The Learning Styles Myth: What the Research Actually Shows

Ask a room of thirty adults how they learn best and you will get thirty confident answers. I am a visual learner. I have to hear it. I need to do it with my hands. The confidence is striking, and it is worth taking seriously, because people are usually reporting something true about their experience. What is not true is the mechanism they have been handed to explain it.

I want to be careful here, because this idea reached most people through teachers who were doing their best with the evidence they were given, and through training courses that presented it as settled. Believing it was reasonable. Continuing to design instruction around it is what the research argues against.

What the claim actually is

The popular version, usually labelled VAK or VARK, sorts learners into visual, auditory, reading and kinesthetic types. The important part is not the sorting. It is the second step, known as the matching hypothesis: that a learner assigned to a category will learn more when material is delivered in their category and less when it is not.

That is a strong, testable prediction, and it is the part that matters practically. If matching does nothing, then the labels are personality decoration, and the effort spent producing four versions of one lesson is effort taken from things that work.

How you would test it, and what happened

To support the matching hypothesis, a study has to do something specific. It must classify learners by style, then randomly assign learners of each style to different instructional formats, then test everyone with the same assessment. A crossover interaction is what you are looking for: visual learners doing better with visual instruction while auditory learners do better with auditory instruction.

In 2008, Harold Pashler, Mark McDaniel, Doug Rohrer and Robert Bjork surveyed the literature for studies using exactly that design. They found remarkably few. Of those with an adequate design, the results contradicted the hypothesis. Learners tended to do better or worse according to the material and the format, not according to their assigned style. Later reviews have kept finding the same shape of result, including studies where students were taught their style and told to study accordingly, then showed no benefit.

The honest summary is not that styles were disproven in some final sense. It is that a claim with enormous influence over classroom practice has never been able to clear a straightforward test, despite decades of opportunity.

Why the myth is so durable

Preference is confused with performance

You genuinely prefer diagrams. That preference is real and it is measurable. But preference predicts what you enjoy, not what you retain, and the two often diverge. Fluent, comfortable study feels productive precisely because it is easy, which is the trap behind the illusion of knowing.

It flatters everyone involved

The framing is generous. Nobody is bad at learning, they were simply taught in the wrong channel. That is a kind story, and kind stories survive.

It is unfalsifiable in everyday use

If a visual learner does well with a lecture, the label absorbs it. If they do badly, the label explains it. Nothing about ordinary classroom experience can dislodge the belief, which is why the correction has to come from controlled studies.

What actually improves learning for everyone

The useful move is not to replace one label with another. It is to spend the effort on the interventions that have survived testing across ages, subjects and formats.

• Retrieval practice. Pulling information out beats putting it in again, and the effect is large and consistent.
• Spacing. The same total study time distributed across days produces better long-term retention than one block.
• Dual coding. Words and relevant images together outperform either alone, for everybody, not for a visual subgroup.
• Elaboration. Asking why something is true, and connecting it to what you already know, builds the links retrieval depends on.
• Interleaving. Mixing related problem types trains the harder skill of choosing an approach, not just executing one.

The first of those is the one I would defend most fiercely, and it is set out in full in our guide to retrieval practice. The third is worth separating from the styles idea carefully, because it sounds superficially similar and is not: dual coding says combining channels helps all learners, whereas learning styles says restricting to one channel helps some learners. The evidence supports combining.

Match the format to the content, not the learner

Here is the principle I would put on a staffroom wall. Some content has an inherently visual structure: anatomy, map projections, the composition of a painting. Some content is inherently auditory: pronunciation, musical intervals. Some is procedural and demands the hands: pipetting, knife skills, chords.

In those cases the format is not a preference, it is the subject matter. Teaching pronunciation through a diagram fails for every learner regardless of label, and no amount of style-matching rescues it. That single reframing does most of the work people hoped learning styles would do, without the classification exercise.

It also explains why our own app leans on narrative plus imagery rather than on style questionnaires. I should disclose that MindSnap is our app, so read this as an example rather than a recommendation: the story carries the causal structure while the artwork carries the spatial detail, which is dual coding applied to content that genuinely has both. It is the same reasoning behind treating drills and knowledge as separate problems, which we looked at in brain training versus real learning.

If you have organised your studying around a style label for years, nothing here means you learned badly. It means the label was doing less than you thought, and the time you spent converting material into your preferred format was probably paying off through elaboration rather than through matching. Keep the effort. Point it at retrieval instead.

Ask what the material demands, not what channel you prefer.