Concept formation is hierarchical
Links from semantic and episodic memory, inputs from senses and feedback form hierarchical mappings. These, in turn, combine to form mappings of frequently used mappings and mappings of frequently used mappings of frequently used mappings.
Chess is a good starting point for understanding mapping hierarchies because it has been researched intensely. We know that beginning players see individual pieces, such as the bishop and its diagonal potential moves. Each piece is part of a collection of possible chess moves. More experienced players collapse these individual possibilities into sequences of opening moves together in multi-move patterns or mappings. For example, when experienced players see the board six moves in, they know which opening sequence each player used, even though they were not there during the beginning of the game.
How does this happen? Our brains not only perceive and interpret patterns, but store their relationships. The more frequently two events occur sequentially, the stronger the neural pathway that associates those two events in the mind. This is the well-known stimulus-response, ding-a-ling-and-the-dog-salivates Pavlovian/Hebbian phenomenon that occurs predictably after you ring a bell several times just before you feed an animal. The brain, in essence, records sequences of actions and the feedback it receives by strengthening and lengthening neurons to build associations.
The power of association, routine and practice magnifies as it continues, through a process called “chunking.”[i] Chunking involves linking a set of features with a concept, like “car”, and then subsuming cars with trucks and fire engines under a higher concept “vehicles.” Chunking information into modules increases the speed and power of our brains. Modules of thousands of elements can be called with a single link. Links can encompass modules of modules. Modules can overlap and intertwine in vast networks.
Master chess players memorize around 50,000 chunks. Then they associate these chunks into even higher order chunks, or “templates”, with “slots” for optional move and countermove sequences that they can try out in their minds.[ii] This means that the necessary 50,000 chunks for mastery can be manipulated far more rapidly because they have been subsumed into, perhaps, 10,000 templates. In fact, the “10,000 hours of practice”, popularized in Malcolm Gladwell’s Outliers as the base requirement for mastery of any complex skill, includes much chunking.