Pattern or Calculation? Evaluating Problem Solving in Non-Expert Chess Players

Abstract

Chess has a long history as a model domain for studying expert decision-making. In this study it will be used to study how domain-specific knowledge and cognitive skills interact and influence decision-making. De Groot’s influential two-stage model [1] proposes that in chess, decisions are made in two phases. First, the player identifies the key elements or “motifs” in the chess position based on their pattern knowledge. Second, they perform a search of move sequences or “calculation of variations” to verify and realize the idea of the recognized motif. The aim of this study is to examine the importance of these two stages at different chess skill ratings.

Bratko, Hristova, and Guid [2] compared the influence of motif recognition and calculation on solving tactical chess problems concluded that calculation was the dominant feature to discriminate between players in terms of their success. The key limitation of the study was the relatively high Elo chess rating of participants (1845-2279) who were all able to detect the large majority of relevant motifs. As a result, it remains unclear whether motif recognition plays a larger role as a predictor of success in problem solving among players with lower Elo scores. Our hypotheses are as follows: (1) motif recognition is the dominant discriminating problem solving step of lower rated players (<1800), and (2) there exists a clear threshold beyond which calculation overtakes motif recognition as the primary discriminating factor in solving tactical chess problems.

To test these hypotheses, we will conduct an experimental study involving chess players with an Elo rating of 1000-1800. A sample of approximately 30 participants will be recruited through local chess clubs. Each participant will attempt to solve 12 tactical chess problems collected from the Chess Tempo database. After each problem the participants will explain which motif(s) they identified and what sequence they calculated. While examining the problem, participants’ eye movements will be tracked to support the verbal reports and help verify motif recognition.

Based on the experimental design, we expect to find that motif recognition plays a more significant role than calculation in the tactical problem-solving success of lower-rated chess players. Such findings would support the view that at earlier stages of chess skill development, pattern knowledge makes a more important difference between success and failure to solve the problem. Additionally, we expect our results to identify an Elo rating threshold, beyond which calculation overtakes motif recognition as the more decisive factor. These findings would add a developmental layer to De Groot’s two-stage model, showcasing how cognitive strategies shift with increasing expertise.

References

[1] A. D. De Groot, Thought and Choice in Chess. The Hague, The Netherlands: Mouton Publishers, 1978.

[2] I. Bratko, D. Hristova, and M. Guid, “Search versus knowledge in human problem solving: A case study in chess" in Model-Based Reasoning in Science and Technology: Studies in applied philosophy, epistemology and rational ethics, L. Magnani and C. Casadio, Eds. Cham, Switzerland: Springer, 2016, pp. 569–583. doi: 10.1007/978-3-319-38983-7_32.