Clark Quinn's Learnings about Learning
(The Official Quinnovation blog)

31 December 2008

Rethinking Learning Styles

Clark @ 9:52 AM

I’ve pointed out the problems with learning styles in the past, but I want to rethink them with you, as we took quite a positive out of them in a unique way.  This was back in 99-2000, when I led a project developing an intelligently adaptive learning system (Intellectricity ™; inspired by Joe Miller‘s vision of a system that respected who you were as a learner).  The system took a unique approach, adapting on the basis of who you were as a learner instead of your demonstrated domain knowledge (though it did that, too, though not like an intelligent tutoring system).

To do this, I looked long and hard at learning styles, including Jonassen & Grabowksi’s uncritical compendium, and (the other) John Carroll’s Cognitive Factor Analysis research.  I decided then what I still do now, that essentially all of the learning style instruments are garbage.  It’s not just me saying this, but so does a commissioned study by the Learning and Skills Research Centre.  And, as I previously reported, psychologist Daniel Willingham says we shouldn’t adapt learning to styles. So, is there anything to salvage?

I want to say yes.  The obvious reason is to recognize that learners do differ, and help learning designers be mindful of that.  And there are some insights.  For example, take Howard Gardner’s Multiple Intelligences.  When I was investigating learning styles, I didn’t like having bimodal dimensions, say where you’re either an introvert or an extrovert. The problem is, those are so context-dependent that you could be an introvert in the classroom, and an extrovert in the lunchroom.  Or even topic by topic.  I liked that Gardner assesses how social (interpersonal) you are separate from how independent (intrapersonal) you are. That became one of our principles when I challenged our psychometrician and our senior cognitive scientist to argue contrary, and they agreed with me.

So we took a different approach.  Starting from a premise of how learners differ in regards to learning, we made it more a competency than a characteristic: “how good are you at learning socially” (and I’d now add Marcia Conner‘s distinction of small group versus large group) separate from “how good are you at learning on your own”.

We ended up developing 31 different characteristics to evaluate, and chose the 9 we expected to have the most leverage into the first version of the system (which we got up and running).  These were across cognitive, affective (read: personality, e.g. the big 5 psychological traits), and conative (motivation, anxiety or ‘safety’, etc).  We had the system adapt on the basis of these competencies, not in changing the media to accommodate styles, but looking at different sequencing between (what I argue are the important characteristics) of example, concept, practice, etc.

We also believed that many if not all of these learning competencies could be improved, and designed strategies to develop skills over time. The premise did require a long-term relationship with the system, but that was our goal anyway.

The point here being that if, instead of fixed characteristics, we think of a suite of malleable learning competencies as a way in which our learners can differ, we gain two things.  First, we find ways we can support learners who have weaknesses in particular learning competencies (dealing with visual data representations, for example), and second, we can develop them in those competencies as well (which goes hand in hand with Michelle Martin & Tony Karrer’s Work Literacy).

It’s also a tangible investment in organizational competency, and potentially the only real leverage an organization can have, going forward.  Think: learning skills instead of learning styles, and develop your learners accordingly!


  1. This is refreshing. Learning styles always looked clearer when seen as a set of perspectives, rather than a blueprint of characteristics. I always found it difficult to address the issue when learners said “I’m really a visual learner” in the same way as people say, for example, “I’m a Leo” and I tended, perhaps a little obtusely, to focus on what they weren’t. If you can do that, let’s work on the rest. Isn’t that a teachers responsibility?
    Competencies looks like a useful way of looking at it. How do you link in learners own understandings of their “learning style” to that? Is that kind of reflection one more competency in the list, part of the metacognitive section perhaps, or is it transversal?

    Comment by Nick Kearney — 31 December 2008 @ 5:11 PM

  2. Hi Clark,
    Many thanks for your insights on learning styles. You post here has provided another important direction on research relating to learning styles and competencies. I have written a few posts on learning styles and am interested in conducting further research that relates to neuroscience findings. I have been thinking about organizational competency, but have found a fundamental “issue”: what is required by organisation may not be the same as that by an individual and vice versa. How could one resolve the conflict between the two? Besides, under network learning, the focus seems to be on autonomy and openness, whereas under organisation learning, the focus is to comply with “group’s decision” and one must be a team player, with “closeness” or “membership” under teamworking. Do you see it as an issue?
    Great to learn your views.
    Happy New Year.

    Comment by Sui Fai John Mak — 31 December 2008 @ 5:36 PM

  3. There are two ideas here that I find very interesting: first, thinking of styles as unimodal, rather than bimodal dimensions. As you point out, there is not reason that the endpoints for a dimension need to be inversely correlated. (Although I wouldn’t think of Gardner’s MI theory as an example of that principle, although it may have been the inspiration, as it is an abilities theory rather than a styles theory).
    The other intriguing idea is using this principle in an adaptive system as a way of determining empirically what “learning styles” might mean. In other words, letting the data show you the extent to which there is consistency of style across tasks, occasions, etc., rather than starting with a theory and using a couple of one-shot tests to verify or disconfirm.
    Really interesting stuff.

    Comment by Dan Willingham — 2 January 2009 @ 6:27 AM

  4. Nick, Joe’s concept was helping learners understand themselves as learners, and I reckon that if we know things about them, we should share. Having them know themselves as a learner is a meta-cognitive thing that can help them look at their own learning and consequently, by having it visible, have them able to reflect and improve it.

    John, I like to think it’s in an organization’s interest to have learners improve, so the two should be coincident. To me it may be an issue of the organization saying, in effect: we’ll invest our effort in improving you, and in return you’ll invest your effort in our goals.

    Dan, yes, the unimodal came more from an abilities perspective, but I think there’re reasons to decouple, or at least consider. And yes, in a longterm relationship in a system you do get some evidence. We were going to put our assessment in a ‘game’, where your decisions indicated your actual processing characteristics. More useful than self-report, certainly. But the ability to track and datamine is an interesting proposition, if you know what to track (just saw an announcement for an educational datamining conference). One of the opportunities of suitable meta-tagging and data-logging.

    Thanks for the feedback!

    Comment by Clark — 5 January 2009 @ 10:44 AM

  5. As someone who did not fit the teaching “style” in a traditional classroom, I can say that a change in instructional design does make a difference for a student. I like that you term them learning “abilities” rather than “style” as style seems to me to be something that is a preference that can be turned on and off. My daughter has auditory processing problems and as such is a “visual” learner which can be tested. This means she is able to take visual cues and convert that to “learning”. Oral stimulus, such as lecture, is difficult for her, especially when there are other noises as she converts all the sounds into one big message (not able to tease out one sentence from another from different speakers).

    I was wondering how you have approached the differences in media. It seems that all of the current research on information overload, brain waves and parts of the brain that is being used for different types of thinking and stimulus would also help inform your system.

    Comment by Virginia Yonkers — 6 January 2009 @ 11:17 AM

  6. Virginia, I’m trying to separate out using the right medium for the message (e.g. a diagram for static relationships, animation for dynamic) versus a flexible ordering of elements: example first for some, problem first for others, concept first for yet others. This is initially for typical learners; I’m not attempting to account for accessibility, e.g. processing deficits at this point, but I’m definitely sensitive and think there are interesting ways to consider providing such support. Have a look at John Sweller’s excellent work on Cognitive Load Theory and media for some great thoughts on differences in media and learning effects. Much of it is summarized in Clark, Nguyen, & Sweller’s Efficiency in Learning.

    Comment by Clark — 6 January 2009 @ 4:20 PM

  7. Exactly, giving up “Learning Style” and think about “Learning Skill” instead. I found that it’s impossible to identified exactly individual learning style because it is changed overtime and depended on the several aspects. Moreover, it’s not the good idea to adjust the theme of learning pedagogy suit to individual learning style. The thing that we should do is to find the way of appropriate learning skill development.

    Comment by Sasalak — 21 January 2009 @ 5:47 AM

  8. Just a very late thought, since I’ve just found “here”. There’s an assumption that matching instructional methods to learning style will result in greater learning, but there’s actually research on both sides — that learning is best when matched, and also when there’s a deliberate mismatch.

    Different models predict and explain each finding.

    Comment by Robert Bacal — 10 June 2009 @ 1:43 PM

  9. That’s an excellent suggestion, Nick, to consider learning “skills” rather than learning “styles”.

    I think another perspective (though not mutually exclusive) is to consider so-called learning styles in terms of learner “preferences”. In the corporate sector, I see learner engagement as a serious challenge. If we can better accommodate the learner’s preferences, perhaps we can boost their enjoyment of the learning experience?

    Comment by Ryan Tracey — 15 August 2010 @ 7:45 PM

  10. Check out the 4MAT Instructional Design System and the research that supports it by Dr. Bernice McCarthy and her perspective on Learning Styles. I believe you will find it interesting. See more at Aboutlearning.com

    Comment by Leslie LeMaster — 15 October 2010 @ 2:57 PM

  11. Thanks for the feedback. Yes, issue of match or mismatch, the problem is reconciling them into a bigger picture. I believe the additional factor will be their confidence and anxiety that will determine which to do when, but am waiting for the research to catch up. Read the UK study for a good analysis of why folks don’t have a vested interest in open research. Preferences matter, tho’ what a learner prefers may not be best (at least until we start explicitly addressing learner capabilities). And 4MAT is one I do not like: both for the basis (Kolb’s core model’s been questioned), and for her proprietary approach. Again, see the UK study (Learning and Skills Research Centre link above). And the new study led by Hal Pashler.

    Comment by Clark — 18 October 2010 @ 11:29 AM

  12. […] Rethinking Learning Styles […]

    Pingback by CIVNET | Blog | Learnlets.com — 27 January 2011 @ 12:38 PM

RSS feed for comments on this post. TrackBack URI

Leave a comment

Powered by WordPress