Unusual Pictures: An Analysis of Some Abnormal Pictorial Structures in a Painting by Juan Gris Author(s): John Willats Source: Leonardo,

Vol. 16, No. 3, Special Issue: Psychology and the Arts (Summer, 1983), pp. 188-192 Published by: The MIT Press Stable URL: http://www.jstor.org/stable/1574911 Accessed: 02/05/2009 05:32
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=mitpress. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

The MIT Press is collaborating with JSTOR to digitize, preserve and extend access to Leonardo.

http://www.jstor.org

Leonardo, Vol. 16, No. 3, pp. 188-192, 1983 Printed in Great Britain

0024-094X/83$3.00+0.00 Pergamon Press Ltd.

UNUSUAL PICTURES: AN ANALYSIS OF SOME ABNORMAL PICTORIALSTRUCTURES IN A PAINTING BY JUAN GRIS

John Willats*
Abstract-A distinctionis made betweenpictures whichare unusualbecause of their subjectmatter (for example, some surrealistpaintings), andpictures whichare unusualbecause they are based on unusualpictorial structures. Two kinds of pictorial structures are described: transformation systems, such as oblique projection and perspective; and denotation systems, such as the use of lines to denoteedges. It is arguedthatJuan Gris and other Cubistpainters used both these pictorial systems in unusualways. This argumentis illustratedby analysing Gris' 'Breakfast', 1914. The analysis shows that Gris used a number of unusual transformationsystems, including vertical oblique projection, inverted linear perspective and inverted atmospheric perspective. Theanalysis also reveals the use of a numberof unusualpictorial structures related to the denotation systems: the use of false attachments, both withinthe pictureandbetweenthepictureand theframe;the reversalof the normalrulesfor the representation of occlusion; and the use of real surfaces en collage. It is suggested that Gris used these unusual structures deliberately in order to point to, and investigate, the nature of normal pictorial structures.

There are two kinds of unusual pictures:pictures which astonish us because of their unusual subject matter, and pictures which, although their subject matter may be quite ordinary, surpriseor puzzle us because of their unusual structure. Pictures of the first type, such as Salvador Dali's 'The Great Masturbator' (1929) may contain objects or scenes which are themselves outrageous or unusual, or they may, like Rene Magritte's 'Time Transfixed' (1939) be made up of ordinary objects combined in incongruous ways. In pictures of this first type, the composition and technique are nearly always realistic, and for a good reason: the more the artist can persuade us to look through the picture surface to the objects themselves, the more likely we are to be shocked or intrigued by the subject matter of the picture. This 'looking through' is also characteristic of the way in which we use most ordinary pictures: we concentrate on what the artist, draughtsman or photographer is saying, rather than how it is being said. With ordinary pictures, we forget that they are contrived, and their surface becomes so transparent that we imagine that we have the real world itself in front of our eyes. What spoils this illusion is when we come across pictures which deal with quite ordinary scenes, but which are based on pictorial structures which are unusual or unfamiliar. Chinese and Japanese paintings, which are based on various kinds of oblique projections rather than perspective, must have seemed extraordinary to Europeans of the seventeenth and eighteenth centuries;and, even today, when we are so used to seeing them in reproduction, they still seem strange to us. Pictures of this kind provoke in us the most banal responses: was there something wrong with the artists' eyesight, or were they just not very good at perspective? Similar responses are often provoked by modern paintings, especially perhaps by Cubist paintings like Juan Gris' 'Breakfast' (1914) (Fig. 1). This painting depicts a scene which is both ordinary in itself, and is made up of ordinary objects: cups, glasses, a coffee pot and similar everyday objects on a breakfast table. But although we can identify these objects without much difficulty, they are so jumbled that the picture at first seems almost incomprehensible. Why should anyone want to paint such a confused picture, we wonder? Are our legs being pulled, or does the artist really see the world that way?
*Honorary Research Fellow, North East London Polytechnic, Faculty of Art and Design, Greengate House, Greengate Street, London El 5 OBG, U.K.

p" ~ ~ ~~~~~~'

Fig. 1. Juan Gris, 'Breakfast', collage, crayon and oil on canvas, 80.9 x 59.7 cm, 1914. Collection, Museumof ModernArt (Lille P. Bliss Bequest), New York.

For psychologists, or for people like myself who are interested in pictorial structures, the great merit of pictures of this kind is that they draw our attention away from the subject matter, and make us realise that even the most ordinary pictures must have some kind of a structure. Thus, although unusual pictures may be intriguing in themselves, their chief value for us lies in what 188

UnusualPictures
they can tell us about ordinary pictures. I want to argue that Gris' motive was very much the same. What he was investigating was the nature of pictorial structures; and the way in which he did this was to experiment with, and sometimes reverse, the ordinary rules of painting. Of all the different kinds of pictorial structures, the best known are the various drawing systems [1]. Of these, the best known (though perhaps the least understood) is perspective;but there are a number of other systems, such as orthographic projection and oblique projection, with which we are almost equally familiar. Perhaps a better name for these systems would be transformationsystems (the name by which they are known in mathematics), because they transform spatial relationships in the real world into corresponding spatial relationships on the picture surface. All these different systems can be seen as different ways of coming to grips with the problem of representing the three-dimensions of the real world on the flat surface of the picture, and they all have various advantages and disadvantages according to the purposes for which they are to be used. In orthographic projection, for example, which is the simplest of all the projection systems, vertical directions in the real world are represented by vertical directions on the picture surface, and horizontal directions in the real world are represented by horizontal directions on the picture surface. Front-to-back directions in the third dimension are simply ignored. Orthographic projection is an ideal system for engineers, because it shows the true shapes and dimensions of rectangular objects, and the disadvantage of the absence of the third dimension is compensated for by showing more than one view. For rounded objects in paintings, however, the system is often less suitable. An orthographic projection of the side view of a cup, shown in Fig. 2a, does not bring out its roundness; while the top view fails to reveal its true shape. An alternative to orthographic projection, which solves the problem of the third dimension, is oblique projection. In this system front-to-back directions in the real world are represented by oblique directions across the picture surface. This formed the basis of most Chinese and Japanese paintings before the impact of nineteenth-century Western culture. Oblique projection is still widely used for technical illustrations, and its near relations, such as isometric projection and axonometric projection, are much in demand for architects' drawings. Children who were asked to draw a scene consisting of a number of objects on a table, and then shown drawings of the scene in various other drawing systems, chose oblique projections as the most realistic, clearest and best liked, regardlessof age; and this was true even for those children who had produced drawings in perspective [2]. Again, however, this system is less suitable for rounded objects: the cup shown in oblique projection in Fig. 2b appears distorted.
orthographic projection

189
is vertical oblique projection. Here,

A system intermediate between oblique projection and vertical directions across the picture surface are used to represent front-to-back directions in the real world. In the West the system is relatively unfamiliar, but vertical oblique projection formed the basis of most traditional Persian and Indian paintings, and in children's drawings it is the most used of all the projection systems [3]. Drawings of rectangularobjects in this system tend to be ambiguous, since one direction across the picture surface is used for both vertical and front-to-back directions in the real world, but for rounded objects the system is ideal. Figure 2c shows a drawing of a cup in vertical oblique projection, and another glance at Fig. 1 shows that this was the system which Gris used for the cups, the glasses and the various other rounded objects in his picture. A tracing of the table on which these objects are standing (Fig. 3) shows that it is drawn in axonometric projection, a version of vertical oblique projection in which objects are shown turned through an angle. By this means the ambiguity inherent in ordinary vertical oblique projection is avoided; and since this system shows the tops of objects as true shapes, it is ideal for rectangular objects where the plan or top view is the most important. This is the reason why the system is often used for architects' drawings, and Gris would presumably have been familiar with it since it was introduced by Auguste Choisy at the end of the nineteenth century. In Gris' Breakfast the top of the table is its most important feature, so axonometric projection is wholly appropriate. Gris, then, chose a transformation system for his picture which gave the clearest and most realistic possible view of the scene. Pictures, however, do not just depend on transformation systems; and what Gris gave us in clarity and realism with one system, he took away with the other.

Fig. 2. Drawing of a cup in (a) orthographic projection, (b) oblique projection, and (c) vertical obliqueprojection.

Fig. 3. Cupsand otherroundedobjectsin verticalobliqueprojection,with a tracing of the table showing that it is in axonometricprojection(a version of vertical oblique projection).

190

John Willats
marks in the picture stand for, and the picture is almost reduced to a flat pattern. In 1912, only two years before Gris painted his picture, the critic Jaques Riviere had pointed out the crucial importance of the position of the spectator, and the weakness of perspective as a basis for painting because it shows us objects from just one particular point of view. False attachments can hardly ever happen in the real world, because as Riviere [9] said: "... in reality, we can change position: a step to the right and a step to the left complete our vision. The knowledge we have of an object is, as I said before, a complex sum of perceptions". In paintings and drawings we cannot do this, so the position of the object must be carefully chosen. As Huffman said: "One assumption we shall make ... is that all pictures are taken from a 'general position'; that is, that a slight change in the position from which the picture is taken would not change the numberof lines in the picture, or the configurations in which they come together" (p. 298). Writing in the same year as Huffman, Clowes pointed out that pictures for computer analysis ought always to be enclosed by a frame, otherwise the computer would be unable to distinguish between the real world and the marks in the picture. Within the frame is a symbolic world, in which marks stand for or denote edges and corners, or objects such as cups and tables. Outside the frame is the real world of cups and tables. Between these two worlds is the frame, which prevents the real world from leaking into the symbolic world. All these three kinds of pictorial structures-hidden line elimination, false attachment and the picture frame-can be regarded as aspects of the picture's denotation system. Hidden line elimination restricts the reference of the lines to just those features of the scene which can be seen from a particularpoint or direction of view. Avoidance of false attachment prevents one pictorial mark from standing for more than one feature of the scene. And the pictureframe separatesthe symbolic world of the picture from the real world to which it refers. In his Breakfast Gris has chosen a transformation system which shows a breakfast table with the utmost clarity and realism, but a denotation system which destroys this clarity and realism. To begin with, the normal rules for hidden line elimination are inverted or ignored. Not only are various features included in the picture which ought to have been hidden, but a number of other features are eliminated which ought to have been included. Figure 5 shows some of these features. Parts of the far edge of the table ought not to have been shown, because they would have been hidden by the coffee pot; in Fig. 5 these edges are shown using a zig-zag line. On the other hand, part of the saucer in the middle of the picture, parts of the glass and parts of the coffee pot have been eliminated, although they would have been visible in the scene. In Fig. 5 these edges have been replaced, using a dotted line. Secondly, there are numerous instances of false attachment in the picture. The rim of the cup in the centre is falsely attached to the tablecloth; the rim of the cup on the right is falsely attached to the side edge of the table; and the edge of the coffee pot and the far edge of the tabreare falsely attached together to an area of collage; these false attachments are marked with asterisks in Fig. 5. Finally, the distinction between the real world and the symbolic world of the picture has been blurred by the false attachments between the marks in the picture and the picture frame. The corner of the table on the right, and the corner of the area of collage at the top are falsely attached to the frame, and there are other instances or near instances of false attachments between objects in the picture and the picture frame. Two of these instances are marked with double asterisks in Fig. 5. At the simplest level, then, the picture can be seen as a battleground in which straightforwardtransformation systems -the good guys-are pitted against inverted denotation systems-the bad guys. The transformationsystems show us the

Denotation systems say what the marks in the picture stand for [4]. Since pictures, like sentences, can never literally contain the objects they represent, even the most realistic pictures are, in reality, only made up of blobs of ink or patches of paint which stand for, refer to, or denote the tables and cups of the real world. Just as there are various transformation systems, characteristic of different periods and cultures, so there are various denotation systems. Before the advent of photography, probably the most familiar system was the one in which lines are used to stand for edges: either true edges, like the edges of a table, or smooth edges (or occluding contours, as they are sometimes called) like the edges of a cheek or an apple [5]. In photography, on the other hand, and in Impressionist and Pointillist painting, there are no lines as such, but only small areas of colour or tone which denote the intercepts of small bundles of light rays. Impressionist painting was greeted with cries of outrage when it first appeared, but photography is only the same system, using (as a rule) smaller areas of colour or tone. Gris' painting revertsto an earliersystem, using (mostly) lines to denote edges-but with a difference. In 1971 two writers, Huffman in the U.S.A. [6] and Clowes in England [7], working in Artificial Intelligence, analysed a denotation system for line drawings of rectangularobjects. One of the things that Huffman pointed out was that in normal line drawings the lines stand for edges which we can see from a particularpoint of view. If all the edges are put in, irrespectiveof whether they can be seen, the drawing becomes very confused. This is what happens as a rule when line drawings are produced using a computer, and the removal of these hidden edges (a process known as hiddenline eliminationor HLE) is a long and tedious part of the program. Compare, for example, the drawings shown in Figs 4a and 4b. Figure 4a is almost incomprehensible, simply because the hidden lines have not been removed. Another important contribution which Huffman made was to point out that the clarity and realism of a drawing depend very much on whether or not it contains what painters call false attachments, or what Guzman [8] in an earlier paper called "nasty coincidences". Compare the three views of the same object shown in Figs 4b, 4c and 4d (all shown in trimetric projection). Figure 4b contains no false attachments and gives a clear and realistic view of the object. Figure 4c contains a false attachment between two edges, and the clarity and realism have already begun to disappear. Figure 4d contains a false attachment between two corners, and here the effect is even more marked. Because the junction on the right can stand for more than one corner, it becomes difficult to tell what the other

Fig. 4. Four drawingsof an object, with (a) no hiddenline removal,(b) the hiddenlines removed,(c) a false attachment betweentwo edges and (d) a false attachment between two corners.

UnusualPictures
** u

191

Fig. 5. Parts of the scene which wouldhave been hiddenhavebeenincluded in the picture (shown here using zig-zag lines). Parts of the scene which would have been visible have been left out (dotted lines). There are a numberof instances offalse attachment,bothwithinthepicture(asterisks) and between the picture and the frame (double asterisks).

scene as clearly as possible, but the denotation systems spoil the effect by confusing us about what the various marks stand for. Who wins out? Well, the transformationsystems seem to win on recognition, because we can still see reasonably clearly what the various objects are, but the denotation systems seem to win out on spatial relationships, because it is difficult, in the original picture, to see where the various objects go. This is perhaps slightly surprising, since denotation systems, as a rule, tell us about what the marks stand for, and it is transformation systems which tell us where the marks go. The fact is that neither transformation nor denotation systems act on their own: it is the interaction between systems which is important. But, in any case, the picture is not quite as simple as I have suggested. To begin with, two other transformation systems are used in the picture in addition to vertical oblique projection. In ordinary pictures in perspective the edges of objects below eyelevel converge to a vanishing point towards the top of the picture. In Gris' picture the edges of the table are parallel rather than converging; but other lines are added which confuse the issue. The picture contains a number of rectangularforms with no particular reference, and these fan out from the bottomof the picture, giving an effect of invertedperspective, and disguising the underlying system of vertical oblique projection (Fig. 6). But something even more extraordinary is happening. In all ordinary projection systems, vertical directions in the real world remain vertical in the picture. But in this painting the vertical axes of the coffee pot and the bottle incline to the left, while the axes of the egg cup, and a cup which can just be seen in the top centre of the painting, incline to the right. Since these axes seem to converge upwardstowards the top of the picture, the effect is to introduce a suggestion of normal linear perspective (Fig. 6).

ig. 6. Therectangularformsfanoutfrom the bottomof the picture,giving an effect of invertedperspective. The vertical axes of the coffee pot, the bottle, the egg cup, and a thirdcup convergetowardsthe top of thepicture, giving an effect of normalperspective.Normal atmosphericperspectiveis reversed, swinging the backgroundof the scene at the top of the picture towards the observer.

In normal atmospheric perspective, the warmest and most fully saturated colours appear in the foreground of the scene, shown towards the bottom of the picture. Most of Gris' painting is in dull browns, greys and blues, all recessive colours; the only areas of strong warm colour appear in the parcel and, to a lesser extent, in the coffee pot lid, both of which are in the background of the scene at the top of the picture. This has the effect of swinging the top of the picture, where one would expect to find the greatest depth, towards the spectator. Thus, while the primarytransformation system is used in a straightforwardway, two of the three secondary systems are reversed, and the third used in an extraordinary way to give an effect of normal linear perspective (Fig. 6). As if to redress the balance, Gris introduces a secondary denotationsystem which ought to make for the maximum clarity and realism. In the primary denotation system, lines in the picture are used to stand for edges in the real world. This is a relatively abstract system, since the lines in the picture are physically quite different from, for example, the smooth rounded sides of the cups and glasses. In the secondary denotation system, in contrast, areas of real wallpaper, real woodgrain paper, and real newspaper are used to stand for real areas of wallpaper, wood and newspaperin the real scene. Since, as indicated before, pictures can never literally contain the objects they represent, using these fragments of the real world within the frame of the picture further confuses the distinction between the real world and the symbol system of the picture, like the false attachments between the picture and the picture frame.

192

John Willats first the truth and then the nature of their own symbol systems, rather than just using those systems to give descriptions of the external world [12]. The emergence of linguistics as a separate discipline at the beginning of this century is perhaps the supreme example of this tendency. As if to illustrate this link with other symbol systems, Gris signs his name with a fragment of
newspaper: "... ourn..." (Juan) "Gris". Acknowledgements-I would like to thank the School of Graphic Arts, Royal College of Art, for supporting my attendance at the International Conference on Psychology and the Arts and the North East London Polytechnic for help in the preparation of the manuscript.

Thus the picture as a whole can be seen as a very carefullyand delicately controlled balancing act, in which both transformation and denotation systems are used both normally and in reverse. But what is the point of this balancing act, if all we end up with is a rather confused picture? I would like to suggest at least three possible motives. Some writers [10] have suggested that pictures, like languages, depend on symbol systems which are purely conventional, and that realism in pictures is a matter of habit. That is, pictures which at first seem unintelligible may appear quite realistic when we get used to them. Sixty years on, this has still not happened with Cubist paintings. Although we are by now quite familiar with Cubist paintings, and even though we may understand some of the rules on which they are based, they still do not seem as realistic as pictures in perspective. On the other hand, it is no longer possible to claim, as other writers have suggested, that pictures are just 'natural', and that pictures in systems other than perspective are just naive or incompetent. By using mixtures of systems, Gris shows us what each system can and cannot do; but far more important than that, he shows us that pictures are symbol systems, not just fragments of the real world. The second of Gris' possible motives is that by using some of the rule systems in a normal way and some in reverse, he is able to show us that there are different kinds of pictorial rules. The motive here seems very similar to that of the linguist Chomsky [11], who invented the sentence "colorless green ideas sleep furiously" to illustrate the difference between syntactic and semantic rules. In this sentence the syntactic rules-the rules of grammar-are used in the normal way, and the sentence as a whole is perfectly grammatical. But the semantic rules-the rules of meaning-are used in an abnormal way. Colourless things cannot be green; ideas cannot be either green or colourless; ideas cannot sleep; and so on. As in Gris' painting, using one set of rules in a normal way and another set in an abnormal way draws attention to the fact that there are different kinds of rules. Finally, by using painting to investigate its own nature, Gris brought it firmly into line with other disciplines. By the middle of the nineteenth century the received truths of perspective and Euclidean geometry had begun to be questioned, and both mathematicians and painters began, increasingly, to investigate

References 1. F. Dubery and J. Willats, Perspectiveand OtherDrawing Systems (London, The Herbert Press, 1983; New York, Van Nostrand Reinhold, 1983). 2. E. A. Saenger, Drawing Systems: a Developmental Study of Representation (unpublished doctoral thesis, Harvard University, 1981). 3. J. Willats, How Children Learn to Draw Realistic Pictures, QuarterlyJournal of ExperimentalPsychology 29, 367 (1977). 4. J. Willats, What do the Marks in the Picture Stand for?The Child's Acquisition of System of Transformation and Denotation, Review of Research in Art Education 13, 18 (1981). 5. J. M. Kennedy, A Psychologyof PicturePerception(San Francisco, Jossey-Bass, 1974). 6. D. A. Huffman, Impossible Objects as Nonsense Sentences, in B. Meltzer and D. Mitchie (eds.), Machine Intelligence, Vol. 6 (Edinburgh: Edinburgh University Press, 1971). 7. M. B. Clowes, On Seeing Things, ArtificialIntelligence2, 79 (1971). 8. A. Guzman, Decomposition of a Visual Scene into Threedimensional Bodies, Proceedings of the Joint Fall Computer Conference(Washington, Thomson, 1968) pp. 291-306. 9. J. Riviere, Sur les tendances actuelles de la peinture, Revue d'Europeet d'Amerique1, 384 (1912). [Cited in E. F. Fry, Cubism (London: Thames & Hudson, 1966)]. 10. R. Wollheim, Representation: the Philosophical Contribution to Psychology, in G. Butterworth(ed.). The Child'sRepresentationof the World(London: Plenum Press, 1977). 11. N. Chomsky, Aspects of the Theoryof Syntax (Cambridge, Mass., M.I.T. Press, 1965). 12. E. Nagel and J. R. Newman, Godel'sProof (New York: University Press, 1978).