An example of human chess play in the light of chess playing programs.

Abstract : This paper is concerned with the use of chess programs to study human thinking. The work on chess programs has produced a collection of mechanisms sufficient to play chess of modest caliber. Independently of their detailed characteristics, they help understand what must be done in order to play chess. The approach used was to examine in some detail the behavior of a man deciding what move to make in a specific middle game position. Having available a protocol, a transcript of the verbal behavior of the man while he is analysing the board and making his decision. Previous work with protocols in other tasks (proving theorems, guessing sequences, learning concepts) has aimed at constructing computer programs that match the behavior in detail. In this paper the authors undertake only the first stages of such an analysis, laying bare the reasoning the subject employed, by examining his protocol in detail. The analysis draws upon ones general knowledge about reasoning mechanisms and how to organize information processing.