I argue that there is no distinction between allographic and autographic representations. One consequence of this is that replicative forgeries have the same aesthetic and artistic value as originals, and are accurate records of actions. I end with some reflections on the pragmatic structure of forgery.

I.

In Languages of art, Goodman formulates what he calls ‘the critical question’ with respect to authenticity: is there an aesthetic difference between two pictures if I cannot tell one from the other? Or, is it possible that some difference I do not discern by simply looking at two pictures constitutes an aesthetic difference between them? (Goodman 1990, 138) I agree that this is the critical question concerning replicative forgery, in the sense that aesthetic, ontological or semantic considerations of authenticity should be guided by a requirement to answer it. Otherwise, the issue will no longer be the authenticity of works in particular as distinct from, e.g., objects auctioned because they are the personal effects of personalities. The bottom line for grasping this difference is that among the properties an object happens to have, not all are relevant to its functioning as an artefact, or else some may allow only exapted functions.

How does this question square with the well-known distinction between autographic and allographic works and arts? The distinction claims that a work is “autographic if and only if the distinction between the original and the forgery of it is significant”, allographic if otherwise (Goodman 1990, 147). This would be circular as a definition, not to mind an explanation, of why forgery matters and is possible in some areas: it does not answer the ‘critical question’ above, and does not tell us why fakes matter. What we need is to know whether the distinction between original and forgery is significant. The autographic-allographic distinction is in fact a heuristic device that leads to an inquiry into density and discreteness in representation, which is Goodman’s main concern in the area of authenticity and replication. If the difference between original and forgery turns out to be significant, then there is an aesthetic difference between two pictures even when I cannot tell one from the other. If it is not significant, then there is no aesthetic difference.

Specifically, if the inquiry can be used to claim that forged paintings do not have the same value as originals, the claim will have to be grounded on a denial that original and fake are synonymous, since according to Goodman all functions and cognitive effects of works either boil down to meaning or can be described as forms of meaning. This is not quite as restrictive as it sounds. Languages of art stretches meaning beyond what we have in mind when we take natural language semantics as our model, to include such things as demonstrations, exemplifications and various other unconceptualizable and ‘je ne sais quoi’-type experiences (as Leibniz called certain nonconceptual categorizations), which we want the forgery to be able to cause.

Beyond this, we can break out of the restriction to meaning in another way. I will try to capture it with an imaginary example. If I attempt to make Madame de Recamier’s portrait look ugly by variously tweaking her facial features, I will necessarily change the picture’s meaning: on any attempt, the picture will mean different facial features. But the converse does not hold: although I change the features, I might not manage to turn a beautiful face-picture into an ugly one. Changes in perceptible shape do not necessarily result in changes at the level of aesthetic or stylistic properties, but they necessarily result in changes of meaning. In other words, synonymy (replicability of meaning) is a good test case for replicability of other functions, such as aesthetic and stylistic functions.

II

Now, let’s try to answer the question whether the difference between original and forgery is significant—given an extremely wide definition of meaningful syntactic difference, and therefore an extremely strict definition of synonymy. The closest Goodman comes to substantially answering the question is his denial that pictures “not only practically but theoretically” can be finitely differentiated (Goodman 1990, 172). Put in the simplest terms possible, this is a way of stating the familiar idea that ‘in a picture, any change matters’. But for present purposes, we have to understand finite differentiation a little better than this. Finite differentiation can be thought of as the existence of zones of indifference within which type-indifference can be established for the purpose of grouping pictures together under a single value (in the same way that we can group graphic variations of “EXIT” under a single type as synonymous). If we cannot make such groupings, no two pictures will be strictly equivalent.

As I see it, two obstacles prevent the absence of finite syntactic differentiation for pictures from amounting to a stricture on synonymy. One militates against Goodman’s claim that synonymous pictures are theoretically impossible, the other against the claim of practical impossibility. Here is the first. On a substantivalist conception of space, for any two given points that do not coincide, there is a region of space between them occupied by points. To apply this to pictures, imagine that two outline drawings A and B with different sizes are superposed on a sheet of paper, so that different possible outlines can be drawn in the space between their outlines. Then take any outline picture C between the two; this will be different to both A and B, allowing (on the same principle) a picture D which is between A and C and a picture E between B and C, and so on. Differentiation of the possible pictures into picture-kinds will be regressive as long as any two pictures are different by stipulation. However, none of this entails that two numerically distinct outline shapes on different sheets of paper cannot have identical shape or length—independently of how, or whether, they are perceived or measured. Therefore, lack of differentiation into types does not entail theoretical impossibility of replication. If replicability is claimed to be theoretically impossible, it will have to be for reasons of a different kind to those Goodman claims.

The second obstacle is epistemic, and more practical in scope. We have to be able to describe any differences between the spatial properties of pictures as syntactic differences. The concept of a syntax is already epistemic and implies that spatial differences can be picked up as such by a system, so that they can be converted into differences of function or value. Either we can think of a system as something that gives in to the lack of differentiation which putatively characterizes the ontological level, in which case there will be no meaning and no system; or we can think of the system as picking up information, which means picking up segments coarser than anything we can imagine slicing logically. Thus, in addition to the theoretical feasibility of replication, two numerically distinct pictures will not necessarily have different meanings. This does not fall foul of the non-transitive nature of indiscriminability. Since in replicative forgery we are concerned with the forger’s perception of two pictures, one of them the original, we are in the case described by McDowell as indiscriminability “from the indicated sample”, and not in the case of indiscriminability “from something else that counts as having” the same properties as the indicated sample (McDowell 1996, 170). Indiscriminability “from the same sample” has also come in for criticism, on the grounds that it cannot serve as a ground for individuation of concepts (Dokic & Pacherie 2001). But in the case of replicative forgery, we do not need to individuate concepts: all we need to do is ensure identity relative to a context.

It may appear that problems raised by fineness of colour grain could be solved in a way similar to the problem raised by fineness of spatial grain. However, it is possible that there is no objective (physicalistic) sense in which two colours are the same, in the way that two lines can be said to be the same length or shape. If this is so, then the argument presented in favour of the theoretical possibility of replication does not apply to colour. On the other hand, the epistemic points concerning perceptual discrimination and identity relative to a context still apply.

Neither the epistemic point, nor the previous ontological point, redresses the lack of finite spatial differentiation in pictures that Goodman points out. Instead, it is just that lack of finite differentiation does not imply lack of replicability either in principle or in practice, albeit for different reasons. (I stress that non-finite differentiation itself remains intact, because it is also necessary for the specific kind of meaning pictures can possess.)

III

Once the obstacle to forgery presented by non-finite differentiation is overcome, many other obstacles collapse: those obstacles presented by the replication of any properties dependent upon the perceptual traits of the object. This includes aesthetic properties (for example, the gracefulness of a sculpture of the Three Graces) and stylistic properties (for example, the kind of gracefulness that is shared by a given series of frescoes of the Three Graces by the same artist, or by his assistants). Changes in such properties supervene on changes in perceptual traits: no aesthetic or stylistic properties can change without changes in the picture’s perceptual traits, taken as the base on which the former properties supervene. (See the Madame de Recamier example in section I.)

This allows us to re-draw the basic conceptual distinction concerning forgery. There are no autographic arts or representations, and therefore there is no sense in talking of allographic arts or representations either. Instead, there are just dense and discrete representations. Now to the ‘critical question’ of whether there is an aesthetic difference between two pictures if I cannot tell one from the other. Here, the answer has to be that there is no aesthetic difference. There is probably a difference in shape between the two pictures (in outline, occlusion and shading shapes, barring some extraordinary coincidence), but it is not a syntactic difference, so it cannot prevent synonymy, and it is not a perceivable difference, therefore it cannot prevent both pictures from causing the same aesthetic and stylistic effects.

IV

What kind of terminology could capture the relative status of forgeries and originals? Perhaps appropriate concepts can be found in another area that analyses artefacts: philosophy of language. An initial, but inadequate, approximation consists in describing an original and a replicative forgery as different tokenings of the same work. This is true, but, as it stands, unfair to the forgery. A replicative forgery is a perfect record of the original work taken as the result of a sum of actions carried out by an agent in a given situation. For it to be a perfect record, it has to exemplify all the formal properties on which representational functions, aesthetic effects and art-historical evaluations can supervene, something we have tried to ensure. But if the forgery works as a record of an action, then it is not just an occurrence coming under the same type as the earlier token, but something closer to the original tokening—since a tokening of an expression is, precisely, an action. This can make it tempting to describe the forgery as a quotation of the original work. Whether forgeries of paintings are pictorial quotations depends on whether we take quotations to exemplify types or tokens. I will assume that at least some quotations exemplify tokens, and therefore refer to a prior tokening. Then, presenting a forgery is like saying (1), but it is not like saying (2):

(1) In the context of production, the artist did this ‘___’.

(2) After you left, he made this gesture ‘___’.

In both cases, ‘___’ is a case of the same gesture previouly made. But in (2), ‘___’ also refers to the prior occurrence of the gesture. In (1), it is clear that ‘___’ cannot refer to what the artist did: for the replicative forgery to refer to the original, the forger would have to admit that the forgery is not the original. Of course, the forgery can be used to make a quotation in the full sense of (2): reference plus replication. But if it is so used, the forgery no longer functions as a forgery but as a replica. Therefore, what defines replicative forgery is the highly unusual pragmatic predicament in which the forger finds himself. Although he presents an object that is a quotation (that is type-identical to a prior token and refers to that prior token), he has to present the object as if it is that prior token.

I hope that this helps to clear the moral ground of objections to forgeries. Objections to replicative forgery cannot be based on a claim that a forgery necessarily produces either inaccurate records of actions, or somehow inferior aesthetic experiences. So they have to find some more basic ground, independent of artistic and aesthetic value.

REFERENCES

Dokic, J. and Pacherie, E. (2001) “Shades and concepts”, Analysis 61(3), pp. 193-201

Goodman, Nelson (1990) Langages de l’art, Nîmes: Ed. J. Chambon

McDowell, John (1996) Mind and world, Cambridge, Mass.: Harvard University Press