Imagine a simple, hypothetical choice. You need to learn one topic before class. You can open a diagram, play a recording, read a chapter, or work through a practical example. A quiz has called you a visual learner, so the diagram seems like the obvious choice.
You might choose the diagram simply because you prefer studying that way. But once the label leads you to expect that the diagram will improve your understanding or memory—and to dismiss the recording or chapter—it is making a prediction about how you learn.
That prediction is at the center of the debate over the “learning styles myth.” Research has not established learning-style labels as a reliable way to choose which study format will work best, although some experiments have found benefits from matching instruction to a preference. A more useful starting point is to let preference break ties between suitable options, while giving more weight to the task, your prior knowledge, and what you can do afterward.
A useful label has to help you make a better choice
A quiz that asks which formats you like measures a preference. It does not test how well you can interpret a graph or how much you will remember from a recording. Whether that preference predicts better learning is a separate question.
Researchers call this prediction the matching hypothesis: people should learn more when instruction matches their measured preference.
Harold Pashler and colleagues set out a way to test the claim. Suppose one group prefers diagrams and another prefers spoken explanations. Within each group, researchers randomly assign some students to a diagram and others to the same material in spoken form. Everyone then takes the same test.
If diagram-preferring students learn more from the diagram while listening-preferring students learn more from the recording, the quiz may be identifying a useful difference. Now suppose both groups score higher after studying the diagram. In that case, the diagram is the better option for both groups, whatever their quiz results say. Knowing who prefers diagrams and who prefers listening would not help us choose between the two formats.
The VAK learning styles model groups preferences into three categories: visual, auditory, and kinesthetic. VARK adds reading and writing as a separate category. In VARK’s current definitions, “visual” mainly means diagrams, charts, maps, and symbolic relationships, while “kinesthetic” includes practical examples and experience—not movement alone.
VARK’s current website says it does not recommend matching teaching to a learner’s style or treating the categories as fixed labels. The question here is whether a quiz result can reliably guide your choice of study material.
One experiment did find a match
In a study by Janina Lehmann and Tina Seufert, 42 German psychology students were selected after a larger preference screening. They were randomly assigned to read or listen to the same short text about volcanism. Here, “visual” meant written words, not a diagram.
On the test given immediately after the lesson, students who strongly preferred written material scored higher after reading than after listening. The other group had less clear-cut preferences, leaning toward listening overall. Their average score was higher after listening, but the study did not establish a clear listening advantage for this group. The strongest evidence for matching therefore came from the students who preferred reading.
The result was also narrow. The comprehension measure contained only three retained questions, the same pattern did not appear on the recall test, and there was no delayed follow-up. The study shows that a matching-consistent result can appear under a particular set of conditions. It does not tell a new student which format will work better for another topic or a later exam.
Is matching better than using one format for everyone?
A 2024 meta-analysis combined 21 studies with 1,712 participants and found a small average advantage for instruction classified as matched rather than mismatched.
That comparison is informative, but it is not always the choice a student faces. Consider an invented example.
Hypothetical example. The figures below are invented average scores on the same 100-point test for two equal-sized groups. Both formats are assumed to be equally accessible and equally costly.
| Reported preference | Diagram-led instruction | Text-led instruction |
|---|---|---|
| Prefers diagrams | 80 | 60 |
| Prefers text | 70 | 65 |
Matching each group to its preference produces an average of 72.5. Giving each group the opposite format produces 65, so matching looks better.
But giving diagram-led instruction to both groups produces an average of 75. Even the group that preferred text scored higher with diagrams: 70 rather than 65. In this hypothetical example, matching would give the text-preferring group the less effective lesson. To justify using the quiz, we would need evidence that its recommendations improve on choosing a format that works well for both groups.
The 2024 reviewers also checked whether outcomes showed the directional pattern the matching hypothesis requires: different preference groups benefiting from their corresponding formats. That pattern appeared in a minority of the outcomes they examined, and the studies varied greatly in their questionnaires, materials, tasks, and tests. Many also excluded at least some participants who could not be placed clearly into a category. The reviewers concluded that the evidence was too inconsistent to justify widespread matching.
Across studies, the rule remains unreliable
A 2025 analysis of 17 meta-analyses reached a more negative assessment after separating studies that tested matching from studies that merely linked style scores with achievement. It covered a broader and different set of learning-style constructs, so its numerical results should not be treated as a direct rerun of the 2024 analysis. The practical conclusion is similar: occasional associations or task-specific matches have not produced a dependable way to assign instruction.
A larger study tested the idea with 222 students training to become teachers. They studied course material presented either as text or as figures. Their visual and verbal preference scores did not predict which version produced better test performance. Matching also did not affect how well students expected to do or how confident they felt.
The study used actual course material and a preregistered plan. Its preference questionnaire did not produce highly reliable scores, however—a limitation the authors identified as their biggest concern about the data—and the outcomes were measured immediately in one university context. The result deserves attention without being treated as the final word.
The two experiments do different work. The volcanism study shows that a matching-consistent result can appear. The larger teacher study shows that the promised pattern did not emerge under another, more extensive test. Across the wider record, a learning-style quiz has not predicted the better option reliably enough for a new student and task.
Preference can line up with expected success
A preferred format can still feel like the one through which you learn best.
In another study of verbal and visual preferences, participants studied pairs of words and pairs of pictures, then predicted how likely they were to remember the material. Immediately after studying word pairs, people who reported a stronger preference for words tended to expect better recall. A stronger preference for pictures was linked to higher expectations for picture pairs.
The memory test did not show corresponding benefits associated with those preferences. Stronger word or picture preferences were also not linked to more accurate predictions.
This does not explain every reason people identify with a learning style. It establishes a smaller but useful distinction: preference can agree with the result you expect without agreeing with the result you later achieve.
Let the task guide the choice
Questioning learning-style labels does not make format irrelevant. The task may require a particular representation or activity. If your exam asks you to interpret graphs, your practice needs to include interpreting graphs. Pronunciation requires access to sound, and a physical procedure eventually has to be performed.
What you already know can also change the kind of help that is useful. A 2025 meta-analysis of 60 experiments found that more instructional assistance helped learners with less prior knowledge on average, while learners with more prior knowledge generally did better with less assistance. The size of these effects varied across educational levels, subject areas, and ways of measuring prior knowledge. For a student, the useful question is whether the material supplies background or guidance you still need—not whether it matches a sensory label.
Accessibility comes before preference. Captions, transcripts, text alternatives, and compatible formats make information usable. CAST’s Universal Design for Learning guidance recommends alternatives for barriers to perception; it does not present them as evidence for matching instruction to a learning style.
When several options are suitable and accessible, preference and convenience can be sufficient reasons for choosing one. Suppose your exam asks you to interpret a graph. After studying an explanation in whichever format you find useful, put it away and try interpreting a new graph. That gives you a more relevant check than whether the material felt right.
Use preference to choose among suitable options, but do not use a learner label to reject a format the task requires. Judge the method by what you can do afterward: can you explain the idea, recall it later, solve the problem, or perform the procedure?
References
CAST. (2024). CAST Universal Design for Learning Guidelines version 3.0: Support multiple ways to perceive information (Consideration 1.2). https://udlguidelines.cast.org/representation/perception/ways-perceive-information/
Clinton-Lisell, V., & Litzinger, C. (2024). Is it really a neuromyth? A meta-analysis of the learning styles matching hypothesis. Frontiers in Psychology, 15, 1428732. https://doi.org/10.3389/fpsyg.2024.1428732
Hattie, J., & O’Leary, T. (2025). Learning styles, preferences, or strategies? An explanation for the resurgence of styles across many meta-analyses. Educational Psychology Review, 37, Article 31. https://doi.org/10.1007/s10648-025-10002-w
Knoll, A. R., Otani, H., Skeel, R. L., & Van Horn, K. R. (2017). Learning style, judgements of learning, and learning of verbal and visual information. British Journal of Psychology, 108(3), 544–563. https://doi.org/10.1111/bjop.12214
Lehmann, J., & Seufert, T. (2020). The interaction between text modality and the learner’s modality preference influences comprehension and cognitive load. Frontiers in Psychology, 10, 2820. https://doi.org/10.3389/fpsyg.2019.02820
Melzner, L., & Kappes, C. (2025). Testing the meshing hypothesis in prospective teachers: Are there effects of matching learning style and presentation mode on learning performance and on metacognitive aspects of learning? Instructional Science, 53, 365–389. https://doi.org/10.1007/s11251-024-09689-1
Pashler, H., McDaniel, M., Rohrer, D., & Bjork, R. (2008). Learning styles: Concepts and evidence. Psychological Science in the Public Interest, 9(3), 105–119. https://doi.org/10.1111/j.1539-6053.2009.01038.x
Tetzlaff, L., Simonsmeier, B., Peters, T., & Brod, G. (2025). A cornerstone of adaptivity—A meta-analysis of the expertise reversal effect. Learning and Instruction, 98, 102142. https://doi.org/10.1016/j.learninstruc.2025.102142
VARK Learn. (n.d.-a). The VARK modalities: Visual, aural, read/write & kinesthetic. Retrieved September 25, 2026, from https://vark-learn.com/about-vark/the-vark-modalities/
VARK Learn. (n.d.-b). What VARK is (and isn’t): A catalyst for better learning. Retrieved September 25, 2026, from https://vark-learn.com/about-vark/what-vark-is-and-isnt/