Abstract

Where direct experimentation is impossible, astronomy, meteorology, evolutionary biology, and other sciences have long relied on the simulation as a tool for testing theories. Conjectures about the origins of certain human behaviors can be tested using carefully constructed simulations with nonhuman animal subjects. The adequacy of such simulations depends on a number of criteria, but even successful simulations that satisfy all of these criteria are not sufficient to prove the original conjectures. Rather, successful simulations provide mere "plausibility proofs." The Columban simulations are a set of studies of varying degrees of adequacy in which complex, novel behavior in human subjects has been investigated with pigeons. Though computer simulations are, in general, invaluable, computer simulations of cognition are, as simulations, inadequate. The computer metaphor of human intelligence can be traced to a faulty syllogism that pervades the cognitive science literature.