Mnemonics and Chunking

Short-term memory holds roughly 7 items at a time, give or take two. Chunking does not expand that slot count. It changes what counts as one item. The string 1-4-9-2-1-7-7-6 is 8 separate digits, too many to hold easily. Read as 1492 and 1776, two historical dates, it is 2 chunks. Same information, same number of memory slots used, far less strain, because each chunk is one meaningful unit instead of four meaningless ones.

The method of loci turns a flat list into a story by attaching each item to a familiar location, like rooms along a walk through a house: milk balanced on the front doorknob, eggs cracked into the umbrella stand, bread loaf sitting on the hallway bench. Walking through the house in memory later retrieves each item because the location itself becomes a retrieval cue. A hierarchy does similar work without a physical path: instead of 20 unrelated facts in a random order, group them into categories, then subcategories, so recalling one branch pulls the related items with it. Both tricks replace meaningless, isolated bits with a structure that already has connections built in.

This matters because encoding depends on hooking new information onto what is already organized in long-term memory, the same idea behind semantic networks and deep processing in Unit 2's earlier material. Chunking fails when material carries no familiar meaning, which is why a string of digits in an unfamiliar language or an unfamiliar number system stays hard to chunk: without known categories or patterns to group by, each symbol remains its own chunk, and capacity limits reappear in full force.