Chunks and Templates in Semantic Long - term Memory : The Importance of Specialization

The papers that Bill Chase wrote on skill in chess (with Herb Simon) and in the digitspan task (with Anders Ericsson) count among the most influential publications in the literature on expertise. Yet, the data discussed in these papers are based on the results of very few subjects, whose performance is essentially analyzed as case studies. In this chapter, I argue that case studies and individual data analysis can provide powerful information for developing and testing theories. I describe a 2-year study that, in the spirit of Chase and Ericsson (1981), followed the improvement of a chess master trained to memorize as many chess positions as possible, each presented for 8 seconds. The data are analyzed with a focus on the question of specialization of expert knowledge, and are compared to computer simulations carried out with the CHREST cognitive architecture. The question of specialization is further investigated with recent data that focus on chess players specializing in different openings. In line with Chase and Simon’s chunking theory, CHREST, and other theories based on the concept of chunking, the results show a clear effect of specialization. Implications are drawn not only for the study of expert behaviour but also for the very applied question as to whether the quality of scientific papers can be evaluated reliably.