Copyright © 1999 Massachusetts Institute of Technology
Historically, chess has been one of the leading fields in the study of EXPERTISE (see De Groot and Gobet 1996 and Holding 1985 for reviews). This popularity as a research domain is explained by the advantages that chess offers for studying cognitive processes: (i) a well-defined task; (ii) the presence of a quantitative scale to rank chess players (Elo 1978); and (iii) cross-fertilization with research on game-playing in computer science and artificial intelligence.
Many of the key chess concepts and mechanisms to be later developed in cognitive psychology were anticipated by Adriaan De Groot's book Thought and Choice in Chess (1946/1978). De Groot stressed the role of selective search, perception, and knowledge in expert chess playing. He also perfected two techniques that were to be often used in later research: recall of briefly presented material from the domain of expertise, and use of thinking-aloud protocols to study problem-solving behavior. His key empirical findings were that (i) world-class chess grandmasters do not search more, in number of positions considered and in depth of search, than weaker (but still expert) players; and (ii) grandmasters and masters can recall and replace positions (about two dozen pieces) presented for a few seconds almost perfectly, while weaker players can replace only a half dozen pieces.
De Groot's theoretical ideas, based on Otto Selz's psychology, were not as influential as his empirical techniques and results. It was only about twenty-five years later that chess research would produce a theory with a strong impact on the study of expertise and of cognitive psychology in general. In their chunking theory, Simon and Chase (1973) stressed the role of perception in skilled behavior, as did De Groot, but they added a set of elegant mechanisms. Their key idea was that expertise in chess requires acquiring a large collection of relatively small chunks (each at most six pieces) denoting typical patterns of pieces on the chess board. These chunks are accessed through a discrimination net and act as the conditions of a PRODUCTION SYSTEM: they evoke possible moves in this situation. In other respects, chess experts do not differ from less expert players: they have the same limits in memory (a short-term memory of about seven chunks) and learning rate (about eight seconds are required to learn a chunk). In chess, as well as in other domains, the chunking theory explains experts' remarkable memory by their ability to find more and larger chunks, and explains their selective search by the fact that chunks evoke potentially good actions. Some aspects of the theory were implemented in a computer program by Simon and Gilmartin (1973). Simulations with this program gave a good fit to the behavior of a strong amateur and led to the estimation that expertise requires the presence of a large number of chunks, approximately between 10,000 and 100,000.
A wealth of empirical data was gathered to test the chunking theory in various domains of expertise. In chess, five directions of research may be singled out as critical: importance of perception and pattern recognition, relative role of short-term and long-term memories, evidence for chunks, role of higher-level knowledge, and size of search.
Converging evidence indicates that perceptual, pattern-based cognition is critical in chess expertise. The most compelling data are that EYE MOVEMENTS during the first few seconds when examining a new chess position differ between experts and nonmasters (De Groot and Gobet 1996), and that masters still play at a high level in speed chess games where they have only five seconds per move on average, or in simultaneous games where their thinking time is reduced by the presence of several opponents (Gobet and Simon 1996).
Research on MEMORY has led to apparently contradictory conclusions. On the one hand, several experiments on the effect of interfering tasks (e.g., Charness 1976) have shown that two of Simon and Chase's (1973) assumptions -- that storage into long-term memory is slow and that chunks are held in short-term memory -- run into problems. This encouraged researchers such as Cooke et al. (1993) to emphasize the role of higher-level knowledge, already anticipated by De Groot (1946/1978). On the other hand, empirical evidence for chunks has also been mounting (e.g., Chi 1978; Gobet and Simon 1996; Saariluoma 1994).
Attempts to reconcile low-level and high-level types of encoding have recently been provided by the long-term WORKING MEMORY (LTWM) theory (Ericsson and Kintsch 1995) and by the template theory (Gobet and Simon 1996). LTWM proposes that experts build up both schema-based knowledge and domain-specific retrieval structures that rapidly encode the important elements of a problem. The template theory, based on the chunking theory and implemented as a computer program, proposes that chunks evolve into more complex data structures (templates), allowing some values to be encoded rapidly. Both theories also account for aspects of skilled perception and problem solving in chess.
Recent results indicate that stronger players search somewhat more broadly and deeply than weaker players (Charness 1981; Holding 1985), with an asymptote at high skill levels. In addition, the space searched remains small (thinking-aloud protocols indicate that grandmasters typically search no more than one hundred nodes in fifteen minutes). These results are compatible with a theory based on pattern recognition: chunks, which evoke moves or sequences of moves, make search more selective and allow better players to search more deeply.
While productive in its own terms, computer science research on chess (see GAME-PLAYING SYSTEMS) has had relatively little impact on the psychology of chess. The main advances have been the development of search techniques, which have culminated in the construction of DEEP BLUE, the first computer to have beaten a world champion in a match. More recently, chess has been a popular domain for testing MACHINE LEARNING techniques. Finally, attempts to use a production-system architecture (e.g., Wilkins 1980) have met with limited success in terms of the strength of the programs.
The key findings in chess research -- selective search, pattern recognition, and memory for the domain material -- have been shown to generalize to other domains of expertise. This augurs well for current interests in the field: integration of low- and high-level aspects of knowledge and unification of chess perception, memory, and problem-solving theories into a single theoretical framework.
Charness, N. (1976). Memory for chess positions: resistance to interference. Journal of Experimental Psychology: Human Learning and Memory 2:641-653.
Charness, N. (1981). Search in chess: age and skill differences. Journal of Experimental Psychology: Human Perception and Performance 2:467-476.
Chi, M. T. H. (1978). Knowledge structures and memory development. In R. S. Siegler, Ed., Children's Thinking: What Develops? Hillsdale, NJ: Erlbaum, pp. 73-96.
Cooke, N. J., R. S. Atlas, D. M. Lane, and R. C. Berger. (1993). Role of high-level knowledge in memory for chess positions. American Journal of Psychology 106:321-351.
de Groot, A. D. (1978). Thought and Choice in Chess. The Hague: Mouton Publishers.
de Groot, A. D., and F. Gobet. (1996). Perception and memory in chess. Heuristics of the professional eye. Assen: Van Gorcum.
Elo, A. (1978). The Rating of Chess Players, Past and Present. New York: Arco.
Ericsson, K. A., and W. Kintsch. (1995). Long-term working memory. Psychological Review 102:211-245.
Gobet, F., and H. A. Simon. (1996). Templates in chess memory: a mechanism for recalling several boards. Cognitive Psychology 31:1-40.
Holding, D. H. (1985). The Psychology of Chess Skill. Hillsdale, NJ: Erlbaum.
Saariluoma, P. (1994). Location coding in chess. The Quarterly Journal of Experimental Psychology 47A: 607-630.
Simon, H. A., and W. G. Chase. (1973). Skill in chess. American Scientist 61:393-403.
Simon, H. A., and K. J. Gilmartin. (1973). A simulation of memory for chess positions. Cognitive Psychology 5:29-46.
Wilkins, D. (1980). Using patterns and plans in chess. Artificial Intelligence 14:165-203.
Binet, A. (1966). Mnemonic virtuosity: a study of chess players. Genetic Psychology Monographs 74:127-162. (Translated fom the Revue des Deux Mondes (1893) 117: 826 - 859.)
Calderwood, B., G. A. Klein, and B. W. Crandall. (1988). Time pressure, skill, and move quality in chess. American Journal of Psychology 101:481-493.
Charness, N. (1989). Expertise in chess and bridge. In D. Klahr and K. Kotovsky, Eds., Complex Information Processing: The Impact of Herbert A. Simon. Hillsdale, NJ: Erlbaum, pp. 183-208.
Chase, W. G., and H. A. Simon. (1973). Perception in chess. Cognitive Psychology 4:55-81.
Frey, P. W., and P. Adesman. (1976). Recall memory for visually presented chess positions. Memory and Cognition 4:541-547.
Freyhoff, H., H. Gruber, and A. Ziegler. (1992). Expertise and hierarchical knowledge representation in chess. Psychological Research 54:32-37.
Fürnkranz, J. (1996). Machine learning in computer chess: the next generation. International Computer Chess Association Journal 19:147-161.
Gobet, F., and H. A. Simon. (1996a). The roles of recognition processes and look-ahead search in time-constrained expert problem solving: evidence from grandmaster level chess. Psychological Science 7:52-55.
Gobet, F., and H. A. Simon. (1996b). Recall of rapidly presented random chess positions is a function of skill. Psychonomic Bulletin and Review 3:159-163.
Goldin, S. E. (1978). Effects of orienting tasks on recognition of chess positions. American Journal of Psychology 91:659-671.
Hartston, W. R., and P. C. Wason. (1983). The Psychology of Chess. London: Batsford.
Newell, A., and H. A. Simon. (1972). Human Problem Solving. Englewood Cliffs, NJ: Prentice-Hall.
Pitrat, J. (1977). A chess combinations program which uses plans. Artificial Intelligence 8:275-321.
Robbins, T. W., E. Anderson, D. R. Barker, A. C. Bradley, C. Fearnyhough, R. Henson, S. R. Hudson, and A. D. Baddeley. (1995). Working memory in chess. Memory and Cognition 24:83-93.
Simon, H. A. (1979). Models of Thought, vol. 1. New Haven: Yale University Press.
Simon, H. A., and M. Barenfeld. (1969). Information processing analysis of perceptual processes in problem solving. Psycho logical Review 76:473-483.
Copyright © 1999 Massachusetts Institute of Technology