Learning styles: fact and fiction

When I first read about learning styles my wife, who received her B.Ed. in the 1980s, responded “Ah, visual, auditory and kinaesthetic learning.” Certainly, VAK (and VARK, VAKT, etc.) has been a staple of educational workshops and conferences at the school and post-secondary levels for decades. It is not alone: other entries under the “learning styles” umbrella include Dunn & Dunn’s model (stimuli & elements), Kolb’s Learning Style Inventory (convergent/divergent thinking) and, recently, Multiple Intelligences (MI). It was, therefore, surprising to attend a workshop that questioned this educational orthodoxy. Further reading led to research that not only questioned the literature on learning styles — it positively denigrated them.1-3 Indeed, numerous cognitive psychologists and neuroscientists have stated quite emphatically that learning styles don’t exist.4,5

More of a shock, however, was a 2011 blog post that stated in part6:

“I first learned back in the 1980s [that], while self-reported learning style preferences do exist, designing instruction to accommodate them has no basis
-Guy W. Wallace, eLearn Magazine, November 2011, emphasis added.

Clearly, a strong disconnect exists between promoters of the importance of accommodating learning styles in education, and those who maintain that teachers waste considerable time and effort pursuing such things. Further, this has been so for at least 30 years! A quick glance at any conference or professional development program suggests that this is unlikely to end soon. This article is therefore an attempt to summarize what those of us “in the trenches” need to know about it all. I will focus on VAK since it is arguably the most familiar; similar comments apply to other learning style theories.

Inventing inventories

Learning style theories start with the observation that students appear to learn differently. Typically, what this really means is that individuals appear more engaged, and score better on tests, when they engage with material in a specific way: some prefer to go away and read, others prefer a visual presentation, while others seem to “get it” only through hands-on physical activity. Attempts to quantify such preferences involve survey instruments with a series of multiple-choice, multiple-answer, or Likert-scale questions; scores are calculated from the answers, which are taken as measures of preference on various scales. There is a free on-line example on the VARK website for you to try — go ahead and take it; I’ll wait for you!

Done? Did the questions describe you? How hard or easy was it to pick answers? Did you get the same (or similar results) to any surveys you’ve completed before? Were the questions unambiguous, or did you find yourself thinking, “Well, it depends”?

If you had mixed feelings about the inventory, you are not alone. This highlights a problem with psychometric instruments, the age-old issues of validity and reliability: does it measure what it claims to, or is it really measuring something else? Does it measure correctly, both within a population and for different individuals on different occasions? Are the scales truly independent? There are ways to address such questions, and some instruments perform better than others; many, however, lack basic psychometric validity. This is not to say that we don’t have preferences for the way we learn material — we all do. The issue is whether these preferences are fixed, mutually exclusive, stable over time, and properly quantifiable. The key point is that it is a preference, and not an innate inability to learn any other way.

Intuitively obvious — and wrong?

Individuals, then, self-identify as having a preferred way of learning, while psychometric instruments exist to quantify such preferences (with varying success). Well-constructed instruments lend an air of legitimacy to the underlying theory because we recognise ourselves in the results — they resonate with us. What then? It seems intuitively obvious that matching the presentation of content and nature of activities with students’ individual preferences should result in better learning outcomes — technically, the meshing hypothesis. Certainly, many articles and workshops have been devoted to this idea, along with a great deal of commercial activity, but does it actually work? Again, many claim that it does, but proving this turns out to be much harder than you might think.

Diagram for testing different learning style theories.

Fig. 1: Experimental designs for testing learning style theories. (a) Preferences matched to methods and (b) mixed preferences for each method. Signs indicate the expected relative learning gains if the meshing hypothesis is correct. Design (a) does not assess whether non-auditory students nonetheless benefit from auditory instruction materials.

Given the importance of this issue (not to mention the financial implications), the journal Psychological Science in the Public Interest commissioned a review of the literature by a team of cognitive psychologists.7 One of the key questions asked was, what would constitute a valid and reliable assessment of the meshing hypothesis in a classroom context? Since it is well known that a key determinant of what a student learns is what they already know, one criterion is that the assessment be based on learning gain using a pre/post-test structure. The second criterion is more subtle: many of the studies reviewed had split a class into groups based on their identified “learning style”, which were then taught using materials developed specifically with that preference in mind (Fig. 1a). This doesn’t address the core concept, however, which is that students should demonstrate clear gains when the method of teaching is matched to their learning preference and not some other modality.

To really answer the question therefore requires a comparison between students of all learning preferences taught using different modalities. If the meshing hypothesis were correct, the average learning gain for each type of learner should be different from other types when compared across all modalities (Fig. 1b). In reviewing the literature, Pashler et al. comment:“Remarkably, despite the vast size of the literature … we found only one study that could be described as even potentially meeting the criteria [and] even that study provided less than compelling evidence”5 (emphasis added).

In accounting for the perception that the meshing hypothesis is correct, Pashler et al. draw attention not only to flaws in experimental design, but one simple confounding factor: taking time to develop materials specifically for visual, auditory and kinaesthetic learners might well generate insights and ideas that would lead to better teaching and student outcomes anyway! It is unclear how well any of the studies reviewed took this into account.

To VAK or not to VAK?

If evidence for the meshing hypothesis is inconclusive at best, and we accept that while learning preferences exist, learning styles don’t, how do we account for students who cheerfully announce that they are visual learners, and work best with images, diagrams and animations? Do cognitive psychology and neuroscience have anything to say to the student who claims to learn more from reading than attending classes? And, should we as instructors invest time and energy developing materials and activities that do not fit with our own preferred way of learning?

Ironically, one of the reasons why VAK remains so embedded in educational culture is likely what cognitive psychology and neuroscience do have to tell us about the way our brain works. You are probably familiar with the concept of working memory (Fig. 2), which describes how we process sensory inputs (sight, sound, smell, and so on) in short term memory, deciphering their information content through retrieval, matching and storage involving long term memory.8 Just as some sensory information is visual or aural, so we have visual and aural channels in working memory, and store information visually and aurally in long term memory. (Likewise, athletes and musicians commit physical actions to muscle memory through repeated practice.) Daniel Willingham points out that this stored information is not as important as the meaning associated with it.5

Diagram outlining cognitive processes and working memory in brain.

Fig. 2: Cognitive processes and working memory (some features omitted for clarity). Note that detailed, highly connected schemata allow more efficient use of working memory by ‘chunking’ information; new, unconnected information is not as easily processed.8,9

Take the word ‘atom’, for example: it is a specific collection of symbols (letters) with a corresponding sound that we have associated with a particular abstract concept. If a grade six student were to describe an atom, you would likely get a fairly rudimentary definition (which may or may not be accurate); a university professor could provide greater detail, drawing on multiple related concepts. As we learn, we go from recognising and sounding out individual letters, to recognising a whole word by its shape or sound, to associating that word with a specific definition, to employing it as a place-holder for increasingly complex and highly inter-related constructs. In cognitive terms, we develop progressively richer schemata; neurologically, ever larger numbers of neurons are interconnected through a complex web of synaptic links.9

Don’t strain the brain!

Humans demonstrate a range of capacities when it comes to working, visual and aural memory (within limits); some are better at visualisation and have better image recall than others, for example. Such students prefer visual presentations because, at a cognitive level, they don’t have to work quite so hard to process, store and remember the information. That does not make them visual learners however, since they still need to construct meaning from this information through the process of elaboration and rehearsal.5 In short, the process may be more enjoyable, but learning still takes work!

Understanding how the brain works has important implications for instructors. First is the recognition that you and your students don’t all think the same way, and this is perfectly fine; no one is prevented from learning because they “are a visual person” and you are not. Acknowledging preferences, however, and incorporating varied instructional modes and activities into your classes for all can lead to broader student engagement. Further, simply thinking about how to do this, or reviewing resources and activities you can use, provides an opportunity for you to refine your own conceptual understanding.

Secondly, the nature of working memory needs to be taken into account, particularly when designing mixed-media/multimedia presentations.5 You’ve probably been challenged at some point to pat your head while rubbing your tummy; now imagine doing this while reciting your multiplication tables… Working memory is finite in nature and easily overloaded by new material, especially if information is presented from multiple sources simultaneously — or if it has to compete with social media for your attention! Simple design guidelines can, however, make your presentations more efficient and effective.10

Done and done!

In summary, while learning preferences exist, there is no evidence that actively matching teaching approach to individual student preference (the “learning styles” dogma) has any scientific validity. There are, however, benefits to employing different teaching approaches for the whole class, recognising that these preferences are real and individual. Practically speaking:

  • Don’t waste time with “learning style” questionnaires
  • Do discuss the role and limits of learning preferences
  • Don’t overload your students’ working memory in presentations
  • Do explore different ways of communicating hard concepts

Finally, encourage your students to use things they’re good at to scaffold their weaknesses — using strong visual elements to represent concepts, for example, or rapping a description of a dynamic process.

References (all online references accessed September 2014)

  1. John G. Sharp, Rob Bowker and Jenny Byrne, “VAK or VAK-uous? Towards the trivialisation of learning and the death of scholarship.” Research Papers in Education200823(3), pages 293-314.
  2. Frank Coffield, David Moseley, Elaine Hall and Kathryn Ecclestone, Learning styles and pedagogy in post-16 learning: A systematic and critical review. Learning and Skills Research Centre, London UK, 2004.
  3. Frank Coffield, David Moseley, Elaine Hall and Kathryn Ecclestone, Should we be using learning styles? What research has to say about practice. Learning and Skills Research Centre, London UK, 2004.
  4. Paul Howard-Jones, Introducing neuroeducational research: Neuroscience, education and the brain from contexts to practice. Routledge, Abingdon UK/New York NY, 2010. See also http://www.oecd.org/edu/ceri/neuromyths.htm
  5. Daniel T. Willingham, Why don’t students like school? Jossey-Bass, San Francisco CA, 2009. See also “Learning styles don't exist”, https://www.youtube.com/watch?v=sIv9rz2NTUk
  6. Guy W. Wallace, “Why is the research on learning styles still being dismissed by some learning leaders and practitioners?” eLearn Magazine, November 2011.
  7. Harold Pashler, Mark McDaniel, Doug Rohrer and Robert Bjork, “Learning styles: Concepts and evidence.” Psychological Science in the Public Interest2008, 9(3), pages 105-119.
  8. Alex H. Johnstone, “Chemistry Teaching — Science or Alchemy?” Journal of Chemical Education199774(3), 262-268 (Citing A. Baddeley, Working Memory, University Press, Oxford UK, 1986.)
  9. K. Patricia Cross, “Learning is about making connections”, The Cross Papers, number 3, 1999. Education Resources Information Center (ERIC) document number ED432314. https://eric.ed.gov/?q=ED432314
  10. Donald Clark, “10 Brilliant design rules for e-learning”. http://donaldclarkplanb.blogspot.ca/2013/01/mayer-clark-10-brilliant-design-rules.html