Internal Versus External Representation DilworthJ JAAC04

7/25/2019 Internal Versus External Representation DilworthJ JAAC04

1/14

The Journal of Aesthetics and Art Criticism 62:1 Winter 2004

JOHN DILWORTH

Internal versus External Representation

I argue that there are two kinds of representa-tion, whose joint recognition promises, amongother things, to provide a relatively clear and

unified account of both pictorial representationand fictional reference.

To begin, I shall show by demonstrating anapparent ambiguity in the concept of represen-tation that it has two distinct aspects. One stand-ard test for ambiguity of a concept C is to find acase in which C both is and is not applicable tosome object or objects; then one may proceed torepair the damage by distinguishing differentsenses, aspects, or varieties of the concept C.Here is a pictorial test case of this kind for theconcept of representation.

First, it is surely true that a picture P mightrepresenta man, even if there is no actualmanabout whom it is true that he is the man thusrepresented by picture P. For the person whopainted A might have intended merely to repre-sent a man, without having any particularactual man in mind.

However, if there is no actualman representedby picture P, then there is a clear sense in whichP does notrepresent a manat least not in thesense in which it would represent a man if

there were some actual man whom it represented.At the same time, though, to repeat, clearly

picture P might still represent a man in thefirst sense above, in that, if the artists paintingis successful, then it willrepresent a man, inspite of the fact that it does not represent aman in the sense of an actualman.

Now to be sure, the ambiguity here might beargued to be located primarily in the wholephrase representation of a man, or in differentsenses of a man in each case, rather than inthe concept of representation itself. Nevertheless,the case at least shows the need to distinguish

two fundamentally different aspectsor categoriesof representationeach of which is genericallydescribable as representation by a picture P of

some thing X. We confuse these two funda-mentally different aspects at our peril. Thus thelabel ambiguous will remain descriptivelyconvenient for such phrases as P is a represen-tation of a man, and thus derivatively as appliedto the concept of representation itself, even if itis only (very) different kinds of use, or differentaspects, of that concept that are actuallyinvolved.

As for the evidence I have used to establishthe ambiguity in question, it is arguably notcontroversial at all, since it is recognizedalready. For example: Richard Wollheim distin-guishes representations ofparticularobjects, orevents, from representations of objects or eventsthat are merelyof a particular kind, and thusof aman, and so on, rather than of some particu-lar man;1 whereas Kendall Walton uses thefact that a picture can represent a man, eventhough there is no actual man thus represented,as part of his case that the concept of represen-tation is primarily to be understood not inobject-centered ways, but rather in terms of his

own props used in imaginative gamesapproach.2Nelson Goodman, in the face of suchexamples, distinguishes apictureof a man froma man-picture, in cases where no actual man isrepresented.3

However, by focusing on the fact thatexamples such as the above show that there isan apparent, or at least formal, ambiguityin theconcept of representation, I hope to facilitate theraising of issues and approaches that have notyet been adequately investigated.

First, I shall distinguish the relevant distinctconcepts or aspects of representation as external


2/14

24 The Journal of Aesthetics and Art Criticism

versus internal(or outer versus inner) represen-tation. External or outer representation by apicture P is about cases in which there is someactual object X, normally externalto the repre-

senting object P,4

that is represented by P. Thus,for example, picture P externally represents aman just in case there is some actual man,distinct from P, who is represented or portrayedby P.5

However, the formulation of a correspondingcharacterization of internalor inner representationis inevitably more controversial. Here is an initialattempt: internal or inner representation is aboutwhether there is an objectX, that is internaltothe object P, and that is represented by P.

I. ALLAYING ANXIETIES

That the above characterization of internal rep-resentation is inevitably controversial can beseen by giving a substitution-instance of it: pic-ture P internally represents a man just in casethere is a man, who is internal to P, and who isrepresented by P.

This instance is likely to raise logical andmetaphysical hackles on at least two grounds:

first, it clearly involves reference to, andreidentificationof, the man in question (sinceit seems to be said that there is such a man andthat that same man is both internal to P and rep-resented by P); second, metaphysical worriesabout how a man could possibly be internalto a picture P.

Indeed, it seems likely that these, or related,logical and metaphysical obstacles have provedformidable enough to dissuade previousattempts at distinguishing an internal aspect, orkind, of representation from its more conven-tionally accepted external aspect or kind.

Here is a preliminary attempt to allay the abovelogical and ontological anxieties. The issuesrevolve around the status of the phrase there isin the definition. On my account, this phrase ismerely part of a claim that there is an objectof referenceXthat is internally represented bya picture. In other words, this there is has nospecifically ontological implications as to thestatus of any given such object of reference;instead it functions to indicate that statements of

inner representation are indeed referential, inthat they involve a reference to some object.6

A more specifically ontological anxiety is aconcern that an account of internal representa-tion might be committed to the existence ofsuch internal objects, since, of course, the idea

of an actually existing man as somehow beinginternal to a picture of him seems confused oreven self-contradictory. However, my reply isthat a general account of internal representationmust inevitably include cases of internal objectsthat do not exist, since they are fictional in abroad sensesuch as with pictures of unicorns,or Santa Claus, or of a character in some fic-tional novel or play. This again shows that thereferential form of statements of internal repre-sentation does not in itself have any ontologicalimplications as to the status of any particular

objects that are thus referred to.7

Nevertheless, this referential view is stilldistinctively different from a nonreferentialview such as Goodmans, which distinguishes agenuine picture of a man from a man-picture,which term he invokes when no actual man ispictured, and which description merely classi-fies the picture as being of a certain kind, whoseextension is the set of man-pictures.8Thus, onmy account, but not on Goodmans, picture ofX is always genuinely relational.

I shall further discuss the validity and onto-logical harmlessness of reference to fictionalentities in Sections VIII and IX.

II. MORE ON REPRESENTATIONAL AMBIGUITY

Returning to the suggested ambiguity in theconcept of representation, recall that at least theunanalyzed ambiguity, and the evidence onwhich it rests, is not controversial: that a picturecan represent a man, without there actuallybeing such a man, cannot be denied. Thus, theinitial issue is, of course, as to how linguisticcontexts such as picture P represents a manshould be analyzed, in cases when there is notany actual corresponding man represented by P.

I take it it would usually be assumed that thisinitial case just shows that there is a unitaryconcept of representation, which, among otherthings, is such that P represents X does notentail there is an actual X represented by P.However, that nonentailment can be explained

equally well by postulating two distinct con-cepts of representation, on one of which,


3/14

Dilworth Internal versus External Representation 25

namely, internal representation, the entailmentdoes not hold, while on the other (external rep-resentation) the entailment does hold.9For it isclear that standard examples of the nonentail-

ment in question would, on my account, becases of internal rather than external representa-tion, and, hence, be adequately explained bythat concept of internal representation.

Moreover, the issue of ambiguity is not soeasy to evade, because, for instance, one couldseeor recognizethat a picture P represented aman, whether or not there was, independentlyof ones act of perception, some actual manwho was represented by picture P. Thus, it ishard to deny that one could, on the one hand,have conclusive perceptual evidence that a pic-

ture P represents someonewhich someoneone would describe or recognize as a manquite independently of whether one alsohas anyevidence that something else, an actual man, isrepresented by picture P. Thus from an epi-stemic perspective the relevance or applicabilityof two distinct concepts of representationorof two distinct kinds of object of the con-ceptseems to be an inescapable part of the

joint evidentiary situation in such a case.Hence, I would claim that, in spite of any ini-

tial difficulties or obstacles in defining internalrepresentation, some analysis should be attemptedof that concept, that adequately captures orallows for such epistemic differences, as well asbeing acceptable on logical and metaphysicalgrounds.

III. POSSIBLE OBJECTIONS

Before giving my account of internal represen-tation, there is a major issue that needs to bediscussed, namely, the function or purpose of aconcept of internal representation in caseswhere there isan external representation by pic-ture P of object X. I shall argue in Section VIthat in such cases internal representation stillhas an important epistemicrole to play, in that itis at least partly in virtue of seeing what a pic-ture P internally representsits object Xthatone can justifya claim that P externally repre-sents object X.

But there is a potential basic objection to

such an approach, namely, that when externalrepresentation obtains, there would be no such

theoretical need, or room, for a concept of inter-nal representation in addition to that of externalrepresentation, so that its general postulation forall cases of representation is ad hoc.10

This objection could take two forms. First,that in cases of external representation of anobject X by picture P, the object X that onesees when looking at P isX itself, so that there isno distinct object X available, which couldserve as the object of an independent kind ofinternal representation. I shall call this theseeing through objection, in that on thisaccount one sees through an external repre-sentation P of object X to its actual object X.

And second, it might be argued that whetheror not the seeing through view is correct in

some or all cases, epistemic claims that pictureP externally represents object X can be justifiedon other, nonreferential grounds, such as byclaiming that P has a group of properties thatare its representational contentthose prop-erties it represents X as havingwhich explainPs ability to externally represent X withoutintroducing any independent object X that isinternally represented.11 I take it that this sec-ond, more general objectionthe nonreferen-tial objectionembodies a view that is the

common core of most extant views of represen-tation, whatever else they disagree on.

Since my purpose in this paper is not primarilypolemical, I shall address such nonreferentialviews only indirectly, via the above mentionedepistemic justification argument, which at leastwill show that nonreferential views could notprovide a complete account of representation.But I shall attempt to refute the first, seeingthrough view of external representation, sinceversions of itholding that any genuine repre-sentational reference could only be to an externalobject of referenceconstitute a more immedi-ate threat to my approach.

IV. THE IMPOSSIBILITY OF SEEING THROUGH

I shall show that perceptual access to a picture Pof actual object X does not give the perceiverperceptual access to X itselfso that, even ifthere is somethingX-related that one sees whenlooking at picture P, that something cannot be X

itself. Call this the no seeing through, or NSTthesis.


4/14


To begin, the obvious point should be madethat a picture P of some actual object X is notitself identical withthat object X.12Thus, thereis a straightforward, veridical perceptual sense

in which, if one sees picture P of an actualperson X, one does not therebythat is, simply invirtue of seeing picture P itselfalso see personX, because, of course, P is not X, and thereforeit could not correctly be seen to be X. This pointremains valid whether or not object P is a pic-ture or representation of X.

Second, I want to extend this point to cover,not just direct seeing of P itself, but any kind ofperceptual access to, or perceptual involvementwithP in which a viewers gaze is directed uponpicture P, including cases in which the viewers

interpretation of what she sees is primarilyX-relatedrather than P-related, so that on someanalyses she would, strictly speaking, no longerbe seeing P itself. Thus if the general formulaViewer A sees picture P of object X is con-sidered, I wish to also consider cases in whichsubstitution-instances of that formula do notentail Viewer A sees picture P.

For example, a trompe loeil picture P ofobject X is such that, when a viewers gaze isdirected upon that picture P under normal con-

ditions, the viewer is perceptually aware onlyofsomething X-related, rather than of picture Pitself, so that in some sense she does not per-ceive picture P itself.

Nevertheless, I want to claim for theseextended cases too that, since picture P is notitself identical with object X, then Viewer A,merelyby virtue of having perceptual access toP, does not therebygain perceptual access to Xitself. Thus, I would claim that the conjunction of:

(1) Viewer A sees picture P of object X,

and

(2) Viewer A does not see object X independ-ently of her perceptual access to P

entails

(3) Viewer A does not see object X.

Statement (2) is needed to rule out a case inwhich, for instance, Viewer As perceptual fieldincludes both a picture P of X, and, independ-ently of that picture, X itselfbecause, of

course, our interest is only in what is entailed byperceptual access solely to picture P itself.

More informally or intuitively, my point isthis: there is no magical way in which, merelyby looking at a picture P of some actual objectXwhich picture is not identical with Xone

can thereby get to see X itself, no matter howone explains or analyzes what is involved inhaving perceptual access to picture P, and ofwhat is involved in Ps having representationalqualities such that it represents X. For percep-tual claims are veridical, no matter what inter-pretive juggling or rearranging goes on withrespect to what one sees in such a case. Sothat if P is not X, then one cannot correctlybesaid to see X, solely in virtue of ones looking atP, whether or not P is a representation of X, andwhatever is the correct analysis of P is a repre-

sentation of X.Next, I shall discuss an attempt to escape this

conclusion via a claim that some pictures arevisual prostheses, that is, that they are likedevices such as eyeglasses or binoculars,through which one is indeed able to see actualobjects such as X.13However, I shall show that,insofar as it is true that one is able to see X itselfwhile looking at or through such a device, itfollows that the device in question is not arepresentationof X.14

Some who argue for the visual prosthesisview, such as Walton and Lopes, do so byinvoking the concept of counterfactual depend-encethat if some visual properties of theactual object X had been different, then the cor-responding properties of the prosthesis wouldhave been different.15However, that backward-looking, past tense concept is arguably irrele-vant to issues about actually seeing X through agenuine prosthetic device. What is insteadneeded for genuine seeing of X is a forward-lookingcounterfactual dependence, or an FCDconcept: that if some visual properties of theobject X were to becomedifferent, then the cor-responding properties of the prosthesis wouldbecome different. For it is characteristic of actu-ally seeing an object X through such a devicethat changes in the object itself result in changesin the properties one sees it to have in real time(in the future or the continuous present).

But if one were to record such changes onvideotape, that recording itself would no longerbe able to change its properties to reflect any

current and continuing changes in X. Thus, inviewing or seeing that videotape, one would no


5/14


longer be seeing X itself, but only a pictureofX, since that picture is not forward counterfac-tually dependenton X. Thus, on this analysis,genuine seeing of X through a prosthesis simply

is a seeing of X itself rather than of some repre-sentation of X; whereas when one does see arepresentation of X, it does not have the FCDrelation to X itself, and, hence, does not countas an actual seeing of X.

Because of the importance of the currentissue, I shall also provide another, noncounter-factual analysis of the differences betweenprosthetic and representational seeing, which inaddition involves only changes in relationalrather than inherent properties. A useful bridgeconcept in the discussion is that of (what could

be called) environmental stability. A representa-tion R of object X is environmentally stable(hereafter: stable) in the sense that R will con-tinue to be a representation of X through normalchanges in its position or environment, such asits being moved from room to room, or if otherobjects or persons change their position relativeto the current position of representation R. Or,in more usual metaphysical terms, to say that anobject R is an external representation of X isto attribute some nonrelational properties to it, in

virtue of whichit is able to externally represent X,and which nonrelational properties are independentof its other relational properties, such as its positionrelative to other objects in its environment.

However, visual prostheses are instead envir-onmentally unstable, in that, obviously, whatone sees through eyeglasses or binoculars willchangeas the seen objects are moved around, orif other objects are moved around relative tothem. Thus, the price of being able to actuallysee the object X itselfthrough a visual prosthe-sis is that ones doing so depends on complexrelationalproperties of the prosthesis, such asits having the right kind of visual alignment toones eyes and to the actual object X. Hence,prostheses have a property, namely, instability,which stable representations necessarily lack,and so representations cannot be prostheses.

In more intuitive terms, the reason why aprosthesis cannot be a representation of X isbecause it could not stablyrepresent X, as opposedto any other things: it would represent what-ever objects were appropriately aligned with it,

and in thus representing anything, it wouldactually represent nothing.

Thus, I conclude, on the basis of both of theseanalyses, that the prosthesis defense of theclaim that some pictures both represent someactual X, and are such that in looking at them

one can see X itself, must fail, and, hence, that itis never true that perceptual involvement with arepresentation of X involves a perceiving of Xitself. Thus the no seeing through, or NST,thesis is true for external representations.

V. A WIDER CONTEXT

The argument of this paper has been primarilyfocused on pictorial representation, but beforecontinuing it will be useful to situate the discus-

sion in a wider context. I would claim that anykind of external representation, includingconventional linguistic kinds of signification,also involves an appropriate case of internalrepresentation. Thus, broadly speaking, internalrepresentation is concerned with internalreferential aspects of meaning, intension, orcontent,16 whereas external representation isinstead concerned with extension, or external,worldly reference.

In the arts, the internal representation thesis

is primarily relevant to topics such as that ofpictorial representation, and of fictional refer-ence, and my announced epistemic argumentfor internal representation is primarily relevantto pictorial representation cases. Nevertheless,it is still useful to examine such cases in thebroader context of both representational con-ventions and artistic intentions to representactual subjects, even when such attempts end infailure, that is, with no external pictorial repre-sentation or depiction having occurred.17

I shall consider three putative kinds of repre-sentation, to be labeled signification, which is abare reference to a subject, as in baptizingsomeone with a name; description, whichcharacterizes something in some way, even ifincorrectly; and standard or unqualified repre-sentation, our primary topic of interest. Thethree examples to be used all concern cases ofputative portraiture of a subject B by artist A, sothat our primary interest will be in the species ofstandard representation referred to as pictorialrepresentation, or depiction. (The examples are

brieffully adequate discussion of them wouldof course require much more space.)


6/14


Case 1. Subject B has paid artist A to painthis portrait. At the end of the session the artisthands him her painting P, which consists of anapparently random collection of blobs of paint.

Does painting P externally represent or portrayB, and if so, in what sense?Arguably, this is a case of signification only

(no description or picturing)P is of B, oraboutB, only in the sense in which a name is ofor about its bearer. Interestingly enough, thiswould count as a case of signification of B,even if artist A had had no concurrent intentionsto portray B at all while painting the picture,because the social conventions regarding por-traiture, as with naming baptisms, are performativein nature: if one performs the requisite actions

(painting) in the right social setting of agreed-on portraiture, the resultant artifact (the paintingP) thereby conventionally counts as signifyingthe subject. (The internally represented item insuch a case would presumably be a pure, non-descriptive pointer to the relevant individual B.)

Case 2. The same initial situation, in which Ahas been hired to paint Bs portrait. This time,As resultant picture P resembles a slim, shortwomanquite unlike the tall, heavy male sitterB. Again, does painting P externally represent

or portray B, and if so, in what sense or senses?As before, P does signify B for the previous

reasons. In addition, some appropriate conceptof descriptiverepresentation mightnow becomeapplicable as well. Intuitively speaking, slim,short woman is a possible descriptionalbeita significant misdescriptionof a heavy, tallman, whereas random blobs of paint as incase 1, is not a possible misdescription of B.

However, in this case, unlike case 1, theintentions of painter A are of crucial import-ance. Unless she intendedher picture P to pic-torially represent B, there would be no pointinsaying that picture P descriptively represents ormisrepresents B; it would not provide any kindof painterly description of B itselfas opposedto some other object, or no object at allnomatter how closely the two descriptively resem-ble each other. The significant degree of mis-representation in the present case tends to showthat artist A did not have the requisite inten-tionfailing special pleading on her part, suchas a claim perhaps that the painting metaphoric-

ally depicts the inner feminine side of herthickset subject B.

But then exactly what is descriptive represen-tation? I would claim that (1) no genuine exter-nal representation of any kind is involved insuch cases, and, hence, that (2) the relevant

internally represented object, namely, the slim,short woman in this case, is just an internalrepresentation of awoman, rather than of anyparticular actual woman or man. Also, the char-acteristics of the internally represented womanare only descriptively relevant to B at all ifpicture Ps contents include an intention bypainter A to depict B.

Thus, presumably, the main utility of a con-cept of descriptive representation, in caseswhere depiction has not occurred, would be toduly acknowledge the existence of near-miss

cases in which both appropriate intentions todepict an object, plus a reasonably good des-criptive match, may be inferred or seen to havebeen achieved. But even so, such cases strictlyspeaking are not failed depictions of B, sincethey do not depict (pictorially represent) B inany sense, though they must involve an intentionto depict B.

A related case is as follows. In the philoso-phy of language, referential uses of definitedescriptions are acknowledged, which achieve

reference in spite of incorrect descriptionssuch as someone saying that man over there issmoking when the smoker is in fact awoman.18

However, there are at least two reasons whythere are no analogous referential descriptivemisrepresentations among putative pictures.First, there is no close pictorial analogue of thedemonstrative linguistic element that man,supplemented by contextual factors such as agesture or head turning, which can establishreference independently of the description.And second, descriptive terms in language canfunction either to identify (as in a definitedescription) or to predicate a property of a thus-identified item. But in the case of putativepictures, there is not enough structure to distin-guish those semantic functions; such a lackwould also thwart attempts to establish thatthere are misrepresenting referential pictures ofthis specific kind.

The closest quasi-pictorial analogy to refer-ential linguistic misdescriptions would be a dis-

torting mirror, which does establish contextualreference to the objects it shows, and does


7/14


misdescribe the characteristics of the relevantobjects. However, because of the forward coun-terfactual dependence of such images on theiractual subjects (see Section IV), the relevant

objects are not representedat all in such cases;instead, the viewer simply sees the actualobjects through the mirror, as distorted by itsoptical characteristics.

Case 3. A paints Bs portrait with the inten-tion of depicting B. But this time, both A and Bare happy with the result, in that each can rec-ognizeB in the finished portrait. This finally, Iwould argue, is a genuine case of pictorial ordepictive external representation. It would alsoqualify as a case both of significative and ofdescriptive representation of B. However,

depiction, unlike descriptive representation, iscompatible with fairly radical mismatches indescriptive characteristics between picture andsubjectthat is, with genuinely depictivemis-representationas long as recognizability ispreserved.19

Thus, I claim that a necessary condition ofPs externally depicting an actual object X isthat P be recognizableas X,20with that locutionitself being explained in terms of recognizing orseeing the relevant internally represented item

X. While at the same time, near-miss cases in

which recognition is not achieved can beacknowledged to have some representationalstatusbut, nevertheless, be explained away ascases of descriptive, nondepictive representation.

However, recognition by itself does not pro-vide a sufficient condition of external depiction,because false positives are possiblepicturesthat would be generally recognized as X, butthat were not intended to externally depict X.Thus, both recognition and intention to depictare required to achieve a sufficient condition ofexternal depiction.

Now recognition can be achieved simplyfrom perception of a picture itself. But canknowledge of the relevant intention to depict beacquired in the same intrinsic manner, or areexternal or contextual information also required?

To be sure, contextual information is oftenavailable in the forms of titles of paintings,catalog descriptions, and so on, rendering theissue moot in those cases. But in the absence ofsuch aids, here is a brief suggestion as to how

knowledge of relevant intrinsic artist intentionsmight be achieved in a central class of cases.

There are stereotypical characteristics associatedwith well-known people, such as Lincolns tall,gaunt, and bearded profile. Anyone making pub-licly available a picture having such characteris-

tics is prima facie implying that the intentionwas to depict Lincoln, in the absence of anycontextual evidence to the contrary. Thus insuch cases, the same factors that enable one torecognizethat Lincoln is internally representedby the picture also establish artistic intentionstodepict Lincoln, which factors taken together area sufficient condition of the pictures externallydepicting Lincoln.

In the next section, I shall use this analysis ofpictorial representation in the presentation ofthe previously sketched (in Section III) epi-

stemic argument for the referential status of theconcept of internal representation.

VI. INTERNAL REPRESENTATION

My strategy with regard to defining and defend-ing internal representation will be threefold: (1)to accept my initial, tentative definition of theconcept as a working model, in spite of itsapparent difficulties; (2) to use the veridicality

of seeing and recognition to justify its referen-tial logic; and (3) to use the briefly sketchedepistemic approach to it to justify its conceptualindispensability.

To begin, my initial attempt at a characteriza-tion of internal representation was this: internalor inner representation is about whether there isan object X, that is internal to the object P andthat is represented by P. For example, picture Pinternally represents a man, just in case there isa man, who is internal to P and who is repre-sented by P.

As announced, I shall accept this character-ization and justify it as follows. First, if a personwith normal perceptual abilities,21under normallighting conditions, sees a picture of a man, hewill be able to see a manwhen he looks at it, orequivalently, he will be able to recognizewhathe sees as a manassuming the picture issuccessful in representing a man, that is, thatsimilar normal people under normal conditionscan see or recognize a man when they look atit.22

Thus, onesprimary evidenceas to whether apicture A is successful in representing a man


8/14


(whether or not it externally represents a particu-lar man) is derived from evidence regardingthe general abilities of people to seeor recog-nizea man when perceiving that picture A. For

whatever the intentions of the person makingthe representation, she will only have succeededin representing X if those general recognitionalconditions are satisfied.

However, the concepts of seeing and recogni-tion are veridicalconcepts, such that a claim tosee or recognize X is true only if there isan Xthat one sees or recognizes. Thus, if it is truethat one sees a man X when perceiving pictureA, then there must be a man X whom oneseesor one did not, after all, seeor recognizesuch a man X when perceiving the picture.

Hence, short of a massive error theory ofordinary reports of seeing or recognizing pic-tures, which would deny that we ever really dosee or recognize items when seeing pictures, ourordinary epistemic claims to see or recognize Xwhen perceiving A should be respected, asrequiring us to accept that there isan internalXthat is thereby seen or recognized, and, hence,to accept the current referentialcharacterizationof internal representation being argued for.

Now, it might be thought that this conclusion

is premature, on the ground that seeing or rec-ognizing X when perceiving a picture might bea case of externalrather than internal represen-tation. However, it has already been establishedin Section IV that strictly speaking one cannotsee the actualobject X, in seeing a picture of Xthat externally represents X. Hence, all one cando instead is to see or recognize the internalobject Xa case of internalrepresentation.

Now, as already pointed out, the ability to seea man while looking at a picture, or the recogni-tion of what one sees as a man, is a skill inde-pendent of whether or not the picture alsoexternally represents some particular actualman. Nevertheless, there is an important con-nection of such internal abilities with externalrepresentation as well, in that, I shall brieflyargue, it is in virtue of seeing what a picture Pinternallyrepresents (its object X) that one canjustify a claim that P externally representsobject X23given the failure of justifications interms of seeing-through approaches.

The basic reason why external representation

claims are epistemically dependent on internalrepresentation claims is because of the conjunc-

tion of the following three points that havealready been presented separately; namely,(1) claims of successful representation requireperceptual or recognitional evidence, since

mere evidence of artistic intentions to representX does not demonstrate success in doing so;(2) the only perceptual or recognitional evidenceavailable is of internal representational kinds,since strictly speaking one cannot see or recog-nize a picture as externallyrepresenting X; and(3) as claimed in the previous section, recogni-tion of an internally represented object is anecessary condition of external representationof its actual counterpart, and it becomes a suffi-cient condition when reinforced by intent toexternally represent X, which normally follows

as a matter of course via the prima facie inten-tions of those making their pictures publiclyavailable.

To be sure, one might believe that one seesthat a picture externally represents Napoleon,but one cannot actually see it thus even if thepicture isan external representation of Napoleon,for the reasons already given (because of noseeing through plus the veridicality of the con-cept of seeing). Hence, such claims of percep-tion or recognition of external representation

need to be reworked as objective, perception-independent claims about what a picture itselfexternally representsbut the only perceptualor evidential support available for such objec-tive claims is from internal representationalsources.

Now, a full justification of a claim that pic-ture P externally represents an actual man Xwould presumably require much more recogni-tional evidence than merely that one can inter-nally see picture P as aman, or see a man whenone looks at itpresumably, one also must beable to see it as a man of a certain sort, or as aman having further, sufficient identifying char-acteristics. But the ability to internally see aman when one looks at the picture remains as aprimary, evidentially necessary condition forexternal representation of the actual man X.

I take these points as showing that the epi-stemic foundations of internal representation areat leastas secure as those of external represen-tation, so that any potential attacks on myaccount of internal representation as somehow

being unfounded or subjectivist have alreadybeen defused to a significant extent.


9/14


Next, I claim that the overt or intuitivelynatural logical form of statements of internalrepresentationthat there is an object X, that isinternal to the object P, and that is represented

by Pcan itself be given an epistemic defense,in that the evidential chain justifying claims ofexternal representation itself requires that theinternal epistemic evidence have such a logicalform, that is, a logical form in which a particularinternal entity is referred to and reidentifiedas being represented by P.

To begin with, it has already been establishedthat a claim to see a man when looking at pic-ture P is, on the face of it, unavoidably referen-tialthat is, implying a claim that there is aman whom one sees. An additional reason for

accepting that point is that statements reidenti-fying that same man play an essential role in theepistemic justification of claims of externalrepresentation by a picture.

This is so, because justification has a socialdimension: in order to justify a claim that pic-ture P externally represents an actual man X,there must be intersubjective agreement on theevidence for that claim. This requires that dif-ferent persons must each be able to see picture Pas internally representing the same man X,

since otherwise the evidentiary basis for theexternal representation claimthe fact that aman X, that is, the same man Xin each case,can be seen by qualified observers when look-ing at picture Pwill fail to be established.

For example, suppose that two people, A andB, look at the same picture P, and that eachclaims, on the basis of what he sees, that thepicture externally represents Napoleon. If askedfor the evidentiary basis of his claim, A repliesthat he can see, recognize, or identify the man inthe pictureas Napoleonthat is, that the maninternally represented by the picture is recog-nized by him as being Napoleon. And supposethat person B gives the same reply to thequestion.

Now, my claim is that for B to be corroborat-ing As claim, it must be assumed or presup-posed that each is talking about the sameman inthe picture. For if A were to recognize one manX in a picture, and B were to recognize anotherman Y in the same picture, then there wouldbe no single man Z such that both A and B

recognize him as being Napoleon, and, hence,no secure ground for social agreement that

the picture therefore externally representsNapoleon.

Indeed, so basic is this point that one canhardly even make sense of the claim that A and

B might be recognizing different men in the pic-tureother than in the everyday sense that thepicture might represent twodistinct men, so thatA and B might be talking at cross-purposesabout those different men, which are visible indifferent areas of the picture.

To be sure, nonreferential analyses of what Iam calling internal representation might attemptto deny all of this; but my point is that the pos-sibility of reidentification of the same internalobject of reference is a cornerstone of an epi-stemicallydefensible account of representation.

Hence, I would claim, the overt or intuitivelogical form of statements of internal represen-tation cannot be overturned without at the sametime overturning our whole present scheme of

justification of claims that some pictures doindeed externally represent certain actual things(again, given the inadequacy of seeing-throughviews for this task). Thus, the referential logicalform of statements of internal representation isarguably at least as secure as is the referentiallogical form of statements of external represen-

tation, under our present scheme of justifica-tion.

VII. ACTUAL VERSUS REPRESENTATIONAL TRUTH

On the present view, statement-forms such aspicture P represents object X are ambiguousbetween internally versus externally representedobjects. But this also means that descriptions ofthe relevant objects are similarly ambiguous.Thus, we need to distinguish actualtruths aboutexternally represented objects, from internallyrepresented or representational truths, whichare specifically about internal objects as such.

Thus, for instance, a picture of Napoleon issuch that it is true of the person as internallyrepresented in the picture that he is Napoleon,since he can be seen to be Napoleon by any cul-turally competent person looking at the picture,so that he is Napoleon is a representationaltruthabout the person thus seen. But this repre-sentational truth about the object that can be

seen in the picture should not be confused withan actualtruth about the real person Napoleon


10/14


himselffor, as argued in Section IV, theperson thus seen in the picture could not beidentical with the actual person Napoleon.

Nevertheless, there can be actual truths about

internally represented objects, of a broadlyextrinsic kind concerning the relations of inter-nally represented objects to the real world. Forexample, it is actually true of the object seen inthe picture as Napoleon that it is represented bythe picture as beingNapoleon, and that it is notidentical with the real person Napoleon.

The distinction between representational ver-sus actual truths is in part already a familiarone, for there is an important proper subset ofrepresentational truths, namely, fictionaltruths,that have been widely discussed in the litera-

ture.24For, although the category of representa-tional truths in general applies to any kind ofinternally represented object, whether or notthere are any corresponding externally repre-sented entities, the subcategory of fictionaltruths applies only to internal objects that haveno corresponding actual object that is externallyrepresented.

I shall discuss the status of such fictionalobjects and fictional truths about them in thefollowing two sections.

VIII. THE LOGICAL STATUS OF FICTIONAL ENTITIES

I shall now show how the current view enablesa fresh, and arguably improved, account to begiven of the logical status of discourse involv-ing talk about fictionalentities. The basic ideais this. First, there are, of course, no such actualentities, so that their ontological status, asthings that do not exist, is already established.25

Second, since none of these entities actuallyexist, there are no legitimate cases of externalrepresentation of them. Third, consequently,any apparent references to such entities shouldbe construed as being cases of internal repre-sentations of such entitieswhether of linguis-tic, pictorial, or other representational kindsinwhich some representing object internally rep-resents, thus refers to, an internal object.26Andfourth, an account of fictional truthasopposed to actual truthfor such internal fic-tional entities is then potentially available,

along the lines sketched in the previous sectionfor the broader concept of representational truth.

This treatment has several related advantagesover more conventional accounts. First, in sup-port of its account of fictional truth, a relateddistinction between internal versus external

statementsand between differing kinds oftruth-conditions for eachis often made in dis-cussions of fictional characters. For example, aninternalstatement such as Hamlet is the princeof Denmark is fictionally true, while anexternal statement such as Hamlet is one ofShakespeares most complex characters is exter-nallyor actuallytrue.27But, on my account, it ispossible to construe both kinds of statement asbeing about the same fictional character: inter-nal statements describe its intrinsic characteris-tics, whereas external statements describe its

extrinsic characteristics that link it to the actualworld. Also, this account is able to preserve thereferential form of such fictional discourse,since, as argued previously, internal representa-tion is and must be logically referential.

In addition, the possibility of genuine refer-ence to specifically fictional entities is guaran-teed by the necessity of our being able togenuinely refer to (internal versions of) anykinds of entities whatsoever. Thus, no specialpleading is required to legitimate cases of fic-

tional reference, which, as referents of the onto-logically neutral kind of reference involved ininternal representation, are in no way anoma-lous, or in need of special explanation.

A related point is that what characterizesfictional entities as specifically or essentiallyfictional is something about their externalrep-resentational status, namely, that there are noexternal representations of them. But their inter-nal representational status, as just noted, is onethey share with every other kind of entity that isinternally representable.

IX. MORE ON THE STATUS OF FICTIONAL ENTITIES

As noted in the previous section, it is generallyagreed that fictional entities do not exist. Butthis basic fact is often held to be fatal, or at leasttheoretically intractable, for a broadly referen-tial account of fictional entities such as mine.For on a conventional theory of representation,according to which all representation is external

representation, there are only two possibilities ifthere are no objects to externally represent:


11/14


either one must give a nonreferential account ofthe representation of such objectswhichwould be fatal to my theoryor one mustappeal to a special class of fictional objects

that would serve as the referents of external rep-resentations of them, in spite of the fact thatsuch objects are agreed not to exist.

Such theories could be described as broadlyMeinongian,28 in that they hold that someobjects, in spite of not existing, can neverthelesssubsist, or have being, and so on. But,although other such theories have been con-structed,29 it is unclear to me at least that thetheoretical complexities involved in suchtheories have any decisive payoff in terms ofincreased theoretical understanding, or explan-

ation, of the nature of fictional entitieshence,my comment that this kind of approach is theo-retically intractable or unrewarding.

Now I shall introduce another basic feature ofthe distinction between external and internalrepresentation. In the external case, the objectrepresented is of course both logically and onto-logically independentof the object that is exter-nally representing it. It is this fact that, I wouldargue, is the basic reason why Meinongiantheories are so theoretically elusive or unre-

warding, because such entities are freestanding,independent entities whose status as objects ofreference, or as objects having an ontology of acertain kind, has no obvious support of any kindbeyond the need to postulate such entities, ifthis approach is to be adopted.

On the other hand, in the case of internalrepresentation, a case could be made that therelevant objects of reference are in some appro-priate sense referentially dependent on theactual objects that internally represent them, inthat the actual existence of those representingobjects in some way underwrites or guaranteesthe availabilityas objects of reference of appro-priate internally represented objects.

As evidence for this claim, it is perhaps obvi-ous that one cannot see a man, as internally rep-resented by a picture P, without directing onesgaze upon that same physical picture P, so thatones epistemic access to an object that P inter-nally represents is epistemically dependent onones having access to P itself30from whichepistemic dependence one may make appropri-

ate inferences to a corresponding kind of inter-nal referential dependence. Or, as a related

point, one may appeal to the representationalpowers of a given physical object, as providedby its physical design or configuration, as anexplanation of how it is possible for it to inter-

nally represent some appropriate object ofreference.This approach may also be used to explain

apparent cases of artistic creation of fictionalcharacters and events in fictional works. First, itwould seem to be true in some sense thatShakespeare created the character Desdemonain his play Othello. But, if Desdemona wascreated, then how do we avoid the conclusionthat therefore Desdemona existsor at least hassome kind of Meinongian being?

On the current approach, we can instead say

that what Shakespeare actually created, andwhat does indeed exist, is a textual representa-tion of Desdemonaas part of his originalmanuscript of his play Othellorather thanthe character herself. But that representationnevertheless made Desdemona available as anobject of internal reference of that createdrepresentation.

This account is not yet complete, for the pos-sibility of there being many distinct physicalrepresentations of the play, each being able to

make reference to the same internal characterDesdemona, still needs to be explained. In thisconnection, I have elsewhere invoked the con-cept of an originative representation,31which isone that is intended by its author both to be adefinitive representation of the play and itscharacters and to be usable by him or others tomake authorized copies of that representation,which are, in turn, intended to represent thesame characters.

Thus, on this account, there is an irreduciblyintentional element involved in any reidentifica-tion of the same character, as internally repre-sented by distinct representations of herandthis is a necessary feature of a satisfactory account,for it is generally acknowledged that merequalitative identity of representing objects, asin cases of two texts being word for wordidentical, is not sufficient for the identity of anyassociated characters or artworks.32

At the same time, this possibility of reidenti-fication of an internal object across differentrepresenting objects is arguably itself a neces-

sary feature of any genuinely referential accountof fictional or other objectssince without that


12/14


kind of objective, or at least intersubjective,status of internal objects made possible by thisaccount, talk of reference to, or reidentifica-tion of, such entities by various people in dif-

ferent situations would be a mere sham.33

There are at least two other ways of arguingthat reference to fictional characters is possible.An ordinary language approach would arguethat, since references to, and reidentificationsof, fictional characters are common in everydaylinguistic practices, such uses need no add-itional justification beyond their bedrock lin-guistic status.

A related but more sophisticated approachwould distinguish a surface level or theory offictional reference from possible deeper theories,

which might or might not maintain genuinereference to fictional entities. (I argued for sucha surface approach as one form of legitimatetheorizing about reference to fictional entities ina previous paper.34)

Intuitively, the idea is that such a surfacetheory could be genuinely true, and explanatory,as applied to surfacephenomena, such as every-day linguistic references to fictional characters,whether or not a deeper or more reductivetheory might explainor explain awaysuch

apparent references in other terms. On such asurface view one can both acknowledge what isvalid in ordinary language approaches, yetalso allow the possibility of more fundamentallevels of theorizing.

But now that the current account of referenceto fictional entities in terms of internal represen-tation is available, I would claim that it providesthe only necessary deeptheory of fictional enti-tiesto complement, underlie, and explain the

just-discussed surface theory. Thus, on myaccount, fictional reference is genuine in bothsurface and more fundamental theoretical ways.

X. CONCLUSION

To conclude, here is a brief summary of theargument of the paper. In the introduction theapparent ambiguity of statements such as pic-ture P represents object X was introduced,along with a preliminary definition of internalrepresentation as P internally represents object

X. Section I allayed concerns about reference

to an internal object, in that such references are

not existence-entailing, while Section II pointedout the prima facie epistemic unavoidability oftwo distinct concepts of representation, in caseswhere a perceiver is ignorant of the external

representational status of a picture.Section III sketched another epistemic argu-mentspelled out in Section VIaccording towhich, even in a case where a picture P doesexternally represent an object X, the concept ofinternal representation still has an importantepistemic role to play in establishing that fact.Section III also introduced two possible objec-tions to my account: first, a seeing-throughobjection that would replace apparent cases ofinternal representation with perception of actualexternally represented objects, which account I

refuted in Section IV. And second, an alterna-tive nonreferential account of apparent internalrepresentation cases was sketched, which Ilater indirectly rebutted in Section VI viaarguments there that imply that it could not pro-vide a complete account of all representationalcases.

Section IV argued for the impossibility ofseeing through a representation to the objectit externally represents, for example, becausegenuine representations, unlike visual prosthe-

ses, are not forward counterfactually depend-ent on externally represented objects. Section Vmaintained the internal versus external ambigu-ity of representation in a broader context of con-ventional and intentional factors, distinguishinggenuine representation from mere name-likesignification and a supposed descriptive repre-sentation, while also claiming the necessityand joint sufficiency, for external representa-tion, of recognition of an internally representedobject plus intention to externally represent.

Section VI drew on previous results, plus theveridicality of the concepts of seeing and recog-nition, to argue both that recognition is of aninternally represented object X, and that it is invirtue of seeing such an object Xthat one canjustify a claim that P externally represents thecorresponding actual object X, so that the epi-stemic and logical status of internal representa-tion is at least as secure as that of externalrepresentation.

Section VII distinguished actual from repre-sentational truths, while Section VIII described

the advantages of a view of fictional objects asinternally represented objects. And finally,


13/14


Section IX argued for the referential andepistemic dependence of such objectswhichhelps to distinguish them from theoreticallysuspect Meinongian objectswhile also arguing

that the current internal representation viewprovides a deeper theoretical justification forprevious views of fictional reference.35

JOHN DILWORTH

Department of Philosophy

Western Michigan University

Kalamazoo, Michigan 49008

INTERNET:[email protected]

1. Richard Wollheim, Painting as an Art(Princeton Uni-versity Press, 1987), pp. 6771.

2. Kendall L. Walton,Mimesis as Make-Believe: On theFoundations of the Representational Arts(Harvard Univer-sity Press, 1990), chap. 3.

3. Nelson Goodman,Languages of Art: An Approach toa Theory of Symbols (Indianapolis: Bobbs-Merrill, 1968),chap 1.

4. Although there may be special cases of an objectexternally representing itself, or parts of itself, that straddlethe distinction being made here, I shall ignore such cases inwhat follows.

5. Thus, I shall not consider more generic kinds of repre-sentation in which, for instance, a picture of a man in an

encyclopedia might be considered as making plural refer-ence to men, or to anyactual man, rather than to a particularactual man.

6. I distinguish my account from a Meinongian accountof reference to nonexistent objects in Section IX.

7. Thus, my account has some affinities with Brentanosconcept of the intentional inexistence of intentionalobjects, though such connections will have to be exploredelsewhere.

8. Goodman,Languages of Art: An Approach to a The-ory of Symbols, pp. 2126.

9. Interestingly enough, Goodman himself can be con-strued as arguing that picture of a man is ambiguousbetween a relational versus a nonrelational, sortal (man-

picture) interpretation. But his view still supports a unitaryview of representation, according to which only externalrepresentation is genuinely relational.

10. However, arguably, the standard view of representa-tion, based as it is on external representation, is itself ad hocin an inverse way, in that it cannot be generalized to covercases when there is no externally represented object.

11. This is not to attack the concept of representationalcontent itself, but only an appeal to it as supposedlypreempting the postulation of such an object X.

12. And in general, a picture of anX is not itself an X:representations are not instances of the things they represent.

13. Kendall Walton holds such a view for photographs inTransparent Pictures: On the Nature of Photographic Real-

ism, Critical Inquiry11 (1984): 246277, while DominicLopes extends it to pictures generally in his book Under-

standing Pictures(Oxford University Press, 1996), chap 9.For a critique of such views independent of mine, see NigelWarburton, Seeing through Seeing through Photographs,

Ratio1 (1988): 6474.14. Or, more precisely, under these conditions it is not

functioning as a representation of X: there might, forinstance, be cleverly constructed binoculars that both exter-nally represent X when one looks at rather than throughthem, but that nevertheless enable one to see X itself whenone does look through them.

15. Walton, Transparent Pictures: On the Nature ofPhotographic Realism, and Lopes, Understanding Pic-tures.

16. A thesis that allaspects of content or meaning maybe explained in terms of internal representation will not bedefended here, though, on my view, it is defensible.

17. My thanks to the editor, Susan Feagin, for suggestingthe relevance of such cases, and of analogous cases in thephilosophy of language such as the one discussed below.

18. See, for example, Keith Donnellan, Reference andDefinite Descriptions, Philosophical Review 75 (1966):281304.

19. Antonia Phillips provides a convincing account ofthe centrality of recognition in portraiture in her paper TheLimits of Portrayal, in Philosophy and the Visual Arts, ed.A. Harrison (Dordrecht, The Netherlands: Reidel, 1987),pp. 317341.

20. Arguments for this view are given in my paper Rep-resentation and Resemblance, The Philosophical Forum12(1980): 139158, while Lopes, Understanding Pictures, andPhillips, The Limits of Portrayal, provide other argumentsfor recognitional views of representation. The followingsection also provides relevant arguments.

21. By normal here I mean only the basic perceptualcompetencies possessed by almost everyone, independentof more uncommon additional abilities to make sophisti-cated artistic perceptual discriminations.

22. Lopes, Understanding Pictures, gives a usefulaccount of some relations between recognition and repre-sentation, though he assumes a unitary concept ofrepresentation.

23. Lopes, Understanding Pictures, provides a survey ofvarious factors possibly relevant to justification.

24. See, for example, Walton, Transparent Pictures: Onthe Nature of Photographic Realism, and the referencestherein.

25. However, an account is still possible of the sense in

which it is true that Hamlet exists in the fictional world ofthe play Hamlet. See my paper The Fictionality ofPlays, The Journal of Aesthetics and Art Criticism 60(2002): 263273. Also see my Theater, Representation,Types and Interpretation, American Philosophical Quar-terly39 (2002): 197209.

26. See my The Fictionality of Plays.27. See, for example, my The Fictionality of Plays,

and the included references.28. Alexius Meinong, On the Theory of Objects, in

Realism and the Background of Phenomenology, ed. RoderickChisholm (Glencoe, IL: Free Press, 1960), pp. 76117.

29. For example, Terence Parsons,Nonexistent Objects(Yale University Press, 1980).

30. In my The Fictionality of Plays I provide ancillarydiscussions of the epistemic authority of representations in


14/14


helping to determine the internal truth-conditions forfictional statements.

31. In my A Representational Theory of Artefacts andArtworks, The British Journal of Aesthetics 41 (2001):353370, and The Fictionality of Plays.

32. As Borges showed, on one interpretation of hisstory, with his well-known example of a fictitious workby one Pierre Menard, the text of which was word forword identical with the text of a section of Cervantess

Don Quixote, but whose aesthetic qualities were quite

different, hence, ensuring its identity as a literaryartwork distinct from the relevant part of Don Quixoteitself. Borges, Pierre Menard, Author of the Quixotein Labyrinths, (Harmondsworth, Middlesex: Penguin,1985).

33. See also Section VI on this point.34. The Fictionality of Plays.35. My thanks to the editor, Susan Feagin, and to an

anonymous referee of The Journal of Aesthetics and ArtCriticismfor very helpful comments.

Documents

Internal Versus External Representation DilworthJ JAAC04