Harnad, S. (1998) Beyond Object Constancy. Institution of Electrical Engineers (IEE) Seminar on "Self-Learning Robots II: Bio-Robotics." Commentary on Pfeifer (1998) "Embodied Cognition: Dynamic and Information Theoretic Implications of Embodiment"


Stevan Harnad
Cognitive Sciences Centre
University of Southampton
Highfield, Southampton

It is surely not irrelevant that such a large proportion of the brain is devoted to the body and to sensorimotor processing (Jeannerod 1994, Kaas 1995, Zhang et al. 1997). That should already serve to alert us that embodiment is likely to be an important factor in cognition.

Add to that the well-rehearsed advantages of letting the world serve as its own model (Steels & Brooks 1995), Gibson's (1979) influential findings on the role of sensorimotor interaction in the detection of invariants, and perhaps also my own writing on the symbol grounding problem (Harnad 1990, 1996) and there would appear to be a good deal of support for Pfeifer's (1998) position.

I would accordingly like merely to amplify one of his points about the "fundamental problem of learning category distinctions" (Pfeifer 1998). Object constancy -- the capacity to see an object as being one of constant size and shape despite variation in retinal position and distance from the observer -- is certainly one of the most fundamental category invariants that the brain is capable of detecting. (Moreover, so important is it, that it is unlikely that object constancy is entrusted to learning: most of it has probably been "prepared" innately by evolution.) But the principle of detecting and extracting invariants under sensorimotor transformations during sensorimotor interactions with objects can be generalised to more abstract invariants, extracted under more abstract transformations.

I am not just referring to the dissociations between self-centred and object-centred spatial perception, or to higher-order spatial invariants that have been revealed by neuropsychological testing (the fact that a patient with left half-field inattention neglects not only the left half of his retinal field, but also neglects the left half of objects in his right retinal field -- and, presumably, so on through as many higher-order fields within fields as the neuropsychologist might care to test; Corballis 1986). Rather, I am referring to the higher-order invariants that depend on how we choose (or are constrained by our physical and social environment to choose) to sort and label all the kinds of things to which we assign a category name or a descriptive phrase. Here it is not invariance under sensorimotor transformations that the brain is detecting, but invariance under within- and between-category variance.

If we must learn to call some things "X" and other things "Y," and Xs and Ys are at first highly interconfusable, then we must somehow modify our internal representations of Xs and Ys so as to be able to sort and label them reliably. When this involves changing either the values or the weighting of the dimensions of the internal representations of Xs and Ys, the process is called "categorical perception" (Harnad 1987; Livingston et al 1998; Tijseeling et al. 1997). It is no longer sensorimotor in the gross motor sense; but as even calling or not calling something by the same name is a motor act (as surely as grasping or pointing at it is), naming is still based on a very high-level sensorimotor invariance when it is based on direct sensorimotor interaction with Xs and Ys.

When our interactions with objects become still more abstract, being based only on the interactions between names and descriptions -- on "hearsay," so to speak -- then we have arrived at the full power of natural language, the power of symbolic "theft" over sensorimotor "toil" (Cangelosi & Harnad, in prep.; Greco, Cangelosi & Harnad, in prep.). Yet even at such abstract cognitive heights, embodiment is never escaped, for the power of names and propositions is completely parasitic on the meanings of those names, and those must all eventually be grounded in the sensorimotor interactions with the kinds of objects they designate, and the sensorimotor invariants on the basis of which the names are assigned (Harnad 1996).


Cangelosi, A & Harnad, S. (in prep) On the Virtues of Theft Over Honest Toil: Grounding Language and Thought in Sensorimotor Categories: Grounding Language and thought in Senosimotor Categories. http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad96.language.theft.html

Corballis, M.C. (1986) Fresh fields and postures new: A discussion paper. Brain & Cognition 5: 240-252.

Gibson, J. J. (1979) An ecological approach to visual perception. Boston: Houghton Mifflin

Jeannerod, M. (1994) The representing brain: neural correlates of motor intention and imagery. Behavioral and Brain Sciences 17(2) in press.

Kaas, J. H. (1995) The reorganisation of sensory and motor maps in adult mammals. In: The cognitive neurosciences.; Michael S. Gazzaniga, Ed. MIT Press, Cambridge, MA, US. 1995. p. 51-71.

Harnad, S. (1987) Psychophysical and cognitive aspects of categorical perception: A critical overview. In: Harnad 1987. http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad87.cpreview.html

Harnad, S. (1987) The induction and representation of categories. In: Harnad 1987. http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad87.categorization.html

Harnad, S. (1990) The Symbol Grounding Problem. Physica D 42: 335-346. [Reprinted in Hungarian Translation as "A Szimbolum-Lehorgonyzas Problemaja." Magyar Pszichologiai Szemle XLVIII-XLIX (32-33) 5-6: 365-383.] http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad90.sgproblem.html

Harnad, S. (1992) Connecting Object to Symbol in Modeling Cognition. In: A. Clark and R. Lutz (Eds) Connectionism in Context Springer Verlag, pp 75 - 90. http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad92.symbol.object.html

Hayes, P., Harnad, S., Perlis, D. & Block, N. (1992) Virtual Symposium on Virtual Mind. Minds and Machines 2: 217-238. http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad92.virtualmind.html

Harnad, S. (1993) Grounding Symbols in the Analog World with Neural Nets. Think 2(1) 12 - 78 (Special issue on "Connectionism versus Symbolism," D.M.W. Powers & P.A. Flach, eds.). [Also reprinted in French translation as: "L'Ancrage des Symboles dans le Monde Analogique a l'aide de Reseaux Neuronaux: un Modele Hybride." In: Rialle V. et Payette D. (Eds) La Modelisation. LEKTON, Vol IV, No 2.] http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad93.symb.anal.net.html http://cwis.kub.nl/~fdl/research/ti/docs/think/2-1/index.stm

Harnad, S. (1993) Artificial Life: Synthetic Versus Virtual. Artificial Life III. Proceedings, Santa Fe Institute Studies in the Sciences of Complexity. Volume XVI. http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad93.artlife.html

Harnad, S. (1994) Levels of Functional Equivalence in Reverse Bioengineering: The Darwinian Turing Test for Artificial Life. Artificial Life 1(3): 293-301. Reprinted in: C.G. Langton (Ed.). Artifial Life: An Overview. MIT Press 1995. http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/

Harnad, S. (1994) Computation Is Just Interpretable Symbol Manipulation: Cognition Isn't. Special Issue on "What Is Computation" Minds and Machines 4:379-390 [Also appears in French translation in "Penser l'Esprit: Des Sciences de la Cognition a une Philosophie Cognitive," V. Rialle & D. Fisette, Eds. Presses Universite de Grenoble. 1996] http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad95.computation.cognition.html

Harnad, S, (1995) Does the Mind Piggy-Back on Robotic and Symbolic Capacity? In: H. Morowitz (ed.) "The Mind, the Brain, and Complex Adaptive Systems." Santa Fe Institute Studies in the Sciences of Complexity. Volume XXII. P. 204-220. http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad95.mind.robot.html

Harnad, S. (1995) Grounding Symbolic Capacity in Robotic Capacity. In: Steels, L. and R. Brooks (eds.) The Artificial Life Route to Artificial Intelligence: Building Embodied Situated Agents. New Haven: Lawrence Erlbaum. Pp. 277-286. http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad95.robot.html

Harnad, S. Hanson, S.J. & Lubin, J. (1995) Learned Categorical Perception in Neural Nets: Implications for Symbol Grounding. In: V. Honavar & L. Uhr (eds) Symbol Processors and Connectionist Network Models in Artificial Intelligence and Cognitive Modelling: Steps Toward Principled Integration. Academic Press. pp. 191-206. http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad95.cpnets.html

Harnad, S. (1995) Grounding symbols in sensorimotor categories with neural networks. In: IEE Colloquium "Grounding Representations: Integration of Sensory Information in Natural Language Processing, Artificial Intelligence and Neural Networks" (Digest No.1995/103). http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad95.iee.html

Harnad, S. (1996) The Origin of Words: A Psychophysical Hypothesis In Velichkovsky B & Rumbaugh, D. (Eds.) "Communicating Meaning: Evolution and Development of Language. NJ: Erlbaum: pp 27-44. http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad96.word.origin.html

Tijsseling, A. & Harnad, S. (1997) Warping Similarity Space in Category Learning by Backprop Nets. In: Ramscar, M., Hahn, U., Cambouropolos, E. & Pain, H. (Eds.) Proceedings of SimCat 1997: Interdisciplinary Workshop on Similarity and Categorization. Department of Artificial Intelligence, Edinburgh University: 263 - 269. http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/

Harnad, S. (1998) Turing Indistinguishability and the Blind Watchmaker. In: Mulhauser, G. (ed.) "Evolving Consciousness" Amsterdam: John Benjamins (in press) http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad98.turing.evol.html

Livingston, K., Andrews, J., & Harnad, S. (In press). Categorical Perception Effects Induced by Category Learning. Journal of Experimental Psychology: Learning, Memory, and Cognition. http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/

Steels, L. and R. Brooks (1995) (eds.) The Artificial Life Route to Artificial Intelligence: Building Embodied Situated Agents. New Haven: Lawrence Erlbaum.

Schyns, P., Goldstone. R. & Thibeaut P. (1998 in press) The development of features in object concepts. Behavioral and Brain Sciences 21. http://www.cogsci.soton.ac.uk/bbs/Archive/bbs.schyns.html

Tijsseling, A. & Harnad, S. (1997) Warping Similarity Space in Category Learning by Backprop Nets. In: Ramscar, M., Hahn, U., Cambouropolos, E. & Pain, H. (Eds.) Proceedings of SimCat 1997: Interdisciplinary Workshop on Similarity and Categorization. Department of Artificial Intelligence, Edinburgh University: 263 - 269.

Zhang, J. Riehle, A. Requin, J. & Kornblum, S. (1997) Dynamics of single neuron activity in monkey primary motor cortex related to sensorimotor transformation. Journal of Neuroscience 17(6): 2227-2246.