Inflexibility of experts—Reality or myth? Quantifying the Einstellung effect in chess masters

Why good thoughts block better ones: The mechanism of the pernicious Einstellung (set) effect

“…Similarly to the studies on transfer, Bilalić et al (2008) used a control group of weaker players (Class A, Class B, and Class C) and showed them the 1-solution problem only. The Einstellung effect was quantified by seeing how much weaker a player had to be, when only the optimal solution was present, to show comparable performance to that of a better player when the distracting effect of the familiar solution was present (for similar measures of transfer see the first chapter of Singley & Anderson, 1989).…”

“…All the experts found the shorter solution in the 1-solution problem, showing that they were capable of finding it when not distracted by the familiar one. To quantify the impact which the familiar solution had on the performance of expert players in the 2-solution problem, Bilalić et al (2008) exploited one of the advantages of chess as a domain for studying problem solving -the presence of an interval scale for measurement of skill (see Footnote 1).…”

Transactions on Computational Collective Intelligence XX

“…With regard to decision-making, chess masters demonstrate patternrecognition frequently: key features of certain board configurations are recognised extremely quickly [10] and typical, good moves are used when better, relatively uncommon moves exist [6,35]. This indicates that when an adequate knowledge base exists for a domain, pattern-recognition is the preferred modus operandi for human decision-making (examples in other domains are given in [6]). However, given that there may exist many possible solutions for a particular situation, how exactly is a decision regarding what to do reached?…”

“…Tension between use of problem-solving and pattern-recognition to solve problems has been identified by many [20,47], resulting in the proposal that domain experts are, in some cases, inflexible problem-solvers, since they are so entrenched in established paradigms [39,41]. This has been proven to be true, but only to a certain degree of expertise; once an above-average level of knowledge has been acquired about a domain, the so-called Einstellung Effect (a hallmark of expert inflexibility where satisfactory solutions block better ones) is removed [6]. A quantitative, scientific analysis of the potential effects on agent performance by weighting the usage of these systems and configuring their constituent components differently in particular complexities of a stochastic environment is lacking, to our knowledge, in the literature.…”

“…Implementation details of these mechanisms that are of interest to our simulations are discussed in the following section. 6 Discrimination, Familiarisation and Production Creation.…”