Learn better

Learning styles: preferences are real, matching instruction is not proven

Learners often have genuine preferences, but the learning-styles matching hypothesis makes a stronger claim: that teaching someone in their preferred mode produces better learning than another mode. Reviews have not found the specific experimental pattern needed to support that claim. Choose representations for the subject and combine them purposefully instead.

Key takeaways

  • Preference is not proof of better learning.
  • Multimodal teaching is different from matching fixed labels.
  • The content should determine the representation.
  • Retrieval and feedback offer better practical starting points.

The matching hypothesis, precisely stated

A fair test first groups people by a measured style, then randomly assigns them to different teaching modes, assesses everyone with the same outcome, and looks for a crossover: each group should learn best from its supposedly matched mode. Pashler and colleagues found virtually no credible evidence of this pattern.

This does not mean everyone is identical. People differ in knowledge, interests, aptitudes, sensory access and preferences. It means a visual, auditory or kinaesthetic label is not a sound prescription for all teaching.

Multimodal teaching is not style matching

Learning in practice

Preference, representation, performance

  1. Preference

    “I like diagrams.”

    Describes comfort or choice.

  2. Teaching representation

    A map for spatial relationships; spoken modelling for pronunciation.

    Describes how the content is explained.

  3. Demonstrated learning

    Retrieve, explain or perform without the support.

    Provides evidence about the learning goal.

A preferred format does not establish better learning. Choose the representation for the content, then check the outcome.

A map can make spatial relations visible, spoken modelling can reveal pronunciation, and physical practice is essential for a motor skill. Using several modes because each carries relevant information is sensible. Repeating the same decorative message in every mode, or assigning different lessons to labelled learners, is not the same thing.

Replace a common claim with a better choice
Common claimWhat evidence supportsWhat it does not supportBetter practical choice
I am a visual learnerYou may prefer diagramsThat diagrams improve all your learningUse diagrams for spatial relationships and test recall
They need to hear itExplanation and discussion can help argumentAn auditory identityPair explanation with examples and feedback
Children learn by doingPractice matters for performanceDiscovery without guidance always worksModel, guide, then release practice
Teach every lesson three waysComplementary representations can helpRedundant media always improves learningChoose the clearest representation for the idea

A constructive replacement for parents and teachers

Ask what the learner needs to know or do. Use worked examples for novices, visuals where relationships are spatial, demonstrations for procedures, discussion for argument, and hands-on practice where performance requires it. Then ask the learner to retrieve, explain or perform without the support.

There is no need to ridicule a learner who identifies with a style. A preference can open a useful conversation. The important move is to prevent the preference from becoming a ceiling: “Let us start there, then practise the form the task requires.”

A better decision for one actual task

Suppose a learner needs to explain why seasons occur. Start with a labelled diagram of Earth’s tilt and orbit, accompanied by a concise explanation. Then ask the learner to explain the mechanism without copying the diagram. If the learner confuses distance from the Sun with tilt, address that misconception rather than prescribing more visual material because they prefer pictures.

My recommendation is to choose the representation for the concept and judge it by an explanation or application. A spoken explanation, accessible description or tactile model may be necessary for access; providing that support does not depend on accepting a learning-style theory.

Pashler and colleagues specify that style matching needs an interaction: different labelled groups must do best under different teaching methods on a common outcome. A learner enjoying one lesson more, or one class improving after instruction, does not demonstrate that pattern. Do not calculate a learning-style benefit from either observation.

What remains uncertain

Not every possible aptitude-treatment interaction has been tested. Evidence can also vary by age, subject and setting. The practical conclusion is narrower than “differences do not matter”: fixed modal matching is not justified as a general education policy, while responsive, accessible and subject-appropriate teaching remains essential.

Sources and further reading

  1. Pashler, H. et al. (2008), Learning Styles: Concepts and Evidence (accessed 23 September 2026)
  2. Education Endowment Foundation (2021), Cognitive Science Approaches in the Classroom (accessed 23 September 2026)

Access dates are recorded beside each source. Please report a source or factual concern.