Cognitive Load

Our brain has a small area where it briefly holds and processes things. Experts call it working memory. Astonishingly little fits in there: roughly four to seven individual pieces of information at the same time. Cognitive load is the measure of how heavily this area is currently being used. If the load is too high, mistakes happen, intermediate steps get forgotten, or a task is abandoned altogether. The term comes from learning research in the 1980s and is now used everywhere people work with complicated systems.

Why overloaded interfaces slow people down

The limit of working memory cannot be shifted through effort. Someone who concentrates harder does not gain more storage. That’s why the design of a task determines whether it feels solvable or overwhelming. A form with forty fields on one page creates a high load. The same form split into five steps suddenly feels manageable, even though the amount of data is identical.

For companies, this is a tangible cost issue. If users abandon an order because the page demands too much at once, revenue is lost. In industry, something similar applies to control panels on machines. Investigations into accidents in cockpits and control rooms repeatedly point to the same issue: the person had to keep too many displays in mind at the same time.

A common misconception is that cognitive load is inherently bad. That’s not true. Learning requires a certain amount of effort, otherwise nothing sticks. The problem is only the load that has nothing to do with the actual subject matter. A poorly labeled menu costs mental capacity that is then missing for the subject itself.

The three types of load in working memory

Research distinguishes three components. The first is the difficulty of the subject matter itself. A quadratic equation is simply more demanding than an addition. This component cannot be conjured away, only broken down into smaller portions.

The second component arises from poor presentation. A text that refers to a graphic two pages later forces flipping back and forth. This back-and-forth consumes capacity without aiding understanding. Experts call this extraneous load. This is exactly where good design comes in: labeling directly on the graphic, fewer options per screen, a clear sequence.

The third component is the useful part. It arises when someone combines individual pieces of information into a larger pattern. This is exactly what happens during learning. A chess player doesn’t see thirty individual pieces, but a few familiar configurations. Because this bundling saves space, experts in their field appear so effortless. They don’t have more storage, just better-packed bundles within it.

Chatbots, dashboards, and everyday screen use

In discussions about AI, the term currently comes up especially often. A language model that accepts a request in everyday language relieves the user of having to learn menu structures. This significantly reduces the load. At the same time, a new load emerges: one must check whether the answer is correct. Experts call this verification effort.

Software development also uses the term. Code sections that require keeping five files in mind at once are considered expensive to maintain. Teams therefore deliberately write simple code, even when a clever solution would be shorter.

In everyday life, one constantly encounters this principle. A navigation app announces one instruction at a time instead of reading out the entire route at once. Exam questions are broken down into sub-tasks. And anyone who reads messages while studying quickly notices that doing both at once doesn’t work. Working memory is simply full.