About CA
Contact CA

Search Journal

Editorial Board

Permission to Reprint


Site Map


Realism and the Riddle of Style

  Catharine Abell

1. Introduction

My concern in this paper is what, in Art and Illusion, Gombrich calls "the riddle of style".[1] This is the problem of why people at different times and in different cultures have depicted objects in very different ways. An adequate solution to this problem will comprise an explanation of why depiction has a history.

The problem seems intractable because of three common assumptions about the history of depiction that, while independently plausible, are inconsistent. First, we assume that this history is a history of realism. Artists from a wide range of cultures and ages seem to have shared the common goal of capturing the visual appearances of the objects they depicted. Secondly, we assume that depictive styles differ from context to context in part because of features internal to the contexts in which they emerged. For example, we tend to think that the Haida would never have developed the styles of depiction prevalent in The Netherlands in the seventeenth century.

Robert Davidson's Split Beaver (1975) This is an example of the Split Style of the Haida people, indigenous to Canada. Used by permission of Robert Davidson.

Finally, we are loathe to accept the claim that differences of style necessarily result from differences in technical capacity. However, these assumptions are in conflict. If artists throughout history sought to capture the appearances of the things they depicted and did not differ markedly in their technical capacities, then surely the pictures each produced should not depend on features internal to the contexts in which they emerged and should instead look more alike.

Some attempts to solve the riddle of style reject one of these three assumptions. Others argue that the three assumptions are in fact compatible. Below, I discuss a version of each of these approaches and argue that neither is completely successful. I argue that the history of style is not, in its entirety, a history of realism. Nevertheless, this does not solve the riddle of style, since it simply re-emerges as the riddle of realism: the problem of why those pictures that do aim to capture their objects' appearances differ from one another. I argue that the riddle of realism can be solved because styles produced with this aim can be appropriate to their contexts by providing information that is relevant to viewers in those contexts.

2. Styles and Purposes

Gombrich construes the riddle of style as the problem of why artists, all of whom claim to copy what they see, have copied the visible world in very different ways.[2] He assumes, therefore, that the history of style is a history of realism. He explains why different attempts at realism should produce different results despite parity of technical skills by claiming that depiction involves the use of conventional schemata. It is a matter of convention what schema is chosen to depict an object in any given historical and cultural context because such choices are arbitrary: there is nothing about a schema itself that constrains what object it can be used to depict. Different choices of initial schemata result in different ways of depicting things.

Nevertheless, Gombrich claims that conventional schemata can be corrected better to match their objects' appearances. While he denies that the eye is innocent and therefore that objects have an appearance independent of the beliefs of particular viewers, he holds that pictures match their objects' appearances when the pictorial schemata they employ match the perceptual schemata used to interpret visual experiences of their objects. Pictures that achieve such a match are realistic. By correcting their schemata, artists can improve how well their pictures copy the visible world. Although all may aim to produce faithful copies of the visible world, however, the pictures they produce will vary because the different schemata with which they begin will constrain how their pictures look even after they have been corrected.

As Dominic Lopes notes, Gombrich's solution to the riddle of style requires him to deny that particular styles are appropriate to their historical and cultural contexts. Because he takes the choice of initial schemata to be arbitrary, he cannot claim that there is anything about the schemata that are initially chosen in any given context that makes them appropriate to that context rather than to any other.[3] Moreover, Lopes argues, because Gombrich holds that the way in which these schemata are subsequently corrected depends on our perceptual schemata, it is perception, rather than historical and cultural context, that determines which corrections are made.[4]

Lopes rejects Gombrich's solution since it limits the art historian's and the anthropologist's task to that of cataloguing stylistic changes and precludes them from explaining why those changes have occurred. He sets out to provide a response to the riddle which explains what makes particular styles appropriate to their contexts of use.[5] He proposes a 'non-matching' perceptual theory of depiction, according to which different ways of depicting objects capture different visual aspects of those objects, none of which needs match our perceptual experience of them. He notes that our ability to recognise objects is dynamic: we can recognise objects under circumstances and from viewpoints from which we have never seen them before. Each set of such circumstances comprises a recognisable aspect of that object. He claims that pictures present specifically pictorial recognisable aspects of objects that need not match any of the aspects under which we have recognised them before.

Coupled with an account of what styles are, this provides a solution to the riddle of style. Lopes argues that styles are individuated by the kinds of aspects they present, which can be understood in terms of the types of properties they depict their objects as possessing, as lacking, and as neither possessing nor lacking. On this understanding, styles can be more or less appropriate to certain contexts, since the purposes for which pictures are used in a context will dictate which properties need to be represented. By looking at the purposes for which pictures were used in certain historical and cultural contexts, therefore, we can explain why particular styles were appropriate to those contexts.

For Lopes, the history of style is a history of realism because realism is a matter of how well pictures perform the purposes their viewers expect them to serve. A picture is realistic if it serves its purpose well, by informing its viewers about just those properties of its object that they expect it to inform them about. Realism, on Lopes's construal, is informativeness in a context of use.[6] Thus, a picture is realistic to the Haida if it informs them about the anatomical structure of its object while preserving symmetry, whereas Rembrandt's drawing is realistic to us because it informs us about how a boat on a river looks.

Rembrandt, A Canal with a Rowing Boat, pen and ink drawing.
This image is used with the permission of Chatsworth Photo Library. Any form of reproduction, transmission, performance, display, rental, lending or storage in any retrieval system without the written consent of the copyright holders is prohibited.

Robert Davidson's Split Beaver (1975) This is an example of the Split Style of the Kwakiutl people, indigenous to Canada. Used by permission of Robert Davidson.

Lopes presents his solution as one which resolves the apparent tension between the three assumptions that lead to the riddle of style. He resolves this tension by construing realism as a measure of the extent to which pictures achieve their intended purposes. However, this account of realism is implausible. Some pictures serve their purposes perfectly well, without thereby being realistic to those for whom they serve them. It no more likely that the Haida consider split-style pictures realistic than that European audiences around the turn of the last century considered cubist pictures realistic. What prevents these pictures from being realistic is not that they do not serve their purposes well, but rather that the purposes they serve are not of the appropriate kind.

The history of style is not, in its entirety, a history of realism. Gombrich is right to hold that realistic pictures are those that are successfully intended to convey their objects' visual appearances, even though he is wrong to think that all pictures aspire to realism. The difference in their purposes explains why Rembrandt's drawing is realistic while the split-style drawing is not. Lopes is therefore right that understanding the various purposes for which pictures have been used will help explain why certain styles have been used in particular contexts. His account provides a partial solution to the riddle of style. However, the riddle threatens to emerge again in a different guise, this time as the riddle of realism: the problem of how different styles can emerge in contexts in all of which pictures serve the purpose of conveying their objects' appearances.

Lopes himself notes, "Though they differ stylistically, the Audubon print shares with a painting by Giotto the aim of replicating a set of properties, defined by the rules of perspective projection, which record versions of occurrent visual experience." It is pictures governed by this aim that, to the extent that they realise it, are realistic. Without appeal to differences in technical capacity, Lopes cannot explain why attempts at realism differ, since the purpose they serve is the same. Moreover, while it may be plausible to claim that the Audubon print manifests realistic techniques that were not available to Giotto, it is not plausible to claim that realistic styles vary only with the technical skills of their makers. In what follows, I argue that styles can be suited to one among several contexts in which pictures serve the same purpose because they are more informative in one of these contexts than in the others.

3. Informativeness and Relevance

Lopes argues that a style is more realistic the better it serves its purpose. To serve its purpose well, a picture must inform its viewers about just those aspects of its object they expect pictures to provide information about. Because he construes realism as a matter of how well pictures serve their purposes, Lopes holds that a picture's realism depends on its informativeness in its context of use.[7] I have denied that realism is relative to purpose. Nevertheless, I believe that the key to solving the riddle of realism lies in the claim that pictures serve their purposes well if they inform their viewers about the required aspects of their objects.

While Lopes does not unpack the notion of informativeness, others do. In particular, Dan Sperber and Deirdre Wilson argue that being informative involves making a set of assumptions manifest or more manifest to an audience.[8] An assumption is manifest to an individual to the extent that she is capable of representing it mentally and accepting its representation as either true or probably true.[9] For example, the assumption that you are reading this is manifest to you now, as is the assumption that you did not eat a saucer of mud yesterday, although the latter is less manifest because you are less likely actually to entertain it. The assumptions that are manifest to an individual at any time comprise her cognitive environment. As Sperber and Wilson note, an individual's cognitive environment is a function of her physical environment (in a broad sense encompassing her social environment) and her cognitive abilities (which depend not just on intelligence, but on existent assumptions, including memory).[10]

Informing someone about something therefore involves modifying her cognitive environment. Any piece of information that the individual does not already possess and can represent and accept as probably true will effect such a modification. Nevertheless, some information is more efficient at effecting such modifications than other information. In particular, information is more efficient the more relevant it is.

Information is relevant to a person if it is connected with the assumptions already in her cognitive environment such that she can use it, together with the latter, as the basis for inferences that she could not otherwise have made.[11] The more relevant information is, the more efficient it is at modifying her cognitive environment, that is, the more assumptions it makes manifest or more manifest to her. For example, if you believe both that all swans are white with red beaks and that I saw a swan yesterday, telling you that the swan had a red beak will modify your cognitive environment by making your belief that swans have red beaks more manifest to you. However, it isn't a very efficient way of modifying your cognitive environment since it is unlikely to lead you to make many further inferences. Information about the colour of the swan's beak is therefore not relevant to you. Contrarily, information that the swan's feathers were black will be relevant to you, since it will lead you to infer that not all swans are white.

I have argued that the purpose of realistic pictures is to convey their objects' appearances.[12] Coupled with Lopes's idea that pictures serve their purposes better the more informative they are about those aspects of their objects they are used as sources of information about, this leads to the claim that pictures are realistic to the extent that they inform their viewers about their objects' appearances. A picture that is more informative about its object's appearance will be more realistic than one which is less informative about this aspect of its object. Moreover, because informativeness depends on relevance, pictures will be more realistic the more relevant the information they provide about their objects' appearances.

Cognitive environments are properties of individuals: no two people have identical physical environments or cognitive abilities. Relevance is therefore relative to the individual. However, there is a notion of relevance that is relative to a community. To the extent that the various individuals in a group have cognitive environments that overlap, they have a shared cognitive environment.[13] Their shared cognitive environment includes all the assumptions that are manifest to each of them and excludes any that are not. The individuals comprising any given historical and cultural community will have a shared cognitive environment, since their physical environments are similar and their cognitive abilities will overlap. Information is relevant to a community when it is connected with information already manifest to its members such that it can be used, together with the latter information, as the basis for further inferences that could not otherwise have been made. Different communities will have different shared cognitive environments, since the cognitive environments of their individual members will differ. Consequently, different information is relevant to each community.

If pictures are more realistic the more relevant the information they provide about their objects' appearances, how realistic a picture is to a given community will depend on how relevant the information it provides about its object's appearance is to that community. A style can provide information about its object's appearance that is relevant to one community but is irrelevant or less relevant to another. Consequently, that style can be appropriate to one community and less appropriate to others, even if each community uses pictures to provide information about their objects' appearances. For example, in a community in which objects' colours are relatively uniform - people's hair and skin colours do not differ, the landscape is uniformly coloured, and so on - information about objects' colours will not be especially relevant. Contrarily, in a community in which things exhibit a wide variety of colours depending on their exact identities, information about objects' colours will be much more relevant, since it will help members of that community to identify those objects. A style that provides accurate information about objects' colours will therefore be more realistic to the latter community than to the former.

The information that is relevant to a community depends on the physical environment and cognitive capacities shared by its members. Investigating the physical environment and cognitive abilities shared by the members of a given historical and cultural community will therefore help to tell us what information was relevant to that community. For communities that used pictures as sources of information about their objects' appearances, such investigation will tell us why certain styles were realistic to some communities, while different styles were realistic to others.

4. Conclusion

I have argued that the history of style is not, in its entirety, a history of realism. Throughout history, pictures have been used to inform viewers about a wide variety of things other than their objects' appearances. Investigating the purposes for which pictures were used in particular historical and cultural contexts will help us to understand why certain styles were appropriate to those contexts. Nevertheless, it will not help us to understand why certain styles were appropriate to just one among a number of contexts in which pictures were used for a single purpose. To understand this, we need to know what cognitive environment was shared by the members of the relevant community. This will enable us to understand why, given the purpose for which pictures were used in that community, a certain style was more informative than another. In addition to cataloguing stylistic changes, therefore, the historian of art has two further tasks: to investigate both the various purposes for which pictures have been used and the cognitive environments of those who used them.


[1] Ernst Gombrich, Art and Illusion (London: Phaidon, 1977).

[2] Ibid.

[3] Dominic Lopes, Understanding Pictures (Oxford: Clarendon Press, 1996), p. 10.

[4] See Dominic Lopes, "Pictures, Styles and Purposes," The British Journal of Aesthetics 32, (1992).

[5] Ibid.

[6] Lopes develops this account of realism at greater length in Dominic Lopes, "Pictorial Realism," The Journal of Aesthetics and Art Criticism 53, (1995).

[7] Ibid.

[8] Dan Sperber and Deirdre Wilson, Relevance (Oxford: Blackwell, 1986), p. 58.

[9] Ibid. p. 39.

[10] Ibid.

[11] Ibid. p. 48.

[12] For a detailed defence of this claim, and for my own account of realism, see Catharine Abell, "Pictorial Realism," The Australasian Journal of Philosophy, (forthcoming).

[13] Dan Sperber and Deirdre Wilson, Relevance (Oxford: Blackwell, 1986), p. 41.

Catharine Abell

Department of Philosophy

University of Manchester

Manchester M13 9PL, UK


Published January 18, 2006