The Imagery Debate, Filozofia, Filozofia - Artykuły
[ Pobierz całość w formacie PDF ]
Philosophy of Science Association
The Imagery Debate
Author(s): Kim Sterelny
Source: Philosophy of Science, Vol. 53, No. 4 (Dec., 1986), pp. 560-583
Published by: The University of Chicago Press on behalf of the Philosophy of Science
Association
Stable URL:
Accessed: 26/10/2008 14:46
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.
Philosophy of Science Association
and
The University of Chicago Press
are collaborating with JSTOR to
digitize, preserve and extend access to
Philosophy of Science.
THE IMAGERY DEBATE*
KIM STERELNYt
Department of Philosophy
University of Maryland College Park
One central debate in cognitive science is over imagery. Do images constitute,
or constitute evidence for, a distinctive, depictive form of mental representation?
The most sophisticated advocacy of this view has been developed by Kosslyn
and his coworkers. This paper focuses on his position and argues (i) that though
Kosslyn
has not
developed
a
satisfactory account
of
depiction,
there is
nothing
in principle unintelligible about the idea of depictive neural representation, but
(ii) Kosslyn's model of imagery rescues the intelligibility of pictorialism at the
cost of its explanatory power.
in the belief-storeplays a differentcausal
role in the political economy of the mind than a representation in the
desire-store.
What is the nature of these representations?Within cognitive science
there is a lively debate on this issue. Zenon Pylyshyn has suggested that
representationis propositional.' Others have argued for a more
pluralist
account. Stephen Kosslyn
in
particular has developed theories
finding
room for a variety of sentential codes together with a pictorial mode of
representation(See, for example, Kosslyn and Pomerantz 1978; Kosslyn
1980; Kosslyn et al. 1979; and Kosslyn 1981.).
It's not just difficult to see who is right in this debate; that's to be
expected. It's a difficult debate to understand, for this requires disentan-
gling a complex of theoretical, empirical, and conceptual issues.
I
will
attempt
to
disentangle
some of
these
in this
paper.
Kosslyn's view has great initial plausibility. For we seem to be aware
of images-pictures in the mind-playing an importantrole in thought.
*Received December 1984.
tThanks to David Armstrong, Ned Block, Michael Devitt, and Bill Lycan for their
comments
on earlier versions of this
paper.
'Roughly, a system of representation is propositional or sentential if its formulae share
the distinctive syntactic and semantic features of natural language: a generative syntax;
syntactically structuredformulae; semantic properties like truth and reference.
Philosophy of Science, 53
(1986) pp.
560-583.
Copyright ? 1986 by the Philosophy of Science Association.
560
I.
Introduction. In other papers I have defended representationalistac-
counts of cognition (Sterelny 1981, 1983, 1984; also Fodor 1975, 1978;
and Lycan 1981). That is, I have defended the claim that our proposition-
al attitudes are relations to intemal representation tokens. Different atti-
tudes are distinguished by the different functional roles an internal rep-
resentationplays: a representation
THE IMAGERY DEBATE
561
Moreover, the appeal of mental pictures is not based on introspection
alone. I will briefly survey some other sources of the appeal of pictori-
alism.
1. Format. Pictures are a convenient format for various cognitive tasks.
Informationcan be "readoff" visual displays-information that could be
extractedonly laboriously from sentential coding. For instance, it is much
easier to see if three cities lie on a straight line by looking at a map than
by consulting a list of cities filed with their locations. So (the idea runs)
if you want to know if three
cities are on a straight line, your natural
recourse is to construct a mental map, and read off the result. Pictorial
inner representationswould be functional.
2. Imagery and Visual Perception. No one denies our capacity to imag-
ine objects and scenes. The problem is to explain that capacity. A central
feature of our pictorial imagination is its link with visual perception. We
cannot image what we cannot
see. This connection supportspictorialism,
for on other accounts of imagery, the restriction is prima facie mysteri-
ous. If an image is a network of propositions brought to short-termmem-
ory (as Pylyshyn suggests), then there need be no restriction on the kinds
of predicates that are used in the network. But if imagination involves
the same mechanisms as visual perception, this restriction emerges au-
tomatically.
3. The Process of
Imagination.
The attractionsof pictorialism have also
been boosted by a range of experimental results. Two paradigms concern
rotationand scanning imagined objects. I will returnto these experimental
programmes, but I will briefly sketch the main themes here.
Rotation. In his 1970 paper, Shepard announced a suggestive experi-
mental result. Subjects were presented
with
two complex
forms. In some
of
the presentations,
the forms were identical in shape but in different
orientations. Subjects were asked whether the forms were identical. In-
terestingly, the time taken to respond increased linearly with angular dis-
tance between the forms, that is, with the degree through which one
form
would have to be rotated to bring
it into
congruence
with the other. This
prototypicalrotationexperiment
has been taken to imply that images, like
physical objects, are rotatable.
Scanning. Kosslyn (and others) have required subjects to memorize
pictures, for example, maps. Once memorized, the subject calls the map
to consciousness as an image. They
are then instructed to focus on a
particularpoint,
then are asked to confirm or deny the presence of fea-
562
KIM STERELNY
tures on the map. The key result: the furtherthe feature from the focused
point, the longer it takes to respond. It is, therefore, natural to suppose
that images are spatial displays that can be scanned.
In the opinion of many, results like this suggest that visual imagination
represents spatial relations by in some way manifesting spatial relations.
For these results would be readily understood if images were photo-
graphic slides in our heads. Thus, if a slide needed to be rotated in order
for some matching process to take place, then the further a slide needed
to be rotated, the longer rotation would take. Similarly, if a scanner were
like a reader under which the slide passed, the further a point from the
initial focus, the longer it would take
to
reach the scanner.
While no one believes images are slides, it seems as though these re-
sults fit naturally into a pictorial account of imagery. But they can be
accommodated by a propositional account only with the help of ad hoc
stipulations.
The view that we have some pictorial system of representation does
therefore have initial plausibility. But it is a view that obviously raises
problems as difficult as it solves.
(i) We might have sentences
in
our heads. For sentences are not
re-
stricted to any particularmedium. Sentences can come as sound waves,
marks on paper, electrical pulses, punched cards, and so on. Why not
then as patternsof neural firings as well. But pictures seem to be medium
specific. Not just anything can be a picture of my big toe. So we might,
literally, have sentences encoded in our brains, but we do not have pic-
tures in our brains. So the pictorialist thesis is initially diffuse; we have
an internalsystem of representationthat
in
some way(s) is similar to some
pictorial system
of
representation.
But in
which way(s)
and to what sys-
tem. Consider,
for
example,
the differences
between photographs,
stick
figures, diagrams, maps.
(ii) This problem leads to a second. Pictorialism moves from the thesis
that we have pictures in the head to the thesis that we have something
like pictures in the head. But this weakening may not preserve the ex-
planatorypower of pictorialism. That power may depend covertly
on
the
naive
thesis: the view that we have real
pictures
in the head.
In this paper I concentrateon Kosslyn's pictorialism, since his account
is the most developed. In the next section, I outline his position. I then
discuss the intelligibility and explanatory power of the pictorialist posi-
tion. The basic message of this paper is that Kosslyn's model rescues the
intelligibility
of
pictorialism
at the cost of its
explanatory power.
II. Kosslyn's Model of Imagery. Kosslyn's basic idea is to develop an
illuminating initial metaphor of imagery: a spatial display generated on
cathode ray tube.
THE IMAGERY DEBATE
563
Visual images might be like spatial displays provided on a cathode
ray tube by a computer programmeoperating
on
stored data. That is
we hypothesized that images are temporaryspatial displays in
active
memory that are generated from more abstractrepresentationsin long
term memory. (Kosslyn et al. 1979, p. 536)
Nothing literally like a display is activated in the mind of course. That
would involve commitment
to
real though evanescent pictures.
But
im-
ages
are like
displays.
1. Mental images are semantically like displays, in that they depict
situations ratherthan describing them. Depiction is not explained
by appeal to similarity. Rather, Kosslyn gives a cluster of criteria
that will be discussed in the next section.
2. A display is generated from information in long-term store that is
in a quite different form. Similarly, images are generated from
information
in
long-term memory that is coded discursively.
3. A visual display on a CRT is transformablein various ways.
Sim-
ilarly, a mental image is transformablein various ways. A series
of experiments by Shepard, Kosslyn, and others have shown, or
seem to have shown, that images can be rotated, scanned, scaled
up and down in size, in something ratherlike the ways in which
real pictures, especially those on a display, can be.
4. Finally, we can use CRT displays. Informationcan be "readoff"
displays. Similarly, we can read information off images: in virtue
of their
explicit representationof various properties and relations
only implicitly representedin propositional format. Of course, this
"reading-off" is not a primitive function of the mind's eye: for
that would introduce a fully fledged homunculus. The interpre-
tation of images will have to be broken into a series of subfunc-
tions
until we reach a level that is plausibly primitive.
How can this
metaphor
be transformed into an
explicit theory
of im-
agery?
This transformation is not
complete,
but
Kosslyn
and his co-
workershave constructeda runningsimulation of human imagery. A sim-
ulation is theory-like in proceeding at a suitable level of abstraction. A
psychologist need not be concerned with how psychological functions are
realized
in
brainwave though he
must
ensure that
the functions he
posits
can
be so
realized. Similarly,
the
programmer
need not be concerned with
the hardware
implementation
of his
programme.
A simulation is
theory-
unlike in having various theory neutral characteristics, characteristics it
may or may not share with the modeled domain.
The model has three elements. One is the
image itself,
the surface
display. The second is long-term memory store, the information from
[ Pobierz całość w formacie PDF ]