxw
2. Peacocke’s Scenario Content as a Form of Nonconceptual
Content
In giving a more detailed characterization of nonconceptual content it is
useful to start from Cussins’ (1990) discussion. He begins with a distinction
between conceptual and nonconceptual properties. A property is a conceptual property iff it is canonically characterized, relative to a theory, by
means of concepts which are such that a creature must have those concepts
in order to instantiate the property. As Cussins points out, properties are
characterized relative to a theory, and by ’canonical characterization relative to a theory’ he means characterization in terms of properties that that
theory takes to be essential to it. The notion of a nonconceptual property
follows easily. A property is nonconceptual iff it is canonically characterized, relative to a theory, by means of concepts that a creature need
not have in order to instantiate the property. The ‘need not have’ is
important here. Even if the creature acquires the relevant concepts, that will
not transform the property into a conceptual property. The specification of
a nonconceptual property requires only that possession of the relevant
concepts is not necessary.
The difference between conceptual and nonconceptual properties is
straightforward. Consider the two different properties of (a) having a
central nervous system, and (b) understanding the functioning of the central nervous system. Clearly, the first is instantiated by innumerable creatures that lack the concept central nervous system. Equally clearly, it is hard
to see how a creature could be ascribed the second unless it had the
concepts of understanding, functioning and central nervous system. Much
depends, however, on the theory relative to which the property is being
canonically characterized. There are certain theories on which possession
of those concepts would be sufficient-perhaps ’naive biology’ would be
an example here. On the other hand, there are other theories (physiology,
for example) on which possession of those concepts would be necessary,
but far from sufficient.
Once the distinction between conceptual and nonconceptual properties
is granted, it is only a short move to Cussins’ distinction between conceptual and nonconceptual content. Conceptual content is ’content that consists of conceptual properties’, and by the same token, nonconceptual
content is ’content that consists of nonconceptual properties’. In both cases
0 Basil
Blackwell Lld. 1994
zyxwvu
zy
zy
The Autonomy of Nonconceptual Representational Content
407
the properties are being canonically characterized relative to the theory of
content. This is how Cussins fleshes out the basic idea that nonconceptual
content is content that can be ascribed to a creature without thereby
ascribing to it possession of the concepts involved in specifying that
content.4
So, a particular content is nonconceptual iff it can be specified wholly
in terms of nonconceptual properties, where a nonconceptual property is
a property that can be canonically characterized relative to a theory by
means of concepts none of which need to be possessed by a creature
satisfying that property.
With the general notion of nonconceptual content sharpened up we can
move on to a specific proposal that has been put forward for understanding
the nonconceptual content of perceptual experience. Christopher Peacocke’s conception of scenario content explains the nonconceptual representational content of an experience in terms of a spatial type (1989,
pp. 8-9):
zyxw
The idea is that specifying the content of a perceptual experience
involves saying what ways of filling out a space around the origin
with surfaces, solids, textures, light and so forth, are consistent
with the correctness or veridicality of the experience. Such contents
are not built from propositions, concepts, senses, or continuant
material objects.
Such spatial types are specified in terms of a labelled origin and axis
egocentrically centred on the perceiver. Within the framework yielded by
the origin and axis, specifying how space is ’filled out’ is done in terms
of points (pp. 9-10).
We need, for each point (strictly one should say point-type) identified by its distance and direction from the origin, to specify
whether there is a surface there, and if so what texture, hue,
saturation, brightness, and temperature it has at that point,
together with its degree of resistance to touch. The orientation of
the surface must be included.
There remains an ambiguity of scope. Should the definition of a nonconceptual
property be taken in a strong or weak sense, where the strong reading means that
the subject is not required to possess any of the concepts involved in characterizing
the property, and the weak reading means that the subject is not required to possess
all of the concepts involved in the characterization. I take it that the strong reading
is correct, and that requiring the possession of even one of the concepts involved in
the canonical characterization of a property is sufficient to make that property count
as a conceptual property. I also take it that the strong reading is equally appropriate
at the level of content. If the specification of a content involves just one conceptual
property, it will ips0 facto count as a conceptual content.
0Basil
Blackwell Ltd. 1994
zyx
408
zyxwvuts
zyx
zyx
z
Mind 6 Language
The conceptual resources with which such a specification of the spatial
type is carried out are not, of course, attributed to the perceiver, and this
is what makes it an account of nonconceptual content. The account is
completed as follows. This spatial type, which Peacocke terms a scenario,
yields the nonconceptual content of the experience when its axes and
origins have been assigned places and directions in the real world, and
a time has been all~cated.~
This content is correct iff the volume of the
real world around the perceiver, specified according to the same constraints
as the scenario, instantiates the type specified by the scenario.
The specific use to which Peacocke puts his notion of scenario content
is in specifying the possession conditions of concepts. Such a specification
takes two parts. The first part of the specification for a particular concept
yields the constraint that the subject be in a perceptual state with the
appropriate scenario content, and that he be willing to apply the concept
when in that state. The second part is a version of the Generality Constraint
first suggested by Evans (1982, §4.3)-namely, that the subject be capable
of a generalized concept application. Broadly speaking, the thought is that
if a creature is properly to be credited with the thought a is F, then that
creature must be capable both of thinking a is G for any property of being
G of which he has a conception, and of thinking b is F for any object b
of which he has a conception.
3. Peacocke’s Argument Against AT
From the description in the previous section it should be clear that Peacocke is committed neither to accepting nor denying AT. Nonetheless he
does deny it, in favour of the claim that ’conceptual and nonconceptual
content must be elucidated simultaneously. The most basic elements of
the scheme form a local holism’ (1992, p. 91). I will take the liberty of
schematizing his argument, which is presented in a very condensed form
(1992; p. 90):
(1) Scenario content is genuine spatial representational content.
(2) The attribution of genuine spatial representational content to a
creature is only justified if that creature is capable of identifying
places over time.
(3) Identifying places over time involves reidentifying places.
zyxw
zyxwvutsr
Or rather: it does so in conjunction with a further level of nonconceptual representational content, which Peacocke terms ’protopropositional content’ (1992, ch. 3). The
protopropositionalcontent of experience specifies the relational aspects of experience.
I am not mentioning it specifically because it does not affect the argument of this
paper. However, when the term ‘scenario content’ is employed below, this should
not be taken to exclude protopropositional content.
@ Basil Blackwell Ltd. 1994
zyxwvuts
zyxwvu
zyx
zyxwv
zyxw
The Autonomy of Nonconceptual Representational Content
409
(4) Reidentifying places requires the capacity to identify one's current
location with a location previously encountered.
(5) Reidentifying places in this way involves building up an integrated
spatial representation of the environment over time.
(6) Neither (4) nor (5) would be possible unless the subject possesses
at least a primitive form of the first-person concept.
(7) Therefore, nonconceptual content cannot be autonomous.
Several of the stages in this argument seem relatively unproblematic.
(1) can be provisionally accepted, although it would be helpful to have a
clearer sense of what is to count as spatial representational content, given
the claim in (2). (3) is a familiar claim, associated in particular with Strawson (1959). However, (4) and (5) could do with further discussion.
It is important to realize that (4) is making a substantive claim, rather
than just explaining what it is to be capable of reidentifying places.
Peacocke is claiming that genuine spatial content rests on a particular type
of place reidentification, and, conversely, that there are capacities that
some theorists might describe as capacities for place reidentification but
that a creature could have without it being appropriate to attribute genuine
spatial content to that creature. As he stresses, 'spatial content involves
more than just a sensitivity to higher-order properties of stimulation
patterns' (1992, p. 90), and the type of place reidentification implicated in
genuine spatial content is precisely one that cannot be explained in terms
of such sensitivity. Of course, this all depends upon how sensitivity to
higher-order stimulation patterns is understood, but a plausible suggestion
here would be that behaving in a way that merely reflects such a sensitivity
is behaving in a way that could be explained in terms of stimulus-response
(S-R) theory. (4)should be read as claiming that there is more to the sort
of place reidentification implicated in genuinely spatial content than a
minimal capacity to find one's way back to a given place, because an SR account could be given of such a minimal capacity (as it often is in
accounts of the spatial abilities of animals). Place reidentification requires
the exercise of certain cognitive as well as navigational abilities. It involves
something like a conscious registering of particular locations as locations
that have been previously encountered.
It is because so much is built into the idea of place reidentification that
Peacocke claims in (5) that an integrated representation of the environment
over time is also involved. An argument from Sense and Content (Peacocke,
1983) can shed light here. In Sense and Content his argument against
genuine spatial content being attributable to a creature whose behaviour
is explainable in S-R terms is that, however complicated and apparently
purposeful S-R behavioural response might be, allowing genuine content
to be ascribed on the basis of those responses would be to admit the
possibility that a creature might bear propositional attitudes to just one
place in their environment, while not bearing propositional attitudes to
any others (1983, p. 65). This Peacocke holds to be unintelligible because
zyxwvu
@ Basil B h c k w e / / Ltd. 1994
410
zyxwvuts
zyxwvu
zy
zyxw
Mind 6 Language
places are essentially related to other places and place identification is a
holistic phenomenon. Genuine spatial content involves a grasp of the
connectedness of space and that is why it involves some form of integrated
representation of the world over time.
The crucial step in the argument is (6), the claim that place reidentification involves a primitive grasp of the first-person concept. There seem
to be two ways in which this could be argued. The first would be that
the conscious registration of places requires the capacity to entertain
thoughts or proto-thoughts of the sort 'I have been to this place before'.
It is in virtue of this capacity that place reidentification is to be distinguished from simply finding one's way back to a place, and it seems
plausible that entertaining such proto-thoughts requires possessing the
first-person concept. A second line of argument here would be that one
can build up an integrated representation of the environment over time
only if one has a grip on one's own location in it. An integrated representation of the environment is only possible if one is capable of a representation of one's actual and previous locations, as well as of understanding
the role that changes in one's own location contribute to one's changing
perceptual experience. None of these would be possible, Peacocke would
argue, in the absence of some form of the first-person concept.
There might seem to be a difficulty here, however. The defence of (4)
that has been attributed to Peacocke rests on a claim about propositional
attitudes-namely, that it is impossible to bear propositional attitudes to
just one place. But why, one might ask, should this be thought relevant
to a discussion of scenario content? Scenario content is supposed to be
nonconceptual content, and it is often thought (surely correctly) that one
of the distinctive features of nonconceptual content is that the constraints
appropriate at the level of conceptual content just do not apply.6 There
seems to be no contradiction in granting that one cannot have attitudes
to just one place, without this carrying any implications for the possibility
of ascribing genuinely spatial scenario content.' In which case, then, there
zy
zyxwvut
zyxwvu
'
This is certainly the way that Evans saw the matter in his seminal discussion of
nonconceptual content in The Vurieties of Reference. 'It is one of the fundamental
differences between human thought and the information-processing that takes place
in our brains that the Generality Constraint applies to the former but not the latter.
When we attribute to the brain computations whereby it localises the sounds it
hears, we ips0 fucto ascribe to it representations of the speed of sound and of the
distance between the ears, without any commitment to the idea that it should be
able to represent the speed of light or the distance between anything else.' (Evans,
1982, p. 104, n.22) Admittedly, Evans is discussing subpersonal nonconceptual content,
whereas Peacocke's nonconceptual content is a personal-level notion, but the point
holds in virtue of the type of content, rather than the level of processing. The
Generality Constraint is a condition upon concept possession, not upon any type of
mental representation-but see Davies, 1986, pp. 145-6.
Although the traditional notion of content holds it to be whatever is specified in a
'that' clause when attributing a propositional attitude (from which it would seem to
follow that constraints on propositional attitudes are ips0 fucto constraints on content),
Peacocke characterizes content in terms of states that have 'correctness conditions'-
Q Basil Blackwell Ltd. 1994
zyxw
The Autonomy of Nonconceptuai Representational Confenf
zyx
411
seems to be a confusion of levels involved in the argument against ATa constraint appropriate to the level of conceptual content is being applied
at the level of nonconceptual content. Why does this not amount to making
nonconceptual content into a form of conceptual content?
I suspect that Peacocke would respond that this is not a confusion of
levels at all. Rather, it is precisely a sign of how conceptual and nonconceptual content need to be elucidated simultaneously. Because scenario content
is genuine spatial content, its conditions of possibility involve the perceiver’s being capable of holding propositional attitudes to more than one
place. In the absence of this capacity there can be no genuine spatial
content at all. And that, Peacocke would claim, is a further reason why
AT is unacceptable. Rejecting AT in this way, he would stress, does not
involve tacitly transforming scenario content into a type of conceptual
content. The explanatory distinctiveness of nonconceptual scenario content
is preserved, even though its autonomy is denied.
I think, however, that although this objection does not work, it is on
the right lines, and that a successful response to Peacocke’s argument
against AT will proceed by showing that Peacocke’s argument obliterates
the distinction between conceptual and nonconceptual content. In the next
section I would like to outline such a response.
zyxw
zyxwvut
4. Is Scenario Content Nonconceptual Content?
NonconceptuaI content is a type of content that can be attributed to a
creature without thereby attributing to that creature a mastery of any of
the concepts involved in specifying that content (I will henceforth abbreviate this to ’relevant concept mastery’). What I would like to suggest,
however, is that Peacocke’s argument against AT effectively makes it a
condition of attributing scenario content to a creature that that creature
should have mastery of at least one of the concepts involved in canonically
specifying any possible scenario content. In which case, then, there can
be no nonconceptual scenario contents.
Consider stage ( 6 ) in the argument against AT. The claim there is that
the reidentification of places requires a rudimentary form of first-person
thought. The conscious registration of places as places that have already
been encountered requires the capacity to entertain thoughts or protothoughts of the form ‘I have been to this place before’, and this is conditional upon mastery of the first-person concept, as is the reasoning about
one’s own location involved in building up an integrated representation
of the environment. It might initially seem tempting to assume that this
mastery of the first-person is itself sufficient to contravene the requirement
zyxwvu
zyxwvutsrqp
where having correctness conditions is a sufficient condition for representing the
world in a particular way.
0B o d
Blackwell Lid. 1994
412
zyxwvuts
zyxwvu
Mind 6 Language
that scenario content attribution not involve attribution of relevant concept
mastery. If scenario content is specified in terms of an egocentrically
centred origin and axis, then, it might appear, the first-person concept
will be involved in that specification. If so, however, scenario content
would be conceptual rather than nonconceptual content, because it would
only be available to creatures on the condition that they possessed one
of the concepts involved in the specification of any possible scenario
content. This would not be quite right, however. The fact that the origin
and axis are egocentrically centred on the perceiver does not mean that
they have to be specified egocentrically. It is perfectly possible to specify
an egocentric spatial framework without employing any first-person
notions. As Evans puts it, an egocentric spatial vocabulary refers to points
in public space (1982, p. 157). The distinction between egocentric and
public space is at the level of sense rather than of reference, and so it
remains possible to give a specification of egocentric space that does not
involve first-person notions. It might, for example, proceed with reference
to impersonally specified body parts.
This does suggest a further line of argument, however. Suppose, for
example, that we stress other aspects of the putative proto-thoughts
involved in place reidentification. It has been suggested that such a protothought will be of the form ‘I have been to this place before’. Clearly, the
possibility of this type of thought requires attributing to the subject not
just the capacity for primitive first-person thought, but also a mastery of
the concept of ‘place’. The first-person component here is relatively
innocuous, but the notion of a place seems more problematic. The concept
of a place is part of a family of related concepts, such as ’location’, ’point’,
‘position’, etc. Mastery of any one of them cannot be divorced from mastery
of the others, so that attributing to a thinker the concept of one of them
is effectively to attribute to him the concept of any of them. Any specification of scenario content will employ at least one of these concepts, most
prominently the concept of a point, as should be clear from the outline
of scenario content given in Section 2.
This creates a significant tension within Peacocke’s characterization of
scenario content. According to the argument against AT, if a subject is
to have experiences with scenario content he must be capable of place
reidentification, and that requires possessing the concept of a place, or
one of the related concepts in the family. At the same time, however, it
follows from Peacocke’s description of how scenarios are fixed that one
or more of these concepts must feature in any specification of a scenario
content. This makes scenario content into a form of conceptual rather than
nonconceptual content, because any subject who is in a state with scenario
content will be required to possess at least one of the concepts required
to specify that content.
If this is so, then the rejection of AT creates problems for Peacocke’s
position. On the one hand we have the general requirement on nonconceptual content that it can be attributed to a creature without thereby attribu0 Basil
Bfackwell Lid. 1994
z
zy
zyxwvu
zy
zyxwv
The Autonomy of Nonconceptuaf Representational Content
413
ting to that creature mastery of any of the concepts required to specify that
content. On the other hand, however, we have the specific requirements
for scenario content to be genuine spatial content, and these specific
requirements entail that the subject possess at least one concept involved
in specifying any possible scenario content. This tension will not be an
easy one to resolve. Nonetheless, there are several lines of defence available
here, which need to be considered before the conclusion is drawn that
Peacocke’s rejection of AT effectively transforms scenario content into a
form of conceptual content.
5. Objections Considered
(1) An initial response might be to deny that place reidentification
requires the concept of a place. The sort of proto-thoughts implicated in
place reidentification do not, the defence might run, necessarily involve
mastery of a concept from the relevant family. It is worth pointing out,
however, that this strategy is a dangerous one for anybody concerned
with denying AT, because what must at all costs be avoided is making
the proto-thoughts so innocuous that they no longer require possession
of a primitive form of the first-person content, and hence do not permit
the move to (6) in the argument against AT. This certainly rules out the
option of construing place reidentification as a practical capacity manifested in sufficiently sophisticated navigational abilities. Bearing this in
mind, then, it would seem that the only way of construing the protothoughts would be via some sort of demonstrative-perhaps of the form
’I have been here before’. The difficulty with this, however, is that it
seems to contravene Peacocke’s claims about the holistic nature of place
identification. The argument against AT has depended crucially upon the
idea that it is unintelligible that a subject should have thoughts about
just one place in his environment, because places are essentially spatially
related to other places. Consider, however, a subject only capable of
demonstrative thought about places of the form ‘I have been here before’.
That subject would be incapable of having thoughts about more than one
place in his environment at any given moment, thereby falling foul of the
intelligibility requirement.
(2) It was argued in Section 4 that Peacocke’s argument against AT
made the possibility of being in states with scenario content dependent
upon possession of a concept necessarily involved in the specification of
any possible scenario content. The conclusion drawn from this was that
that was sufficient to make scenario content conceptual content. The previous reply attacked the initial claim. It could also be objected, however,
that the conclusion rests upon a crucial equivocation in the conception of
conceptual content.
zyxw
zyxwvu
zyxwvutsr
When it was argued in Section 7 that scenario content was conceptual
@ Basil Blackwell Ltd. 1994
414
zyxwvuts
zyxwvutsr
zyxwvu
zyx
Mind 6 Language
content, the operative notion of conceptual content was this: a given
content is conceptual just in case that content can only be canonically
specified by means of concepts at least one of which must be possessed
by any subject who is in a state with that content. The notion of possession
here is ambiguous, however. On one reading the requirement could just
be that the subject has to possess the concept, irrespective of whether or
not it is employed in the relevant experience. Alternatively, it could mean
not only that the subject has to possess the concept, but also that his
possession of the concept must play an explanatory role in that experience
being the experience that it is. With this distinction in mind, then, it
could be pointed out that the argument in Section 4 depends upon the
first reading-whereas
if the second reading is adopted it fails to go
through. The point is this. The necessity of possessing the concept of a
place for scenario content to be possible should not be taken to imply
that any experience with scenario content will involve the subject’s
employing the concept of place. Rather, the thesis is much more general.
What it amounts to is the idea that experiences with scenario content
are only available to creatures capable of more general forms of spatial
representation (including, for example, thought about places). Although
these more general forms of spatial representation involve the concept of
place, this does not entail that the concept of place is involved in scenario
contents. Because of this scenario content remains nonconceptual content.
The following passage seems to suggest that Peacocke does not clearly
distinguish between these two readings (1992, p. 68):
[I]t is crucial to observe that the fact that a concept is used in fixing
the scenario does not entail that the concept itself is somehow a
component of the representational content of the experience, nor
that the concept must be possessed by the experiencer.8
What makes content conceptual on this account (as opposed, for example,
to Cussins’ 1990, p. 383) is that its appearance in a specification of the
scenario entails its possession by the subject. This is neutral, however,
on the issue raised by the second reading, which makes the stronger
demand that the possession of the relevant concept play an explanatory
role in the representational content of the experience being as it is.
We can begin by considering Peacocke’s position as specified in the
passage quoted. As it stands, Peacocke‘s position does not rule out my
argument for scenario content being conceptual content. Consider the
following reformulation of the argument in Section 4. The fact that the
zyxw
zyxwvutsrq
And also the following: ’There is no requirement at this point that the conceptual
apparatus used in specifying a way of filling out the space be an apparatus of concepts
used by the perceiver himself. Any apparatus we want to use, however sophisticated,
may be employed in fixing the spatial type, however primitive the conceptual
resources of the perceiver with whom we are concerned.’ (1992, p. 63).
@ Basil Blackweti Lid. 2994
zyxwvutsr
zyxwvu
zy
zy
The Autonomy of Nonconceptual Representational Content
415
concept of a place/point can be employed in fixing a scenario entails that
the scenario content is genuine spatial content. The fact that a perceiver
is in states with genuine spatial content entails that the perceiver possesses
the concept of a place/point. This argument certainly seems to satisfy his
conditions upon a content being conceptual content.
But the question remains: can Peacocke retreat to the second reading
of conceptual content in the manner suggested earlier? The additional
requirement that this second reading imposes is that the possession of
the relevant content should play an explanatory role in that content’s being
the content that it is. It would seem, however, that this requirement is
also met by the argument in the previous paragraph. Presumably it is
part of what it is for a given scenario content to be the scenario content
that it is that it should be genuinely spatial content.’ But the central claim
of Peacocke’s argument against AT was that a given scenario content was
genuinely spatial content in virtue of the perceiver’s being capable of
place reidentification, and hence in virtue of their possessing the concept
of place. If so, it seems clear that possession of the concept of point/place
is playing the sort of genuine explanatory role required for conceptual
content.
(3) This last response suggests a final defence that a defender of Peacocke might adopt. It is open to him to modify the first premise of his
argument against AT. That premise, it will be remembered, was that
scenario content is genuine spatial content. This could be modified to the
claim that, although scenario content itself does not qualify as genuine
spatial content, only creatures that are capable of being in states with
genuine spatial content can be in states with scenario content. What might
make this move seem appealing is the idea that, once it is denied that
scenario content is genuine spatial content, the argument that scenario
content is conceptual content could be defused along the following lines.
Certainly, it might be argued, scenario content is only available to creatures
capable of being in states with genuine spatial content, and, moreover,
creatures capable of being in such states must possess the concept of a
place. But it is nonetheless important to distinguish between the levels
of scenario content and genuine spatial content. Although the level of
genuine spatial content does indeed require the possession of certain
concepts, this does not entail that the level of scenario content requires
And presumably part of what it is for something to be genuinely spatial content is that
there should be a determinate answer to the question Peacocke poses in Transcendental
Arguments and the Theory of Content: ’Why is it correct to take an experience as
representing the existence of various features, surfaces and solids as at various
distances and directions, rather than as representing the same features etc. as at
angles rotated around the direction of straight by a certain quantity?’ (1989, p. 13)
Clearly, however, the existence of an answer to this question is part of what it is
for the scenario content to be the content that it is.
0 Basil
zyxwvutsr
Blackwell Ltd. I994
416
zyxwvuts
zyxwvut
zyxwvuts
Mind b Language
the possession of those concepts. In which case, the argument that scenario
content effectively becomes conceptual content fails to go through.
One question that anybody arguing in this direction must answer,
however, is precisely what the connection is between the two levels of
content: why, that is, should experiences with scenario content only be
available to creatures capable of experiences with genuine spatial content?
What is needed is a substantive reason for the interdependence of scenario
content and genuine spatial content. And it would seem that the only
non-question-begging way of doing this would be to argue that scenario
content would be impossible in the absence of genuine spatial content.
This, however, amounts to the claim that there are aspects of scenario
content that are made possible by genuine spatial content. Or, in other
words, that certain features of genuine spatial content make possible
certain features of scenario content.l0 Amongst those features of genuine
spatial content will presumably be some that implicate grasp of the concept
of place/point. If this is granted, however, then it becomes possible to
run the earlier argument that possession of the concept of point/place
is playing the sort of genuine explanatory role required for conceptual
content.
6. Conclusion
This places Peacocke in a difficult position. His argument against the
Autonomy Thesis places such strong conditions of possibility upon scenario content that it effectively transforms scenario content into conceptual
content. Apart from making AT rather nugatory, this endangers the project
of grounding the possession conditions of certain primitive concepts in
nonconceptual scenario content. He seems to have a choice between giving
up his attack on AT, on the one hand, and giving up his suggestion that
scenario content is nonconceptual, on the other. Given the crucial role
that nonconceptual scenario content plays in his account of possession
conditions, as well as the more general theoretical uses it has in both
philosophy and psychology, the wisest course of action would seem to be
to abandon his attack on the Autonomy Thesis and admit the possibility
of a creature being in states with nonconceptual content even though not
possessing any concepts at all.
Rejecting the argument against AT, however, means rethinking some
of the issues opened up by that argument. In particular, the existing
account of place reidentification will have to be modified. There seem to
be two theoretical options open to Peacocke here. The first would be to
lighten the constraints that he places upon genuine place reidentification.
lo
Perhaps the answer to the question mooted in n. 11.
0 Basil
Blackwell Ltd. 1994
zyxw
zyxwvu
zy
zyxw
The Autonomy of Nonconceptual Representational Content
417
It is the reflective and disengaged elements in his understanding of spatial
representation that are creating difficulties, and so it might seem sensible
to get rid of them. He could do this by withdrawing the requirement that
any creature capable of genuine place reidentification be capable of spatial
reasoning, and placing much more stress on the importance of navigational
abilities and practical capacities." In terms of his argument against AT,
this would be to reject at least premise (S), thereby blocking the move to
(6) and hence to the conclusion that a primitive form of the first-person
concept is required. And clearly it would also have the effect of not making
possession of the concept of a place a necessary condition of being capable
of genuine place reidentification, thus keeping nonconceptual content
independent of conceptual content. One problem with this approach,
however, is that it might be thought to beg the question of why we are
talking here about content at all. Certainly some sort of account would
have to be given of the correctness conditions of spatial representation.
What would it be to reidentify a place correctly as opposed to incorrectly,
if the connection between spatial representation and spatial reasoning is
severed?
The second general strategy open here would be to try to preserve
this connection between spatial representation and spatial reasoning, but
without making it dependent upon possession of the appropriate concepts.
This would require developing a theory that would allow for a nonconceptual analogue of proto-thoughts like 'I have been to this place before', and
that would explain how such nonconceptual proto-thoughts could feature
in primitive forms of spatial reasoning to build up an integrated representation of the environment over time. In particular such a theory would
have to explain how a creature can represent itself first-personally without
that requiring possession of even a primitive form of the first-person
concept.12 It would also have to explain the possibility of a creature's
reasoning and reflecting about places even though it did not possess the
concept of a place, or any of the associated concepts. In terms of the
argument against AT, if this could be done then it would be possible to
preserve all the steps except for the last two. Although adopting this
strategy would concede the autonomy of nonconceptual content, it would
nonetheless preserve the nonconceptual nature of scenario content.
What both these strategies have in common, however, is that they leave
open the possibility that a creature might be in states with content despite
not possessing any concepts at all, and this, as pointed out earlier, greatly
widens the explanatory scope of the notion of nonconceptual content
within both psychology and philosophy. Peacocke's notion of scenario
content has a vital role to play in the (relatively) narrow project of
zyxwvutsrq
zyxwvutsr
This would be in the spirit of the notion of causal indexicality developed in Campbell,
l2
1993.
I have attempted to make some headway on this problem in my forthcoming
'Ecological Perception and the Notion of a Nonconceptual Point of View'.
@ Basil Blackwell Ltd. 2994
418
zyxwvutsr
zyxwvut
Mind 6 Language
zyx
zy
explaining what it is to possess a concept. But if its potential autonomy
is conceded then it can have a n equally important role to play elsewhere
in the philosophy and psychology of conceptual content.
Faculty of Philosophy
Sidgwick A v e n u e
Cambridge CB3 9DA
Bibliography
Bermudez, J.L. (forthcoming): Ecological Perception and the Notion of a Point
of View. In J.L. Bermudez, A.J. Marcel and N. Eilan (eds), The Body and
fhe Self. Cambridge MA.: MIT Press.
Campbell, J. 1993: The Role of Physical Objects in Spatial Thinking. In N. Eilan,
R.A. McCarthy and M.W. Brewer (eds), Spatial Representation: Problems in
Philosophy and Psychology. Oxford: Basil Blackwell.
Cussins, A. 1990: The Connectionist Construction of Concepts. In M. Boden
(ed.), The Philosophy of Artificial Intelligence. Oxford University Press.
Davies, M. 1986: Tacit Knowledge and the Structure of Thought and Language.
In C. Travis (ed.), Meaning and Interpretation. Oxford University Press.
Dickinson, A. 1988: Intentionality in Animal Conditioning. In L. Weiskrantz
(ed.), Thought Without Language. Oxford University Press.
Evans, G. 1982: The Varieties of Reference. Oxford University Press.
McGinn, C. 1989: Mental Content. Oxford: Basil Blackwell.
Peacocke, C. 1983: Sense and Content. Oxford University Press.
Peacocke, C. 1989: Transcendental Arguments in the Theory of Content. Inaugural
Lecture. Oxford University Press.
Peacocke, C. 1992: The Study of Concepts. Cambridge MA.: MIT Press.
Premack, D. 1988: Minds With and Without Language. In L. Weiskrantz (ed.),
Thought Without Language. Oxford University Press.
Strawson. P.F. 1959: Individuals. London: Methuen.
zyx
0Basil
zyxwvutsr
Blackwell Ltd. 1994