Analysis and Computation
Summary. Contemporary psychology assumes that human thought processes are characterizable in computational terms. This suggests support for the thesis that analytical philosophy, if occupied with the analysis of human reasoning, is concerned at bottom with the structure of computational competence in each major area of intellectual activity. Computational analogues can in fact be developed both for semantic ascent and for semantic descent. Computerization is thus comparable with logical formalization as a technique of analysis that may be useful in the treatment of certain problems. But there are important reasons why this fact does not afford any general characterization of analytical philosophy.
THIRTY or forty years ago, when computers were a novelty and much less seemed to be known about them than about human thinking, it was natural for the question 'Can computers think?' to be among those in the forefront of discussion. This question soon dissolved very fruitfully, under the pressure of enquiry, into a vast number of more specific issues in the various sciences of computer hardware and software. As a result we have eventually come to have rather more extensive and accurate knowledge--knowledge that is still growing rapidly--about what computers can do and how they can do it than we have ever had about the actual mechanisms of human thinking. For the last decade or two, therefore, the guiding question in the science of thought has tended to point in the opposite direction. The question raised has not been so much 'Can computers think?' but rather 'Do humans compute?'
The computational hypothesis, which has come to dominate cognitive psychology, assumes an affirmative answer to this question, not in the sense of a metaphysical claim that the essence of mind is computation but in the sense of a scientific methodology. It expects the computational analogy to be more successful than any other in generating a variety of theories that are not only testable, but also