York Gunther has
edited a useful collection of essays on a hot topic in the philosophy of mind: 
the notion of nonconceptual content.  As several of the writers included do
well to remind us, there is no consensus over what exactly nonconceptual
content is.  One rough formulation is that nonconceptual content belongs to
certain inner states of humans and other creatures, states that have content in
so far as they represent other states of affairs, but are such that the subject
does not, or cannot exercise the concepts that would be used in articulating
such content.  Much, though not all of the initial impetus for introducing the
notion of nonconceptual content arose from considerations about perception and
its role in our mental economy.  Perception involves the processing of
information from the environment, is essential for concept acquisition, and
seems on the face of things to be a process enjoyed by creatures that either
lack concepts, or, on some interpretations, have only very rudimentary, "low-level"
concepts. 

Following a
general introduction, the collection is divided into four sections, which
overlap to some extent in the issues covered.  Each section is introduced by a
further brief set of comments.  The first two sections contain most of the
central articles on nonconceptual content that have helped to shape the contemporary
debate.  Part I, Preliminaries, contains selections from Fred Dretske’s Knowledge
and the Flow of Information on sensation and perception; extracts from
chapter six of Gareth Evans’s The Varieties of Reference, on
demonstrative identification, though nothing from his equally important chapter
seven;  and chapter three of John McDowell’s Mind and World.  Part II,
Naturalism and Computation, contains a more recent paper by Robert Stalnaker, ‘What
Might Nonconceptual Content Be?’, Christopher Peacocke’s ‘Scenarios, Concepts
and Perception’, parts of Adrian Cussins’s ‘The Connectionist Construction of
Concepts’, together with a recent postscript, Andy Clark’s ‘Connectionism and
Cognitive Flexibility’, and Jose Bermudez’s ‘Nonconceptual Content: From Perceptual
Experience to Subpersonal States’.

Part III is
entitled ‘The Nature of Experience’ and is headed by a recent paper by Sean
Kelly on the implications of the fact that experiences involve fineness of
grain.  Also included in this section are Tim Crane’s ‘The Waterfall Illusion’,
Michael Martin’s ‘Perception Concepts and Memory’, David Hamlyn’s ‘Perception,
Sensation and Nonconceptual Content’, Michael Tye’s ‘A Representational Theory
of Pains and Their Phenomenal Character’ and one new paper by York Gunther, ‘Emotion
and Force’.  The final part IV contains the debate between Bermudez and Peacocke
on the Autonomy Thesis, that it is possible for a creature to be in states with
nonconceptual content, even though that creature possesses no concepts at all.

I begin with some
general comments about approaches to the question of attributing states with nonconceptual
content, and then offer a few specific comments on some of the more recent
material.

In his useful
introduction, Gunther attempts to clarify the general idea of nonconceptual
content,  and spells out some of the ambiguities and alternatives covered in
the notion.  Gunther does a good job of sorting out several of the distinctions
that can be made in clarifying exactly what determines a state as having
conceptual content, and what it therefore means, precisely, to say that a state
does not conform to those requirements, and hence does not qualify as a
conceptual state.  As he also points out, we need to distinguish between claims
about whether being in a state with nonconceptual content entails that the
subject does not exercise the concepts that would be involved in
articulating the conceptual content, or cannot exercise them.  The
issue, as he expresses it, is what precisely is meant by the ‘non-‘ in nonconceptual
content. 

But there are
deeper problems with the whole notion of nonconceptual content, which some of
the contributors address.  How does our external conception of such perceptual
processes relate to the conception that depends essentially on our first-person
perspective?  Take the case of vision.  Considered from the external,
third-person perspective, there are good reasons for holding that visual
perception involves a series of stages.  Some papers in this collection in
effect treat vision in this way.  One important tradition in cognitive
psychology holds that there is a key intermediate level of processing in which
the sensory input in vision is represented in the form of a low-level map-like
array;  the information at this level is taken up selectively by the higher
level attentional processes, and links with the cognitive processes of memory,
planning, goal choice and the selection of action types.  If an account
like this is broadly correct, then the question arises as to which states
in the information processing chain have nonconceptual content — low-level
sensory arrays, or subsequent, possibly higher level, action-guiding states, or
both? 

Certainly, the
conception of an inner sensory experience that is arrived at from this third-person
perspective is very different from than that derived from considerations
underpinned by the first-person perspective, which appeals to the phenomenology
of perceptual experience.  Dretske, for example, treats perception largely from
the external third person standpoint in his selection.  Admittedly, at one
point Dretske refers to the impression of seeing more than can be ‘consciously
noticed or attended to’, but there is no attempt to show how the two different
conceptions of experience are integrated.  For the main part, his treatment of
seeing is, as it were, "from the outside", considering a person as
one might consider a sophisticated robot.  From this external perspective on
perception, both the low-level sensory array, and higher level certain action
guiding states can be held to involve different kinds of nonconceptual
content.  Such contents contribute, in different ways, towards explanations of
both perceptual learning, and also targeted behaviour.

First-person
considerations about perceptual experience suggest rather different motives for
introducing nonconceptual content.  Hence the appeal to what is distinctive
about perception is not straightforward.  This point is worth spelling out more
fully. 

Suppose, for
example, I see someone catch a ball I throw to them.  I can see that they have
a visual experience, just as I can see that a kettle is boiling.  Such seeing
is in one sense direct, in that my perceptual belief need not be inferentially
based upon any prior belief:  I simply form the perceptual belief that the
other person is visually aware of the ball coming towards them.  If asked
subsequently why I formed the belief I had, I could cite facts about how that
person’s eyes was oriented, how they moved and so on;  but at the time I may not
have formed any conscious beliefs of this kind.  Such perceptual beliefs about
the other’s experience are defeasible;  it is possible, from my epistemic
standpoint, that they did not in fact have any visual awareness of the ball. 
The access we have here to another’s experience is from a third-person
perspective.

We also have a
very different kind of access to perceptual experience, the direct first-person
access to what is going on in conscious experience in our own case.  I may
decide to take up an introspect attitude, and attend to what it is that my
consciousness involves when, for example, I have a visual experience while I am
looking at an object, or while engaged in some activity such as riding a
bicycle, or juggling.  But even without my adopting any special introspective
attitude, there remains the fact that when I perceive an object in my
surroundings, my awareness involves a first-person perspective that contrasts
with the different kind of awareness I have when I perceive the perceptually guided
activity of another person. 

What I am aware of
in my own case involves a distinctive phenomenology that is lacking when I find
out about the perceptual states of another through observing their behaviour. 
Because of the subjective phenomenology, available from the first-person
perspective on experience, I come to understand perceptual experience as
comprising the vivid conscious state that it is.  As Gunther points out,
perceptual experience has a ‘specificity of content, a richness of detail and a
fineness of grain’.  Although such aspects are potentially cashable in terms of
the subject’s ability to behaviourally discriminate and otherwise react to its
environment, the grasp we have of the distinctive nature of perceptual
experience is arrived at primarily because we have this subjective, immediate,
access to our conscious states.

The first person
access we have to experience suggests a certain line of argument for nonconceptual
content .  Just as with perception, we have two different perspectives on our
own beliefs and other intentional states.  As is well-known, there are
competing accounts about the exact processes by which we come to have beliefs
about another person’s beliefs and other propositional attitudes.  But on any
account, how I arrive at direct knowledge of my own thoughts, etc., differs
from the way that I come to know about the thoughts of another.  Again,
first-person and third-person perspectives differ. 

It seems
reasonable to claim that part of my grasp of the content of the concept
of belief involves an understanding of the role of beliefs as connected with
reasoning, planning and action.  But, importantly, I understand pure
intentional states, such as thought and belief, to be very different in kind
from perceptual states and processes.  There may be some degree of overlap: 
experiences, in the wide sense of the term, involve some kind of
conceptualisation, at least at a low level.  When I am conscious of seeing
something, I necessarily see it as belonging to a certain kind, even if the
classification is broad and indeterminate, perhaps merely involving a
disposition to react to what I see as "a roughly round shape".  So in
this inclusive sense, all experience involves low-level classificatory content,
which is manifested in my expectancies, and involves propensities to act in
specific kinds of ways.  But — and this is the crucial point –we abstract the
representational content from the context of the subjective experience.  We
consider the organism from the outside, as a physical mechanism navigating
through its surroundings, by means of inner representations of some kind.  In
so far as we grasp the content of representational states as connected with
patterns of potential behaviour in this objective way, we grasp such content from
an external, third-person perspective.  This purely external notion of content
is what many theorists have in mind when they speak of nonconceptual content.

The key point,
however, is that perceptual experiences involve something more than the
(low level) classificatory content.  Seeing a red patch of colour in front of
me, or hearing a noise as situated on my left is, from a phenomenological point
of view, fundamentally different from merely thinking about a patch of red in
front of me, or thinking that there is some noise source situated on my left
(which I may now no longer hear because I have inserted ear-plugs).  Experiencing
things is fundamentally different from merely being inclined to react to
features of my environment in ways appropriate to them.  There is therefore a
compelling prima facie argument, from the first-person standpoint, for the view
that perceptual experiences involve a form of sensory nonconceptual
content that does justice to the phenomenal aspects of conscious experience. 
(This line of thought is suggested by Wilfrid Sellars’s seminal writings,
especially in his later papers, on the place of sensory states in our
conceptual scheme, of which there is no mention in this collection.)  In
discussions of nonconceptual content it is not always made clear in what way
externally based motivations for introducing the notion of nonconceptual
content are connected with this first-person, subjective perspective.  Indeed,
the external conception of nonconceptual content, content arising from considerations
about the explanation of actions, might less misleadingly be labelled ‘low-level
conceptual content’, to indicate the continuity between such content and
high-level fully conceptual states.  This would help to distinguish it from sensory
nonconceptual content (it is interesting that Peacocke has on occasions
referred to low-level conceptual content of this kind as ‘proto-propositional
content’).

One matter where
the distinction is important concerns the types of content that are in play. 
Many writers seem to equate representational content with intentional content. 
But this equation can be questioned;  it is arguable that intentional content
only arises when there is some kind of conscious classification of experience,
and where such a classification can be directly utilised in the formation of
plans for medium and long term action.  Intentional content is
forward-looking.  In an interesting paper related to this issue, Stalnaker
develops the idea that the representational content of nonconceptual states is
a form of informational content, and can be construed as attaching to
many different kinds of states, including beliefs:  he concludes that ‘different
kinds of states have the same kind of content, but …it is nonconceptual all the
way up.’  However, informational content is backward looking, to be explicated,
if at all, in terms of input under ideal circumstances.  This conception may be
relevant to the content of the sensory aspect of perception, but it is doubtful
whether any clear sense of the notion can be articulated for belief states,
which can be caused in too many different ways.

 The distinction
of perspectives is also relevant to the question of the fine grained character
of visual experience, a feature that provides one motive for introducing nonconceptual
content.  In his contribution, Kelly discusses the debate between Peacocke and
McDowell on fineness of grain.  This originated with Evans’s claim that we can
discriminate more colour shades in experience than we have concepts for.  Our
general colour concepts seem to be more restricted than our experiences. 
McDowell countered Evans’s claim by appealing to the employment of
demonstrative concepts that could be expressed by usages such as ‘that shade’. 
Peacocke’s objection to this idea, as set out in his 1998 comments on McDowell
in Philosophy and Phenomenological Research, rests in part on the fact
that there are too many demonstrative concepts potentially available.  In a
reply to this particular objection, Kelly defends the idea that we could
specify on a principled basis the demonstrative concepts that are relevant to
identifying the fine shade differences in experience.  He does not, however,
deal with the more detailed and persuasive arguments against McDowell’s
position, which Peacocke develops in his 2001 Journal of Philosophy
paper (not included in this collection). 

Kelly then goes on
to produce an interesting argument of his own against McDowell, based upon the
phenomena of colour constancy.  In brief, the thrust of his argument is that
demonstrative references to the objective colour shades possessed by objects
fail to capture the experience of colour, since the same surface can
appear differently in experience under different lighting conditions.  I may
rightly take a surface I perceive to be uniform in colour, even though my
experience of different regions of the surface varies because the surface is
poorly lit in some areas.  There are subtleties here about the extent to which
our beliefs influence how we experience things, but it seems right to say that,
in some sense, experience does not correspond directly with the simple colour
properties of the objects, when these are considered independently of the
surroundings.  Thus Kelly claims, ‘the complete and accurate account of my
perceptual experience of an object must contain some reference to the lighting
context in which that colour is perceived.’  In a further argument, following Merleau-Ponty
(and also Peacocke), Kelly argues on phenomenological grounds that that colour
experience is dependent upon the nature and type of object that the colour is
perceived to be a property of.  Kelly concludes that a demonstrative concept
like ‘that shade’ is inadequate to account for experience, and that therefore
perceptual content is nonconceptual.  

It is not clear to
me whether it might be possible for McDowell to rescue his argument here by
appealing to some more complex kind of demonstrative, one that takes the
context into account.  But in any event, Kelly’s argument on this issue does
not go far enough.  He appears to accept the basic externalist view of
experience defended by McDowell.  Experiences are to be identified in terms of
the actual properties objects (and also perhaps their relations in the context)
surrounding the perceiver.  Yet reflection on further cases such as double
vision shows that this now widely accepted view cannot be correct.  Suppose
I see an apple as double.  For reasons which are independent of any external
features, it may be that the shades I am subjectively aware of, in experiencing
the two images of that object, are different.  The image of the apple on the
left may appear a yellower shade of red than the image on the right (I have a
some kind of jaundice which is affecting the two eyes differently, say).  Hence
it is not true to say that we must identify the shades I am aware of by
reference to any external objects.  I am able to identify them, on a particular
occasion, directly.  I appeal to subjective, phenomenological considerations. 
Of course this is compatible with a much weaker form of externalism, on which
it is held that this general ability is supported by the general
connections that normally hold between external objects and my experiences. 

The only new
complete paper here is Gunther’s own contribution.  He considers the status of
the content of emotional states.  According to Gunther, emotions involve
intentionality, and are irreducible; however, they involve nonconceptual
content, since the emotional content violates the principle of force
independence that is associated with the full blown logical complexity of
conceptual contentful states.

In a postscript
added for this volume, Peacocke revises his views on the autonomy thesis.  Peacocke
now argues that there can be spatial contents that, while having objective
significance, are still nonconceptual — they need not lead on to full
judgements, which would involve the capacity to reflect on the reasons or
grounds for their justification.  There can, therefore, be creatures with
states involving nonconceptual content, but who lack conceptual capacities.

In another added
postscript, Cussins elaborates the idea that nonconceptual content is connected
with our actions.  He spells out the way the notion of how the world can be nonconceptually
presented, as a realm of mediation that enables a subject to negotiate along
activity trails.  These trails are structures in the world that guide us in
carrying out our tasks;  such content is a form of practical knowledge (that
would appear to connect in interesting ways with the idea of the mastery of "sensorimotor
contingencies" that O’Regan and Noë appeal to, in a number of recent
publications).

Unfortunately, Cussins
is unable here to respond to the arguments of Andy Clark, set out in the latter’s
valuable recent paper, ‘Visual Experience and Motor Action: Are the Bonds Too
Tight’, in the Philosophical Review (2001), a paper which discusses
semi-automatic responses involving the on-line control of action.  (As is
inevitable with collections of papers on cutting edge research, some important
papers emerge to late to be included.)  Clark argues on empirical grounds, derived in part from the research of
Milner and Goodale, that there are very different conceptions of nonconceptual
content currently being debated.  He argues forcefully for a distinction
between the action-guiding — apparently non-conscious — notion of nonconceptual
content, and a different notion that reflects our conscious experience. 
Combining Clark’s line of thought with the distinction argued for above on more
a priori grounds, between the representational content of sensory nonconceptual
states and the nonconceptual content of higher level classificatory states,
this all suggests that there are at least three distinct notions of nonconceptual
content that should be taken into consideration in theorising about experience
an action.

Nevertheless,
despite these criticisms, overall this is a valuable collection of articles on
an important issue in the philosophy of mind.  Gunther has done a useful
service to his fellow philosophers in bringing them together.

 

© 2003 Paul Coates

 

Paul Coates,
Philosophy, University of Hertfordshire

Categories: Philosophical