Philosophical Review
The Imagery Debate. by Michael Tye
Review by: Kathleen A. Akins
The Philosophical Review, Vol. 103, No. 1 (Jan., 1994), pp. 172-175
Published by: Duke University Press on behalf of Philosophical Review
Stable URL: http://www.jstor.org/stable/2185886 .
Accessed: 18/09/2012 19:08
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
.
Duke University Press and Philosophical Review are collaborating with JSTOR to digitize, preserve and extend
access to The Philosophical Review.
http://www.jstor.org
BOOK REVIEWS
The Philosophical Review,Vol. 103, No. 1 (January 1994)
THE IMAGERYDEBATE. By MICHAEL TYE. Cambridge: MIT Press, Bradford
Books, 1991. Pp. xiv, 172.
The other day a friendof mine was recountinga dream-a classic traveller's
nightmareabout havinga suit made in Hong Kong. Going fromshop to shop,
he found himselflost in a maze of alleys and sidestreets,unable to ask directions or, on findinga tailor,to explain what he wanted. "By the way," I said,
kiddinghim, "what color of suitdid you finallydecide on?" "Color!?" he said,
puzzled. "I don't dream in color; in fact,I don't even dream in black and
white. You see, for me, dreaming isn't visual-there aren't any images,any
picturesat all." Then he paused and a slysmile came over his face. "Or perhaps
I shouldsaythatitdoesn'tseem to me thatI dream in images-at least not after
I've woken up!"
In The Imagery
Debate,Michael Tye attemptsa quick traverseacross the
difficult
of mental imagery.His organizingprincipleis the current
territory
dispute in experimental psychology-between those researchers who advocate a "pictorialist" account of mental imagery and those who favor a
"linguistic" or "descriptionalist"view. In the firsthalf of the book, the
reader is given an admirably clear survey of that landscape. The firsttwo
chapters offera short historicalintroductionto the problem of mental
images (plus a denunciation of introspection as a reliable means of access
to theirnature); the third and fourthchapters explain and evaluate the
empiricaladequacy of the leading theoriesfromboth camps, namelyKosslyn's "pictorial" hypothesis and the "linguistic" hypotheses of Pylyshyn
and Hinton. In the second half of the volume, Tye presents his own theory
of mental imagery (a hybrid account based loosely upon Marr's theory of
vision) and then turns to a number of outstanding philosophical issues.
Tye's view of mental images is that they are "interpreted symbol-filled
arrays"-"functional pictures" thatoccur in a "visual buffer"shared with
normal visual processing to which sentential interpretations (for example,
"This is an F") have been appended.
More specifically,Tye's notion of a functionalpicture turns on three
interrelatedideas: a functionalpicture is pictorial,functional,and representational.First,in explaining what makes a neural representationpictorial,Tye improviseson Kosslyn'snotion of a "quasi-picture": a representation of an object, as seen froma particularviewpoint,is pictorialif and
number of spatiallyadjacent pointson the object (visible
onlyifa sufficient
fromthisviewpoint)are representedby partsof the representation(pixellike structurescalled "cells") thatare themselves"adjacent." (Unlike Kosslyn, however, Tye does not require that all parts of the representation
correspond to the spatial parts of the object, for his model includes symbols, placed withinthe cells, thatrepresentsurfacepropertiesof objects.)
172
BOOK REVIEWS
Here, "adjacency" between the cells of the array-and spatial relations
as implying
betweensuch image partsin general-is not to be read literally,
that the cells themselvesstand side-by-sidein neural space or that they
instantiateany other particularspatial relations.Rather,such spatial reladuring image transfortions are said to obtain if the cells are manipulated,
mations,as iftheystood in such relations.This is the second characteristic
of a functionalpicture.Third, thisrough definitionof a functionalpicture
relationin virpresupposes a more primitiveconcept of representation-the
tue of which any of the elements of the array,cells or symbols,represent
Tye's views on mental representationare a bit thin
at all. Unfortunately,
on the ground here, but he does brieflyreferto "registration"relations
betweensymbolsand externalproperties;additionally,he suggeststhatthe
contentof a representationis fixedby those propertiesthatunder normal
viewingconditions (relativeto an appropriategroup of individuals)would
be causally sufficientto bring about an instance of that representational
type.Let us say,then,thatfunctionalpicturesrequire, first,a one-way
mapand, second,
visiblepartsoftheobjectand cellsoftherepresentation
ping between
represented
a manipulation
ofthosecells"in accordancewith"thespatialrelations
in thearraystandin a particularkindofregistration
and, third,that theelements
to bringthemabout.Prima facie,
sufficient
normally
relationto thoseproperties
then, the definitionof a functionalpicture contains three verydifferent
kinds of criteria-syntactic,"formality,"and semantic constraints.AlthoughTye does not say how these criteriaare interrelated,note here that
trouble will arise if an independent notion of syntaxis not forthcoming.
If we cannot assume that neural syntacticelements are simply"plain for
all to see"-that is, if individuatingneural representationalpartsrequires
semanticor formalconsiderations-then therewill be a complex interdependency between the three criteria.Certainly,then, a good deal more
needs to be said about these constraintsif the notion of a functionalpicture is to become clear.
As I said above, Tye postulates that the functionalpictures of mental
arrays"which share the common
imageryare "interpretedsymbol-filled
machineryof normal visual processing. (In fact,given Tye's view of representationalcontent qua perceptual registration,there seem few other
options.) Appealing to Marr's theoryof vision,Tye associates these arrays
with representationsat the level of Marr's 2.5-D sketch (albeit a slightly
modifiedone), and the visual "buffer"itselfwithwhatevervisual processing site(s) the 2.5-D sketchoccurs. On Marr's theory,recall, there are four
levelsof representationin image processing:first,an image thatrepresents
lightintensityat points in the retinalimage; second, the primalsketchthat
makes explicitintensitychanges and theirspatial organization;third,the
2.5-D sketch that representsedges, depth, and orientationinformation;
and fourth,the 3-D model that representsshapes and their spatial orga173
BOOK REVIEWS
nization in termsof stickfiguresand volumetricprimitives(for example,
generalized cylinders).The visual array,according to Tye, contains Marrlike symbolsfor orientation,depth, the presence of an edge or a ridge,
plus some other symbolsof Tye's own making-symbols for color, light
and texture,plus one special symbol,'O', thatspecifiesthe presintensity,
ence of an object surface.Note thatbecause Marr's 2.5-D sketchdoes not
contain any informationabout object segmentationor object identity,Tye
"This is an F," mustbe appended to
believes thata sententialinterpretation,
on his view, either
produce a given mental image. Such interpretations,
come withthe arrayas generated frommemoryor else are produced "on
line," by scanning the arrayand, on that basis, assigninga new interpretation.
To see a central problem with Tye's association of visual images with
interpreted2.5-D sketches,note that Marr's (unadulterated) theoryof vision makes a fairamount of sense when viewed as a theoryof shape recognition-and this is so even if,at the end of the day, it is not correct.As an
explanationof visualphenomenology,however,Marr'stheorypresentsus with
a kind of paradox. When we experience visual phenomenology,say,when
we look out the windowto the garden below, we do not see littledots and
arrowsaffixedto "cartoon" outlines (the primitivesof the 2.5-D sketch),
nor do we see stickfiguresor generalized cylinders(the primitivesof shape
recognition);on the other hand, we are aware of "pixel by pixel" intensity
information,informationthat is no longer explicitlyrepresentedeven in
the second stage of processing,the primal sketch. Starting with the photographic image, that is, the moreprocessingsteps taken-each one servingto make
fromtheenvironment-theless "like" our visual phenomexplicitmoreinformation
seemsto become.How then could such a process posenologytherepresentation
siblygive rise to the knownphenomenology?What is the relationshipbetweenthe representationalprimitivesof the variouslevelsof representation
and what we see? This is the paradox (call it "Marr's paradox") that arises
ifone simplyappropriatesa theoryof shape recognitionforuse as a theory
of visual phenomenology.In Tye's case, much the same paradox will arise
forthe phenomenologyof imagery.If a mental image is identifiedwithan
interpreted2.5-D sketch,how can the representationalprimitivesof the
2.5-D sketchpossiblyaccount for the kind of visual imagerythatwe experience?
Tye's response to Marr's paradox-or at least his implicitresponse, for
he does not discuss the relation between the vehicles of representation
and conscious experience-is to incorporate everyrepresentationaltype
that is intuitivelypart and parcel of our conscious phenomenology into
the "new and enriched" 2.5-D sketch,while leaving out all those representationalprimitivesthatseem to bear littleobvious relation to our phenomenal experience. Thus, the intuitivepropertiesof visual phenomenol174
BOOK REVIEWS
ogy that,on Marr's scheme, are representedexplicitlyonly at levels prior
to the 2.5-D sketchnow reappear, representedby symbolsin the individual
cells (hence the inclusion of texture,intensity,and color symbols)while
the lower-levelrepresentationalprimitivesof intuitively
"nonvisual" propertieshave simplybeen omittedfromthe new 2.5-D sketch(Marr's "blobs"
or "bars"). Similarly,those intentionalaspectsof images thatare (or would
be) discerned furtherupstreamon Marr's theoryare now merelyappended in the formof "sententialinterpretations."In thisway,the embarrassing machineryof shape recognition(forexample, the cylinders)and object
segmentation (whateverit might be) are discreetlyhidden behind the
scenes. In effect,Tye's answerto Marr's paradox is to collapse togetherthe
various levels of visual representationand therebyproduce a representational constructthat, according to our intuitions,bears an intuitive"resemblance" to our visual experience. The paradox of vision is not answered.It is merelyavoided.
Recall the epigraph at the beginning of this review.It demonstrated
much of what is puzzling about mental imagery-that the phenomenology
variesgreatlyfromperson to person; thatthere is no clear inferencefrom
"how it seems" to the subject to the formof the underlyingrepresentational structure;thatmental imagery,even when it does seem clearly"visual," has manypeculiar, "un-picture-like"qualities,ones thatare not explained by advertingto the propertiesof ordinarypublic images nor even
to the properties of our ordinaryvisual perceptions. Whateverthe true
nature of the beast, understandingmental imageryrequires us to address
most of the persistentproblems in the philosophyof mind-the problem
of intentionality,
the natureof consciousnessand itsrelationto the vehicles
of mental representationand theircontents,the role of introspection,and
so on. In sum, while Tye's book givesa clear and intelligentsurveyof the
psychologicalliteratureon mental imagery,it does not providea satisfying
philosophical account of the most basic issues surrounding mental
imagery.
KATHLEEN
University
ofIllinois,Urbana-Champaign
A. AKINS
ThePhilosophical
Review,Vol. 103, No. 1 (January1994)
SPRINGS OF ACTION: UNDERSTANDINGINTENTIONAL BEHAVIOR. By
Press, 1992. Pp. ix, 272.
ALFRED R. MELE. New York: Oxford University
SpringsofActionis a denselyargued, extremelyvaluable essayon the theory
of action. In the firstplace, it serves a badly needed function.It brings
assesses a range of the best work,done over the
togetherand instructively
175