Split attention

When your eyes have to do the explaining’s job

A diagram with its labels in a separate legend forces a particular kind of mental work. The reader must hold the diagram in mind, scan to the legend, retrieve the relevant label, and then return — not once, but repeatedly. Each round trip consumes capacity in working memory that could otherwise go toward understanding the content itself.

John Sweller and his colleagues identified this as the split-attention effect in the early 1990s, working within the framework of cognitive load theory. The argument is precise: when two sources of information are physically or temporally separated but meaningfully interdependent, the learner must mentally integrate them. That integration is not free. It draws on the same limited working memory that is supposed to be processing the new material.

Effect sizes are real, though they vary with how far apart the sources are and how complex the material.

The practical prediction is equally precise: integrating the legend into the diagram should improve comprehension. The research broadly supports this. Studies comparing integrated and separated formats have found measurable differences in both performance on subsequent tests and ratings of mental effort, the latter typically measured with simple self-report scales.

There is a boundary condition worth noting. The split-attention effect applies when the two sources are mutually referential — each incomplete without the other. A diagram that is fully self-explanatory, and a caption that adds genuinely new information rather than translating the diagram, do not produce the same cost. Redundancy introduces its own problems; split attention is specifically the cost of forced integration across a spatial or temporal gap.

The effect also interacts with expertise. A reader who already knows the material can fill in the gap from long-term memory without taxing working memory, which is part of why the same format that strains a novice barely inconveniences an expert. This is the pattern described elsewhere under the expertise-reversal label.

What the split-attention research ultimately demonstrates is that layout is not cosmetic. Placing a legend outside the figure is a design decision with a cognitive consequence: it moves integration work from the page into the reader’s head, where it competes with understanding. The information is all present; the cost is in the distance.