"As we ascend the scale of complexity from sim-ple thermostat,
through sophisticated robot, to human being, we discover that our eftorts to
design systems with the requisite behavior increasingly run foul of the problem
of combinatorial explosion. Increasing some parameter by, say, 10 percent -
10 percent more inputs, or more degrees of freedom in the behavior to be controlled,
or more words to be recognized, or whatever - tends to increase the internal
complexity of the system being designed by orders of magnitude. Things get out
of hand very fast and, for instance, can lead to computer programs that will
swamp the largest, fastest machines. :Now somehow the brain has solved the problem
of combinatorial explosion. It is a gigantic network of billions of cells, but
still finite, compact, reliable, and swift, and capable of learning new behaviors,
vocabularies, theories, almost without limit. Some elegant, generative, indefinitely
extendable principles of representaion must be responsible. We have only one
model of such a representation system: a human language. So the argument for
a language of thought comes down to this, what else could it be?"
Daniel Dennett, "True Believers: The Intentional Strategy" reprinted
in Lycan, Mind and Cognition, 2nd Edition, Oxford, 1999, Blackwell, p.85,6.