Unity and Objectivity in Perception: A Corrected Russellian Theory of Demonstrative Visual Belief

[Comments to: david_austin@ncsu.edu]

In my book (1990), I argued against the traditional, commonsensical analysis of belief as a two-term relation between a believer and a proposition, and I suggested that the traditional analysis needs to be supplemented with a third term, but I gave no theory of that term. I propose to develop such a theory, beginning with the limited domain of visuo-spatial belief, which provides the best examples of knowledge of contingent particulars.

Kaplan (1977) was the first contemporary philosopher to argue that beliefs typically expressed by demonstratives are best understood as three-term relations among the believer, an individual proposition and way of believing the proposition. More specifically, he expresses a strong preference for a Corrected Fregean Theory of Demonstrations and derides the alternative Indexical Theory of Demonstrations. I have argued (Austin 1990, 115-134) that Kaplan's preferred view is unsatisfactory in two ways: it fails to say how visual experience determines associated demonstrative belief, and it is inconsistent anyway with a proper understanding of its own motivating intuitions. I then indicated that a Corrected Russellian (Indexical) Theory would succeed in the second way, if only it were developed fully enough to succeed in the first.

Such a theory of demonstrations, and indeed any three-term theory of belief, must of course explain particular instances of believing [for example, Puzzling Pierre (Kripke 1979 repr. in Salmon and Soames 1988), Two-Tubes Smith (Austin 1990),
[1] the Augmented Messy Shopper (Perry 1979 repr. in Salmon and Soames 1988)]. Since, however, the three-term relation does not reduce to any sort of belief about the third term, the theorist cannot rely on arguments that assume some sort of belief about the third term. Consequently, we need some other way of choosing from among the candidates for the third term, which range from the neurophysiological up to (but of course not including) the doxastic. Furthermore, this must be done with the following five constraints, concerning knowledge, communication and explanation, in mind.

First, beliefs are such that they can be attributed, and the truth-value of belief attributions is variously sensitive to ways of believing and propositions believed. The truth-value may be exquisitely sensitive, as, say, in the context of a psychological experiment, legal proceeding, or so-called de se belief, where the use of "I" or "he himself" may force such sensitivity; or it may be insensitive, as in so-called de re belief attributions, where, in effect, the attributer declares her lack of interest in ways of believing. Degree and kind of sensitivity may depend on factors outside the believer; for example, who is doing the attributing and for what purpose, to whom the attributer is speaking, what the audience presupposes, what the audience believes other members of the audience presuppose, etc. Any adequate semantics for belief attribution must take all this into account. A framework for beginning this task is provided by Richard (1990). His view introduces the varied class of restrictions--to match the varieties of context-dependence that common sense finds in English. This view, in which (restricted) correlation function and representation are key, makes it urgent to develop a theory of representations (or modes of demonstrating, as Richard called the relevant subclass of third terms in Richard [1983]) to which belief attributions may in some contexts be sensitive. Richard argues (Richard 1990, Chapter 4, Section 5) that we ought to retain (from Salmon 1986) the following necessary condition for true belief attribution: One attributes belief truly only if one refers (somehow) to the proposition believed. He observes that in attributing beliefs to believers who lack the attributer's language (for example, nonhuman animals, infants and speakers of radically different languages whose concepts are askew to ours), we regard failure of referential agreement as sufficient for falsity of the attribution, and we regard success in referential agreement as contributing to truth of attribution. I will argue that there are other contexts in which referential agreement is not required and that, consequently, the distinction between semantics and pragmatics is best viewed as a difference in degree rather than in kind.

Second, it must be possible for a believer to know the individual proposition believed and to know that he himself believes that individual proposition - where the individual in question is the object visually demonstrated. The best reason for thinking that we sometimes believe individual propositions is the fact that we sometimes know such propositions. And any three-term theory that made knowledge of one's own beliefs impossible would be a nonstarter and false.

Third, it must be possible for someone to know others' third terms: x is known by Jones to be such that the believes relation holds among the believer Smith, the third term x, and the proposition that p.

Fourth, because objects are given in visual demonstrations, there is no need for the third term to be a mediating individual concept, such as would be expressed by a definite description of the object's (actual and veridical) appearance. It is here that the Corrected Russellian Theory differs most sharply from the Corrected Fregean Theory t in some mental states, the very object perceived is 'hooked' in a visual demonstration of it: having one's (cognitive) hooks in an object is necessary for demonstrative knowledge of that object. The 'correction' comes in the rejection of Russell's definite-descriptive version of sense datum theory with its overly restricted notion of acquaintance; but sense-datum theories do offer an insight worth saving t that there is an irreducibly demonstrative element in perception t and this insight is not given its full due in the Corrected Fregean Theory.

Fifth, to explain normal, typical belief, one must have a classificatory scheme that also allows for abnormal, atypical belief. A good theory of the best examples ought to have the resources to explain why they're the best and how those that are less than best differ from the paradigmatic. Exploration of this constraint has independent interest since many philosophical accounts of cognitive functions come up short in their attempts to draw a crucial distinction between normal and abnormal function t so the more general question has great methodological significance.

This last constraint also suggests an ideal context in which to address the preceding ones and in which to develop the Corrected Russellian Theory: recent experimental and theoretical work in the neuropsychology of vision presents numerous examples of abnormal visual demonstrations. The field is in just the sort of conceptual mess that cries out for careful philosophical cleanup.

Subject to the aforementioned constraints, I propose therefore to find answers to the following questions: What is a normal or typical visual demonstration? How can we make use of information from neuropsychology about abnormal or atypical visual (and spatial) cognition to characterize normal or typical demonstrations? How is the metaphor of a language of thought to be cashed in for visual (and spatial) cognition? What is the role of mental imagery in visual demonstrations? Is imageless visual demonstration possible (as is alleged in cortical blindness, or 'blindsight')? Is the language of vision a productive and systematic visual language? And if so, what are the constituent structures underlying its syntax and its semantics? Can recent work on visual programming languages (Chang 1990) help us to answer this question? Is the computer model that underlies the metaphor adequate to characterizing the distinction between normal and non-normal visual demonstrations? How is the clinically and theoretically important distinction between normal and non-normal visual function to be drawn? How should this distinction be applied to cases of brain lesions associated with visuo-spatial agnosias?

In the past two years, new monographs, anthologies and articles have become available that address these questions from computational and neuropsychological perspectives. [Many of the 'older' items cited below did not become available until quite recently because the library has been slow in getting them. The most important of them, Farah 1990, became available November 1990.] I will use this material in developing a taxonomy of abnormal visual demonstrations to help show what is truly essential to (successful) visual demonstration. This will form a chapter of A Body of One's Own: Spatial Cognition and Self-Knowledge. I will argue that every visual demonstration has an object; that it is possible (because actual) to demonstrate an object without knowing where it is, even in one's own perceptual space; and that necessarily, visual demonstrations normally have perceptual unity (where damage to the hippocampus, the engine of attention, can destroy the capacity to have visual representations with unity). [to be added: discussion of FICTIONAL OBJECTS, MODAL KNOWLEDGE AND UNITY]

The latter echoes a point stressed by Russell (following Moore) in "Meinong's Theory of Complexes and Assumptions," where he observes that propositions are complexes that have unity. A major difference between Russell and Moore, on the one hand, and their Idealist teachers, on the other, is their disagreement over the source of unity in thought and perception. But they agreed about its central importance in the analysis of thought [(Griffin 1990) (Hylton 1990)]. As I will also argue, they were right in what they agreed about. The historical work of Clemenson (1991), Patricia Kitcher (1990), Hatfield (1990) and others demonstrates that debate over the sources of unity in perception and in apperception, as well as their relationships, has been active for many centuries. Indeed, a unified history of cognitive science may be written with unity at its focus. This project seeks to contribute to the contemporary discussion in this continuing debate.


Allport, A. 1989. "Visual Attention." In MI Posner ed. Foundations of Cognitive Science, 631-682. Bradford Books/The MIT Press.

Austin, DF. 1990. What's the Meaning of "This"? A Puzzle about Demonstrative Belief. Cornell University Press.

Block, N. 1990. "The Computer Model of the Mind." In DN Osherson and EE Smith, eds. An Invitation to Cognitive Science Vol. 3: Thinking, Epilogue, Ch 3. The MIT Press.

Brown, JW, ed. 1989. Neuropsychology of Visual Perception. Lawrence Erlbaum Associates.

Chang, S-K. 1990. "Principles of Visual Languages." In S-K Chang, ed. Principles of Visual Programming Systems. Prentice Hall.

Clemenson, David. 1991. "Intuition of Nonexistent Objects in Early Seventeenth Century Scholasticism." Ch. 2 of: The Influence of Jesuit Cognitive Psychology on Descartes' First Causal Proof for the Existence of God. Unpub. History of Science PhD Dissertation. Harvard University.

Cooper, LA. 1988. "The Role of Spatial Representations in Complex Problem Solving." In S Schiffer and S Steele, eds. Cognition and Representation, 53-86.Westview Press.

Damasio, H and AR. 1989. Lesion Analysis in Neuropsychology. Oxford University Press.

Farah, MJ. 1990. Visual Agnosia. Bradford Books/The MIT Press.

Finke, RA. 1989. Principles of Mental Imagery. Bradford Books/The MIT Press.

Furth, M. 1988. Cambride University Press

Griffin, Nicholas. 1990. Russell's Idealist Apprenticeship. Oxford University Press.

Hatfield, G. 1990. The Natural and the Normative: Theories of Spatial Perception from Kant to Helmholtz. Bradford Books/The MIT Press.

- - - . 1989. "Computation, Representation and Content in Noncognitive Theories of Perception." In S Silvers, ed. Rerepresentation: Readings in the Philosophy of Mental Representation, 255-288. Kluwer Academic Press.

Hildreth, EC and S Ullman. 1989. "The Computational Study of Vision." In MI Posner ed. Foundations of Cognitive Science, 581-630. Bradford Books/The MIT Press.

Hoffman, DD and BM Bennett. 1988. "Perceptual Representations: Meaning and Truth Conditions." In S Schiffer and S Steele, eds. Cognition and Representation, 87-128.Westview Press.

Humphreys, GW and MJ Riddoch. 1987. To See But Not to See: A Case Study of Visual Agnosia. Lawrence Erlbaum Associates.

- - - and V Bruce. 1989. Visual Cognition: Computational, Experimental and Neuropsychological Perspectives. Lawrence Erlbaum Associates.

Hylton, P. 1990. Russell, Idealism and the Emergence of Analytic Philosophy. Oxford University Press.

Kaplan, David. 1977. "Demonstratives" (Draft #2), dittograph. UCLA Philosophy Department. Reprinted, with a postscript, in Themes from Kaplan, J. Almog et al, eds. Oxford University Press, 1989.

Kitcher, Patricia. 1990. Kant's Transcendental Psychology. Oxford University Press.

Kosslyn, SM, MH Van Kleeck and KN Kirby. 1990. "A Neurologically Plausible Model of Individual Differences in Visual Mental Imagery." In PJ Hampson, DF Marks and JTE Richardson, eds. Imagery: Current Developments, 39-77. Routledge.

McNaughton, BL. 1989. "Neuronal Mechanisms for Spatial Computation and Information Storage." In L Nadel, LA Cooper, P Culicover and RM Harnish, eds. Neural Connections, Mental Computation, 285-350. Bradford Books/The MIT Press.

O'Keefe, J. 1989. "Computations the Hippocampus Might Perform." In L Nadel, LA Cooper, P Culicover and RM Harnish, eds. Neural Connections, Mental Computation, 225-284. Bradford Books/The MIT Press.

Olson, Kenneth. 1989. An Essay on Facts. University of Chicago Press.

Osherson, DN, SM Kosslyn and JM Hollerbach, eds. 1990. An Invitation to Cognitive Science Vol. 2: Visual Cognition and Action. The MIT Press. (Chs. 1 - 5 Visual Cognition, 3 - 148.)

Pinker, S. 1988. "A Computational Theory of the Mental Imagery Medium." In M Denis, J Engelkamp and JTE Richards, eds. Cognitive and Neuropsychological Approaches to Mental Imagery, 17-32. Martinus Nijhoff Pub.

Pylyshyn, Z. 1988. "Here and There in the Visual Field." In Z Pylyshyn, ed. Computational Processes in Human Vision: An Interdisciplinary Perspective, 210 - 236. Ablex Publishing Co.

Ramachandran, VS. 1990. "Visual Perception in People and Machines." In A Blake and T. Troscianko, eds. AI and the Eye, 21 - 77. John Wiley and Sons.

Richard, M. 1983. "Direct Reference and Ascriptions of Belief." Journal of Philosophical Logic 12: 425 - 452. (Reprinted in Salmon and Soames 1988.)

- - -. 1987. "Quantification and Leibniz' Law." Philosophical Review XCVI 4 (October): 555 - 578.

- - -. 1990. Propositional Attitudes: An Essay on Thoughts and How We Attribute Them. Cambridge University Press.

Russell, B. 1904. "Meinong's Theory of Complexes and Assumptions." Reprinted in D Lackey, ed. Essays in Analysis by Bertrand Russell, 21 - 76. George Braziller 1973.

Shallice, T. 1988. From Neuropsychology to Mental Structure. Cambridge University Press.

Salmon, N. 1986. Frege's Puzzle. Bradford Books/The MIT Press.

- - - and S Soames, eds. 1988. Propositions and Attitudes. Oxford University Press.

Soames, S. 1984. "Direct Reference, Propositional Attitudes and Semantic Content." Philosophical Topics, 15, 1 (Spring): 47 - 87. (Reprinted in Salmon and Soames 1988.)

- - -. 1985. "Lost Innocence." Linguistics and Philosophy 8, 1: 59 - 71. (Reprinted in P Grim, CJ Martin and P Athay, eds. 1987. The Philosopher's Annual, Vol. 8, Ridgeview Press.).

- - -. 1987. "Substitutivity." In JJ Thomson, ed. Being and Saying, 99 - 132. The MIT Press.

Stiles-Davis, J, M Kritchevsky and U Bellugi, eds. 1988. Spatial Cognition: Brain Bases and Development. Lawrence Erlbaum Associates.

Sunderland, A. 1990. "The Bisected Image? Visual Memory in Patients with Visual Neglect." In PJ Hampson, DF Marks and JTE Richardson, eds. Imagery: Current Developments, 333-350. Routledge.

Zangwill, O and MA Wyke. 1990. "Hughlings Jackson on the Recognition of Places, Persons and Objects." In C Trevarthen, ed. Brain Circuits and Functions of the Mind: Essays in Honor or Roger W. Sperry, Ch 16. Cambridge University Press.


[1] Here is one version of the example: Smith is the subject of a psychological experiment designed to test his ability to focus his eyes independently. The experimental apparatus consists of a large, opaque screen with two eye-holes in it; two tubes lead from the eye holes and can be oriented in many directions to point at a uniformly red, uniformly illuminated sheet of plastic in front of the screen. Smith is to look through the tubes and to report what he sees. He knows the set-up and knows that he is ignorant of the tubes' orientations. At 11:59 a.m., Smith closes his left eye and looks through the right tube with his right eye. A rather cautious and pedantic fellow, he thinks to himself, "this = this". For the next two minutes, he continues to look through the right tube with his right eye, and his belief persists until 12:01 p.m. At 12:00 p.m., he opens his left eye, looks through the left tube, and, for the next minute, thinks to himself, "that = that". Throughout the second minute (12:00 - 12:01), he also wonders, "Is this not identical to that?" (and neither believes nor disbelieves what he could express by "this = that" or by "this = that"), a reasonable thing for him to wonder, since he does not know whether the tubes point at two different spots, or the same one. In fact, they are both pointed at the same spot. To be faithful to the intuitions that underlie Kripke's arguments for direct reference, it must be maintained that (i) the propositions Smith believes here are true if and only if the very spot he actually sees is identical to that spot; and (ii) his uses of the demonstratives "this" and "that" are not simply short for (possibly egocentric) definite descriptions (e.g., "the thing that now actually appears to me to be on the right"). To be faithful to the facts of the case, it must also be that his right eye belief persists for the full two minutes, and that he has a different left-eye belief during the second minute. Taken together, the intuitions suggest that the proposition Smith believes and expresses using "this = this" cannot be accurately re-expressed without devices of direct reference (including the indexicals "I" and "now"). If the traditional analysis is to be maintained, then what Smith believes must instead be an individual proposition, aptly represented by the ordered triple: <T, =, T>, with the spot seen and the identity relation as its constituents. Unfortunately for the traditional analysis so construed, the very same proposition (call it "Spot") is also expressed by Smith's uses of "that = that" and "this = that", given the directly referential nature of the demonstratives used. Hence, on the traditional analysis, Smith already believes what "that = that" expresses before the beginning of the second minute, because "that = that" expresses Spot; and Smith does not wonder, "Is this = that?", for that would be to wonder if Spot is true, and he already believes it. The traditional view therefore seems unable to account for the facts of this case while maintaining the Kripkean intuitions. And I know of no extant theory which accommodates both these intuitions and the facts of the case; in (Austin 1990), I argue the latter claim at the necessary length, supporting it with detailed criticism of the views of Kaplan, Perry, Schiffer, Russell, Lewis, Chisholm, Stalnaker, Plantinga and Ackerman, among others.
[2] The Corrected Russellian and Corrected Fregean views are alike in holding that singular terms are not synonymous with definite descriptions that may be associated with a term's use. Here I am concerned to emphasize the differences between the views.


This page was last revised on Tuesday, August 15, 2000