ABSTRACT
T he game of chess has been of interest to cognitive scientists for over five decades in part because proficiency at playing chess recruits both expert cognitive and perceptual knowledge (De Groot, 1978). Much of the research in chess cognition has been directed at understanding the cognitive mechanisms deployed by chess masters when faced with specific chessboard configurations. For example, De Groot (1978) examined differences between weak players and master players and found that both categories of players mentally search approximately the same number of possible moves, but that the experts are capable of selecting better moves more quickly. One putative mechanism for this observed difference may lie in the fact that chess masters are very accurate at reconstructing chess piece positions, even after viewing chessboards for durations as short as 5 seconds. In a seminal study, Chase and Simon (1973) investigated this phenomenon in detail. Their goal was to characterize the perceptual and cognitive strategies used by expert chess players when encoding chessboards into memory. To address this question they created two main tasks: a perceptual task, asking subjects to reconstruct chessboard configurations in plain view using glances as a measure of chunking, and a memory task, asking subjects to reconstruct chessboard configurations from memory, using the clustering and accuracy in recall as a measure of chunking. To assess the importance of previous chess knowledge/expertise, Chase and Simon used both valid chessboard configurations, subconfigurations of which were likely to be familiar to the experienced players, and scrambled or “invalid” chessboard configurations, created by randomly placing the game pieces on a chessboard. They found that the more experienced the player, the larger the size—in terms of the number of pieces—of the encoded mental “chunks,” even if the total number of chunks remained roughly consistent with the typical 7 ± 2 working memory span (Miller, 1956). Chase and Simon also observed that this experience-predicated advantage mostly disappeared with invalid chessboard configurations, presumably because the experts did not have any useful knowledge regarding the nominal game positions encountered in this condition (although later research showed that experts maintain a small but reliable advantage with random positions; Gobet & Simon, 1996a, 1996b).
