Chunking

Chunking is the strategy of recoding several individual items into a single, meaningful unit so that working memory's limited slots can hold more effective information. Miller's (1956) landmark paper on the capacity of short-term memory argued that people can reliably hold about seven, plus or minus two, independent chunks, but that the size of each chunk is flexible: a chunk can be a single digit or an entire familiar phrase, depending on how much prior structure the person has learned to compress into one unit. This is why an expert can hold far more raw information in mind than a novice within the same fixed number of chunk slots, they are not born with more capacity, they have learned to pack more into each chunk.

Reference: Miller, 1956. The NeuroRank implementation holds the canonical form and scales interference via task-irrelevant stimulus density.

1. Recalling a full item build order as one learned sequence instead of six separate purchase decisions.
2. Reading a multi-enemy team-fight formation as a single recognized composition rather than five independent positions.
3. Reproducing a practiced grid or callout pattern as one remembered shape instead of tracking each cell separately.

The Memory module's 3x3 grid recall task is a direct chunking-relevant measure: players who recode the flashed sequence into a recognizable shape or path, rather than memorizing each cell independently, tend to reproduce longer sequences correctly, and sequence length combined with correct-count feeds the Working Memory dimension score.