A study commissioned by Psychological Science in the Public Interest called “Learning Styles: Concepts and Evidence,” by Harold Pashler, Mark McDaniel, Doug Rohrer, and Robert Bjork, finds convincingly that, at this point, there is no evidence that teaching to different learning styles—specifically meaning to a student’s apparent preferred modality such as visual or auditory—works. The authors therefore conclude that using scarce school funds toward doing just this doesn’t make sense.

People may be quick to conclude that this invalidates one of Disrupting Class’s central premises—that we all learn differently. That is a faulty conclusion.

Although we did talk about learning styles in the book—which in retrospect appears to be a mistake—we made it clear that we were not experts in cognitive or neuroscience and that although “there is considerable certainty that people in fact learn differently, considerable uncertainty persists about what those differences are. At the moment the only sure thing is no one has yet defined these differences so unambiguously that there is consensus.” (p. 24) The point is that people learn differently—not that we profess to understand what those differences are.

This recent study by Pashler et al says that, in the words of an eSchoolNews article, “those studies that did use an appropriate [research] method found results that flatly contradict the learning-style theory.” It does not, however, contradict that people learn differently.

This isn’t all that stunning. As I’ve been specifically saying in my talks for well over a year now, categorizing these differences in how individuals learn by learning style is among the most discredited in the research. Noted cognitive scientist Daniel Willingham, for example, refuted this notion as well in an article in the American Educator back in 2005 titled, “Do Visual, Auditory, and Kinesthetic Learners Need Visual, Auditory, and Kinesthetic Instruction?

Of course, there appears to still be some disagreement. According to a March 25, 2009 article in The Journal of Neuroscience titled “The Neural Correlates of Visual and Verbal Cognitive Styles” by David J.M. Kraemer, Lauren M. Rosenberg, and Sharon L. Thompson-Schill, there is some evidence that teaching by learning style could make a difference. In this study, the authors used fMRI scans and saw that visual and verbal cognitive styles did map back to specific areas of the brain. They could also see differences in how individuals process information regardless of the content type. (A quote from the summary reads: “Results demonstrated a pattern of activity in modality-specific cortex that distinguished visual from verbal cognitive styles. During the word-based condition, activity in a functionally defined brain region that responded to viewing pictorial stimuli (fusiform gyrus) correlated with self-reported visualizer ratings on the VVQ. In contrast, activity in a phonologically related brain region (supramarginal gyrus) correlated with the verbalizer dimension of the VVQ during the picture-based condition.”)

Moving outside of this particular debate, this doesn’t change the fundamental point that people learn differently. People don’t disagree with this. There is clear evidence that that people learn at different paces. Some people understand a concept quickly. Others struggle with it for some time before they understand it. We know that explaining a concept one way works well for some people, and explaining it another way works for others whereas it baffles the first group. We also know that this can differ from person to person depending on subject area. One of the key reasons online learning seems to be better on average than face-to-face learning is because time can become variable in an online learning environment so that students can repeat units and lectures until they master a concept and only then move on to the next concept.

As Ken Koedinger, a cognitive science researcher at Carnegie Mellon University, says in the eSchoolNews article: “There is lots of evidence that individualizing instruction based on students’ prior knowledge of a topic leads to more effective and efficient learning. Students entering an algebra class vary much more on their prior knowledge—for instance, their skills with fractions and negative numbers—than they do in their learning styles. And the instructional strategy is clear: Move on for concepts and skills that a student knows well. Slow down or double back for ones they do not.” In the same article he also says that individualizing for student interests may make sense, although we need more research: “For instance, students read more, and thus may learn more, when they are given reading assignments selected to match their interest areas.”

Even some of our supposed critics don’t actually disagree, as at least one has worried that in an online learning world people who would better learn from lectures or in other settings might not have that option.

We still need to do a lot of research in this area. And my hunch tells me that some of our best learning on this topic will come from actual practice in the field. Companies like Knewton and Guaranteach that are building engines with robust data systems so that they can better individualize learning opportunities for students based on a great variety of possible categories not based on any preselected theory (think how Netflix learns over time what movies you are likely to like and Amazon better figures out what books you’ll enjoy) will likely find patterns that lead to breakthroughs in what we know about how different individuals best learn better than we could do in a laboratory. There will be limitations to this approach as well of course—and lab work will be important to complement it. It’s possible that with the assessments we have today, engines like this will miss large chunks of what is actually valuable in learning and lead us astray. But with the ability to do rapid randomized control trials with thousands of people over and over again either consciously or unconsciously, the possibilities to understand what these differences actually are become more real and ever more exciting.

In the meantime, ruling out faulty categorization schemes is vital so that we can improve our understanding. But that doesn’t mean that an overarching premise that doesn’t pretend to be so specific as to have the answer is itself a faulty premise.


  • Michael B. Horn
    Michael B. Horn

    Michael B. Horn is Co-Founder, Distinguished Fellow, and Chairman at the Christensen Institute.