The question has a precise answer, which is unusual in education research, and the reason is that the claim itself is precise enough to test. Matching instruction to a learner's preferred modality should produce measurably better learning than mismatching it. That prediction can be checked directly. When Pashler, McDaniel, Rohrer and Bjork went looking in 2008 for studies that had checked it properly, they found the field had almost never run the experiment.
The Test the Claim Requires
To demonstrate meshing you need a crossover design. Assess learners' styles. Randomly assign them to one of at least two instructional formats, so that some receive their preferred mode and some do not. Teach the same content. Test everyone with the same assessment.
The prediction is an interaction: visual learners should beat auditory learners on visual instruction and lose to them on auditory instruction. Anything less than that design cannot support the claim. A study showing visual learners do well on visual teaching proves nothing without the comparison group, because visual teaching might simply be better for that content.
What the 2008 Review Found
Pashler and colleagues surveyed the literature for studies meeting that standard and reported that the number was vanishingly small relative to the size of the field. Most of the enormous learning-styles literature consisted of instrument development, correlational work and studies without the crossover comparison.
The handful of adequately designed studies they identified produced results contradicting the meshing hypothesis. Their conclusion was not that the idea is definitively false but that a practice adopted at global scale rested on evidence that had, for practical purposes, never been collected.
The Instruments Have Their Own Problems
Beneath the meshing question sits a measurement question. Coffield and colleagues' 2004 review examined the major models and found the field fragmented, with 71 distinct frameworks and many instruments showing weak reliability and validity. Several failed to produce stable results on retest.
That matters because a matching study can only be as good as its classification. If the styles being matched are measured unreliably, a null result is ambiguous — though it hardly rescues the practice, since the same unreliable instruments are what schools and companies are actually using.
Why the Belief Survives the Evidence
The strongest reason is that learning-styles advice smuggles in genuinely good practice. Teachers who adopt it start varying their presentation, adding diagrams to explanations and building in hands-on work. Those changes help, for reasons established independently — combining words with relevant images aids comprehension across the board. The improvement is real; the attribution is wrong.
There is also a personal pull. A style label reframes difficulty as mismatch rather than as difficulty, which is comforting, and comfort is a poor guide to what is working. As Kirschner argued in 2017, the persistence of the idea is itself a case study in how an appealing framework outlives its evidence.
What This Does Not Mean
It does not mean preferences are imaginary, that variety in teaching is pointless, or that everyone learns identically. People differ enormously in prior knowledge, working memory, motivation and interest, and those differences affect learning substantially — they are simply not the same thing as a modality preference.
The practical upshot is to demote the style result from prescription to description. Take the learning style test to know what you find comfortable, then choose study methods on the evidence instead — which is what how to study without learning styles sets out.