The Myth of Learning Styles (And What Actually Works)

Where the Theory Came From

The learning styles hypothesis — that individuals have preferred modes of receiving information (visual, auditory, kinaesthetic, and so on) and learn best when instruction matches that preference — became one of the most widely adopted ideas in education during the 1990s and 2000s. It was built into teacher training programmes, school policies, and corporate learning frameworks. Its intuitive appeal was considerable: people do differ, and those differences presumably affect how they learn.

The most commonly cited model, VARK (Visual, Auditory, Read/write, Kinaesthetic), was developed by Neil Fleming in 1987. Other models proliferated: Kolb's experiential learning styles, Dunn and Dunn's multidimensional model, Gardner's multiple intelligences (often conflated with learning styles despite being a distinct theory). By the 2000s, researchers estimated that the learning styles industry — assessment tools, training materials, and instructional design consultancies — was worth hundreds of millions of dollars annually.

What the Research Actually Shows

The critical problem with learning styles theory is not the observation that people differ in their preferences. It is the lack of evidence for the "meshing hypothesis" — the claim that matching instruction to preferred style improves outcomes. For the theory to be validated scientifically, it is not sufficient to show that people have different preferences. It must be demonstrated that outcomes improve when instruction matches those preferences, and that outcomes suffer when there is a mismatch.

Multiple systematic reviews have failed to find consistent evidence for the meshing hypothesis. A particularly influential 2008 review by Pashler et al., published in Psychological Science in the Public Interest, examined the available evidence and concluded that the hypothesis had not been adequately tested, and that the studies claiming to support it generally failed to meet basic methodological requirements.

What Actually Predicts Learning Outcomes

The research literature points to several factors that consistently predict learning outcomes, none of which involve matching instruction to style preferences. Prior knowledge is among the strongest predictors: learners with more existing knowledge in a domain acquire new knowledge in that domain more efficiently, regardless of how it is presented. Cognitive load matters: instruction that minimises extraneous load — irrelevant complexity that taxes working memory without contributing to learning — consistently outperforms instruction that increases it. Retrieval practice, spacing, and elaborative interrogation all produce robust learning advantages across populations.

The most effective instructional approaches appear to be those that work with how human memory functions generally, rather than how any individual supposedly prefers to receive information. The good news embedded in this finding is that effective learning strategies are not gated behind individual assessment — they are broadly applicable.

Why the Myth Persists

Despite the lack of supporting evidence, belief in learning styles remains widespread among educators and learners. Several factors sustain it. The theory is intuitive and flattering — it tells people that their preferences are meaningful and should be accommodated. It gives teachers a framework for thinking about student differences. And the research literature is not widely read outside academic psychology.

There is also a confirmation bias dynamic at work. When someone who identifies as a visual learner encounters information presented visually and retains it well, that experience confirms the theory. It is harder to notice the many cases where non-preferred formats work equally well, or better.

A More Productive Framework

Rather than matching instruction to style preferences, the evidence suggests focusing on the nature of the content being learned. Some content genuinely benefits from visual representation — maps, anatomical diagrams, mathematical functions with geometric interpretations. Other content is inherently verbal. The question is not what the learner prefers, but what the material requires. This content-appropriate approach is better supported by evidence and is arguably more respectful of learners' capacity to engage with material in whatever form is most informative.