chunking

Chunking is a cognitive strategy in which individual elements of information are grouped into larger units, or chunks, to improve the capacity of working memory. By organizing data into meaningful wholes, people can hold more information in short-term memory than by treating each item separately. The concept is associated with George A. Miller, who in 1956 argued that short-term memory can retain about seven plus or minus two items, a limit that can be expanded through chunking into fewer, but more complex, units. The size and content of a chunk depend on prior knowledge and experience; experts can form larger chunks because they can recognize patterns and structures within the material.

Chunk formation relies on schemas and long-term memory to supply the structure that binds items together. Chunking is not simply repetition; it is the alignment of items with existing representations in memory, allowing quicker encoding, retrieval, and transfer to long-term storage. Examples include remembering a telephone number as three or four chunks, or recognizing a chess position as familiar patterns rather than as individual squares.

In education and skill learning, deliberate practice can foster chunking by teaching underlying rules, hierarchies, or templates. In computer science, chunking refers to dividing data into successive blocks for processing or transmission, as in network packets or file formats; natural language processing uses chunking to identify phrases (shallow parsing).

See also: working memory, short-term memory, long-term memory, mnemonic techniques, pattern recognition.