Connectionists should not lose sight of the fact that the electronic circuit has little in common with the neural circuit in the brain.
"But oh, beamish nephew, beware of the day,
If your Snark be a Boojum! For then
You will softly and suddenly vanish away,
And never be met with again!" (Lewis Carroll)
"Most connectionists seem wary of proclaiming too boldly that their
networks model the actual activity of the brain" (Green 1998)
1. I couldn't help smiling at this statement by Green (1998) in his target article, "Are connectionist models theories of cognition?" Keeping in mind K. S. Lashley's repeated warnings against modelling brain function on inert copper wire and electronic gear soldered into circuits (see Lashley 1963/1929, 1998/1952; Orbach 1982), Green's quote from McClelland, Rumelhart, and Hinton (1986, p.10) that connectionist models "seem much more closely tied to the physiology of the brain than other information-processing models" but their "physiological plausibility and neural inspiration... are not the primary bases of their appeal to us" had me positively giggling.
2. Nobody seems to remember "Lashley's lesson" any longer: that neurons, unlike metal wires, are live, metabolizing cells that change their sensitivity from moment to moment; that they are excited or inhibited by an indefinite number of other neurons via contacts that are variable. The junction at the synapse is anything but a solder-joint. It too varies from moment to moment for a number of reasons, one being the micromovements of end feet on the post-synaptic cell. The all-or-none law on which connectionist models are based has been interpreted incorrectly to mean that the neuron, regardless of its condition, either fires or does not fire to a particular excitatory stimulus. And "Lashley's lesson" includes the view that changes in excitatory potential occur not only at synapses that were excited during learning but at quiescent synapses as well. The evidence for this includes transfer of training, recovery of function, stimulus generalization, stimulus and response equivalence, and learning by similarity. The "Hebb synapse" as the site of learning does not take "Lashley's lesson" seriously.
3. For a historical account, interested readers may wish to consult my forthcoming book, "The Neuropsychological Theories of Lashley and Hebb" to be issued in July of 1998 by the University Press of America, Inc., Lanham MD (Orbach 1998).
Green, CD. (1998) 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
Lashley, KS. (1963/1929) Brain mechanisms and intelligence: A Quantitative Study of Injuries to the Brain. (Introduction: D.O. Hebb. New York: Dover Publications
Lashley, KS. (1998/1952) Vanuxem Lectures. Four lectures delivered at Princeton University, 1952. In J. Orbach (Ed.) The Neuropsychological Theories of Lashley and Hebb: Contemporary Perspectives Fifty Years After 'The Organization of Behavior'" Lanham, MD: University Press of America.
McClelland, J. L., Rumelhart, D. E., & Hinton, G. E. (1986) The appeal of parallel distributed processing. In: Parallel distributed processing: Explorations in the microstructure of cognition (vol. 1), ed. Rumelhart, D. E. & McClelland, J. L., MIT Press.
Orbach, J. (1998, in press) The Neuropsychological Theories of Lashley and Hebb. Lanham MD: University Press of America.
Orbach, J. (Ed.) (1982) Neuropsychology After Lashley: Fifty Years Since the Publication of "Brain Mechanisms and Intelligence." Hillsdale, N.J.: Lawrence Erlbaum Associates