The butterfly learner

As I mentioned in my previous post, I’m encouraging Patrick McAndrew to blog. His first proper post is a goodie – it sets out his concept of the Butterfly learner. This is an idea we have discussed over beer/coffee before, but he has never got around to publishing formally. Enter the blog and you have an immediate outlet.

Anyway, the idea is that small changes in the environment can have unforeseen effects on learners. As Patrick says, this may have significant implications for learning design work, since it may be that learning is just intrinsically chaotic. There are too many variables and the inter-relation between them is too complex to model effectively. I have seen something of this in practice – when I chaired the large first level course T171, we had around 12,000 students. We couldn’t house them all on one server at the time, so we split them across two and replicated the forum structure on each. They were studying the same course material, had the same activities, and the same support structure. What emerged was two markedly different groups however. On one server they were very active, loved the course and engaged in all the activities, and on the other server the forums were dominated by complaints, arguments, calls for clarification, and not much activity around the actual course material.

This might not in itself demonstrate that learning is essentially chaotic – one of the influencing factors seemed to be the nature of the early dominant voices in the forums. If they were positive then the cohort tended to follow and if they were negative then we got a largely negative cohort. This is a well known phenomenon in group psychology, and you can see how it would happen. So it is a variable, and one I didn’t control for, but that doesn’t necessarily mean learning is chaotic – just complex.

But if there are enough such effects, then it may as well be chaotic as far as the average educator is concerned, because it becomes impossible to control for all such variables. Take the example above – even if I had thought about it beforehand it would have been difficult to control with such a large cohort. I could have split groups according to the outcomes of psychometric tests maybe, or planted some ‘happy’ stooges in each forum to get it going on the right note, but these all seem rather unlikely (not to mention dodgy from an ethical point of view). And that is assuming that a happy forum is best for learning anyway. What I didn’t do (and regret now) is look at the results between the two sets of students. Did the nature of the dialogue have any effect on outcome? My guess is it probably didn’t in the end.

Back to learning then, and maybe this is why the standard lecture has persevered for so long. Learning may be chaotic, so we should remove as many variables as possible. The lecture is a fairly stripped down educational experience. We also tend to know a good lecturer when we see one. The problem is that even if we wanted the lecture to continue unchanged (and I don’t) the nature of learning has changed – when I am online I have access to people, resources, my own notes, tools, services, instantly. Learning becomes much more embedded in our everyday lives and not something we parcel out to a specific room. We are going to have to learn how to learn in this environment. And if it is chaotic, then we will need some ways of controlling that chaos (I’m mixing definitions of the term chaos in that sentence, but never mind for now).


  1. What an interesting post – not least because I was one of the T171 students in (I think!) the active, ‘happy’ group. I was curious about your question that “Did the nature of the dialogue have any effect on outcome?” and I have the feeling that maybe the outcome you can study – the course results – isn’t the long term outcome which was most powerful for the students involved. For me, it changed the way I thought about learning and engaging with learning. It was refreshing, interesting, exciting, frustrating, fascinating, complex – and gave rise to a longer term interest in online education.
    Maybe the change isn’t one that can be constrained within the measurable bounds of the course itself? Maybe the change learners experience is something deeper and more long lasting? Sometimes if you’re concentrating on measuring the splash a stone has made when dropped into water means that you miss seeing how far the ripples went…

  2. Hi Sarah,
    yes you were in the happy group – and one of the reasons it was happy I think. Yes, you’re absolutely right – measuring academic performance would be a very crude measure. Which only adds to the butterfly thing – we don’t know the starting variables, are unsure of the process and have a multitude of outcomes. So we tend to simplify it to lectures, notes, exams.

  3. Nice to see your example Martin, there always seem to be personal experiences that support the Butterfly Learner view but not measured data. Actually this feels right maybe a related effect is the Heisenberg uncertainty principle where if we try to measure something we change it! Hmmm – I remember making a slide about this for a talk maybe I can reproduce a blog version as my next homework :-).

  4. Martin
    The new digital photography course (T189)had to split the forums, I think, so it may be worth talking to some of the CT to see how the dynamics of each group compared there, and whether ‘group happiness’ was in any way correlated to average marks across the group…

Leave a Reply

Your email address will not be published. Required fields are marked *