Hostname: page-component-78c5997874-lj6df Total loading time: 0 Render date: 2024-11-15T03:21:26.535Z Has data issue: false hasContentIssue false
  • English
  • Français

A Semantics for Pictures

Published online by Cambridge University Press:  01 January 2020

Gary Malinas
Gary Malinas*
Affiliation:
University of Queensland, St. Lucia, Queensland, Australia, 4067
Get access Rights & Permissions [Opens in a new window]

Extract

There are a number of semantic predicates which we intuitively employ when we talk about pictures. Pictures which illustrate episodes from the life of Jesus refer to him. An illustration of a Tasmanian Tiger in a zoological catalogue denotes the class of Tasmanian Tigers. Some of the things we say concerning pictures are true in them; e.g., it is true in Rembrandt’s ‘Changing of the Night Watch’ that a guard wears a helmet; it is false in Rembrandt’s painting that one of the guards is female; and it is neither true nor false in the painting that the guards had eggs for breakfast. Rather than talk of what is true I false in paintings, we can speak of what they depict; e.g., Rembrandt’s painting depicts a man wearing a helmet, and it does not depict a female guard. Nor does it depict what the guards had for breakfast. In order to distinguish between what is excluded by a painting’s content — e.g., the presence of a female guard — from what is just not depicted by it — e.g., what the guards had for breakfast — it is important to be able to speak of what is true or false in pictures. Such presystematic talk is both intelligible and useful. Nevertheless, there are pitfalls associated with it. The concept of truth belongs to a family of semantic concepts which also includes entailment. Are the logical consequences of propositions which are ‘true in’ a picture also ‘true in’ it? An affirmative answer to this question conflicts with intuition. That a guard wears a helmet entails that a guard wears a helmet or the moon is made of cheese. Intuitively, this latter proposition is not ‘true in’ Rembrandt’s picture. Further, propositions have infinitely many equivalent formulations. That a guard wears a helmet is equivalent to the proposition that either a guard wears a helmet and the moon is made of cheese or a guard wears a helmet and the moon is not made of cheese. Is this latter formulation of the proposition that a guard wears a helmet ‘true in’ Rembrandt’s picture? Such questions suggest that the concept of ‘true in,’ while sharing some properties of the concept of truth, needs to be distinguished from it. In addition to talk of what pictures refer to/ denote, and propositions which are true/false/neither in them, we classify some pictures by reference to their semantic properties. ]as trow’s duck-rabbit which Wittgenstein cites is an example of an ambiguous picture. Robert Stevens’ picture of a tidal pool which looks like a detailed study of part of the human anatomy is a visual pun.

Type
Original Articles
Information
Canadian Journal of Philosophy , Volume 21 , Issue 3 , September 1991 , pp. 275 - 298
Copyright
Copyright © The Authors 1991

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

1 Peacocke, Christopher ‘Depiction’, The philosohical 96 (1987) 385Google Scholar

2 Ibid.; my emphasis

3 Peacocke, 388

4 Cf. below for a discussion of whether F-related properties are necessary for depiction.

5 Walton, Kendall ‘Pictures and Make Believe,’ The Philosohical Review 82 (1973) 283-319Google Scholar

6 Ibid., 313-14

7 Ibid., 314

8 Color-point descriptions are not uniquely correct. What counts as a minimally discernible point will depend in part on the purposes which the color-point description is meant to serve. The matrix chosen determines how fine-grained the description will be. Similar comments apply to the choice of background color classification. It may be very detailed if the color-point description will be used to generate facsimiles of a picture, or it may be two-valued, if the aim is to print black and white copies.

9 However, not every color-point need belong to a significant region. Some pictures contain what painters call ‘white space,’ i.e., areas which do not have representational content. (Nor need they be white in color.) Such areas are not representational, though they are compositionally important. Further, white space can be put to non-pictorial use: e.g., the use of balloons in cartoons to enclose depicted character’s thoughts and sayings.

10 The individuation of significant regions by reference to their attributes is a matter of considerable complexity. Consider a picture of a table covered by a cloth. The legs of the table are denoted by brown regions, and its shape is implied by properties of the cloth-symbol, the denotation of which occludes the rest of the table. What region of the two-dimensional surface does the table-symbol occupy? Does the table-symbol overlap the cloth-symbol? Or should we say that the surface does not contain a symbol which denotes the table? (I.e., it contains regions which denote table legs; it contains a region which denotes a table cloth; and the picture implies that the legs are connected parts of a table which is occluded by the cloth.) Let us call this latter response the ‘Berkeleyan’ answer. An alternative to this answer takes the table-leg symbols to be whole-table symbols as well. That is, the mereological sum of the discreet regions which they occupy denotes the total table, of which the denoted legs are a part. Regions do not contain points which refer to every part of what they designate. They are two-dimensional areas which refer to three-dimensional objects. It is a necessary feature of regions that they exhibit attributes only of parts of the objects which they denote. On this view, the ‘table-leg’ symbols denote a whole table and exhibit attributes of some of its connected parts — i.e., its legs. These two views employ different criteria for fixing the denotations of regions. The first relies on the concept of pictorial implication to fix the contents of pictorial symbols, and accordingly draws a sharp distinction between what a picture depicts and what it implies. The second emphasizes the fact that depiction is a subrelation of denotation, and that it is a necessary feature of depictions that they only exhibit some of their contents’ attributes. Accordingly, regions which exhibit attributes of parts of a denoted object can denbte the whole to which the parts belong. Cf. below for further discussion of pictorial implications and of exhibited attributes.

11 Cf. Goodman’s treatment of qualities such as size and shape in terms of ‘concreta,’ i.e., phenomenal color-points-at-a-time. The color-points which comprise regions of pictures are neither phenomenal nor time-indexed. In these respects they differ from Goodman’s ‘concreta.’ However, his formal point concerning the properties of size and shape of sums of ‘concreta,’ for example, apply to sums of color-points as well as to ‘concreta.’ In particular, he notes that the system of ‘concreta’ requires supplementation with a further basic predicate — i.e., ‘is of equal aggregate size’ — in order to define further size terms and shape terms. And further basic terms are required to define further properties of properties — i.e., second order properties. (Cf. Nelson, Goodman The Strcture of Appearace [Boston: D. Reidel 1977] 180-257.Google Scholar) The irreducibility of a surface’s significant features to the resources of a color-point description of its surface help to explain why the contents of picture, though effable’, are typically inexhaustible. Their fecundity is at least in part due to the extensiveness of their significant attributes, including second order attributes, and the determinate-determinable relation’ which obtain among predicates which we use to refer to their attributes and their attributes’ denotations. While all of a surface’s significant features supervene on its system of color points, a description of its color points does not contain sufficient resources for identifying its significant features.

12 Further, in doing so, they partially fix their regions’ denotations. A region which exhibits or quasi-exhibits attributes Al … Am, and which has attributes An … Ar that conventionally refer to attributes A’n … A’r, denotes something which is Al … Am & A’n … A’r. While this information is usually logically insufficient to determine a region’s denotation, taken as an input to viewers’ recognitional abilities, background information, and grasp of pictorial conventions, it is typica1ly sufficient for them to identify the kinds which regions denote.

13 The logical system formulated in this section is a weak kleene three-Valued system. Its origins can be traced to Bochvar. For discussion of the system and its matrices, Susan Haack, cf. Deviant Logic (Cambridge, MA: Cambridge’ University press 1974). esp. 169-70Google Scholar. My application of the logic as a logic for scenes and what is true in them was suggested to me by Perry’s applications; John Perry, cf. ‘From Worlds to Situations,’ Journal of philosophical Logic 15 (1986) 83-107.Google Scholar

14 For one sketch of such a study, cf. Erwin, Panofsky Meaning in the Visual Arts (New York: Anchor Books 1955), 33-5Google Scholar. Panofsky concludes his sketch by noting that ‘While we believe that we are identifying the motifs on the basis of our practical experience pure and simple, we really are reading "what we see" according to the manner in which objects and events are expressed by forms under varying historical conditions.In doing this, we subject our practical experience to a corrective principle which may be called the history of style’ (35).

15 Some pictures and some genres indicate multiple points of view on their subject matter — e.g., some Cubist paintings, and some pictures of impossibilia, where the scene which is depicted is not impossible, but the impossibility which they depict lies in the fact that the points of view from which their contents are depicted could not be occupied by a single viewer at a given moment.

16 An ancestor of this paper was read at the Australasian Philosophical Association Conference in Perth in 1988. It has been presented in seminars at Queensland University, La Trobe University, and the Research School of Social Sciences at the Australian National University.I thank the participants, and this journal’s referees, for their comments and criticisms.