Abstract
Maps and realist pictures comprise prominent sub-classes of iconic representations. The most basic, most important sub-class is perception. Other types are drawings, photographs, musical notations, diagrams, bar graphs, abacuses, hieroglyphs, and color chits.
I am indebted to Ned Block for comments on an earlier version.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsNotes
- 1.
For a fine discussion of differences between maps and language, see Elizabeth Camp, ‘Thinking with Maps’ Philosophical Perspectives 21 (2007), 145–182.
- 2.
Representations are not the only sorts of things that can be iconic. In more extensive, forthcoming work on iconicity, I explicate iconicity for information registration. A state X informationally registers state Y if and only if (a) instances of states X and Y statistically co-vary in a significant way, (b) instances of X tend to be caused by instances of Y, and (c) X’s meeting conditions (a) and (b) is functional. Information registration is not representation. In my terminology, truth is propositional veridicality; accuracy is non-propositional veridicality. Representation requires having either accuracy conditions or truth conditions as part of the nature of the state that represents. Initial registration of the retinal image in visual systems does not have, and is not taken in science to have, accuracy or truth conditions. A bacterium informationally registers light. Although the occasional scientist attributes seeing to bacteria and even trees, no bacterium’s states are explained in the statements of laws of any science as having accuracy conditions. Information registrations, however, can and commonly do meet conditions for being iconic. Registrations of the retinal image have a structure and function that map iconically to spatial aspects of the retinal image, and degrees of light intensity. In such cases, the function of the natural mapping is entirely biological, not representational. Non-representational, non-perceptual sensory states commonly bear iconic relations to sensed aspects of the environment. For more on the distinction between representation and non-representational information registration, see my Origins of Objectivity (Oxford: Oxford University Press, 2010), chapter 8.
- 3.
That the mapping is functional implies that it could fail to match structural elements in the subject matter. So there can be non-veridical and purely fictional iconic representational mappings. Fictional pictorial mappings are parasitic on real mappings. Non-veridical mappings are parasitic on veridical ones.
It is possible to allow minor divergences from strict 1-1 mappings. Perhaps for convenience two representational elements could be mapped to a single represented item, as when two circles occur on a subway map for stops at the same station on different subway lines.
I am not fully convinced by such examples. Commonly, different circles represent different positions within the same station. When they do not, it is commonly possible to regard the two different circles as the same representational element, repeated for convenience–or as occupying different maps (one for each subway line). I owe the example to Ned Block. Although I do not insist on strict 1-1 mappings, I take them to be paradigmatic.
- 4.
I state the first two conditions separately, although condition (2) could be taken to be implicit in what is meant by a natural, systematic mapping in condition (1).
- 5.
This list of sciences is paradigmatic, not definitional. I take the notion of naturalness to be intuitive. The key point is that the mappings are not in themselves representational or intentional. Natural mappings are close cousins of what Grice called natural meaning. See H.P. Grice, ‘Meaning’, The Philosophical Review 66 (1957), 377–388.
- 6.
David Lewis, Convention (Cambridge: Harvard University Press, 1969).
- 7.
Jerry A. Fodor, Lot 2: The Language of Thought Revisited, op. cit., 173.
- 8.
Ibid, 172.
- 9.
Ibid, 174–177. I think that Fodor intends the claim more broadly, to mean that iconic representations lack any structure relevant to being veridical.
- 10.
Ibid, 173.
- 11.
Ibid, 173. The principle and the argument for it are also stated in Fodor’s ‘The Revenge of the Given’, in B. McLaughlin and J. Cohen, Contemporary Debates in Philosophy of Mind (Blackwell, Oxford, 2007), 108.
- 12.
Kosslyn’s idea is expressed in his Image and Mind (Cambridge, Mass.: Harvard University Press, 1980), 33; and Image and Brain (Cambridge, Mass.: London, 1994), 5. Fodor does not credit Kosslyn. See also Kosslyn’s ‘Mental Representation’ in Tutorials in Learning and Memory, J. Anderson and S. Kosslyn eds. (New York: W. H. Freeman and Company, 1984), 105–107. Kosslyn’s syntactical and semantical ideas are vulnerable to the same points I make against Fodor’s. For further expressions of the Kosslyn idea, see M. Tye, The Imagery Debate (Cambridge, Mass.: MIT Press, 1991), 44; D. Braddon-Mitchell and F. Jackson, Philosophy of Mind and Cognition: An Introduction 2nd edition (Oxford: Blackwell, 2012), where, 179ff., they claim, ‘there is no natural way of dividing a map at its truth-assessable representational joints’. Although Braddon-Mitchell and Jackson do not mention Fodor or Kosslyn, they in effect echo the view, basing it on the claim that ‘there is no natural minimum unit of truth-assessable representation in the case of maps’. They present this view as if it were obvious. I discuss minimality of size toward the end of this article.
- 13.
Taken literally, the second sentence in (ARG) implies that all parts of a picture depict the person. This view is clearly mistaken. A tiny picture part that depicts the left side of a mole on the person’s cheek is not a picture of the person. I assume that here Fodor is simply being careless in his formulation.
- 14.
E. J. Green and J. Quilty-Dunn, ‘What is an Object File?’, forthcoming British Journal of the Philosophy of Science. Their principle PP(2) is clearly inspired by Fodor, although they do not present it as a repair of Fodor’s mistakes. They argue for PP(2) in ways nearly identical to the way in which Fodor argues for PP(1)–again using Kosslyn’s example of cutting up a picture in arbitrary ways.
- 15.
N. Goodman and H. Leonard, ‘The Calculus of Individuals and Its Uses’, Journal of Symbolic Logic, 5 (1940), 545–55; N. Goodman, The Structure of Appearance (Cambridge, Mass: Harvard University Press, 1951; 2nd ed. Indianapolis: Bobbs-Merrill, 1966).
- 16.
Citing and explaining in detail why visual psychology routinely takes visual representations to have the format of a picture-like array would take up too much space for this article. For examples of work that either illustrate or help motivate the approach, see S. Murray, H. Boyaci, and D. Kersten, ‘The Representation of Perceived Angular Size in Human Primary Visual Cortex’, Nature Neuroscience 9 (2006), 429–434; M. Silver and S. Kastner, ‘Topographic Maps in Human Frontal and Parietal Cortex’, Trends in Cognitive Sciences 13 (2009), 488–495; T. Poggio, ‘The Computational Magic of the Ventral Stream: Towards a Theory’, Nature Precedings (2011), http://dx.doi.org/10.1038/npre.2011.6117.1.
- 17.
Not all iconic representations represent particulars. Some graphs represent only correlations among properties. Such iconic representations lack referential applications.
- 18.
For more discussion see my ‘Five Theses on De Re States and Attitudes’ in J. Almog ed. The Philosophy of David Kaplan (Oxford: Oxford University Press, 2009); Origins of Objectivity, op. cit., 83–84.
- 19.
Ego-here 1 is an application of an ego-centric index, marking the position of the perceiver. A spatial ego-centric index marks the origin of a spatial mapping from a perception to spatial structures, and does so in a way that privileges the origin as being of special psychological (“ego”) significance. Nearly all perception contains such applications of spatial or temporal ego-centric indexes. See Origins of Objectivity, op. cit., 187, 199, 287. Most commercial, paper maps lack ego-centric indexes and are allocentric. They map space in a way that is independent of the position of the map or the map’s user. Many allocentric maps still have origins established by referential applications. See note 25.
- 20.
I do so in a coming book, tentatively titled Perception: First Form of Mind.
- 21.
Fodor does not explain his notion of ‘logical form’. See note 9. I regard logic as an account of propositional validity by virtue of propositional structure. Pictures and perceptions are not propositional. Regardless of how one uses the term ‘logical form’, there are certainly forms of pictorial and perceptual representation that have veridicality conditions and a semantical structure, together with something analogous to a grammar.
- 22.
Origins of Objectivity, op. cit., 74–75.
- 23.
For detailed discussion of attribution and representation-as in perception, see Origins of Objectivity, op. cit., 379–381; ‘Origins of Perception’, Disputatio 4 (2010), 25–28. Fodor makes the further fundamental error of conflating information registration with genuine representation. He calls both ‘representation’. He assimilates all iconic representation to information registration. See note 2. For detailed discussion of the distinction, see Origins of Objectivity, op. cit., chapter 8; ‘Origins of Perception’, op. cit., 2–5; ‘Perception: Where Mind Begins’, Philosophy 89 (2014), 385–403; reprinted in T. Honderich ed. Philosophers of Our Times (Oxford: Oxford University Press, 2015).
- 24.
Fodor argues that some psychological states, for example those in what he calls an ‘echoic buffer’, are non-perceptual iconic representations with semantic content. He applies (a)–(h) to such states. I will not discuss this argument in detail. But it is as off-hand and unsound as the main argument that I have discussed. To begin with, the argument confuses information registration and representation. (See notes 2 and 23.) Although Fodor does not cite examples of what he means by an “echoic buffer”, one can assimilate what he says about it to information-registrational states like the first registration of the retinal image. The first registration of the retinal image is iconic, but it is not representation. His claim (‘The Revenge of the Given’, op. cit., 113) that such states must have semantical content because categorization (attribution) is extracted from them is clearly mistaken. Perceptual categorizational attribution is extracted from the initial registration of a retinal image. But that registration lacks semantical content. It is purely information registration.
Fodor argues that an “echoic buffer” is not subject to the “item effect”. The item effect is ‘the rule of thumb that, all else being equal, the “psychological complexity” of a discursive representation (for example, the amount of memory it takes to store it or to process it) is a function of the number of individuals whose properties it independently specifies’, Ibid, 110–111.
Fodor does not explain ‘discursive’ clearly, but his explanation, Ibid, 107, takes discursive representation constitutively to have all the properties (a)–(h) that he denies of iconic representations. I believe that his accounts of iconicity and discursiveness are both defective. If Fodor’s description of the “item effect” were correct, one would expect perceptual representation as well linguistically expressed conceptual representation to show it in memories of such representation. Then Fodor’s argument would divide non-representational, information-registrational states (registration of the retinal image, the “echoic buffer”)–which do not show an “item effect”–from representational states–both iconic and non-iconic, both perceptual and conceptual–which do. The argument would then fail to bear on the distinction between iconic and non-iconic psychological states.
But the argument has yet further defects, scientific defects. Limitations on memory, even in retaining complex representational states, including perceptual states, vary with the type of memory, not just the representational complexity of the state. Certain types of very short-term, iconic memory retain virtually the full complexity of perceptions’ representational content. M. Coltheart, ‘Iconic Memory and Visible Persistence’, Perception and Psychophysics 27 (1980), 183–228. Certain types of unconscious iconic long-term memory are also virtually unlimited in their capacity to retain the complexity of perceptual states or of beliefs formed most directly from perception. Cf. T. Brady, T. Konkle, G. Alvarez, and A. Oliva, ‘Visual Long-term Memory has a Massive Storage Capacity for Object Details’, Proceedings of the National Academy of Sciences of the United States of America 105 (2008), 14325–14329. Even visual working memory, the original poster child for the item effect, is not limited in the way that Fodor assumes. Other factors besides the number of items and representational complexity determine even visual working memory’s limitations. For reviews and explanations of why the item effect is not a basic explanatory notion, see C. R. Sims, R. A. Jacobs, and D. C. Knill ‘An Ideal Observer Analysis of Visual Working Memory’, Psychological Review 119 (2012), 807–830; D. Fougnie and G. Alvarez, ‘Object Features Fail Independently in Visual Working Memory: Evidence for a Probabilistic Feature-store Model’, Journal of Vision, 11 (2011), 1–12; G. Bae and J. Flombaum, ‘Two Items Remembered as Precisely as One: How Integral Features Can Improve Visual Working Memory’, Psychological Science 24 (2013), 2038 –2047; K. Hardman and N. Cowan, ‘Remembering Complex Objects in Visual Working Memory: Do Capacity Limits Restrict Objects or Features?’, Journal of Experimental Psychology: Learning, Memory, and Cognition 41 (2015), 325–347; T. Brady and G. Alvarez, ‘Contextual Effects in Visual Working Memory Reveal Hierarchically Structured Memory Representations’, Journal of Vision 15 (2015), 1–24. Fodor’s argument that there are iconic representational states in perception is laced with both conceptual and scientific errors.
- 25.
Allocentric maps can avoid context-dependent referential applications of the maps’ spatial origin. But many allocentric maps use ordinary proper names, which do involve context-dependent determiners. And even non-context-dependent, canonical names must, in the end, be explained in terms of context-dependent referential applications. Ego-centrically anchored maps all involve referential applications to the “home” anchor position.
- 26.
Both Fodor in Lot 2: The Language of Thought Revisited, op. cit., 175, and J. Quilty-Dunn in ‘Iconicity and the Format of Perception’, Journal of Consciousness Studies 23 (2016), 255–263, take a term that symbolizes the singular reference to be required, if an iconic representation (the whole perception or picture) is to represent a particular. Quilty-Dunn writes, ‘…icons lack the representational apparatus to bind features by picking out an object and attributing those features to the object’ (261). But that is exactly what paradigmatic icons, like realist paintings of individuals and perceptions, do.
Green and Quilty-Dunn, ‘What is an Object File?’, op. cit., make much of the supposed non-iconicity of memory index files. Anaphoric applications in memory–in index files—that derive from referential applications in perception may or may not have a ‘symbol’ that effects the anaphora. There being such a symbol would not prevent perceptual memory from being iconic.
Indeed, if symbols occur in relations that bear natural correspondences to relations in a subject matter, the arrangement of symbols is iconic. Note the iconic representation of positions of light-rail stations by the arrangement of their names. As the light-rail map example indicates, many iconic representations have non-iconic symbolic elements. Moreover, being a symbol does not preclude being iconic, as hieroglyphs and some Chinese words show. But such a symbol is not needed in perceptual memory that relies on anaphoric retention of perceptual singular reference, any more than a symbol for a referential application is needed in perception, or indeed natural language. Anaphora can be effected through occurrent events that are causally and functionally connected to the occurrent referential application event involved in the original perception. Such a causal link can underlie the changing iconic perceptual attributives that support the application’s use in the index file. There need be no symbol for the referential application in the index file. But since iconic representations can have symbolic elements that are non-iconic, the presence of such a symbol in index files would not prevent such files from being iconic.
A significant error in Quilty-Dunn’s article is the mis-attribution to me in Origins of Objectivity, op. cit., of the view that the difference between perception and cognition consists (sometimes he says ‘partly consists’, sometimes he does not) in perceptual representations’ being iconic, and in cognitive representations’ being discursive, or language-like. I think that some cognitive, even propositional, representations are iconic. In fact, I think that all propositional beliefs immediately formed from perception are iconic.
- 27.
E. J. Green and J. Quilty-Dunn, ‘What is an Object File?’, op. cit. state that PP(2) and the following principle are ‘the signature markers of iconic format’: (H) (for Holism) ‘Each part of the presentation represents multiple properties at once, so that the representation does not have separate vehicles corresponding to separate properties and individuals’. Their exposition follows Fodor in conflating information registration and representation. Further, not all iconic representation represents multiple properties. So packaging is not a signature marker of iconic format. As noted, a color chit can be used to represent a color shade, and nothing else. As I indicated, it is problematic to claim that when multiple properties are represented in a package, the representation goes by way vehicles. If vehicles are (say, picture) parts or object-like entities, then it is not true that a single vehicle represents multiple properties. The parts are not, strictly, the representing units. If the properties or property-instances count as vehicles, then different “vehicles”, not one, effect representation of different properties: the color of the picture part represents the color; the shape of the part represents the shape (scaled appropriately); and so on. In perception, different aspects of a perceptual state represent different environmental attributes, even though properties are represented in a package. So (H) is mistaken in various ways.
Green and Quilty-Dunn argue that since perceptual working memory sometimes retains different properties to different degrees, perceptual working memory cannot conform to (H). They conclude that perceptual working memory is not iconic. They further infer from considerations of simplicity that perceptual representation is not iconic either. Even apart from the mistakes in PP(2) and (H), this train of reasoning seems to me a reductio of their conception of iconicity and their conception of how representation takes place in pictures and perception. The fact that various properties are iconically represented in a package in perception does not begin to show that they cannot be remembered iconically to different degrees.
- 28.
There is evidence that iconic perceptual working memory in fact retains different features to different extents, as one would expect. Packaging is one factor in retention. But many factors bear on how well different iconically represented attributes are retained in memory. Attention might affect different attributes differently. Facts about how different properties are registered differently in neural coding can also ground differential retention. Relationships among the types of properties retained can affect retention. For papers that bear on these issues, see M. Wheeler and A. Treisman, ‘Binding in Short-term Visual Memory’, Journal of Experimental Psychology: General 131 (2002), 48–64; Y. Jiang, M. Chun, and I. Olson, ‘Perceptual Grouping in Change Detection’, Perception & Psychophysics 66 (2004), 446 – 453; D. Fougnie and G. Alvarez, ‘Object Features Fail Independently in Visual Working Memory: Evidence for a Probabilistic Feature-store Model’, op. cit.; G. Bae and J. Flombaum, ‘Two Items Remembered as Precisely as One: How Integral Features Can Improve Visual Working Memory’, op. cit.; K. Hardman and N. Cowan, ‘Remembering Complex Objects in Visual Working Memory: Do Capacity Limits Restrict Objects or Features?’, op. cit.; T. Brady and G. Alvarez, ‘Contextual Effects in Visual Working Memory Reveal Hierarchically Structured Memory Representations’, op. cit..
- 29.
Here is a definition from Merriam-Webster: a sign (such as a word or graphic symbol) whose form suggests its meaning.
- 30.
See W.S. Geisler, J.S. Perry, B.J. Super, and D.P. Gallogly, ‘Edge Co-occurrence in Natural Images Predicts Contour Grouping Performance’, op. cit.; J. Frisby and J. Stone, Seeing: The Computational Approach to Biological Vision, op. cit., chapter 6.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2018 Springer International Publishing AG
About this chapter
Cite this chapter
Burge, T. (2018). Iconic Representation: Maps, Pictures, and Perception. In: Wuppuluri, S., Doria, F. (eds) The Map and the Territory. The Frontiers Collection. Springer, Cham. https://doi.org/10.1007/978-3-319-72478-2_5
Download citation
DOI: https://doi.org/10.1007/978-3-319-72478-2_5
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-72477-5
Online ISBN: 978-3-319-72478-2
eBook Packages: Physics and AstronomyPhysics and Astronomy (R0)