A similar example I heard was that a chess grandmaster may be able to take a look at a chessboard with a game in play and memorize the entire board immediately. But only if the board "makes sense" - all the pieces are in positions that could actually be reached in a real game.
If you take those same pieces and rearrange them willy-nilly, then this ability to instantly memorize its layout goes away.
I recall that Jeff Hawkins, when talking about his Hierarchical Temporal Memory ML model (which is supposed to be brain-like), said something like "Nature has spatial and temporal locality. Brains evolved to best store information that also has spatial and temporal locality—in other words, to recapitulate and model the natural world. To the degree that some pattern is akin to one that arises in nature, the brain can store and compute upon it easily. To the degree that a pattern is 'arbitrary'—something that cannot arise in nature—the brain finds it hard to hold into."
The moment-in-time arrangement of chess pieces on a board does not exactly have spatial or temporal locality; but if one has learned a set of mental transformation rules that let that board be translated into a narrative for how it got to be that way—then that narrative is itself something quite natural for the brain's architecture to represent.
If you take those same pieces and rearrange them willy-nilly, then this ability to instantly memorize its layout goes away.