Chunking: how experts remember more with the same brain

In 1956, the psychologist George Miller published one of the most cited papers in the field, arguing that working memory holds a strikingly small number of items at once - roughly seven, give or take two. Decades of follow-up research have refined that number downward for some tasks, but the core observation still holds: short-term memory capacity is small and fairly fixed.

The trick that gets around the limit

Miller's paper also contains the more useful half of the idea: the limit is on items, not on the information those items contain. A chess master doesn't remember a board position as thirty-two individual piece-and-square facts. They remember it as five or six familiar patterns - clusters they've seen thousands of times before, each one collapsing many pieces of information into a single "item" that working memory can hold.

This is chunking, and it's not a special talent chess masters are born with. It's the direct result of experience: enough repeated exposure to a domain that certain groupings become automatic and get stored as a single unit instead of several.

Why this matters for studying anything

If working memory can only hold a handful of items regardless of expertise, then getting better at a subject isn't primarily about expanding that limit - it's about making each item you hold in it bigger and denser. A beginner reading a sentence of chemical notation is juggling individual symbols. An expert reading the same sentence is juggling two or three recognised structures. Same working memory, wildly different effective capacity.

This is also why unfamiliar material feels overwhelming in a way that has nothing to do with intelligence: you're forced to hold it at the smallest possible chunk size, which fills working memory fast. Fluency in a subject is, to a real extent, the accumulation of bigger chunks.

What builds chunks

Chunking isn't something you can shortcut by trying harder to group things - it comes from repeated, successful retrieval of the same material over time, which is exactly what spaced review produces. Every correct recall of a fact makes it slightly more automatic, and enough automatic facts around a topic start fusing into a single retrievable unit instead of many separate ones. It's a slow byproduct of practice, not a technique you apply directly.