CONNECTIONIST MODELS CAN REVEAL GOOD ANALOGIES
Commentary on Green on Connectionist-Explanation

Abstract

Green (1998a) argues that distributed connectionist models are not theories of cognition. This is reasonable if it means that the explanatory role of connectionist models is not clear, but Green's analysis seems directed against the wrong target when he applies a realist position to models. His argument confuses models with objects. Models are useful as long as they establish analogies between unknown and known phenomena; but not all details are important. The real problem may concern the explanatory role of connectionist models (which is what Green also seems concerned about), but then it should be formulated on different grounds. If they are intended as cognitive models (and not as mere AI artifacts), their internal operations should be describable (by analogy) using a cognitive vocabulary. This is often not the case with connectionist models. Are they always useless as cognitive models then? I cannot share Green's conclusion that the only hope for connectionism is to model brain activity. On the contrary, because the most attractive feature of connectionist models is that they can perform cognitive tasks using no symbols, they can be useful tools for studying (by analogy) the origin and grounding of symbols.

Keywords

References

• Fodor, J. A. (1975) The language of thought New York Thomas Y. Crowell
• Fodor, J. & Pylyshyn, Z. (1988) Connectionism and cognitive architecture: A critical analysis. Cognition 28: 3 - 71
• Ford, K. M. & Hayes, P. J. Reasoning agents in a dynamic world: The frame problem. PSYCOLOQUY 3(59) ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1992.volume.3/psycoloquy.92.3.59.frame-problem.1.ford+hayes
• Goldsmith, M. (1998) Connectionist modeling and theorizing: Who does the explaining and how? PSYCOLOQUY 9(18) ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1998.volume.9/psycoloquy.98.9.18.connectionist-explanation.15.goldsmith
• Greco, A. (1990a). Some remarks about connectionism in psychological simulation. Cognitive Systems, 2-4, 359-372. http://cogprints.soton.ac.uk/abs/psyc/199804024
• Greco, A. (1990b). What kind of psychological processes can be modelled by a connectionist system? Paper presented at the Workshop "Connectionism: bridge between mind and brain?," Center for Interdisciplinary Research ZIF, Univ. of Bielefeld, Germany. http://www.lettere.unige.it/sif/strutture/1/home/greco/zif90/zif90 .ht m
• Greco, A. (1994) Integrating "different" models in cognitive psychology. Cognitive Systems, 4-1, 21-32. http://cogprints.soton.ac.uk/abs/psyc/199804025
• Greco, A., Cangelosi, A. (1996). Language, categorization, and representation: a pilot study using neural networks. DISA Technical Report n. 96-03, Univ. of Genova, Italy. http://www.lettere.unige.it/sif/strutture/1/home/greco/techrep3.zi p
• Greco, A., Cangelosi, A., Harnad, S. (1998) A connectionist model for categorical perception and symbol grounding. Paper accepted at the International Conference on Artificial Neural Networks (ICANN 98). http://cogprints.soton.ac.uk/abs/psyc/199803023
• Green, C. D. (1998a). Are connectionist models theories of cognition? PSYCOLOQUY 9(4) ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1998.volume.9/psyc.98.9.04.connectionist-explanation.1.green.
• Green, C.D. (1998b) Connectionist nets are only good models if we know what they model. Reply to Lee et al. on Connectionist- Explanation. PSYCOLOQUY 9(23) ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1998.volume.9/psycoloquy.98.9.23.connectionist-explanation.20.green
• Green, C.D. (1998c) Higher functional properties do not solve connectionism's problems. Reply to Goldsmith on Connectionist- Explanation. PSYCOLOQUY 9(25) ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1998.volume.9/psycoloquy.98.9.25.connectionist-explanation.22.green
• Green, C.D. (1998d) The degrees of freedom would be tolerable if nodes were neural. Reply to Lamm on Connectionist-Explanation. PSYCOLOQUY 9(26) ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1998.volume.9/psycoloquy.98.9.26.connectionist-explanation.23.green
• Harnad, S. (1990) The Symbol Grounding Problem. Physica D 42: 335-346. http://cogprints.soton.ac.uk/abs/psyc/199803014
• Harnad, S. (1993) Problems, Problems: The Frame Problem as a Symptom of the Symbol Grounding Problem. PSYCOLOQUY 4(34) ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1992.volume.3/psycoloquy.92.3.34.frame-problem.11.harnad
• Hatfield, G. (1989). Computation, representation, and content in noncognitive theories of perception. In: Silvers, S. (Ed.), Rerepresentation. Dordrecht: Kluwer.
• Hesse, M.B. (1966). Models and analogies in science. Notre Dame, Indiana: University of Notre Dame Press.
• Lee, C., van Heuveln, B., Morrison, C.T., Dietrich, E. (1998). Why connectionist nets are good models. PSYCOLOQUY 9(17) ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1998.volume.9/psyc.98.9.17.connectionist-explanation.14.lee
• McCarthy, J., Hayes, P. (1969) Some philosophical problems from the standpoint of artificial intelligence. In Meltzer B., Michie D. (eds.) Machine intelligence, vol. 4. New York: American Elsevier, pp.463-502.
• Medler, D.A., Dawson, M.R.W. (1998). Connectionism and cognitive theories. Psycoloquy 9(11) ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1998.volume.9/psyc.98.9.11.connectionist-explanation.8.medler
• Nagel, E. (1961).. The structure of science, Harcourt, Brace & World, New York.
• Pylyshyn, Z. W. (1984) Computation and cognition. Cambridge MA: Bradford