Brit. ]. Phil. Sci.

40 (1989), 83-103 Printed in Great Britain

Perception and Neuroscience

GRANT GILLETT

ABSTRACT
Perception is often analysed as a process in which causal events from the

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

environment act on a subject to produce states in the mind or brain. The role of the
subject is an increasing feature of neuroscientific and cognitive literature. This
feature is linked to the need for an account of the normative aspects of perceptual
competence. A holographic model is offered in which objects are presented to the
subject classified according to rules governing concepts and encoded in brain
function in that form. This implies that the analysis of perception must consider not
only the fact that there is an interaction between the perceiving subject and the
perceived object but also that the interaction is shaped by a system of concepts
which the subject uses in thought and action.

Fire may burn our Bodies, with no other effect, than it does a billet, unless the
motion be continued to the Brain, and there the sense of heat or idea of Pain, be
produced in the mind wherein consists actual Perception. Locke [1689].

Locke's account of perception was a simple one. Events impinged upon the
periphery of a human being, they provoked further events in nervous tissue
which relayed these changes in some form to the brain where the simple events
are formed into complexes and result in conscious experience. It has obvious
conceptual relations to the development of mechanistic neurobiology. In an
evolving form the idea that the mind receives a concatenation of elemental or
primitive sensory items and these generate its picture of reality has dominated
thought in the empiricist tradition.1 It has been allied to the thesis that
consciousness involves access to 'inner' experiences with a certain subjective
quality and led to a 'folk psychology' which is under some strain (Churchland
[1986]). Now that our understanding of the brain and neuroscience are both
developing we are in search of new models that will help us make sense of
puzzling facts such as 'recognition of conspecifics is not just one aspect of the
1
A. J. Ayer espouses this view in The central questions of Philosophy (Penquin: Harmondsworth
1973). Elsewhere I have called this the 'Empiricist Representational Theory' {Representation
Meaning and Thought. Oxford: O.U.P., forthcoming).
84 Grant Gillett
general cognitive business subserving recognition of teacups, dogs, trees and
so forth' (Churchland [1986], p. 224). I want to offer an approach to
perception and its relation to the brain that differs from the traditional views in
what are, I think, promising ways.
Neuroscience uses a framework derived from information processing theory
to give an account of the causal transactions that mediate between an
individual's sensory contact with the world and the execution of bodily
movements, and investigates the way that such informational processes are
carried out in the brain. Philosophical accounts have, to date, contributed to
this burgeoning scientific enterprise little more than the notion that complex
perceived entities such as dogs, trees, teacups and persons are present to

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

consciousness as a result of the combination of more simple elements (which,
for some theorists, are preformed according to linquistic types). By exposing
certain problems in that theory and offering a different theoretical orientation I
hope to make certain points about the conceptual relations between sensation,
receptor function, perception, causation, and thought and show how it is that
these can get so muddled in the discussion of theories of perception. I will apply
and illustrate the view I favour by attempting to undermine the conceptual
foundations of a strong physicalist version of the argument from hallucina-
tion.

All experience consists in events in the subject's brain which usually but not
always convey information about the outside world. The fact that in hallucina-
tion there is no input from the outside world means that what the subject receives
in experience is something which stops short of contact with external events.
Therefore (in as much as the difference is not apparent to him) what he is
presented with generally in experience are not the events and objects of the
outside world which he takes himself to directly perceive but rather events within
his brain which give him indirect knowledge of the external world.

II
Neuroscientific approaches to perception begin with incoming patterns of
physical events which impinge upon receptor mechanisms. These send
patterns of neural impulses to the brain where those patterns undergo certain
transformations in a series of centres each of which modifies the way in which
information within the array is organized. There is a definite influence exerted
on the processing of this information by higher functions. This can be
illustrated by something as basic as attentional selection and focussing on
novel or highly significant stimuli (such as an aberrent member of a series or
one's own name among a list of less meaningful stimuli). The evoked electrical
activity to such stimuli is of a different type to that which is produced in
response to the background stimuli among which they appear. This suggests
 Perception and Neuroscience 85
that the higher recognition which is required to register the complex
configuration of such stimuli has an effect on the events at an earlier point in
the processing or input pathway (including the primary sensory receiving
areas of the cortex). Weiscrantz, in discussing the higher processing of visual
cues suggests that inferotemporal areas of the cerebral cortex select such
stimuli for directed and maintained attention (Weiscrantz [1974]). These
centres are, of course, much further on in the informational flow than the
centres in which incoming stimuli are selectively reinforced so as to claim their
share of cortical processing. Pribram calls this a 'gating' function of the infero-
temporal area which acts so as to change receptive fields and their
responsiveness in the primary visual cortex (Pribram [1971]).

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

The problem is that this backwards effect from higher to lower levels of
processing does not have an echo in the philosophical view that we have
enunciated. An interesting task therefore becomes to devise a philosophical
theory of perception which does justice to these facts. This requires us to give
some account of how the organism 'knows' which stimuli to select. That this
has widespread ramifications in perception is suggested by the remark
'selective attention phenomena play a crucial role even in simple and
immediate perceptual tasks' (Ullman [1986]). Theorists have also invoked
hypothesis testing and similar functions to explain the way sensory processing
is done.
To help these problems the distinction between a 'thin' or structural-causal
and 'thick' or epistemological reading of 'information' is needed. 'Thin'
information is considered as information because it involves structured events
which, by interacting in a certain way, play a defined role in a process itself
understandable in epistemic or representational terms. 'Thin' informational
processes are transactions in cause and effect or dispositional terms and they
involve steps which mechanically follow one another or interact with one
another; no notion of validity or groundedness is involved. The latter notions
imply that besides the result of certain processes or mapping transitions there is
a norm, rule or standard which that result may or may not meet. A thin
informational account in which structured events and patterns of events
causally interact cannot allow this epistemic or evaluative presumption to
creep into the analysis and beg certain explanatory questions which the
analysis may not have answered.2
'Thick' information is conceptually meshed with talk of knowledge,
representation (and also misrepresentation) and judgment. It invokes a
panoply of rational constraints characterizing thoughts about the world. In
this conceptual milieu the notion of what ought to follow from what (and not
merely what the system is disposed to do or what tends to happen) is of central
importance. Discussion of thought introduces the fact that information is
2
Which is not to say that such an analysis cannot answer those questions but merely that the
answering of them must explicitly address the normative problem.
86 Grant Gillett
being evaluated for validity or groundedness in accordance with certain
norms. We require an analysis of the relationship between this latter
conceptual apparatus and the 'thin' informational processes so as to be
completely clear just what we are and what we are not claiming about what
goes on in the nervous system and to avoid sliding imperceptibly over
explanatory lacunae by allowing mental attributions to 'take up the slack' in
the scientific account. I shall, as many cognitive scientists do, use epistemic
metaphors where they seem helpful but will also attempt to illustrate their
conceptual dangers.
Incoming informational events provoke a change in the ongoing pattern of
neural activity in the brain in general and the cerebral cortex in particular. The

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

effects produced then result either in the generation of patterns of motor
impulses (which will ultimately be enacted on muscles or glandular activity) or
in an altered readiness to respond in the brain itself. Given this picture one can
attempt to identify the mental processes concerned by correlations with
introspective data. The paradigm provided by Locke suggests that the gradual
concatenation of simple informational elements into complex wholes may well
allow the detection of staging posts in the mind/brain's construction of reality
(notice that no distinctions are made within the notion of information
operative here). Phenomena such as colour impressions or even a well formed
memory image of sitting in grandmother's parlour were reported by Penfield in
a series of historic experiments on surgical patients (Penfield and Rasmussen
[1950]). Our conception of information processing is now much more
sophisticated but we still find puzzles about the point in processing where
information becomes conscious have a lingering fascination.3 Data from brain
lesions make it possible to locate which bits, if any, are essential for certain
kinds of informational transformation—say the matching of an array to letter
shapes, or the naming of colours—but we do not try to locate a cognitive or
introspectible phenomenon in some piece of the brain in this way. We are
aware that to identify a centre necessary for the carrying out of some function
is only a provisional step in understanding the complexity of the task
concerned. When we do begin to unravel some of the cognitive elements of a
given task we often begin to look at what we have intuitively thought about
that task in a new way. This means that the naive picture we have of many
psychological abilities must be acknowledged as being provisional and
requiring further analysis.
Despite this caution and the development of powerful non-linear processing
models, the philosophical implications of the traditional account seem clear;
the brain/mind builds up a picture of the world on the basis of elements of the
perceptual image which the receptor system has received. The mind processes,
in a mental way, the information ('thick' sense) coming in just as the brain
3
Assuming that we know what 'conscious recognition' is. which is in some doubt (see K. V.
Wilkes 'Is consciousness important' in B. ]. Phil Sri 1985).
 Perception and Neuroscience 87
carries out analogous physical processes which are the realization of these
mental phenomena. As certain features and items are detected and assembled
so the mental (and underlying physical) processes contain information (sense
unspecified) of greater complexity and action-relevance (or what we might call
'a more adequate representation of the world'). This information is processed
in the higher cognitive, conative and executive reaches of the system. In view
of this chain of events, it is tempting to claim, with Blakemore, 'We deceive
ourselves if we think that our perceptual world is complete. It is what our
neurones are able to tell us. . . . Neurones present arguments to the brain
based on the specific features that they detect, arguments on which the brain
constructs its hypothesis of perception.' (Blakemore [1977], p. 90.)

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

In the concluding phrase Blakemore is indicating that a salient feature of
visual information processing, as it is understood in perceptual/cognitive
psychology, is that the perceiver sees things in terms of their relations to what
he normally encounters in the world and not in terms of arrays of 'sensory
elements'. Blakemore, following Gregory, claims that in this sense the
perceiver frames hypotheses about what is producing the patterns of
stimulation impinging upon him.4 Some processing views work toward the
conceptually favoured contents by concatenating simple sensory/receptor
events in some way but such an 'assemblist', or 'stepwise internal processing'
theory of perception is, as the appeal to 'hypotheses' makes clear, not adequate
to fulfil the requirements of cognitive psychology and neuroscience (as Carnap
may have suggested if asked). For instance, a human perceiver tends to close
incomplete figures and see notional figures based on the contextual cues in an
array. This has led to explanations which interpose higher functions—
'hypothesis markers'—in the processing of perceptual information. The
presumption is that the way visual data are organized to form a representation
depends on the kind of representation the perceiver guesses or expects to be
appropriate.
Another problem is that the perceiver, in this picture, is in touch with the
world only via the medium of a number of intervening events. Worries arise
from the fact that epistemic attributions require that the perceiver represent
the object as it really is and thus that there are right and wrong ways of
perceiving an object and these stand in a normative relation to the cognitive
and perceptual processes of the indvidual. Mechanisms within an individual
can only involve complex dispositions to process input in certain ways. But
somehow our understanding of perception must reveal how a perceiver detects
what is there in a way not wholly dependent upon his inclination to react thus
and so but according to categories which dictate how an array ought to be
classified. When we have explained that then we have explained how it is that
thick information is available to him. But here the picture which begins with
4
The 'hypothesis of perception' theory is one borrowed from R. L. Gregory e.g. in 'Visual
Perception and Illusions' in J. Miller, («/.) Slates of Mind London: B.B.C. 1983.
88 Grant Gillett
what the individual 'receives' (in some paraphysiological sense) and requires
him to construct a world out of that, provokes a fiendish set difficulties.

If there were an inner man who looked at the retinal image from inside his head,
his perceptual processes would require just as much explanation as those of the
whole man. Indeed, such a homunculus would have a few problems of his own.
The image that he supposedly looks at is upside down, foreshortened, and the
wrong size; it exists in two slightly discrepent versions, and it changes constantly
as the eyes move. Many sophisticated theories have been advanced to solve these
problems, but real perceivers do not have them. We do not see our retinal images;
we see the real environment of objects and events. (Neisser [1976], p. 16)

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

The problems arise because we take the primitive elements of sensory input to
the organism and make the thinker, armed only with the propensities of his
own computational mechanisms, sort them all out into conceptual thoughts.
'Descartes' dilemma' is the label Blakemore coins for this problem of going from
representations to knowledge about the external world and he agrees with
Neisser that an internal rational homunculus will not do. 'For the source of
knowledge, for the solution to Descartes' dilemma, we must look within the
maps, at the nerve cells of which they are made—the neurons of knowledge'
(op. tit., p. 85). He seeks to elucidate the epistemological nature of concepts and
knowledge by a detailed examination of processes occurring (physically) inside
the subject. In a similar vein, Dennett describes, in outline, hierarchies of
homunculi each composed of stupider, or as he says, 'smaller homunculi but
more important, less clever homunculi' until the clearcut epistemic trans-
actions with which one begins are reduced to simple operations easily
explicable in terms of neuronal states (Dennett [1978], p. 80). But both
Dennett and Blakemore ignore the conceptual consequence of assimilating the
two uses of 'information'.
It is true that systematically structured sequences and networks of neural
events contain thin information from the world. It is not clear that they allow
us to understand thick information about the world. I would contend that we
need to clarify just what is involved in thick information in order to
characterize what it is that neuronal networks are out to achieve and how they
do it, and so elucidate how the brain goes about its cognitive tasks. I will argue
that, when we do this analysis, certain features of perception and thought,
even in the brain damaged, become more understandable.
There is no question that networks (particularly parallel networks) of
neurones can detect configurations of events within stimulus arrays but there
remains a question as to how they settle on just the patterns that fit with
intersubjective rules of use so as to allow a thinker to perceive things as being
thus and so according to concepts he shares with others. Such concepts
essentially have a certain independence of individual cognitive processes.
 Perception and Neuroscience 89

III

Neisser realizes that the subject of experience is not merely a locus of processes
which result from environmental stimulation but in fact makes a contribution
to perceptual knowledge and that any account of perception must make it
possible that he is in direct epistemic contact with his world.
In my view, the cognitive structures crucial for vision are the anticipatory
schemata that prepare the perceiver to accept certain kinds of information rather
than others and thus control the activity of looking. Because we can see only
what we know how to look for, it is these schemata (together with the
information that is actually available) that determine what will be perceived, {op.

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

tit., p. 20)

He adds
The schemata that accept information and direct the search for more of it are not
visual or auditory or tactual, but perceptual. To attend to an event means to seek
and accept every sort of information about it regardless of modality and to
integrate it as it becomes available. We pay attention to objects and events not to
sensory inputs, (op. cit., p. 29)

Neisser is emphasizing that it is the objects and features of the world that form
the intentional objects of thought. Thus a full account of perception must
disclose not only the source of the normative features in perception but also
how they are tied to objects, qua objects, rather than to the sensory input to the
individual. There is, of course, no question that attention involves the selective
processing of information but a conceptual account should reveal why and
how significance is given to certain subsets of information and thus what the
directing schemata amount to. Neuropsychological support for Neisser's view
is offered by Luria who also suggests that language is a crucial factor in
perception.
The process of perception is thus evidently complex in character. It begins with
the analysis of the structure perceived, as received by the brain, into a large
number of componants or cues which are subsequently coded or synthesized and
fitted into the corresponding mobile systems. This process of selection and
synthesis of the corresponding features is active in character and takes place
under the direct influence of the tasks which confront the subject. It takes place
with the aid of ready-made codes (and in particular the codes of language), which
serve to place the perceived feature into its proper system and to give it general or
categorical character. (Luria [1973], p. 229)

(His remarks should ring bells for those who have paid any attention to Kant or
Wittgenstein.) What Neisser and Luria want is some idea how to accommo-
date the 'schematizing' or 'synthesizing' (Kant) knowledge that seems to play a
crucial role in perception. This calls for a place to be found for what Kant calls
90 Grant Gillett
the 'rules of the understanding' whereby the world is presented to the subject
in a structured form apt to ground certain judgments. 5 Luria's appeal to
language is attractive because if we could, in some way, link perceived
contents to semantics then the normative constraints in linguistic meaning
would be at our service in attempting to understand perception and the 'thick'
information involved in it.
Kant's rules of the understanding which impose conceptual form on
experience, Neisser's 'schemata' tied to objects in the world, and Luria's codes
linked to natural language all jostle for a place in an adequate account of
perception. We must also bear in mind the suggestive empirical findings about
higher to lower 'feed-down' mechanisms and the role of the infero-temporal

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

cortex in the direction and maintenance of attention (a faculty which I have
elsewhere located at the centre of all conceptual thought (Gillett [1988])).
Lastly, we need to bear in mind our need to fix the source of categorisations
beyond the dispositions of the individual to react thus and so to presented
stimulus arrays. Ideally, we want to include these elements in our understand-
ing of how thought can relate to the obviously intricate structure of the
neurophysiological processes that underlie it. Even if mental description were
irreducible to physical description, we would still regard brain states as the
causal substrate of mental function6.
Neisser must be right to insist that in perception a thinking subject does not
examine retinal images or other neural phenomena and then build up a
picture of the world outside as causing those phenomena. 7 But, apart from
Neisser's target, who may well be a straw man, the normative consider-
ations—rightness and wrongness independent of the dispositions realized in a
given thinker—imply that we need a basis on which to understand 'schemata'
which is not limited to, though perhaps allows for realization by, neuronal
networks. I will attempt to provide such a basis.

IV
We are, in fact, not bereft of a way to understand the relation between
perception and neuroscience merely because the naive 'processing sequence'
picture is inadequate. Pribram, a psychologist working at the interface of brain
5
Kant says that conceptual thought involves concepts (or functions of unity of judgment) and
rules of the understanding, [Critique of Pure reason (trans. N. Kemp Smith) London: Macmillan
1929, e.g. B171]. When one considers the relations between language, selective attention and
perceptual contact indicated by Luria, demonstrative judgments will be seen to be particularly
important.
6
Such a position is taken both by Davidson in Essays on Actions and Events and C. Peacocke in
Holistic Explanation.
7
Even though the brain and its informational processing mechanisms enable the subject to
represent the world this view is to be deplored, because a private language cannot serve as an
adequate basis for meaning and there is no way to learn what terms to apply to one's essentially
private retinal images.
 Perception and Neuroscience 91
function and cognition formulated, at one stage, an analogy between
holograms and the nature of the mind. He failed to develop the fruitful analogy
he produced perhaps because of the vague 'wave-front' picture of neural
activity on which his exposition of it rested and is lack of a good model of brain
processing to supplement his theory. It is now possible to remedy both of these
defects.
A hologram is an image of the world with three dimensional properties. It is
made by producing interference effects (on a photographic plate) from two
interacting laser beams reflected from the array to be pictured. The plate
reveals no clue as to the image it holds unless the specific type of laser light used
to generate the image is used to recover the information encoded therein.

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

Pribram remarks.
The properties of the hologram are just those demanded by us to account for
ordinary perception. I have already made the suggestion that arrival patterns in
the brain constitute wave fronts which by virtue of interference effects can serve
as instantaneous analogue cross calculators to produce a variety of moire type
figures. Now by means of some recording process analogous to that by which
holograms are produced, a storage mechanism derived from such arrival
patterns and interference effects can be envisioned. This is possible, since
reconstructions of images from holograms have many of the attributes of
perceptions. (Pribram [1969])

Leaving aside the talk about wave fronts etc. which no one has been able to
develop in a convincing way, the analogy, as I want to develop it, looks like
this.

the the laser beam a picture/

laser
world holographic representation
plate

is like

the brain thought/

world , processes . . representation
r
interpersonal interpersonal
behaviour behaviour
In this picture or analogy it is interpersonal activity, or interactions between
communicating thinkers (which for us means human relationships and the
human practices of communication constituting a language) that present the
world to the brain in such a way as the brain can, under their influence, serve
as the basis of mental life. My contention is simple; 'rules', 'schemata' or 'codes'
that give form to thought are realized in the brain and have the structure that
92 Grant Gillett
they do in virtue of the way in which brain function has been shaped in an
interpersonal milieu. Thus the brain is a faithful information processor which
records and, under the right conditions, produces meaningful expression and
behaviour, just as the holographic plate 'records' and can be made to
reproduce three dimensional information.

We can begin by looking at the origin of the concepts which figure in

perception.
Grasping a concept is an ability to link a whole series of experiences in virtue
of some shared feature a subject can detect in or impute to them (Evans [1981],
p. 104 ff.). How does a thinker come to acquire such abilities? A human being

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

starts life equipped with a number of very natural even instinctive reactions. In
neurophysiological terms these may be mediated by circuitry which has a
genetic basis in that we are congenitally disposed to react to certain stimuli in
certain ways. Such natural propensities probably include reactions to
movement, the human face, bright colours, the tendency to grasp, suck and
manipulate, emotional responses to pain and pleasure and so on. It is upon
these natural propensities that concepts can be built. This basis, in congenital
aspects of brain function, amounts to a natural 'congruence' of activity shared
by conspecific thinkers and forms a foundation on which shared use or shared
techniques for dealing with the world can be built. Armed with this substantial
biological congruence with the adults with whom she has to do and the ability
to modify that behaviour to meet certain standards, a child can build up
patterns of contact with the world by imitation, shared action, shared
attention and communication. All these aspects of behaviour are interwoven
and as they develop the child's brain processing becomes increasingly
organized by the selective strengthening of some synaptic connections, and the
elimination of others (Easter et al, [1986]). Note that the ability to suspend or
exhibit a natural response 'at will', as the judgments of others and/or the
present occasion warrant, is central and sharply relevant to the issues of
normativity, causality and teleology raised above. In fact, a child builds up a
set of normatively constrained or rule-governed reactions which conform to
certain standards of use and serve certain purposes. For instance, the child
forms the sounds it utters into words and, in forming these words, increases
her ability to respond to the world in the range of ways that human society
make available. Thus, on the basis of shared natural reactions, the child's way
of seeing the world is being informed and structured by those interactional and
verbal abilities which she is mastering in expanding and developing her
thought. 8 She learns to direct her attention to things that those interacting
with her have picked out and appreciate the experiences she is having as being
8
L. Vygotsky, (Thought and Language, Cambridge [Mass.]: M.I.T. Press 1962) reports the central
role of words and the perceived understanding of those words by others in children's problem
solving.
 Perception and Neuroscience 93
composed of elements (such as the colour red or a teddy, bear) able to be found
in other experiences. Her experience is thus conceptual experience and
gradually she acquires a repertoire whereby the way she thinks of things
comes to involve a structured network of concepts or practices of responding,
associating, abstracting and grouping according to rules. The child learns to
expand her informational competence not only by describing her world to
herself and others and inviting correction of those descriptions, but also by
discerning the intentions of others, requesting information and so on.9 The
child's information is presented and refined in terms of human reactions and
interactions and shared patterns of wilfully directed response which reflect the
(informal) rules for communication and thought that she 'latches on to'.

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

Throughout the experience of growing up the brain keeps a cumulative record.
In the human case, as we have noted, the conceptual system with which
'input' is going to engage in order to enter into thought is one that draws
heavily on interpersonal activity. Thus the record is structured in terms of
categories which emerge within the intentional activity of human beings, to
which the child is exquisitely sensitive (Donaldson [19 78]). Both brain activity
and information processing patterns are therefore influenced by the indicators
and markers formed by the words and actions of persons. The world is
presented as comprising items, features and events which are picked out or
'codified' (Luria)/'schematised' (Kant/Neisser) according to the way people
react to them. Because of the natural propensities already discussed, a human
thinker finds it easy to latch on to these ways of systematizing his reactions to
the world. The child's own actions, the objects and properties he encounters,
and the regularities he observes in their relations are all recorded in ways
determined by the human interests and concepts in terms of which they are
experienced and understood (Curran [1980]). We could say that a child
acquires techniques, focussed on the use of words as tools or devices, which
guide and inform his activity within an appreciation of the world: 'Language is
an instrument. Its concepts are instruments.... Concepts lead us to make
investigations; are the expression of our interest, and direct our interest'
(Wittgenstein [1953], 1.569).
This brings us very close to the active, hypothesis testing, purposive and
exploratory nature of perception that cognitive psychology has come to accept.
By acquiring the semantic tools provided by concepts, a knowledge of how
terms fit together and the way in which they can and cannot be combined in
human operations, a child begins to learn 'rationality' so that his thought
makes sense. In this way the elements of perception—each presented
9
This enables an understanding of the role of'egocentric speech' as is discussed by Vygotsky {op.
94 Grant Gillett
according to the human significance they have and in that way 'encoded' in
patterns of brain processing and informational selectivity—are knit together
so that the parts are, to some extent, constituted by the place they occupy in
the whole. For instance, one might see a partially obscured letter in a stimulus
array first as an E and then as an F depending on what letters are around it. In
one presentation it may be part of END and in the other of FIT. But this is only a
simple example of a much wider human tendency to use contextual cues to
disambiguate stimuli. A child gains mastery over this powerful technique as he
learns to draw on widespread processing capacities in responding to what he
encounters. Sometimes this cognitive 'work' will be quite conscious as one
puzzles away at a difficult case but most of the time partial or incomplete

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

stimuli are just smoothly fitted into the coherent picture of the world that is an
intrinsic part of conscious experience. As a child reflects upon the terms he uses
in various situations, the terms he might have been inclined to use, and the
reactions (actual and hypothetical) of others to those uses, he begins to see
how both the world and his thoughts fit together. For an item to be perceived it
must be fitted into this project of understanding and thus articulated with
other experiences according to the concepts and classifications that the child is
using. Perception is part of one's thought life in virtue of the fact that it is
constitutively articulated with this structure. 'I only call an articulated process
a thought: you could therefore say "only what has articulated expression'"
(Wittgenstein [1975], #32).
The development of perceptual abilities and thus the application of concepts
to what one encounters is faithfully reflected in the ways in which the
neurones of the brain come to causally affect one another. Any neuronal firing
can potentially (at least in early infant life) give rise to an indefinite number of
different patterns of transmitted neural activity. The particular patterns that
end up being favoured are a function of experience. Given a basic hard-wiring,
which may be genetically determined, certain synaptic connections and
certain excitation patterns will emerge to define the effective information
processing role of any given neurone or neuronal assembly. This means that,
even though the natural propensities on which concepts are built may be a
function of genetic endowment, the actual information processing structure
and function of :he brain and the patterns of activity laid down within it are
dependent on what it has been exposed to. Thus the microprocessing structure
that comes to be realized in the informational networks of the human brain
bears a human stamp; it is a 'presented' world rather than a merely
encountered world. Each assembly of processing units will be attuned to
significant stimulus configurations which the subject has been directed to
attend to and notice by the way that configurations within that class have
been treated as noteworthy by those who 'teach' her. Therefore the human
brain is affected by the world as humanly structured in an analogous way to
that in which the holographic plate is affected by the world via the reflected
 Perception and Neuroscience 95
laser light. It is the (conceptual) features of the environment as picked out in
the interactions between conspecific thinkers and the weightings given to
those features that explains how and why the brain selects certain patterns of
input so as to realize responses which 'fit' rules guiding activity and thought.
The plate from which we recover a holographic image requires a specific
method of projection in its genesis and in its subsequent use. The specific
medium that produces a hologram is laser light and only the same light can
bring out the features of the image from the apparently cryptic impression on
the plate. In an analogous way only interpersonal experience will imprint a
human brain with that information that allows it to be the causal basis of the
mind as we know it (a 'locus' of articulated and discursive activity). One would

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

expect that the human brain would therefore make maximum use of its
information when what is there is manifest in the context of human
interaction.10 Indeed, given the placticity of brain function one would suspect
that there might be a certain sustaining role played by continued human
interaction in the maintenance of processing efficacy. This is not to say that the
brain is a completely 'plastic' template or Lockean tabla rasa but only that,
within the natural propensities that brain structure make available, detailed
connections reflect human experience and are sustained by it.11
Support for this thesis emerges from current developments in parallel
distributed processing (PDP). A PDP system with feature detectors sensitive to
elements in an array may be 'trained' by a range of simple stimulus
configurations to detect the recurrent patterns that occur within the range.
That is, the system detects certain patterns as significant merely because they
co-occur in its experiential history. However not all arrays are simple and, in
some, elements will occur with about equal frequency in conflicting types of
patterns (for instance a matrix may be used to show horizontal and vertical
lines and there is nothing to distinguish between these groups in terms of
frequency of occurrence of any element (Rumelhart and Mclelland [1986], p.
182 ff.)). Where the elements are combined in varying and even conflicting
ways the system learns much better if it is assisted by cues (or 'correlated
teaching inputs') which 'signal' or 'tag' a given stimulus as belonging to one of
the specified groups. Such cues can be faded out once the system has developed
a preference for grouping according to the 'rule'.
This latter situation approximates much human learning. Human learners
are exposed to a huge number number of different stimulus arrays and must
learn to distinguish quite subtle patterns in which simple cues appear in
bewildering range of permutations and combinations which differ in signifi-
cance. Perhaps the powerful selective cues are provided by the responses of
10
The importance of interpersonal exchanges also suggests why humans take attitudes to
themselves and enter into intra-personal dialogue once they have mastered a system of
discourse.
1
' A feature which we use in the rehabilation of victims of neurological damage.
96 Grant Gillett
conspecifics who themselves know 'what counts as' an instance of the rule
being learned. Their contribution is therefore to impart the bases on which we
'go on in the same way' and so come to grasp those 'rules of the understanding'
which the developing thinker uses to make sense of experience (exactly similar
points will be able to be made about non-human thinkers who use articulated
rules to guide and inform their activity in the world; my continued discussion
of human thinkers should not be seen to be making exclusive claims except
where the nature of the abilities involved entails such). The central place of
conspecific in this account makes it unsurprising that conspecific recognition
should be neuropsychologically different from the 'cognitive business' of
recognition in general (Churchland, loc. cit). It is suggestive that the areas

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

from which the higher influences on stimulus selection and directed attention
discussed above (infero-temporal areas) are closely contiguous with those
areas in which recognition of conspecifics and their facial expressions are
mediated (mesial occipito-temporal cortex (Tranel and Damasio [1985])). One
could plausibly conjecture that we are neurologically organized so as to extract
the maximum help for our cognitive tasks from the responses we obsever in
conspecifics.
The human brains begins with certain natural structural features that
ensure a basic congruence in reactions between human beings and these basic
capacities enable persons to develop shared rules for perceptual judgments.
These rule-governed abilities Inter alia are the basis of the mental life of
thinking beings.
In an analogous way to the way in which the world as presented by the
medium of reflected laser beams changes the structure and responsiveness of a
photographic plate so as to create a holograpic representation, so the world as
presented to the human subject in the interactions and communications of
others changes the responsiveness and capacities of the human brain to create
a unique representation. This 'representation' involves processing abilities
shaped according to a system of human concepts and apt for the cognitive
tasks performed in perception and reasoning.

VI

This approach gives us a way of dealing with the problems with which we
began. Sensations arising from states in the body or from external objects
cause events in the nervous system which form patterns of neural activity
underlying thought and behaviour. The way in which neural events assume a
causal role in the activity of human beings is, however, determined by the way
that recognition and reidentification of objects have been shaped by rules for
the use of concepts and been (contingently) linked to those events. These
concepts, constituted within the domain of public meaning and knowledge,
involve items and happenings as they are presented to beings who fall within
 Perception and Neuroscience 97
12
the scope of a common natural history. The 'schemata' which sort out our
sensory manifold so as to select those patterns that fit some concept or
conceptual specification are therefore determined by human patterns of
activity. Thus the items we perceive are not a function of information content
which is invariant as presented to different 'information receivers'. If the
similarities between two receivers consist solely in physical events and their
patterns of impingement upon receptor surfaces then the crucial structuring of
information which occurs on the basis of natural congruence but involves
elaborate higher order processing activity will be lacking. Therefore the items
we perceive are misleadingly analysed in terms of physical impingements on
sensory organs or 'stimulations of sensory receptors' (Quine [1969], p. 84);

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

they are more accurately seen as figures, objects, events, qualities and
operations as presented to us within the milieu of human activity.
Meaning pervades perception, because perception is a human activity in
which sensory contact with the environment gives us information about the
things that concern (or potentially concern) us. The brain is adapted to deal
with information in the highly specific ways determined within practices
where the subject has learnt rule-governed techniques of search and detection
and not by its status in a causal theory. It is not surprising that we talk not only
about perceiving 'red', 'square' and 'movement' but also, for example, the
romantic character in some of Mozart's later music, the logical necessity of
- (a. — a) or the tension in the air during a difficult meeting. Neisser remarks,
"In the normal environment most perceptible objects and events are mean-
ingful. . . . These perceptions often seem very direct in that we become aware
of the meanings without seeming to notice the physical details that provide
evidence for them.' (Neisser, op cit., p. 70). In other words, it is events and
objects that mean something to me which form the 'bedrock' of my cognition
because whenever I classify something as a this or a that or see it as being thus
and so a higher function imposes form on what I perceive. For this reason the
'subpersonal' level of analysis may miss what it is that a person perceives and
completely fail to detect the crucial influences at work in determining the
elements of experience. To exhibit what is involved in perception we must
reveal the interaction between our sensory contact with the world and the
rule-governed concepts which are 'encoded' in those higher functions
modifying sensory input.
We do not construct a repertoire of responses to the environment (be they
epistemic or pragmatic) on the basis of events which are identical for different
physical detectors but rather we are initiated into a shared structure of
meanings which are enabled by but also which influence the structure of our
brain processes. The input into the human system is in richer terms than those

12
Wittgenstein would say that understanding how these things determine meaning is to
understand the 'grammar of an expression, op. cit.. 1953. 1.90. 371. 373.
98 Grant Gillett
a physical event theorist usually admits into his account. 13 Thought and the
contents of perception are structured by the ways in which those contents
have fallen or could fall within the interpersonal activities where we learn to
perceive and in which our brain takes on the 'informational shape' that will
serve as the causal basis for mental life.
It is clear that if there is a disorder either in the causal interface between our
receptor organs and the world or in the brain which will receive the input from
the world then important perceptual consequences will follow. Firstly, we will
lose the congruence which was a condition of the agreement in judgments
needed to provide us with our repertoire of perceptual and mental acts.
Secondly, if we have the repertoire already, then the informational patterns to

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

which it is attuned will be disturbed and so our thought will lose that vital
enabling contact with the world by which its content is fixed. Of course,
cognition depends upon the integrity of the structured brain processes which
enable both perception and judgment to occur in a regular relation, but mental
acts can only be philosophically elucidated by attention to the criterial or
constitutive exchanges and experiences (ordinary non-Cartesian sense) in
which a thinker has mastered the techniques of using rules to deal with the
world on the basis of his sensory contact with it and so has grasped the
concepts involved. Language, as Luria noted, is important because it is so
prominent in human rule-governed activity and details so many of the
nuances of classification and response that modulate our appreciation of what
we perceive.

VII
We can now return to the physicalist version of the argument from
hallucination. It is argued that in a case of hallucination an apparent
perception can arise purely from brain events (a number of different causal
stories are wheeled in to colour the argument). I recognize and report these
events as perceptual experiences of the world as I take it to be, much in the
same way as I recognize other internal events such as itches and pains. Thus, it
is argued, perception consists in knowledge by inference to an external world
on the basis of the evidence provided by internal (perhaps brain) events.
Against this I have argued that perception and its conceptual elements will
not conceptually reduce to physical events and thus that to understand what it
is to perceive something is not merely to detail internal dispositions, neuronal
events or successions of physical states.14 Mental life may, 'physically

1
' Much of the same point is made by John McDowell in 'Criteria, defeasibility and Knowledge Proc
BritAcad 1983.
14
The point is familiar from the arguments of J. Fodor. Psychological explanation New York:
Random House 1968. D. Davidson 'Mental events' in Essays on Actions and events Oxford:
O.U.P. 1980 and D. W. Hamlyn 'Behaviour' Philosophy xxviii (1953).
 Perception and Neuroscience 99
speaking', comprise certain neural events but we need a glimpse into that
'meaning structure' constituted by persons and their activities to elucidate
mental content, and thus the nature of perceptual competence. Whether we
approach this from the epistemological or neuro-scientific point of view we end
up looking for an understanding of how the brain sorts and classifies the
stream of stimuli with which it is bombarded. Thus we need some account of
how a perceiver learns what counts as this or that (perceptually) or learns the
rules governing perceptual judgment. I have also outlined the way in which
perception-based thought and judgment concerns public phenomena. Some
concepts, particularly those to do with the bodily states, sensations (such as
pain or nausea) or mental acts of one of the interlocutors, will have criteria for

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

their application which involve an acknowledgement of the different epistemic
positions in which users find themselves, but the rules for their use must still
focus on public cues and markers of meaning ('criteria'). Thus, as I have noted,
the way in which physical sensations (such as itches and urges) are 'encoded'
in brain events depends on the intersubjective practices of meaning which
confer content upon them. Therefore, to argue that aberrent causal chains
leading to false perceptual experience preserve the essential subject-involving
features of perception, neglects the fact that mental activity is conceptually
embedded within a public world. The only reason why hallucinations have
content is that they borrow it from instances of genuine (direct and indirect)
perception which have acquired it in a context of rule-governed shared
responses and natural propensities. What such cases do teach us is that terms
dealing with inner sensations should not be assimilated to the case of reported
outer objects of perception but must be regarded as having a less obvious basis
for their meaning (or a less obvious 'grammar'). Quite apart from this, we
should be cautious about an over ready assimilation of the properties of
hallucinations to normal experience for two reasons. Firstly, it is not at all clear
that hallucinations ever completely simulate their perceptual 'cousins', indeed
it is probable that they are rather deviant likenesses (either in experiential
figure or ground) of real experience and they can often be distinguished as such
even by an incompetent perceiver. Secondly, by the very nature of hallucinosis
it is quite unclear what thoughts can actually be attributed to the subject. A
person having an hallucination has an inclination, perhaps irresistable, to
think that she is perceiving something that she is not. We cannot allow that
the only thing that could cause such an inclination is the brain presenting her
with a picture which arises more proximately in the causal chain than a
normal perception. Indeed, when we consider the arguments against it, we see
that that could not possibly be an adequate characterisation of the situation.
But we need not opt for this view. To think one is seeing something need not be
to see something which exists only in thought.
ioo Grant Gillett

VIII

Neisser demanded some account of cognitive structures which would enable

us to understand how experience could be met by anticipatory schemata apt to
suit the variety of instances which a concept may subsume and also apt to deal
with 'the real environment of objects and events'. I have addressed this
demand by pointing to the rules or normative constraints which tie content to
a range of situations and the judgments made by concept users in those
situations. To have a given concept—or structure of 'rules of the understand-
ing' (Kant)—is just to have the ability to apply an intersubjective rule for
linguistic and other behaviour across and within a variety of situations. Of

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

course, this is not to say that we cannot learn anything about the way the
brain builds complexes of information in a form apt to engage with the
conceptual system from studying the processing of patterns of input events; it
would be foolish (and empirically discredited) to do so a priori. The best guess
would be that philosophy of mind, psychology and neuroscience would all
benefit from 'co-evolutionary development' but not quite in the way
Churchland envisages. One can conjecture that the complex centrifugal effects
on receptor function may carry the weightings and structure imposed by our
conceptual experience down to the most basic levels of neural function
subserving perception. For instance, the output from the retinal cells is
modulated by effects which originate in the cerebral cortex and which thus
alter the patterns of firing in the earliest stages of reception of visual
information (thin sense). But details of the actual way that the brain realizes its
informational task are not my concern, except in so far as to insist that its
function will reflect the (public) meaning structure in which it has taken
'shape'. More complex mental abilities will trade in the currency provided by
simpler (though holistically connected) sensory classifications. At each stage
in the mental response the derivation of significance will reflect rules which
operate within the conceptual scheme the subject has acquired.
Perception, the process by which sensory contact engages with the subject's
grasp of the meaning structure and the world it structures, is, of course,
dependent upon the normal causal interaction between receptor surfaces and
the world. That fact, however, does not support a causal theory of perception.
In fact, no sequence of processes gives us the right kind of understanding of
what perception is in that it does not necessarily see primacy where the subject
does (e.g. in mother's face) and it underplays the role of the concept-using
subject and the rules he is following.
Neural complexes combine and react to patterns of input in complex ways
that are sensitive to experience, but, studied apart from the nexus of
interpersonal and meaningful practices where they take shape, they do not
explain the content of thought. They obey causal, physiological laws of
 Perception and Neuroscience 101
neuroscience.15 Given certain causal antecedents, neural complexes produce
certain outputs and thus provide the causal basis for thought and meaning but
the processes involved can neither disclose nor explain the meaningful
structure of perception and thought.
Blakemore remarks, 'We seem driven to say t h a t . . . neurones have
knowledge. They have intelligence, for they are able to estimate the probability
of outside events.' (op. cit., p. 91). But we must demur, it only seems this way
because the neurones are part of people who think. The neurones faithfully
react to impingements on the nervous system and people, who participate in
practices where they have mastered the techniques of judgment that are
enabled by these patterns of reaction 'estimate the probability of outside

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

events'.
No neurones have knowledge, a neurone no more knows that it is firing as a
result of the effect of a red object than a blood vessel knows it is contracting as a
result of a vasomotor reflex. Knowledge talk is just out of place and misleading
as it tends to produce the illusion that the epistemic task of accounting for the
rule-governed nature of thought has been discharged. It is, therefore, just not
true that 'Descartes Dilemma' is to be resolved by more knowledge about what
goes on in a thinker. It is to be resolved by pointing out that what goes on in a
thinker is, at every level, adapted to the way that life is conducted outside him
in a rich, conceptually infused world. The source of knowledge, as Neisser
remarks, is the real environment and the 'schemata' or rules which structure
our understanding of it. We cannot, in fact, understand why patterned or
structured causal interactions in the brain should have the configurations
they do unless we locate them within a project whose general outlines can be
sketched in mental terms.
We can make helpful gains in understanding brain processes by invoking
metaphores based in interpersonal communication (e.g. 'evidence' or 'analy-
sis') but, given the difficult 'grammar' of our talk about mental states, must do
so with caution. We are not quite in the happy position of having a simple
concept like 'spin' to import into the field of study (as can be done in quantum
physics). We are constantly walking in a conceptual maze where the
metaphors we are using relate to workable but complex and non-ostensively
definable models and invoke the very terms we are attempting to elucidate.
More often than not the subtleties, presuppositions and conceptual connec-
tions of epistemic discouse are overlooked or even conflated into an
enthusiastic but muddled 'explanation' of what has puzzled epistemology for
centuries. A plausible but potentially misleading literature has grown out of
the idea that neural components can think, feel and exercise intelligence. They

15
Note that the understanding of these functions itself usually depends upon meaningful thought
tofillout content specifications—by interview, introspective reports, reasoning tasks or tests of
comprehension etc. It will also depend on the theoretical model provided by thick (or
conceptual) information systems.
102 Grant Gillett
do none of these things. People (and, arguably, animals) do these things and in
so doing exploit the amazing capacities of the nervous system.

IX
A major step was taken in science when we realized that inanimate objects
could move without spirits animating them. Unfortunately in neuroscience we
often make the retrograde step of ascribing to inappropriate bearers the mental
and spiritual predicates properly reserved for those who participate in practices
which enable thought and meaning. It is no wonder that as a result of so doing
we become profoundly confused and end up with problems about the nature

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

and contents of perception.
I have tried to use the hologram analogy to show that, even if the meaning
and content of thought is only to be understood in terms of the milieu in which
it has taken shape, we can still appreciate the vital informational role played by
the brain in allowing that milieu to be the way that it is. The holographic
aspect of human brain function is its adaptation to a way of receiving
information in which objects and events fall into certain classifications with
widespread behavioural implications. These classifications are captured by
human concepts and shared by concept-users. The system of classifications
imposes a shape on the way that the brain works and in order to understand
and unlock that shape we need to make extensive appeal to human thought
and interest. The hologram could be said to bring the importance of this
perspective back to the surface so that it can be acknowledged in framing an
adequate theory of perception and thought.
Aristotle said that the form of the human being was to be a rational and
social animal and I would contend that even the processing shape of the
human brain bears the indelible mark of that form.

ACKNOWLEDGEMENTS
I should like to thank Dr K. V. Wilkes (St Hilda's College, Oxford) and Mr P. F.
Snowdon (Exeter College, Oxford) for comments on an earlier draft of this
manuscript.
Magdalen College, Oxford

REFERENCES
CHURCHLAND, P. [1986]: Neurophilosophy. Cambridge: M.I.T. Press.
CURRAN, H. V. [1980]: Cross-cultural perspectives on cognition; in G. Claxton, (ed.).
Cognitive Psychology: new directions, Routledge and Kegan Paul.
DENNETT, D. [1978]: Brainstorms. Cambridge (Mass.): M.I.T. Press.
DONALDSON, M. [1978]: Children's Minds. London: Fontana.
 Perception and Neuroscience 103

EASTER, S. S. JR. PURVES, D., RAKICK, P and SPITZER, N. C. [1986]: 'The changing view of
neural specificity'. Science, 230, p. 507.
EVANS, G. [1981]: The varieties of reference J. McDowell (ed.). Oxford: O.U.P.
GILLETT, G. [1987]: 'The generality constraint and conscious thought' in Analysis.
LOCKE, J. [1689]: Essay concerning human understanding P. H. Nidditch (ed.). Oxford:
O.U.P. 1975 Bk II. ch. IX. #3.
LURIA, A. R. [1973]: The Working Brain. Harmondsworth: Penquin.
NEISSER, U. [1976]: Cognition and reality. San Francisco: W. H. Freeman.
PENFIELD W. and RASMUSSEN, T. [1950]: The cerebral cortex in man: a clinical study of
localisation of function. New York: Macmillan.
PRIBRAM, K. H. [1969]: 'Some dimensions of remembering: steps toward a neuropsy-
chological model of memory' in Pribram, (ed.) Perception and Action, Harmonds-

Downloaded from http://bjps.oxfordjournals.org/ at University of Otago on July 3, 2015

worth: Penguin.
PRIBRAM, K. H. [1971]: Languages of the brain. New Jersey: Prentice Hall.
QUINE, W. V. O. [1969]: 'Epistemology naturalized' in Ontological relativity and other
essays. New York: Columbia U.P.
RUMELHART, D. E. and MCLELLAND, J. L. [1986]: Parallel distributed processing:
Explorations in the microstructure of cognition, vol 1, Bradford: MIT Press.
TRANEL, D. and DAMASIO, A. R. [1985]: Knowledge without awareness: an autonomic
index of facial recognition by prosopagnosics, Science, 228, p. 1453.
ULLMAN. S. [1986]: 'A.I. and the brain: computational studies in the visual system' in
Annual review of neuroscience 9.
WEISCRANTZ, L. [1974]: 'The interaction between occipital and temporal cortex in
vision: an overview' in (F. 0. Schmidt and F. G. Worder (eds). The neurosciences,
third study program. Cambridge (Mass): M.I.T. Press.
WITTGENSTEIN, L. [1953]: Philosophical Investigations 1.569 (tr. G. E. M. Anscombe).
Oxford: Blackwells.
WITTGENSTEIN, L. [1975]: Philosophical Remarks, (tr. R. Hargreaves and R. White).
Blackwells: Oxford.