Expertise requires building sophisticated chunk recoding schemes, which can be achieved through effective practice (e.g. Purposeful practice, after Ericsson and Pool).
The process appears somewhat automatic, insofar as experts aren’t usually thinking “I need to develop better chunks.” Rather, to study at the limits of your capacity means making progress by way of encoding larger chunks—at least in some part.
Ericsson and Pool (2016) discuss this claim at length, but little evidence or explanation is provided for the mechanism.
The ability to recognize and remember meaningful patterns arises from the way chess players develop their abilities … by spending countless hours studying games played by the masters. You analyze a position in depth, predicting the next move, and if you get it wrong, you go back and figure out what you missed. Research has shown that the amount of time spent in this sort of analysis—not the amount of time spent playing chess with others—is the single most important predictor of a chess player’s ability. (p. 56)
Much of deliberate practice involves developing ever more efficient mental representations (p. 59)
… the most successful quarterbacks are generally the ones who spend the most time in the film room, watching and analyzing …” (p. 64)
Ericsson, A., & Pool, R. (2016). Peak: Secrets from the New Science of Expertise (1 edition). Eamon Dolan/Houghton Mifflin Harcourt. Peak - Ericsson and Pool