robinson 1978_two theories of representation

18
8/12/2019 Robinson 1978_Two Theories of Representation http://slidepdf.com/reader/full/robinson-1978two-theories-of-representation 1/18 Two Theories of Representation Author(s): J. Robinson Source: Erkenntnis (1975-), Vol. 12, No. 1, The Philosophy of Nelson Goodman, Part 1 (Jan., 1978), pp. 37-53 Published by: Springer Stable URL: http://www.jstor.org/stable/20010577 . Accessed: 13/07/2013 07:01 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 [email protected].  . Springer  is collaborating with JSTOR to digitize, preserve and extend access to Erkenntnis (1975-). http://www.jstor.org This content downloaded from 89.206.117.115 on Sat, 13 Jul 2013 07:01:09 AM All use subject to JSTOR Terms and Conditions

Upload: 3468who

Post on 03-Jun-2018

229 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Robinson 1978_Two Theories of Representation

8/12/2019 Robinson 1978_Two Theories of Representation

http://slidepdf.com/reader/full/robinson-1978two-theories-of-representation 1/18

Two Theories of RepresentationAuthor(s): J. RobinsonSource: Erkenntnis (1975-), Vol. 12, No. 1, The Philosophy of Nelson Goodman, Part 1 (Jan.,1978), pp. 37-53Published by: Springer

Stable URL: http://www.jstor.org/stable/20010577 .

Accessed: 13/07/2013 07:01

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 [email protected].

 .

Springer  is collaborating with JSTOR to digitize, preserve and extend access to Erkenntnis (1975-).

http://www.jstor.org

This content downloaded from 89.206.117.115 on Sat, 13 Jul 2013 07:01:09 AMAll use subject to JSTOR Terms and Conditions

Page 2: Robinson 1978_Two Theories of Representation

8/12/2019 Robinson 1978_Two Theories of Representation

http://slidepdf.com/reader/full/robinson-1978two-theories-of-representation 2/18

J. ROBINSON

TWO THEORIES OF REPRESENTATION

i

In the first chapter of his Languages of Art Nelson Goodman argues that

pictorial representation is a form of symbolization or reference:

The plain fact is that a picture, to represent an object, must be a symbol for it, stand

for it, refer to it ;and that no degree of resemblance is sufficient to establish the requisite

relationship of reference.1

For the purposes of this paper I am going to assume that Goodman is

right in this claim. What I want to discuss is a further question, namely

what theory of reference best accommodates his insight. David Kaplan

once

suggestedthat there are two

possibletheories of

pictorial reference,one based on the descriptive content of a picture and the other on its

genetic character .2 In this paper I develop this suggestion, by constructing

two theories of pictorial representation, one based on Frege s theory of

naming and the other on Kripke s. According to the Fregean theory, what

a picture represents is a function of its sense , whereas, according to the

Kripkean theory, what a picture represents is a function of its history .

I argue that while both theories illuminate the concept of pictorial repre?

sentation, neither states necessary and sufficient conditions for its correct

application,even

thoughit often

happensthat

Kripkeanand

Fregeanconditions on representation jointly determine the reference of a picture.

However, it seems tome that neither theory, either alone or in conjunction,

can account for metaphoricalcases of representation-as

or for false

pictures.

ii

Assume that a representational picture behaves like a singular term. Then

on aFregean theory

ofrepresentation

apicture has both

senseand

reference and it refers to whatever it represents in virtue of its sense.3

Representational pictures that do not represent any existing thing (such

Erkenntnis 12 (1978) 37-53. All Rights Reserved

Copyright ? 1978 by D. Reidel Publishing Company, Dordrecht-Holland

This content downloaded from 89.206.117.115 on Sat, 13 Jul 2013 07:01:09 AMAll use subject to JSTOR Terms and Conditions

Page 3: Robinson 1978_Two Theories of Representation

8/12/2019 Robinson 1978_Two Theories of Representation

http://slidepdf.com/reader/full/robinson-1978two-theories-of-representation 3/18

Page 4: Robinson 1978_Two Theories of Representation

8/12/2019 Robinson 1978_Two Theories of Representation

http://slidepdf.com/reader/full/robinson-1978two-theories-of-representation 4/18

TWO THEORIES OF REPRESENTATION 39

give us different information about Venus just as the two descriptions The

Morning Star and The Evening Star give us different information about

Venus. In general, it seems that a picture which represents a as a b is a

picture the represented properties of which are those of a b yet determine

the reference of the picture to be a. For example, a picture of Venus as the

Morning Star represents Venus via the represented properties of the

Morning Star. It attributes to Venus certain properties, such as appearing

in themorning which would not be attributed to her by a picture of Venus

as the Evening Star. More problematic, however, as we shall see later, are

metaphorical cases of representation-as such as a picture that represents

Churchill as a lion or the Gillray cartoon that represents Pitt as a parasite

on the crown.

m

There are at least two major objections to our Fregean theory of representa?

tion.

(a) If we hold that the sense of a picture determines its reference in the

way I have outlined, then it follows that a picture represents whatever the

properties represented by it belong to. If, for example, a portrait of my

grandfather gives a true pictorial description of your grandfather, in the

same pictorial symbol system, then the portrait represents your grand?

father as well as mine, in that system. But we would not normally say this.

Rather we would say that this is a portrait of my grandfather and, curiously,

it

happens

to be true of, to describe

correctlyor, if it is in a realistic

symbolsystem to look like your grandfather also. Similarly, a picture that repre?

sents one identical twin would ipso facto represent the other as well. Worse

still, a picture purporting to represent Pickwick that gives a true pictorial

description of my grandfather would, on our Fregean theory, thereby

denote my grandfather, but ex hypothesi a picture of Pickwick fails to

denote.

The chief reason why cases of this sort can arise is that pictures do not

always represent either the essential properties of things or species, or those

propertiesof

thingsor

speciesthat

uniquelyindividuate them. It

might,for

example, be an essential property of Aristotle and part of the sense of the

name Aristotle that he was the most famous pupil of Plato. But being the

This content downloaded from 89.206.117.115 on Sat, 13 Jul 2013 07:01:09 AMAll use subject to JSTOR Terms and Conditions

Page 5: Robinson 1978_Two Theories of Representation

8/12/2019 Robinson 1978_Two Theories of Representation

http://slidepdf.com/reader/full/robinson-1978two-theories-of-representation 5/18

40 J. ROBINSON

most famous pupil of Plato is not a property that is easily picturable. We

could, of course, paint a picture of Aristotle and have him announce

through a think-bubble that he is the most famous pupil of Plato, or we

could entitle our picture The most famous pupil of Plato , but neither of

these manoeuvres is permissible if we want to treat a picture as a truly

pictorial symbol, i.e., as something that functions as a symbol in virtue

of its configuration of colours, lines or brushstrokes. (Of course this is too

simple as a definition of a pictorial symbol since inscriptions of wordsare also lines and patches of colour, but at least we can say that words are

in a linguistic, not a pictorial symbol system, even if the distinction between

the two is somewhat obscure.)9 Similarly, it is presumably part of the sense

of the name Pickwick that he was the creation of Charles Dickens. But,

again, even a picture of Dickens with Pickwick in his think-bubble could

be a picture of Dickens thinking about my grandfather. It seems to be

impossible to rule out this kind of ambiguity.

The essential properties of eagles are not necessarily easily picturable

either. A dictionary picture of an eagle is less likely to give a pictorial

characterization of the internal structure of an eagle, which may be one of

its essential properties, than simply to give a rough idea of an eagle s shape.

Hence it is quite possible that, on the one hand, the picture is not true of

every eagle, since some eagles may have only one leg or a broken bill, and,

on the other hand, the picture is true of some buzzards as well as many

eagles. Thus if we identify the sense of a singular or general term with a

set of essential properties, then the analogy with representational pictures

seems to break down :a picture denoting any member of the set of eagles

will not necessarily give a pictorial characterization of all, or even any, of

an eagle s essential properties, nor will pictures representing Aristotle or

Pickwick necessarily give a pictorial characterization of Aristotle s or

Pickwick s essential properties.

However, we might, with considerable warrant from Frege himself,

choose to identify the sense of a singular term not with the essential

properties of the object denoted by that term, but with the properties of

that object by means of which it can be uniquely individuated.10 For

example, being the most famous pupil of Plato might not be essential to

Aristotle, but itmay indeed be a property that distinguishes him from other

men. Unfortunately, however, even the most detailed realistic picture is

This content downloaded from 89.206.117.115 on Sat, 13 Jul 2013 07:01:09 AMAll use subject to JSTOR Terms and Conditions

Page 6: Robinson 1978_Two Theories of Representation

8/12/2019 Robinson 1978_Two Theories of Representation

http://slidepdf.com/reader/full/robinson-1978two-theories-of-representation 6/18

TWO THEORIES OF REPRESENTATION 41

not always sufficient to distinguish one individual from another. We might

identify my grandfather (and thus distinguish him from your grandfather)

verbally by means of some description specifying his parentage, the exact

time and place of his birth, or even the fact that he ismy grandfather (and

not yours), but there is no unambiguous pictorial way of stipulating such

properties. And similarly when we try to distinguish my grandfather from

Pickwick, or even, perhaps, eagles from hawks.

In short, whether we interpret the sense of a picture to be a set of

essential properties or a set of properties that individuate a referent or

referents, guaranteed uniqueness of reference and guaranteed failure of

reference seems to be impossible. Hence Frege s principle that sense

uniquely determines reference does not seem to hold for pictures.

(b) A second important objection to our Fregean theory is that it fails

to account for representation-as. On this theory, a picture that represents

Churchill as a soldier, for example, is a picture whose represented proper?

ties are those of a soldier and which, in virtue of those represented

properties, denotes Churchill. Hence a picture of Churchill as a lion should

be a picture whose represented properties are those of a lion and which, in

virtue of those represented properties, denotes Churchill. In other words,

its sense is a set of lion-properties but its reference is not a lion but a man,

Churchill.

In fact there seem to be two kinds of pictures that might represent

Churchill as a lion :(1) what we might call pictures of straightforward lions

(straightforward pictures of lions?) and (2) pictures of rather odd lions,

that smoke cigars, harangue the beleaguered troops or lounge on the

Front Bench of the Commons. In neither of these two sorts of case does

the sense of the picture determine its reference. Churchill is neither an

ordinary nor a peculiar lion: he is not a lion at all.

The first sort of picture is clearly recalcitrant to our Fregean analysis. A

picture of a lion is simply not a correct pictorial description of Churchill.

The second sort of picture ismore complex. The picture represents some

leonine properties and some Churchillian ones. For example, it depicts a

lion with a Churchillian face smoking a cigar on the Front Bench. Now,

on the Fregean analysis of representation-as, the sense of this picture

should be a set of lion-properties but in fact some of the properties repre?

sented are properties no lion ever had or could have, such as the possession

This content downloaded from 89.206.117.115 on Sat, 13 Jul 2013 07:01:09 AMAll use subject to JSTOR Terms and Conditions

Page 7: Robinson 1978_Two Theories of Representation

8/12/2019 Robinson 1978_Two Theories of Representation

http://slidepdf.com/reader/full/robinson-1978two-theories-of-representation 7/18

42 J. ROBINSON

of Churchillian lips. Furthermore, the picture should denote Churchill in

virtue of its represented properties or sense, but the very properties that

make us want to say that this is a picture of Churchill are precisely those

properties that no lion ever had or could have. It is not the golden mane

and the four paws represented that makes us want to say this is a picture

of Churchill, but the Churchillian features and the cigar. We might want to

say, perhaps, that the sense of the picture is ambiguous: it could represent

either Churchill or a lion. But this, of course, is simply to misinterpret the

picture. In short, the Fregean theory of representation-as is faced with

grave difficulties in metaphorical pictures.

Another problem for the Fregean theory is the existence of false

pictures . It seems to be possible for a picture to give us false information

about its subject. Thus a picture of the adult Churchill with a lot of blond

curly hair is not (entirely) true of Churchill. A picture of the Parthenon

with the wrong number of columns along the front is not true of the

Parthenon. Yet we might say nonetheless that the pictures represent

Churchill and the Parthenon respectively, although they ascribe to the

objects they refer to properties that those objects do not in fact have. If

this is so, then clearly the Fregean theory cannot account for the fact,

since on that theory a picture represents what it represents precisely by

giving true information about it.

Finally, it is difficult to see how the Fregean theory can distinguish

between a picture of a as a b and a picture of b as an a.What, for example,

would be the difference between a picture of a melon as an orange and a

picture of an orange as a melon ?

IV

Our second theory of pictorial representation is based on Kripke s theory

of names.11 Again, we consider representational pictures as behaving like

singular or general terms, but on our Kripkean theory what a picture

represents is a function not of its sense but of its history.12

Kripke wants to deny the Fregean theory that proper names have sense

and that they refer in virtue of their sense. He claims, roughly speaking,

that a proper name refers to a particular individual in virtue of certain

historical facts about that name. Thus human beings normally get their

This content downloaded from 89.206.117.115 on Sat, 13 Jul 2013 07:01:09 AMAll use subject to JSTOR Terms and Conditions

Page 8: Robinson 1978_Two Theories of Representation

8/12/2019 Robinson 1978_Two Theories of Representation

http://slidepdf.com/reader/full/robinson-1978two-theories-of-representation 8/18

TWO THEORIES OF REPRESENTATION 43

proper names partly in virtue of their parentage and partly via baptism or

some such naming or dubbing ceremony. The name Saul Kripke is

correctly applied to Saul Kripke not because Saul Kripke refers to

whoever instantiates the properties which constitute the sense ofthat name

but because that name was bestowed on Kripke at birth, his parents called

him by that name, other people met him and learnt his name and so on.

In other words, what Kripke calls a chain of communication is set up

and the name is passed from link to link of this chain. Someone on the

far end of the chain may use the name Saul Kripke to refer to Kripke

without knowing anything uniquely true of Kripke or any of Kripke s

essential properties; all he may know, for example, is that Kripke is a

logician. But it is nevertheless quite possible, on the present view, for such a

person to succeed in referring to Kripke by his use of the name Saul

Kripke simply by being on the end of the chain of communication .

Clearly, then, the reference of Saul Kripke is not a function of its Fregean

sense but of its genesis and history.

If portraits behave like Kripkean proper names, then presumably they

too get their reference from an initial dubbing of some sort. I would

suggest that those features of a picture s genesis that are most likely to

determine its reference are itsmodel or sitter, its title and the intentions of

its artist respecting it. In many cases I think there is a considerable

temptation to say that a portrait gets its reference in just this way. Thus

perhaps the main reason why a picture purporting to represent Pickwick

but also giving a true description of my grandfather is not normally said

to be a picture of or to represent my grandfather is that the picture has

had no historical link with my grandfather; he did not sit for the picturenor did the artist intend to represent him etc. In short he did not play any

relevant causal role in the production of the picture. Similarly, where a

picture gives a true description of two identical twins, it is normally said

to represent only that one twin who was intended as the subject of the

painting, who sat for it etc. Of course, the artist may deliberately have

painted a picture designed to be a true description of both twins, but this

would be an unusual case. Normally we assume that a portrait refers

uniquely and that if it gives a true description of more than one person, it

is historical factors that are brought into consideration to determine which

of these persons the picture in fact represents. Similarly, the Kripkean

This content downloaded from 89.206.117.115 on Sat, 13 Jul 2013 07:01:09 AMAll use subject to JSTOR Terms and Conditions

Page 9: Robinson 1978_Two Theories of Representation

8/12/2019 Robinson 1978_Two Theories of Representation

http://slidepdf.com/reader/full/robinson-1978two-theories-of-representation 9/18

44 J. ROBINSON

theory offers an explanation for cases like the Wohlgemuth woodcut

which in the Medieval N?rnberg Chronicle represents Mantua on one page

and Damascus on another. On this account we can say that the woodcut

represents both cities not because it is true of both, which seems unlikely,

but because the woodcut is entitled Mantua on one page and Damascus

on the other, and clearly the artist intended the woodcut to represent

Mantua on one page and Damascus on the other. These examples

strongly suggest, then, that even if the sense of a uniquely referring pictureor portrait may be relevant to determining its reference, it is not the only

determining factor.

Pictures that behave like general terms or indefinite descriptions, with

distributive reference to any object satisfying the description, can also be

handled by analogy with Kripke s theory of the reference of terms for

natural kinds.13 According toKripke, we pick out paradigm cases of some

natural kind such as the eagle and then fix the reference of the general

term eagle by saying that anything of that kind is to be classified as an

eagle. We may say that all those things over there are to be termed eagleson the grounds that they nest in the same area, look alike, intermarry, have

the same internal structure etc., but this does not preclude the possibility

of something s being an eagle without looking like, intermarrying with,

nesting close to or having the same internal structure as our paradigm

eagles. Thus the dictionary picture of an eagle picks out only a paradigm

eagle or set of eagles: it does not claim to give an accurate description of

every eagle, nor does it claim to describe nothing but eagles. In this respect

the Kripkean theory has an advantage over the Fregean: it is in fact

very unlikely that the dictionary picture will be true of all and only

eagles.

What Kripke would say about non-referring pictures, such as pictures

of unicorns or centaurs, is less clear. Since there are no paradigm unicorns,

it is hard to see how there could ever have been an initial dubbing ceremony

whereby the general term unicorn or a picture of a unicorn came to refer

to a particular kind or species. Similar problems beset pictures of Pickwick.

Of course, Phiz s drawings purported to refer to that same person whom

Dickens purported to refer to by the use of the name Pickwick . But

although Dickens makes believe that he has performed a genuine dubbing

ceremony by inventing the name Pickwick and using it as though he were

This content downloaded from 89.206.117.115 on Sat, 13 Jul 2013 07:01:09 AMAll use subject to JSTOR Terms and Conditions

Page 10: Robinson 1978_Two Theories of Representation

8/12/2019 Robinson 1978_Two Theories of Representation

http://slidepdf.com/reader/full/robinson-1978two-theories-of-representation 10/18

TWO THEORIES OF REPRESENTATION 45

referring to a particular individual, the dubbing is illusory: there is really

no-one being dubbed.

Despite these lacunae in the Kripkean theory, it does seem to be able to

account for many important facts about representation. However, it is

not so easy to see how it can cope with representation-as, since according

to this theory how a picture represents an object (what the object is repre?

sented as) is irrelevant to determining what object, if any, that picture

represents. Nevertheless, there is, I think, an interesting way of attempting

to solve this difficulty. Itmight be argued that what a picture represents is

a function of its history, whereas how a picture represents something (what

it represents-as) is a function of the properties it represents as determined

by a mapping from the pictorial properties (what colours go where on the

canvas). In other words, the Kripkean theory explains what a picture

represents and the Fregean theory explains what a picture represents-as.

This view may at first sight appear to combine the best features of both

the Fregean and the Kripkean theories and at the same time to give a con?

vincing account of representation-as. For it follows on this view that if a

represents b as a c, then a is a picture that represents properties of a c but

has b as itsmodel, is intended to represent b, has ? as its title etc. Thus a

picture of Churchill as a lion is a picture whose pictorial properties deter?

mine represented properties such as having a golden mane, having four

paws and having sharp fangs, but whose reference is determined by the

fact that the artist intended this particular picture of a lion to be an inspir?

ing vision of Churchill, that he entitled it Churchill on the Eve of the

Normandy Invasion etc. etc. Moreover, this account has the added advan?

tage of being able to make a distinction between a picture of a lion that

represents Churchill and what we might christen a picture of Churchill-as

a-lion that represents Churchill. A picture of a lion that represents Churchill

is what in ordinary language we would call a picture of a lion but which has

Churchill as its title (not, presumably, its sitter), whereas a picture of

Churchill-as-a-lion that represents Churchill will not be a picture of a lion

tout court but a picture of a lion that is smoking a cigar, haranguing the

beleaguered troops or lounging on the Front Bench of the Commons. In

the first case the properties represented by the picture are entirely leonine;

in the second case they are partly leonine, partly Churchillian. This theory

also allows us to make a clear distinction between a picture representing a

This content downloaded from 89.206.117.115 on Sat, 13 Jul 2013 07:01:09 AMAll use subject to JSTOR Terms and Conditions

Page 11: Robinson 1978_Two Theories of Representation

8/12/2019 Robinson 1978_Two Theories of Representation

http://slidepdf.com/reader/full/robinson-1978two-theories-of-representation 11/18

46 J. ROBINSON

melon as an orange and a picture representing an orange as a melon. As

we have seen, the Fregean theory alone cannot make this kind of distinc?

tion. Finally, it offers an explanation of false pictures. A picture of the

Parthenon with the wrong number of columns along the front is a picture

whose model was the Parthenon but whose descriptive content is in?

accurate.

v

The Kripkean theory of representation does seem to be able to handle some

of the problems that the Fregean theory alone could not. Nevertheless, as a

theory that purports to explain representation entirely it is grossly

inadequate. In this section I shall argue (a) that the theory does not give

sufficient conditions for pictorial representation, (b) that there are good

grounds for thinking that it does not even state necessary conditions on

pictorial representation, and (c) that representation-as remains a problem.

(a)There are at least two

convincingkinds of

counter-exampleto the

claim that the Kripkean theory provides sufficient conditions for pictorial

representation. The first sort of case is that of bad (or mad) pictures. For

example, a picture intended by its painter to be a picture of Queen Victoria,

entitled Queen Victoria and having Queen Victoria as its model may

nevertheless fail to represent Queen Victoria if the painter is incompetent.

Significantly, we judge a painter s competence at representation by his

ability to give a correct, or at least appropriate, pictorial characterization

of his model in the symbol system he is using. For a picture to represent the

queen,there must be a

mappingfrom its

pictorial propertiesto the

repre?sented properties of the queen, as the Fregean theory demands. The

Kripkean theory ignores the crucial Goodmanian requirement that a

picture is a character in a symbol system and that it functions as such in

virtue of its pictorial properties and not, say, its price or, more particularly,

its origin and history. Thus a schizophrenic who claims that a squiggle he

draws represents Queen Victoria is not believed unless he can point to

features of his squiggle that map into represented properties of the queen.

(Perhaps, for example, a constricted squiggle represents the queen as

havinga constricted

personality.)14A second problem for our Kripkean theory is posed by pictures which

This content downloaded from 89.206.117.115 on Sat, 13 Jul 2013 07:01:09 AMAll use subject to JSTOR Terms and Conditions

Page 12: Robinson 1978_Two Theories of Representation

8/12/2019 Robinson 1978_Two Theories of Representation

http://slidepdf.com/reader/full/robinson-1978two-theories-of-representation 12/18

TWO THEORIES OF REPRESENTATION 47

do not purport to represent the objects which served as their models. A

good example is that of the Gainsborough landscape pictures which do not

represent any actual landscapes but which were modelled on small, delicate,

carefully constructed compositions of dry leaves, feathers, bits of broken

glass, and so on. Intuitively we do not want to say that the resulting pictures

represent feathers and bits of glass, nor even is it true that the pictures

represent feathers and bits of glass as trees and lakes. Now, it could be

argued that the historical evidence is ambiguous in such cases: thus

although the model for a Gainsborough landscape may have been a con?

struction out of feathers etc., Gainsborough s intention was to paint a

landscape (although possibly not any actual landscape). This ambiguity

points to adifficulty

in applying theKripkean model to pictures ;for, where?

as a proper name gets its reference via a conventional dubbing ceremony

such as a baptism, a portrait does not seem to bear such a simple conven?

tional relation to its model. Although it is true, as Gombrich has con?

vincingly shown, that the artist does not copy or imitate what he sees,15 it is

nevertheless impossible for just anything to serve as a model for a par?

ticular picture. For example, if Gainsborough had not been able to see

his bits of glass as lakes and his feathers as trees, he would not have used

them as models for a landscape.16 In short, Gainsborough s intention to

paint a landscape is not independent of his choice of model as our objection

seems to assume.

Another example where the model for a picture is not - or not uniquely-

what that picture represents is the picture of the eagle in the dictionary.

The model for the picture may have been a particular eagle, but the picture

does not refer just to this eagle but distributively to every eagle. Similarly

in the case of a picture of amelon (some melon or other). The picture may

have had a particular melon as its model without having that melon as its

sole referent. In this case the Fregean theory seems much closer to the

truth: the picture refers to whatever meets the pictorial description it

gives. It is often said, somewhat paradoxically, that the significance of

works of art lies at once in their uniqueness and in their universality. This

doctrine may be partly explained with respect to representational painting

by the fact that, for example, a picture of a melon by a great artist may

point out certain features of a melon, such as its peculiar colour, solidity

and texture, which are unique to melons but which tend to be true of all

This content downloaded from 89.206.117.115 on Sat, 13 Jul 2013 07:01:09 AMAll use subject to JSTOR Terms and Conditions

Page 13: Robinson 1978_Two Theories of Representation

8/12/2019 Robinson 1978_Two Theories of Representation

http://slidepdf.com/reader/full/robinson-1978two-theories-of-representation 13/18

48 J. ROBINSON

melons, or at least of all the most melon-ish melons, i.e., those that have

sensible properties typical of melons to a pronounced degree.17 So a

picture of a melon, having a particular melon as its sitter , may in fact

refer to all melons having the melon-properties represented in the picture.

(b) I hope to have shown that our Kripkean theory does not state

sufficient conditions on representation. Whether it gives us necessary

conditions, however, is a trickier question, chiefly because it seems to me

that we do not have any strong intuitions about the difficult cases. For

example, Imay intend to paint a picture of Queen Victoria in a realistic

symbol system and I use her as my model. I am an incompetent painter

and do not expect great results, but by some freak the picture turns out to

be an extremely good likeness of Jane Smith (on a realistic symbol

system). Let us assume there is no subtle aesthetic or psychological point

to the fact that my picture looks like Jane; it is a mere accident, a slip

of the brush. Now, my intuition would be to say that despite the history of

the picture it is a picture of Jane because it gives an accurate pictorial

description of Jane. I will exclaim in surprise I ve painted a picture of

Jane . But what if Jane were not even born at the time I painted the picture ?

Suppose my picture languishes in an attic for a century before being

unearthed by a friend of Jane s. There is, I think, a strong inclination to say

in this case that the picture cannot represent Jane, and presumably the

origin of this inclination is that she was not historically associated with the

picture in the relevant way. Thus our intuitions in these two cases seem to

me to lead in different directions.

Another problematic case occurs when a picture gives partly true

characterizations of two different people, only one of which is historically

connected with the picture in the relevant way. Thus Imay intend to repre?

sent Queen Victoria and use her as my model, yet the picture turns out to

look like Jane wearing the Crown Jewels. Even in this case it seems tome

that there is a strong inclination to say that the picture represents Jane

(assuming she is alive and known to me) if the properties of Jane that the

picture represents are more important properties than those of the queen.

Thus in a portrait it is usually the accuracy of the representation of the

features that is crucial to determining who is represented, rather than the

clothes or jewels worn. Again, historical facts about the picture seem tome

to be overruled by Fregean considerations when we try to determine the

This content downloaded from 89.206.117.115 on Sat, 13 Jul 2013 07:01:09 AMAll use subject to JSTOR Terms and Conditions

Page 14: Robinson 1978_Two Theories of Representation

8/12/2019 Robinson 1978_Two Theories of Representation

http://slidepdf.com/reader/full/robinson-1978two-theories-of-representation 14/18

TWO THEORIES OF REPRESENTATION 49

reference of the picture. Nevertheless, I think these cases are arguable.

I would not want to claim that I have shown conclusively that Kripkean

considerations are not necessary to determining pictorial reference.

(c) On the plausibly Kripkean theory of representation-as that I

suggested earlier, what a picture represents is a function of its history and

how it represents that thing (what it represents that thing as) is determined

according to the Fregean theory. However, if I am right and the Kripkean

theory of representation is inadequate, then the Kripkean account of

representation-as also fails. Neither a straightforward picture of a lion nor

a picture of a lion smoking a cigar represents Churchill in virtue of the

history of the picture. On the Kripkean view, a straightforward picture of

a lion represents Churchill in virtue of the fact that the artist intended to

paint an inspiring vision of Churchill and entitled his picture Churchill on

theEve of theNormandy Invasion. But this cannot be the whole story. The

only reason we accept a picture of a lion as a picture that represents

Churchill is that it is both

meaningful

and

apposite

to

regard

Churchill as

a lion. A picture of a vase of flowers entitled Churchill on the Eve of the

Normandy Invasion is likely to be merely puzzling (again, unless we supply

a context in which itmight be appropriate to regard Churchill as, say, a

vase of snapdragons as opposed to a vase of forget-me-nots). Similarly in

the case of the picture of a lion smoking a cigar. In this case it could be

argued that not only did the artist intend to represent Churchill; he also

used Churchill as his model for certain parts of his drawing such as the

parts which represent the Churchillian lips, chin and cigar. Again,

however,

as in theexample

of theschizophrenic s squiggle,

the historical

factors are not decisive. The artist has to get the shape of Churchill s lips

and the tilt of his cigar right in order for us to be able to say that the picture

really represents Churchill and not just an odd-looking lion. In other

words, what is represented is not determined independently of how it is

represented: our Kripkean solution to the problem of representation-as

is theoretically neat but artificial and far-fetched.

Other apparently plausible examples fail for similar reasons. Thus

although it is true that the reference of the Wohlgemuth woodcut is

determinedpartly by

the title and the artist sintentions,

it isonly because

the woodcut represents a city that it is appropriate for it to represent both

Mantua and Damascus. Although the picture is not true of both cities in its

This content downloaded from 89.206.117.115 on Sat, 13 Jul 2013 07:01:09 AMAll use subject to JSTOR Terms and Conditions

Page 15: Robinson 1978_Two Theories of Representation

8/12/2019 Robinson 1978_Two Theories of Representation

http://slidepdf.com/reader/full/robinson-1978two-theories-of-representation 15/18

50 J. ROBINSON

details, it is true of them both in so far as it ascribes cityhood to both. The

Kripkean theory of false pictures is also inadequate. It is not good enough

to say that a picture of the Parthenon with the wrong number of columns

along the front refers to the Parthenon just because the Parthenon served

as its model. It is only if the pictorial description of the Parthenon is

largely, even if not wholly, correct that we accept the picture as a repre?

sentation of the Parthenon. A picture of a vase of flowers would not

normally count as even a false representation of the Parthenon. Finally,

although the Kripkean theory offers a way of distinguishing between a

picture of amelon as an orange and a picture of an orange as amelon, it is

a highly artificial distinction: in fact it is likely that both sorts of pictures

would have to represent some properties of both a melon and an orange.

VI

Ihope

I have established without doubt that neither the

Fregean

nor the

Kripkean theory offers us a sufficient condition for pictorial representa?

tion. Very often, however, a case that cannot be explained by the one

theory alone can be explained if we invoke the other theory as well. For

example, although, according to the Fregean theory, a picture refers to

whatever satisfies the pictorial description it gives, in fact the very same

picture (in the same pictorial symbol system) may refer to different things

in different contexts. Thus whether a picture refers to Pickwick alone, my

grandfather alone or to all jolly Victorian gentlemen having the properties

representedin the

pictureis

somethingthat cannot be determined

just byexamining the pictorial properties and the mapping from these into the

properties represented. We need to know the context in which the picture

occurs, for example where it is- in the pages of a novel, in the portrait

gallery of my stately home, or in an advertisement for Toby Ale- and what

it is used for- to illustrate a novel, to prove how stately my home is, to

demonstrate what jolly people drink Toby Ale- and so on. Now, this

notion of context can be very easily tied to the Kripkean theory, since in

my particular example it is very much amatter of the history of the pictures

whether they represent Pickwick, my grandfather,or

drinkersof

TobyAle.

Similarly, on the Fregean theory, the dictionary picture of the eagle may in

This content downloaded from 89.206.117.115 on Sat, 13 Jul 2013 07:01:09 AMAll use subject to JSTOR Terms and Conditions

Page 16: Robinson 1978_Two Theories of Representation

8/12/2019 Robinson 1978_Two Theories of Representation

http://slidepdf.com/reader/full/robinson-1978two-theories-of-representation 16/18

TWO THEORIES OF REPRESENTATION 51

fact represent or refer to certain buzzards, since they happen to fit the

pictorial description . But if we invoke the Kripkean theory as a supple?

ment to the Fregean, we can rule out the buzzards, since no buzzards

functioned in the appropriate way in the history of the picture.

Moreover, as we have already seen, we cannot explain the reference of a

picture solely by appeal to Kripkean considerations, although very often

we can account for the difficult cases ifwe invoke the Fregean analysis as

well. There can still be doubt about what a picture represents- or whether

it represents anything- even if it is called Queen Victoria , the artist

intended it to represent Queen Victoria and he used Queen Victoria as his

model, but if the picture also gives a true description of Queen Victoria,

this tends to eliminate the doubt.

Now, this discussion may suggest that although neither theory by itself

is sufficient for representation, perhaps both theories set conditions that

are necessary for representation. However, it seems to me that this is not

so. I have already argued that the evidence for claiming Kripkean considera?

tions asnecessary

forpictorial representation

isambiguous.

As we have

seen, part of the trouble is that it is hard to see exactly how a Kripkean

theory of pictorial representation would work: there simply is no simple

pictorial equivalent to a baptismal ceremony. As for the Fregean theory,

it is certainly true that many pictures represent their objects via giving true

information about them, but it is equally true that many pictures do not.

Consider, for example, the extreme case of the squiggle said to represent

Queen Victoria. I claimed that it was not enough to say that the squiggle

represents Queen Victoria for Kripkean reasons; itmust also be the case

that thesquiggle maps

into certainrepresented properties

that it isappro?

priate to attribute to Queen Victoria, such as the possession of a con?

stricted personality. But the trouble is that appropriateness is not the same

as truth. Itmay be false of Queen Victoria that her personality is constricted

and yet in a given context there may be a point to representing her as

having a constricted personality.

Equally difficult for the Fregean theory to handle are metaphorical

cases of representation-as. In order to understand a metaphorical picture

that represents Churchill as a lion or Pitt as a parasite on the crown, we

need to know what makes the metaphor appropriate. Why is that lion

smoking a cigar? Why has that toadstool growing on the crown got a

This content downloaded from 89.206.117.115 on Sat, 13 Jul 2013 07:01:09 AMAll use subject to JSTOR Terms and Conditions

Page 17: Robinson 1978_Two Theories of Representation

8/12/2019 Robinson 1978_Two Theories of Representation

http://slidepdf.com/reader/full/robinson-1978two-theories-of-representation 17/18

52 J. ROBINSON

shape like Pitt s profile? Now, to understand the point of such pictures we

need to know what properties Churchill and a lion or Pitt and a toadstool

have in common, but the relevant properties are precisely those which are

not represented in the picture. Churchill does not have four paws but he is

fierce and courageous; Pitt does not have his neck attached to a crown but

he is overly dependent on the monarch. To rescue the Fregean theory, it

could be argued that the represented properties in the Pitt cartoon are

metaphoricallyrather than

literallytrue of Pitt. But this manoeuvre will

not work for the following reasons. (1) Some of the represented properties

are literally true of Pitt :he does, for example, have a nose ofthat particular

shape. (2) Many of the represented properties neither literally nor meta?

phorically inhere in the object represented: Pitt s neck is not sinuous, nor

is it attached to any crown-like object but to his chest. (3) Even ifwe grant

that the represented properties include being parasitic on the crown and

that this at least ismetaphorically true of Pitt, we are still faced with the

problematic possibility that in fact Pitt was not parasitic on the crown and

that Gillray was simply committing pictorial libel.Thus the reasonable conclusion to this paper seems to be the rather

negative one that although both the Fregean and the Kripkean theories

illuminate the concept of pictorial representation, neither of them offers

either necessary or sufficient conditions for its correct application. One

possible reason for this is that both theories are inadequate theories of

reference. Another perhaps more likely possibility is that representation

and representation-as cannot be entirely explained as species of reference.

University of Cincinnati

NOTES

1Nelson Goodman, Languages of Art (Bobbs-Merrill, 1968), p. 5.

2David Kaplan, Quantifying In , reprinted in Reference and Modality, ed. Linsky

(Oxford University Press, 1971), pp. 131-134.

3For Frege s theory of sense and reference, see Gottlob Frege, On Sense and

Reference , inPhilosophical Writings of Gottlob Frege, ed. Geach and Black (Blackwell,

1966), pp. 56-78.4

See, in particular, The Thought , reprinted as Appendix A in Essays on Frege,

ed. Klemke (University of Illinois Press, 1968), pp. 507-535.

5 For a defence of the view that what a picture represents is a function of what

character it is in what symbol system, see Goodman, Languages of Art, especially

This content downloaded from 89.206.117.115 on Sat, 13 Jul 2013 07:01:09 AMAll use subject to JSTOR Terms and Conditions

Page 18: Robinson 1978_Two Theories of Representation

8/12/2019 Robinson 1978_Two Theories of Representation

http://slidepdf.com/reader/full/robinson-1978two-theories-of-representation 18/18

TWO THEORIES OF REPRESENTATION 53

chapter I. Goodman explicitly denies the Fregean point that for a picture to represent

an object it must give true information about that object.6

Frege makes this point in The Thought , p. 518.7

This is not quite what Frege would say. According to him, a predicate refers to the

characteristic function of a set, not to the individual members of that set.8

Goodman, Languages of Art, p. 21.9

See Goodman, Languages of Art, pp. 225-232, for his account of the crucial distinc?

tion.10

For a discussion of this distinction, see Saul Kripke, Naming and Necessity , in

Semantics of Natural Language, ed. Harman and Davidson (Dordrecht, Reidel, 1972),

for example pp. 258-260.11

Kripke himself claims that it is not a theory , merely a better picture than the

Fregean of what goes in reference. See Kripke, Naming and Necessity , p. 302.12

Kripke, Naming and Necessity , especially pp. 298-299.13

Kripke, Naming and Necessity , pp. 314-323. Note that Kripke does not claim that

all general terms behave like the names of species.14

A constricted squiggle might also express the constricted nature of the queen s

personality, but that is another matter.16

Ernst Gombrich, Art and Illusion (Phaidon, 1959).16

Richard Wollheim has argued that the concept of representation should be ex?

plained in terms of the concept of seeing-as. Art and Its Objects (Harper and Row,

1968), especially pp. 14-18.17One thinks, for example, of Cezanne s pictures of apples.

Manuscript submitted 1April 1976

Final manuscript received 23 July 1976