Efficient chunk schemas usually encode domain-specific attributes

As people develop more efficient chunks for some domain (“Chunks” in human cognition, Recoding can increase chunk size), they often incorporate abstract, domain-specific attributes into their encodings. For instance, a chess master perceives chessboards in terms of “Pawn chains” and “attacking files” (Chase and Simon, 1973, p. 80)—rather than a bunch of positions of pieces. Their mental representation of the state of a game of chess is abstract, high-variance, high-dimensional, and relatively illegible to a novice. An improvising musician might think in terms of “tension” and “anticipation.”

These abstractions make the representations {less general} and {more difficult to transmit}, in exchange for {capturing more information within a particular domain}.

Ericsson and Pool use this observation to make a (likely too strong) generalization (2016, p. 60):

… such mental representations … are very “domain specific” … This explains a critical fact about expert performance in general: there is no such thing as developing a general skill. You don’t train your memory; you train your memory for strings of digits or collections of words or for people’s faces.”


Chase, W. G., & Simon, H. A. (1973). Perception in chess. Cognitive Psychology, 4(1), 55–81. Chase and Simon - Perception in chess

Ericsson, A., & Pool, R. (2016). Peak: Secrets from the New Science of Expertise (1 edition). Eamon Dolan/Houghton Mifflin Harcourt. Peak - Ericsson and Pool