The Logic Structure of Pictorial Representation 
Paul Crowther 
We use the term »representation« in many different contexts. Danto tells us, for example, that the word Icarus »represents« Icarus; Goodman 
mentions, in passing, that an ambassador represents his country; pictures 
represent, maps represent, it is also said that photographs represent. Now what 
tempts us to group all these together initially, is the fact that they all involve a 
semantic function. Even in the case of the Ambassador, we are entitled to 
invoke this function marginally insofar as his activity is a presentation of his 
country's as opposed to his own interests. His authority ultimately »refers 
back« to the powers that granted it. It is the semantic function which has 
thence led Goodman to assert that denotation is the »core« of representation. 
This core has a twofold character. On the one hand a representation as in the 
case of portraiture, can embody a genuine two term relation - there is a picture, 
and there is a person who the picture is »of«. On the other hand, there are 
pictures which involve only one term - the picture itself, insofar as their 
subject-matter - Mr. Pickwick or Pegasus or whatever, are fictional entities. A 
clarification of the relation between one and two term picturing, has I think to 
be the foundation of any coherent theory of representation. 
However, we immediately face a crucial problem. We have found that repre-
sentation is used synonymously with »stands for«, »refers to« or in Goodman's 
case »denotes« i.e. variants of semantic function. But we characteristically 
associate representation, not with just any old semantic function, but one 
where the representation perceptibly resembles the subject it is denoting. I 
mean of course, such things as pictures, maps and photographs. These, we 
might claim, are the classes of artifacts which give the term »representation« 
its ontological potency. It is clear then, that an investigation of representation 
must take as its central task, the clarification of »resemblance« and its role in 
the context of one and two term representations. 
Now by far the most impressive treatment of this whole area is found in Flint 
Fit. vest. /Acta Phil., XV (2/1994), 199-198. 
200 Paul Crowther 
Schier's Deeper into PicturesHis basic approach deals with it in terms of a 
theory of »natural generativity«. The basic outlines are as follows: 
»Pictorial competence in a symbol system is the ability to generate naturally 
interpretations of arbitrarily many novel members of the system. When an 
initial interpretation ofsome symbol does in fact effect an ability in someone to 
interpret novel symbols without further ceremony, then that initial interpreta-
tion was iconic or pictorial, the symbol so interpreted was a picture.«2 
Schier's point here is that picturing is a mode of communication, which, once 
learnt, is decisive. To recogniseXas a picture of>> in the most basic sense does 
not require anything in the way of further ad hoc conventions. Any new 
picture we encounter draws on natural abilities to recognise similarities be-
tween it and that which it pictures. Interestingly, however, whilst this clearly 
involves some question of resemblance, it is the »recognition« aspect which 
Schier gives massive emphasis to. We are told in this respect that 
»... the theory of natural generativity is soaked in causation. Essential to that 
theory are two causal claims' that an interpretation of S as being of 0 is iconic 
or pictorial in so far as it has been prompted by the interpreter's 0 - recognising 
abilities and that a picture of 0 is precisely something which can trigger the 
interpreter's O-recognising abilities. Iconic interpretation and iconicity are 
thus functionally defined, e<r3 
For the present writer, there is a problem here. Schier's analyses are substan-
tially sound but they are embedded in a framework, which emphasises a causal 
theory of perception. This functionalist approach tends to somewhat obscure 
the logical core of picturing, and, indeed, to point in a direction which also 
obscures the more important philosophical ramifications of picturing's relation 
to self-consciousness. Now in this discussion I shall not address the latter 
issue. I shall be content rather, to foreground the logical features of pictorial 
representation through an interpretation of defence of resemblance which 
(unlike Schier's) is orientated toward the pictorial object and the conditions of 
its creation. As a means to this, I will critically engage with ideas from 
Goodman and (to a lesser extent) Joseph Margolis and others. 
First, Goodman's approach to picturing has, as its central feature, a rejection of 
the centrality of resemblance. We differentiate pictorial representation from 
other modes of denotation, by virtue of the fact that it, in common with other 
forms of representation, is »syntactically dense«. 
1 Flint Schier, Deeper Into Pictures, Cambridge University Press, Cambridge 1986. 
2 Ibid, p. 46. 
3 Ibid, p. 195. 
The Logic Structure of Pictorial Representation 201 
»A scheme is syntactically dense if it provides for infinitely many characters so 
ordered that between each two there is a third... [Hence]... no mark can be 
determined to belong to one rather than to many other characters 
We need not detain ourselves over the many general puzzles that this defini-
tion might raise; a more pressing difficulty is presented by the relationship 
between picturing's »dense« character as a system, and its denotative function. 
Specifically, how do the two correlate? One can conceive of ad hoc situations 
where one might say things like »if you should find a picture of Sartre on the 
front door, it means I'm out« - but this would be simply using the picture to 
denote, and not an instance of a picturing relation. The correlation would be a 
case of an arbitrary convention. If, however, the term pictorial representation 
is to have any descriptive potency, we must have non-arbitrary criteria for 
correlating picture and denotation. Goodman, however, does not specify any 
criteria whatsoever. Indeed he has inaugurated something of a tradition for 
rejecting the most plausible criterion of correlation, namely visual resem-
blance. Let us review his objections to this notion. 
First Goodman posits the »naive« theory: 
»'A represents B if and only if A appreciably resembles B\ or 'A represents B 
to the extent that A resembles B'.«5 
As I shall show a little later Goodman's »naive« theory is actually two theories, 
the first of which is valid, and the second not. But first, let me consider his 
objections. Goodman claims that an object »resembles itself« to a maximum 
degree, but rarely »represents« itself i.e. resemblance, unlike representation, is 
reflexive. Additionally; resemblance is symmetric whereas representation is 
not. 
»B is much like A as A is like B, but whilst a painting may represent the Duke 
of Wellington, the Duke doesn't represent the painting.«6 
Now I (unlike many7) am not happy at the idea of reflexivity being ascribed to 
terms such as resemblance, which find their descriptive potency in the context 
of two term relations. And indeed, if one chooses to follow this very dubious 
path, I can think of no reason why an object should not represent itself as much 
as resemble itself. It is also worth noting that Goodman, in the above objec-
tions, has not (as he set out to do) countered the »naive theory« i.e. - that 
resemblance is a necessary condition of pictorial representation, but rather an 
4Nelson Goodman, Languages of Art, Hackett Publishing Co., Indianapolis 1976, pp. 136 and 
137. 
"Ibid, p. 3. 
6Ibid., p. 4. 
202 Paul Crowther 
absurd theory of his own devising which holds that resemblance is a sufficient 
condition of representation. 
Goodman does, however, eventually come up with two plausible objections to 
resemblance as a necessary condition. First, Constable's painting of Marlborough 
castle resembles any other painting more than it resembles Marlborough 
Castle, yet we say the picture of »of« the castle, and not simply of an-other 
painting. Hence resemblance cannot be the criterion for correlating picture and 
denotation. Margolis has objected to this account that 
»Goodman utterly fails to meet the objection that some respect or other may be 
specified in which the Constable perceptually resembles the Castle more than 
it resemble any other picture. «* 
This objection seems to be made redundant, however, by the fact that Goodman 
holds that any painting (no matter what criteria of perceptual resemblance we 
invoke) will always resemble a good copy of itself, more than it will its 
subject-matter. Even so, this would only count against an extreme resem-
blance theorist such as Beardsley, who holds that 
»If P is a design, P depicts an F if and only if P contains some area that 
resembles more closely the visual appearance of F's than it resembles any 
other object. «9 
Insofar as a picture always visually resembles a good copy more than it does 
its subject-matter, this account cannot hold. However, we must remember that 
the »naive« theory which Goodman takes himself to be criticising only con-
tends that for A to be a representation of B there must, to use Goodman's own 
words, be some »appreciable resemblance«. Clearly Goodman is vacillating as 
to what sort of resemblance theory he is wishing to reject. 
This becomes even more apparent when we consider his second objection to 
resemblance as a necessary condition of pictorial representation. Goodman 
entitles the relevant sub-section of Languages of Art as »Imitation« but occu-
pies most of his discussion with a rejection of the »copy theory« i.e. the 
argument that a picture is to be construed as a picture »of« something insofar 
as it depicts that subject with absolute verisimilitude. Against this view he 
holds that phenomenological appearances have a multitude of aspects of which 
the copy theorist is after the »natural« one; hence, for example, he is not out to 
depict the Duke of Wellington 
7 Such as Max Black, Joseph Margolis, and Roger Scruton. 
"Joseph Margolis, Art and Philosophy, Harvester, Brighton 1980, p. 101. 
9 Monroe Beardsley,Aesthetics: Problems in Criticism, Harcourt Brace, New York 1958, p. 270. 
The Logic Structure of Pictorial Representation 203 
»... as he looks to a drunk through a raindrop.ci10 
He is after, in fact, a sure »seeing«; an »innocent eye« that perceives in 
»aseptic« conditions. Goodman, however, shown that perception is by its very 
nature interpretative and creative; and cites Gombrich's Art and Illusion as 
showing how pictorial representation reflects this interpretative quality. Hence, 
given the fact that there are no pure visual »givens« or »facts«. 
»The copy theory is ... stopped at the start by inability to specify what is to be 
copied. «n 
Indeed, the copy theory takes a further beating in that 
»Where a representation does not represent anything there can be no question 
of resemblance to what it represents,«n 
Now I am substantially in agreement with Goodman's notion of perception, 
through (for reasons I shall make clear further on) I do not think he has drawn 
much benefit from reference to Gombrich. However, the question again arises 
as to exactly which resemblance theory Goodman is rejecting. We will re-
member that his initial formulation of the naive theory had two aspects. A 
represents B only if it appreciably resembles B; and A represents B to the 
extent that it resembles B. I think that the »copy theory« which Goodman has 
been rejecting is really a variant of the second aspect (though one can not be 
absolutely sure of this, since Goodman is so grudging in the depth to which he 
outlines alternatives to his own position). Hence, whilst I am in substantial 
agreement with Goodman's contention that the »copy theory« is incoherent 
this still leaves the first aspect of the »naive« theory untouched. All sorts of 
confusion as to Goodman's intention lurk here. Margolis observes that 
»Goodman does not deny that what represents and what is represented may 
resemble one another, only that representation as such does not as such 
depend on resemblance.«13 
But of course Goodman has not established this conclusively, and in the 
absence of criteria for the correlation of pictured being forthcoming from him, 
it is to the notion of »appreciable resemblance« we must return. 
An objection might be launched at the outset. Max Black declares for example: 
»My chief objection to the resemblance view ... is that when pursued, it turns 
out to be uninformative ... The objection to saying that some paintings re-
10 Goodman, op. cit., p. 7. 
11 Goodman, op. cit., p. 9. 
12 Goodman, op. cit., p. 25. 
13 Margolis, op. cit., p. 102. 
204 Paul Crowther 
semble their subject-matter is not that they don't, but that so little is said when 
only this is said.«14 
The fact, however, that »little is said« in saying that picturing presupposes 
visual resemblance, does not of itself disprove such a contention. Unfortu-
nately, all the other arguments which Black proposes are substantially the 
same as Goodman's objection to resemblance as a sufficient condition of 
representation. This, however, still leaves open the possibility of visual resem-
blance as a necessary condition. I shall argue that from it, knowledge of a quite 
informative nature arises. 
However, we must first dispose of one admittedly tautological sense of saying 
that one visual object resembles another. A fried egg for example resembles a 
mountain insofar as they are both »extended« (in Locke's sense of the term). 
However, it is rather empty to say that one thereby resembles or »looks like« 
the other, in that extension is a property possessed by any visual object 
whatsoever by definition. For the term »visual resemblance« to be less than 
empty, then, to ascribe such a relation between objects, involves us specifying 
some more precise way in which their visual aspects correlate. Andrew Harrison 
puts us, in general terms, on the right road as follows: 
»... one thing represents another either if the two can be relevantly held to be 
similarly structured so that in accordance with this structure it is possible to 
pair the unity of one with the units of the other, or else they are themselves 
such units.«*5 
This definition is a start, but will require some modification as we progress. 
For example, in the case of pictorial representation the »units« must be certain 
visual aspects of pictured. Specifically they will be a function of common 
shape, colour, and texture (though texture resolves ultimately into aspects of 
the other two). Harrison, however, rejects this invocation of »visual aspects«. 
For example: 
»... a standard Renaissance drawing of an egg or face will present the viewer 
with a mass of lines and hatching [sic] that certainly represent, but do not at 
all resemble the surface of an egg or the appearance of skin.«*6 
However, Harrison is wrong here, in that, viewed from the right distance and 
angle, masses of line and hatching do resemble visual aspects of eggs or skin, 
and enable us in fact to specify eggs or skin as elements in what is pictured. 
14 Max Black, »How Do Pictures Represent?« in,Art Perception, and Reality, ed. M. Mandelbaum, 
Johns Hopkins University Press, Baltimore 1972, p. 122. 
15 Andrew Harrison, »Representation and Conceptual Change« '^Philosophy and the Arts, Royal 
Institute of Philosophy Lectures, Vol. 6, p. 126. 
16 Harrison, Ibid. 
The Logic Structure of Pictorial Representation 205 
I shall return to both distance and the role of specification later. For the 
moment, I want to show how visual resemblance gets a purchase in pictorial 
representation, by describing some aspects of the creation of pictures. 
A first point is that the artist has a choice of two or three dimensional media -
drawing, painting, and sculpture. Strictly speaking, picturing is a function of 
the first two. In painting, the artist has colour, shape, and texture, available to 
him or her; in drawing, shape and texture alone. Let us consider the example of 
painting. The possibility most closely related to the nature of the medium 
itself, is the creation of exemplification of two-dimensional entities - such 
things as individual shapes or textures, or even distinct areas of monochrome 
colour. Such atomistic elements can be combined to make more complex 
entities, and this leads in turn to all sorts or interesting ontological possibili-
ties. For example, we can paint to red square upon a white background, and 
describe it as instantiating just that relation, or as a white square enclosing a 
red square. Now whichever, of these two descriptions we opt for, it would 
surely not count as »seeing as« in the way that we see a series of marks on a 
canvas »as« a warhorse or nude or whatever. This is because the painting of 
the red and white squares in ambiguous. On the one hand it serves to instanti-
ate certain classes of two-dimensional objects - namely red squares and white 
squares; on the other hand it might be taken as serving the additional function 
of referring to these classes. There is however, nothing in the formal configu-
ration itself which would warrant the assumption of this latter function. Indeed 
we might modify Occam's Razor here, and claim that denoted entities should 
not be unnecessarily multiplied. 
With these points in mind, let us now consider pictorial representation. In this 
case, the artist creates two dimensional entities i.e. marks upon a canvas, but 
configures them in such a way that can be seen as something other than marks 
upon a canvas. What does this involve and imply? Well, we characteristically 
individuate visual objects in the perceptual fields by reference to their visual 
aspects. We are able to say that »This is St. Andrews Bay« or »there is a man« 
not just because we have a language, but because the objects of our judgement 
have unified re-encounterable aspects of shape, texture, and colour, which 
enable us to recognise them at a certain distance and angle as particular objects 
or members of a class of objects. Now I am making no claims here that 
individuation by visual aspects is necessarily the most important part of our 
conceptual scheme (though I think a good case should be made for it); nor am 
I saying that the capacity for visual individuation does not presuppose the 
unified operation of all the senses. My only claim is that we can and do make 
individuations by reference to visual aspects alone. 
This is the starting point for pictorial representation. The artist is concerned 
206 Paul Crowther 
not with any arbitrary correlation of units between picture and subject-matter 
but rather those relevant to visual individuation. By organising paintmarks so 
as to correlate with the subject's visual individuating aspects, the artist's work 
when viewed from frontal position and appropriate distance (i.e. not too near 
to far) will be seen as a picture »of« such and such a thing or things. A 
schematic drawing or painting with no emphasis on texture will generally tend 
to copy the individuating aspects of kinds or types of object, rather than 
particular instances of them (for example, the schematic male/female pictures 
often found on Changing Rook doors). Similarly, whilst a complex and heavily 
textures painting is well suited to picturing the individuating aspect of quite 
particular objects, it need not do so, and indeed will sometimes move on a very 
high level of generality. (The images used in commercial art and advertising, 
for example, are frequently very much of this kind.) 
It is clear, from the foregoing, that whilst a picture is »of« a subject-matter and 
represents its individuating visual aspects at a concrete or more general level, 
it will not be identical with, or part of, that subject-matter. Indeed, that it is 
directly and perceptibly distinct from its subject, is surely a necessary condi-
tion for calling it a »picture of« that subject, as opposed to saying it »is« the 
subject matter. There are of course marginal cases. Suppose for example, that I 
paint a monochromically uniform frontal view, of a child's red plastic building 
brick. All I would have at the end of this is a red square (or square of red). 
Even through copying the brick, I would be working from an aspect which was 
not sufficient to visually individuate it. Hence there would be no criteria 
generated from the painting itself, for saying that it was a picture »of« a brick. 
It would simply be the presentation of a two dimensional entity. However, 
might we not make it a picture by convention? For example we could say: 
»Whenever you see a painting of a red square it is a picture of a red brick 
viewed frontally«. The problem here of course, it that when we want the red 
square to be a picture of a red ceiling, or of a narrow area of a plain red box 
cover, (or a red square pure and simple), we have to re-make the convention 
each time, because whilst the presented aspects resemble views of, or parts of, 
such objects, they do not visually individuate them. Hence each attempt to 
denote such objects would have to carry an accompanying ad hoc stipulation 
to the effect »red-square = [whatever]«. This would give us a kind of hybrid 
meaning, midway between picturing, and linguistic description, but logically 
distinct from both. Let us suppose, however, that I go on to paint the brick 
from an oblique and titled angle. In this case there would be grounds for 
calling my work a picture - but only of a red cube. I have considered visual 
aspects of the brick which individuate it at least as a member of a specific class 
of three dimensional objects. If I now want to go further, and create a picture 
of a (member of the class) plastic brick, I must have recourse to complex 
The Logic Structure of Pictorial Representation 207 
details of texture. If I want to picture just this particular brick and no other, I 
must look for visual aspects that individuate it from other bricks; or else by 
depicting it in the context of its visual surroundings. 
These points lead to the general principle that no two dimensional artifact can 
be called a picture unless its formal elements correlate with enough visual 
aspects of some other object or objects given in three dimensional space, for us 
to specify (from looking at the picture alone) what that object is - either as a 
concrete particular, or instance of a certain kind of thing. 
This account provides us with materials for refuting Goodman's objections to 
the resemblance theory. First we can stipulate which features of his or her 
subject, the artist needs in order to make a picture resemble. He or she is not 
concerned with a blanket reproduction of all his subject's visual aspects, but 
rather those which individuate it as a concrete particular, or as an instance of a 
specific type or kind. It is these aspects which he or she makes the formal and 
material aspects of the picture resemble. The relevant aspects will be chosen 
with necessary reference to the above criteria, but the artist's particular style of 
rendering them will be influenced by his or her own expressive ends, and the 
nature of the medium worked in. Indeed the artist will make use of various 
cultural conventions that surround the medium. I mean here, the kind of thing 
which Gombrich calls a »stereotype«. For example, in Art and Illusion, he 
shows how Wolgemut's woodcuts, purporting to depict different medieval 
cities, turn out in fact to be variants of one stereotype city. Goodman takes 
such things as testimony to the »relativity of vision and representation«.17 
However, this is to misconstrue them. Gombrich says that: 
» Without some starting points, some initial schema, we could never get hold of 
the flux of experience.«™ 
Now it is clear that whilst Gombrich's »schema« are conventional in the sense 
of being models or formulas for picturing, they are not arbitrary constructions 
i.e. purely conventional. Rather they serve to embody minimal visual aspects 
necessary for individuating members of specific classes of things given in 
three-dimensional space. The use of such stereotypes enables picturing to get 
to grips with its subject-matter. They are not an alternative to making artifacts 
with aspects that visually resemble other objects; but rather a generalised 
starting point. Whilst different cultures or individual artists will make use of 
different stereotypes, this gives no ground for postulating the »relativity« of 
picturing except in a very qualified sense. Indeed if such stereotypes were not 
founded on visual resemblance of some basic sort, we would not think of 
17 Goodman, op. cit., p. 10. 
18 Ernst Gombrich, Art and Illusion, Phaidon, London 1977, p. 76. 
208 Paul Crowther 
calling them »pictures« so much as another form of representation. Goodman 
actually criticises Gombrich for not being relativist enough, on the grounds 
that the latter holds »perspective« to be more than a mere convention. Again, 
this point needs to be qualified. Perspective is construct founded upon the 
problem of depicting the visually individuating aspects of objects and relations 
on three dimensional space, on a two dimensional surface. To this extent it is a 
convention. Other conventions can be used to solve such a problem, but what 
is significant is that perspective is the solution that enables the closest general 
visual resemblance between a two dimensional surface an objects and their 
interrelations in three dimensions. Hence, whilst being a convention, it is by 
no means the arbitrary one construed by Goodman. (I shall return to this topic 
at length elsewhere). 
This brings us to Goodman's point that a picture will always resemble other 
pictures, particularly copies of itself, more than it will resemble non-pictorial 
objects, and that picturing therefore cannot be founded on visual resemblance 
to subject-matter. Picturing is, however, a practice which arose and has been 
nurtured on making two-dimensional configurations that resemble aspects of 
other objects in three dimensional space. That is why in looking at pictures we 
never think of seeing them in relation to things they might more closely 
resemble; and indeed why we regard them in only secondary terms as two 
dimensional. But in saying that it is convention which leads us to see the 
picture »as« something, are we not conceding Goodman's case? No. Because 
whilst convention may lead us to pick out pictorial qualities on a two dimen-
sional surface, the fact that we can do so, is because the surface resembles the 
individuating aspects of some other visual thing or things. In other words, 
picturing is founded on a natural phenomenon, namely visual resemblance. 
The convention arises by focusing on and making a practice out an aspect of 
this phenomenon; namely that certain two dimensional configurations can 
resemble the individuating aspects of other objects given in three-dimensional 
space. Once we have learnt the convention, we can read in a general sense at 
least what the picture is »of«, without reference to any further ad hoc external 
convention or context (such as was found in the »pseudo-picture« of the red 
brick viewed frontally). It is this reference by resemblance to individuating 
visual aspects that constitutes pictorial representation's distinctive logical core. 
Let us now consider Goodman's final objection to this view, namely that when 
a picture is of a non-existent, it cannot be said to resemble that non-extent. On 
this issue I am substantially in agreement with David Novitz's approach. 
»... it is simply untrue that a picture cannot resemble a fictional entity. It can 
provided the entity in question has certain imaginary visual attributes. Of 
The Logic Structure of Pictorial Representation 209 
course anything which is entirely non-visual, no matter whether it is real or 
imaginary cannot be picture«.19 
Margolis sees fit to qualify this view on the grounds that imaginary entities 
»... resemble actual entities because, and only in the sense that, their descrip-
tions entail that we take them to resemble natural entities. «2" 
However, this is rather to miss the point of Novitz's position. If I interpret him 
rightly, Novitz is working from something like Hume's theory of imagination 
(detached from the atomistic theoiy of perception). On these terms, a picture 
(like a mental image) of a fictitious object is constructed (with its description 
as a guide) from visual aspects of existent objects. For example, whilst the 
monsters who tempt St. Anthony in the Isenheim Alterpiece are in themselves 
like no creatures who ever existed, their parts at least do resemble the parts of 
such creatures. Hence, to depict a fictitious entity, the picture must resemble 
visually individuating aspects of members (or parts thereof) of some class or 
classes of existing visual objects. By overlooking this, Margolis is led to posit 
visual resemblance as a necessary condition of some types of two term 
picturing (such as portraiture), but not of picturing as such. As he puts it, 
»What picturing (the one term characterisation) does require is that the 
ordered visual features of a picture be capable of being interpreted, fairly, as 
conforming to a description of 'what is pictured' - where 'what is pictured' is 
specified intentionally,«2' 
Hence: 
»Resemblance between pictures that putatively picture (allowing the equivo-
cation) and actual X's inclines us to interpret a picture as picturing X's ... in 
virtue of postulating an intention to picture X's; otherwise, we have only 
resemblance without picturing.«11 
On these terms, to see P as picture of X, entails an inference to the artist's 
intending to picture X. But surely, if a picture resembles an X closely enough 
for us to say »that is a picture of an X«, then its logical status as a picture of 
that kind of item is established without any positing of »intention«. 
Indeed, suppose that an artist paints a picture which is meant to be of an 
effeminate man, but that the female characteristics are so emphatic that it 
simply looks like a woman. In such a case, the artist has failed to communi-
19 Novitz quoted in Margolis, op. cit., p. 100. 
20 Margolis, op. cit., p. 100. 
21 Margolis, op. cit., p. 101. 
22 Margolis, op. cit., p. 101. 
210 Paul Crowther 
cate. Unless we know the causal origins of the work, we take it to be a picture 
o/a woman. And in this we are entirely justified. For, as we have already seen, 
the logical distinctiveness of pictorial representation consists in the fact that, 
once learned, it can be applied without recourse to ad hoc external conventions 
which determine exact denotation. In the present case we would say logically 
that we are dealing with a picture of a woman and, empirically, that it is one 
unsuccessfully created in order to secure reference to a man. 
The problem then, that has really dogged all discussion of pictorial representa-
tion comes down to this. Picturing is intentional in a twofold sense. An artist 
can have some specific intention which is the reason for creating his or her 
picture - say to depict Trafalgar Square or whatever. But irrespective of this 
denotative intention, the taking of a means to an end in accordance with the 
convention called picturing is also intentional. Irrespective of who or what he 
or she intends to picture, an artist must at least take up materials and configure 
them so as to resemble the individuating visual aspects of some specific three-
dimensionally given kind of thing. It is this layer of function neutral inten-
tional activity which defines picturing. It is the logical structure of which two 
term and one term representation (i.e. denotation and fiction, respectively) are 
concrete embodiments.