13 August 2026 · Alex Abramson

What Is Visual Learning, and Why Does It Work?

Most people have heard that they are a "visual learner." Most people have also heard that this is a myth. Both are half right, and the half that survives is the interesting one.

The myth: learning styles

The idea that each student has a fixed style (visual, auditory, kinaesthetic) and learns best when teaching is matched to it is called the meshing hypothesis, and it has been tested repeatedly. It does not hold up. Studies that sort students by their self-reported style and then teach them in matched and mismatched ways generally find no reliable advantage for the match. Believing you are a visual learner does not mean pictures work better on you specifically.

So far, so debunked. But notice what those experiments actually tested: whether visuals help some people more than others. Nobody tested whether visuals help.

What survives: dual coding

They do, and the reason has nothing to do with personality. It has to do with how working memory is built.

The dual coding theory, developed by Allan Paivio in the 1970s, proposes that we process verbal and visual information through two partly separate channels. A word and a picture of the same thing are not stored as one representation but as two, linked. When you later try to recall the idea, you have two routes to it instead of one.

Richard Mayer's work on multimedia learning turned this into something you can design against. A consistent finding across his experiments: people learn better from words and pictures together than from words alone. Not because pictures are more engaging, but because splitting the load across two channels leaves more capacity in each.

Everyone has both channels, and using both beats using one for essentially all learners, which is exactly why the learning-styles version of the claim falls over. The effect is real. The personalisation around it is not.

The load is the whole game

Cognitive load theory adds the constraint. Working memory is small — a handful of items at a time. Anything that spends it on something other than the concept is waste.

So how a visual is presented matters as much as whether there is one:

• The split-attention effect. A diagram on one page and its explanation on another forces you to hold one in memory while reading the other. Integrating the label into the figure removes that cost entirely.
• The redundancy effect. Narration plus identical on-screen text is worse than narration alone. Two channels carrying the same payload is not double the bandwidth; it is one channel plus interference.
• The modality effect. Spoken explanation alongside a visual generally beats written explanation alongside the same visual, because the words arrive through the ear and leave the eye free for the picture.

That last one is specific, testable, and the one most educational material gets wrong. A slide dense with bullet points sitting next to a diagram is competing with itself.

Why animation, specifically

A static diagram shows the end state. Many concepts are not end states at all. They are processes, and the thing you need to understand is the transition.

Consider the difference between a completed geometric proof and watching the construction lines appear in order. The finished figure contains all the same information. But it does not contain the sequence, and the sequence is the argument. The same is true of a sorting algorithm, a chemical reaction, a market finding equilibrium, or a wave interfering with itself.

Animation carries something a static image structurally cannot: what happened, in what order, and what caused what. When the concept is a process, animating it does not make the picture livelier. It makes the picture more complete.

There is a real caveat. Animation only helps when it is paced. A visual that moves faster than the explanation is a visual you cannot follow, and research on animated instruction consistently finds that learner control over pacing matters. Motion for its own sake is decoration, and decoration competes for the same working memory the concept needs.

What this means in practice

If you are trying to understand something difficult, the evidence points somewhere fairly specific. Look for an explanation that shows the process rather than the result. Prefer one where the words are spoken and the picture is watched, rather than both read at once. Prefer one where the labels sit on the thing they label. And be suspicious of anything that moves a lot without saying why.

None of that is about being a visual learner. It is about the fact that your working memory is small and two channels are wider than one.

This is the principle Jentoo AI is built on. Ask a question and it generates an animated explanation with synchronized narration — the figure drawn in the order the argument needs, labels attached to what they describe, and the words spoken while you watch rather than printed next to the picture. It is a fairly direct implementation of what the multimedia-learning research has been saying for thirty years.

Written by Alex Abramson. Published 13 August 2026.