Adaptive expert decision making: Skilled chess players search more and deeper

Abstract

Previous research has suggested that depth of search in chess does not increase much as a function of skill. We submitted players to a problem-solving task with complex positions. We found a strong skill effect in depth of search, rate of search, and number of nodes generated. At the level of strong masters, the absolute values of these variables were much higher than in previous studies (sometimes 10 times higher). Supplementary data on memory, practice, reaction times, and time-constrained decision making (a maximum decision time of 10 seconds) indicated that players’ behaviour was consistent with the behaviour of players previously studied in the literature and with predictions of theories based on pattern recognition. Beyond adding support to the hypothesis that both the ability to search and pattern recognition are relevant aspects of expert thinking, these results are important in showing that previous research has vastly underestimated experts’ search potential. We conclude that long-term memory knowledge allows both extensive search and rapid evaluation when making decisions under time pressure. Players adaptively use either problem-solving method depending on the demands of the task.