Vol. 43 No. 10 Oct. 2002 1 De Groot Simon Chase 1 Cognitive Science Approach to Shogi Playing Processes (1) Some Results on Memory Experiments Takeshi Ito, Hitoshi Matsubara and Reijer Grimbergen In the past, there have been numerous studies into the cognitive processes involved in human problem solving. From the start, games andgame theory have playedan important role in the study of human problem solving behavior. The advantage of using games for the study of cognitive behavior is that games provide a complex but well-defined problem in which evaluation of results is relatively easy. In chess, one of the most well-known cognitive experiments was the study by De Groot on memorizing positions. As a follow-up to De Groot s work, Chase andsimon introducedthe theory of chunking to explain why expert game players perform so well on memory tasks. Chunking is the process of dividing a chess position into smaller parts that have meaning. Chase andsimon showedthat stronger players have bigger chunks of chess knowledge than weaker players. As a first step in our cognitive study of Shogi, we repeatedsome experiments that were conductedin chess. We felt that repeating these experiments was necessary as there are some important differences between chess and Shogi from perceptual point of view (for example, shogi has a 9 9 boardwith all squares having the same color, shogi pieces are two-dimensional and in shogi captured pieces remain part of the game). Because of these differences, it could not be assumed that the results for chess wouldcarry over to shogi. In this paper we give the experimental results of memory tasks in shogi, both with andwithout a time limit. Our results were similar to the ones obtainedin chess. As in chess, there is a correlation between playing strength andthe performance on the memory tasks. From this we can draw a similar conclusion for shogi as for chess: stronger players have bigger chunks of shogi knowledge than weaker players. 1. Department of Computer Science, University of Electro-Communications / 21 Department of Media Architecture, Future University- Hakodate/PRESTO Department of Information Science, Saga University 2998
Vol. 43 No. 10 1 2999 Newell 1) GPS General Problem Solver 1 De Groot 2),3) 2 Chase 4) De Groot Simon 5) 10,000 100,000 6),7) Saito 8) 9) 10) 2. De Groot 2.1 1 2.1.1 8 3 9 CD-ROM
3000 Oct. 2002 Fig. 1 1 An example position of memorization experiment. 1 20 30 40 50 60 2 10 1 (a) (a) 1 (b) EMR-8 2 3 90% 40 90% 36 /40 98 8% 2 1 1 2.1.2 2 20
Vol. 43 No. 10 1 3001 2 Fig. 2 Average memorization time. Fig. 4 4 Average memorization time for each subject. 5 3 1 (a) (b) 0.2 1 3 Fig. 3 Average memorization time for each problem. 30 40 50 3 4 1 3 5 7 9 1 3 96.3% 93.2% 1 3 99.8% 96.7% 0.2 1 5 0.2 6 100% 6 3 1 1 7 1
3002 Oct. 2002 5 Fig. 5 Eye tracking diagram of expert. 6 Fig. 6 Eye tracking diagram of novice. Fig. 7 7 Eye tracking diagram of club player.
Vol. 43 No. 10 1 3003 8 1 3 Fig. 8 Reproduction process of novice for Problem 3. 9 1 3 Fig. 9 Reproduction process of club player for Problem 3. 10 1 3 Fig. 10 Reproduction process of expert for Problem 3. 8 9 10 1 3 8 1 9 2 50 10
3004 Oct. 2002 1 2.1.3 1 1 50 50 4 2.2 2 1 2.2.1 1 5 10 5 3 3 3 3 3 1 1 8 3 9 1 1 CD-ROM 1 20 30 40 50 60 2 10 2 1 20 30 40 50 60 3 3
Vol. 43 No. 10 1 3005 1 2.2.2 11 50% 65% 3 60 90% 12 11 13 2 1 3 1 11 Fig. 11 Accuracy rates for real-game positions. 12 Fig. 12 Accuracy rates for random positions. 13 3 Fig. 13 Eye tracking diagram of novice with 3 seconds limit.
3006 Oct. 2002 14 3 Fig. 14 Eye tracking diagram of expert with 3 seconds limit. 14 100% 3. (1) 3 3 (2) 1 (3) 3 (1) 5 3 3 0.2 7 ±4 7 ± 4 2 1 14 5 4 3 3 (2) 1 1 1 1
Vol. 43 No. 10 1 3007 (3) 50 60 1 60 70 4. 1 1) Newell, A. and Simon, H.A.: Human problem solving, Prentice-Hall (1972). 2) De Groot, A.D.: Thought and choice in chess, Mouton (1965). 3) De Groot, A.D. and Gobet, F.: Perception and memory in Chess studies in the heuristics of the professional eye, Van Gorcum (1996). 4) Chase, W.G. and Simon, H.A.: Perception in chess, Cognitive Psychology, Vol.4, pp.55 81 (1973). 5) Simon, H.A. and Gilmartin, K.J.: A simulation of memory for chess positions, Cognitive Psychology, Vol.5, pp.29 46 (1973). 6) Reitman, J.S.: Skilled perception in Go: Deducing memory structures from inter-response times, Cognitive Psychology, Vol.8, pp.336 356 (1976). 7) pp.207 219, (1997). 8) Saito, Y. and Yoshikawa, A.: The difference
3008 Oct. 2002
Vol. 43 No. 10 1 3009
3010 Oct. 2002 of knowledge for solving Tsume-Go problem according to the skill, Game Programming Workshop 97, pp.87 95 (1997). 9) MDS (1989). 10) (1994). ( 14 3 1 ) ( 14 9 5 ) 1994 4
Vol. 43 No. 10 1 3011 1986 1993 1994 2000 1998 2001 21 NPO Reijer Grimbergen Reijer Grimbergen is an associate professor at the Department of Information Science at Saga University. His research interests are search, game playing and the cognitive modeling of human problem solving. Grimbergen received his M.S. in computer science from Nijmegen University (Holland) in 1989 and his Ph.D. in cognitive science from the Nijmegen Institute of Cognitive Science and Information Technology (NICI) in 1994.