At EBI last weekend, I gave a skillshare on Problem-Based Learning (PBL) - learning based on setting students problems to solve and then reflecting on the solutions they come up with (it does not mean, as the name might imply, that we start with the “problems” with our students’ dancing). We didn’t have enough time to get round to the main activity: actually coming up with good problems to use in dance classes, but we did quite well on experiencing some PBL, reflecting on what made a good problem and starting an overview of what research has found on the subject. In order to complete this overview, I promised to write up some notes. Because PBL isn’t my specialty, but more something I thought was relevant to give the class focus, I’ll also diverge into the broader field of the Learning Sciences. To give an idea of how I tried to use PBL in a class, and to reflect on the input from the participants, I’ll write in chronological order of what we did.

NOTE: I have references and comments about these references for everything, but in the interest of actually getting something out of the door, I will make a first reference-less pass. If you check out “constructivism” on Wikipedia, there are some very specific complaints about how people are (mis-)using it for teaching, and that there is no evidence that it works - this is slightly misleading, especially when transferring to the world of dance - I will detail why at a later point.

Teaching and Learning (the extended version of how I introduced my skillshare during class demos)

There is a mindset we get into as a teacher: we need to break down the subject we want to teach into its right components, find language, movements and metaphors to convey this knowledge, and construct exercises to let students practice what we’ve taught them. In other words, as teachers, we have knowledge and skills, which it is our role to transmit to our students. We then let our students practice these skills and put the knowledge into use and this is how they learn. This view is often called instructionism (although strong proponents of instructionism may argue the details). The major challenge for this in a dance class is connecting the concrete instances of practice we give to our students with the concepts we want to convey. At most, we’ll get through 4 or 5 “moves” that put the concept into practice. How do we get students to connect these moves with all the other times the concept comes into play? How do we get students to understand that these moves are not what they are actually learning (even that is what we spend most time on in class). How do we get over the fact that something students may find challenging about these moves may not be at all related to the concepts we are working on? How do we get students to integrate their knowledge about different dance concepts into an interconnected schema?

The Learning Sciences is an interdisciplinary field which is broadly interested in understanding the contexts in which people learn. Our (I am learning scientist, though I have contributed very little myself to this body of knowledge) research over the past 40 years has provided some powerful ideas on how to move beyond instructionism. Some of these ideas exploit constructivist theory - that learners do not “receive” knowledge from teachers, but must “(re-)construct” knowledge in their minds. Others exploit socio-cultural research - how do learners, teachers and experts fit in a culture of learning and doing? Others still explore the cognitive basis of expertise and knowledge representation - how do we acquire the skills and knowledge and how do we re-use them to deal with new problems?

Of course, as a learning scientist, I know about these things. And what do I do when I teach dance? Pure instructionism all the way down the line. I wanted help to brainstorm on how to improve on this.

One of the ways the ideas from the learning sciences are put into play as a framework for designing productive learning contexts is PBL, which I chose for the focus of class (alternative choices might have been related ideas of productive failure, delayed instruction or worked examples). My goal was to run a PBL class on PBL. We start with a problem, we break into groups and reflect on the problem, we share some of our results, we get more support on what a good solution to the problem might look like, and under what criteria we might consider it good and then we talk about some of the canonical solutions to the problem (i.e. we get round to the instruction part). In fact, it was a double PBL class, as we did a whole mini round of experiencing PBL in a dance setting as a setup to discussing PBL itself.

Experiencing PBL in a dance setting: here’s a music, figure out some ways to dance to it.

Un Jour m’y Prend l’Envie, by Ciac Boum, on their album Vert <div></embed></div>

I broke the class (four people at that point) into two groups of two and let them work on it unassisted for the duration of the song (just under 5mn). I really liked the variety of ideas that they came up with, but was slightly concerned that none of them actually matched the fundamental rhythm of this song in an explicit way. So I then showed some ways that I would dance - one related to an idea from each group and a third that was different and set a new task in new groups (I think a 5th person had arrived at that point): try to figure out what I was doing - and specifically how it related to the rhythm of the music (for half a song duration). We then came back again and the approximations of what I was doing were of quite good quality - but not really any better than what they had come up with on their own in terms of matching the music rhythm. I then clapped out some rhythms and asked them to reflect on those options individually to tell me what the rhythm embedded in the music was. This was slightly inconclusive - they were able to match some of my clapping, but not to conceptualise the actual meter - I’ll spoil it for you in the next paragraph.

We concluded this section by my explaining the meter of the song. It’s 5 count, organised in 3+2, except with the phrasing placing the 2 as a pickup or anacrusis, giving a 2+3 feel. Suggestions for counting and dancing this: short looong, short looong, short looong, which can be step touch, touch step, step step; quick quick slooow, quick quick sloow (12 1&&, 12 1&&), which can be step step steeeeeep, or with a sideways pendulum, swing swing step, swing swing step. Or 12345, danced as step step step step step, with an axis change on the 1, or the 3. In an actual class, this would have led into discussions about meter and phrasing in music, and how they relates to dance. We could then have moved on to the different parts of our body that we can use to express rhythm, in particular how steps can involve an axis change (or not). I would probably have set some more problems specifically focussing on these topics.

PBL about PBL: discuss what makes a good problem in a dance class.

We then broke into two groups of three (yay, another person had arrived) and set the task of discussing how participants had felt about the 3 problems they were set (“in pairs, figure out ways to dance to this music”, “in pairs, figure out the movements I just showed/danced and how they are related to the music”, “on your own, try to figure out the rhythms I just clapped and how they are related to the music and the movements”), and then broadening out to what would make good problems in a dance class. We set a timer for 7 minutes and prolonged it for another 3 because everything was going so well.

These discussions were overwhelmingly amazing (another bonus point about “post-instructionism” is that though the teacher has to put in a lot of work to create and adapt their material, they then get to bask in how awesome their students are). Here are some of the highlights of what was mentioned when we came back together. (SIDENOTE: because of my lack experience in leading this kind of class, I’m not sure that I did a particularly good job, especially in terms of leading this discussion, which took a good 10-15 minutes - also apologies to the participants in terms of taking notes).

The first problem felt very open, producing very different answers depending on participants previous dance experience. Several people felt like they didn’t have the musical tools to deal with this new music: “I’m not musically minded”. In a similar way, they imagined how much more difficult the exercise would have been if they didn’t already know how to dance, or if they weren’t put into groups. It felt like a good exercise to open the brain to new possibilities.

The second problem had similar issues. Participants also touched on the question of learning styles - whether just showing wasn’t too difficult or went to fast for “non-visual learners”. They also felt that it was useful for them to see the movement connected to the music (moving to music being something that dancers can sometimes do easily, even when the rhythmic details are not obvious to them).

The third problem added in more possibilities to help identify the 5-countness, and opened up to other “learning styles”.

More generally, participants discussed

  • whether this kind of open problem wasn’t way too difficult for beginner dancers
  • how do we explain that we keep only a small fraction of the solutions that participants come up with? If all these solutions are “wrong”, why let us battle through them, rather that helping us find the “right” ones.
  • that it was a good way to meet and learn to communicate about dance
  • that it seemed difficult for shy people and introverts
  • that it seemed great for children, who possibly have fewer barriers to being creative and daring to be “wrong”.
  • the relationship between the problems and learning styles
  • whether the teacher has a goal in mind and how hidden this goal is from the students

Back to instructionism: a long lecture on PBL

We then moved to a phase I wasn’t super happy with, where I discussed everything that was said and related it to known aspects of PBL and of the learning sciences in general. We were running short on time, it became apparent that we weren’t going to get round to actually coming up with some PBL ideas for dance classes, and everyone’s eyes started glazing over. On the other hand, I was super happy that I got to discuss everything - nothing was discarded, everything the participants said was relevant.

PBL, delayed instruction and productive failure

One of the core ideas of PBL is that the kinds of problems that we set up are not expected to be solved, either because there is no single “correct” solution, or because the problem is defined such that part of solving the problem is defining what a good solution should look like, or because the canonical solution is beyond the grasp of the participants. A large part of the experience is a “productive failure”, which leads to better conceptualization of what the problem is, reflecting on what a good solution would look like, and evaluating the solutions that we come up with. This experience is then completed with instruction/teaching, identifying some of the canonical solutions and discussing how these solutions are related to the ones the students came up with.

Without the prior problem-solving, the instruction/teaching would make less sense: as it is, we can not only better understand the canonical solutions, but we have a better intuition for why they are “good” solutions, and how they are related to more general concepts. We also improve our problem solving skills, leading us to increasingly understand what a good dance solution looks like and how to find and evaluate candidate solutions.

We also decouple learning (productive and unproductive) from performance (success or failure). In particular we value finding the productivity of failure and avoid the misleading sense of accomplishment of unproductive success - i.e. when students succeed at a task but don’t learn from it.

When we give the problem, we may mask our goal (what concept we are working on) if it would give away too much of the problem, or too much of the solution. But we set up a problem with a specific goal in mind and making sure that we’re not setting up a game of “guess what concept I think you should be taking away from this class”, leading to “what I actually wanted you to work on is stretch”, and “GROAN, why didn’t you say so to start off with”.

Learning styles and reflection

We came back several times to learning styles. Learning styles are the pop psychology of theories of learning: men are from Mars, women are from Venus and learners have a preference for Auditory, Visual or Kinesthetic modalities. Nice ways of looking at the world and wouldn’t it be neat and tidy if they were true? But the research has found very little evidence that they are.

We all learn in all modalities at once. But depending on what is being taught, just like multiple metaphors might convey different aspects of a solution, multiple modalities will convey different aspects. In that respect, teaching in all modalities caters to our learning in multiple modalities. However, there are other issues to consider. Deliberately not teaching in some modalities can be a strategy to set a problem: fill in the missing modalities. Deliberately teaching different aspects in different modalities creates a form of cognitive conflict (see next section) where we set a different problem: figure out how to reconcile the information in these modalities. Trying to show the same thing in all modalities (and more generally, any instruction that aims to help students too much) creates a kind of spoonfeeding - and then, much like a GPS spoonfeeds us directions and we realize that, without the GPS, we would get nowhere, spoonfeeding creates a prepackaged bundle that students will have difficulty learning and/or integrating with their existing knowledge.

The kinds of learning styles that do appear to exist involve the strategies that learners use to memorize information, acquire skills, etc. They exist in so far as we can directly observe that some people try to build deep conceptual knowledge and others are satisfied with more shallow knowledge (deep knowledge ideally has better transfer - we can apply it creatively in new situations - shallow knowledge is the kind most often “regurgitated” on tests and then promptly forgotten). The research I’ve seen seems to operate along the principle that we should encourage everyone to become deep conceptual learners - i.e. to change their learning style.

There may also be some value in the idea that some people have different preferences in terms of order, going either abstract to concrete or from concrete to abstract. Some people may also prefer direct experimentation and hands on experience, and not find value in using reflection to relate this experience to abstract concepts. We see this in many examples of expert behavior (dancing, driving a car, performing a hip replacement operation, speaking a language fluently): often experts do not frame their expertise in abstract, conceptual terms. For example, everybody who speaks English as a native language is capable of choosing the correct tense in this sentence (nobody would say “everybody who is speaking english”), but few people can accurately reflect on why they choose the one and not the other.

PBL is designed to encourage people to reflect on what makes a good solution to a problem, on how this solution is related to abstract concepts and on how to use their existing knowledge and problem-solving strategies to solve new problems. Some people may not find value in this - but this poses the wider problem that they not find value in any class (whether based on PBL or not) that tries to address conceptual issues. They might be better served by being given problems to solve and to let them solve them experientially, without using deeper reflection - which would be fairly similar to learning through immersion.

Cognitive load, cognitive conflict and prior knowledge

Cognitive load is the idea that the constraints on our working memory prevent us from dealing with too many new things at once. We need to form mental representations of knowledge in working memory that allows it to transfer into long term memory.

If this theory is correct, it imposes severe constraints on any “here, solve this problem” kind of teaching. The problem solving is too difficult because of the cognitive load involved. A potential solution to this (and a necessary element of any effective teaching) is to build on prior knowledge. By relating new problems to existing knowledge, we not only relieve the cognitive load, but encourage cognitive conflict - challenge between two different representations of knowledge that need to be resolved. Many researchers argue that we learn by working through cognitive conflict. Solving problems in groups is also a way of generating cognitive conflict - exposing people to conflicting prior knowledge.

A lot of instruction tends to give students “some new concept” that comes in a prepackaged bundle. If this concept is not presented in relationship to prior knowledge, there is not only a danger that it will not create cognitive conflict and will therefore not help “weed out” the incorrect existing knowledge we have, but also that it will not “mesh” well with the prior knowledge - so we will consider it unrelated to all the things we already know.

Another method we can use to reduce cognitive load is called scaffolding - creating various forms of aids which help orient the problem solving process or reduce the size of the problem to make it easier to handle. But in a long term progression, if those aids are constantly there, they become crutches that we cannot deal without. Or if badly chose, they can be like side-wheels for a bike - something that fundamentally changes what riding a bike is like and apart from giving confidence, could be argued to do as much harm as good. So we must be very careful in considering how we frame problems for our students - and how we frame them differently over time.

Identity and community

My current favorite part of thinking about teaching is how teaching relates to identity. Rather than thinking of what distinguishes a dancer from a non-dancer only in terms of knowlege, it’s useful to think of this difference in terms of epistemic frames. An epistemic frame is a way of making sense of and doing things that combines knowledge, skills, identity and values. To become a dancer, we not only need the skills and knowledge of a dancer, but we need to acquire the identity and values of a dancer. The biggest problem we will ever face with new dancers (and even some much more experienced dancers) is that they identify as “non-dancers”. Non dancers cannot do things that only dancers can do, because it fundamentally challenges their identity. Get people to take on a new identity - or even pretend to take on a new identity - and we open doors for them.

Challenge their identity or remind them too much of their identity and we prime them or produce stereotype threat. This is a rather difficult tightrope to tread and I’m not quite sure how to deal with it. For example, during the skillshare, one participant noted that they did not consider themselves a musician and could therefore not identify the rhythmic structures in the music. Although it’s possible they “truly” couldn’t, as soon as they identify as a non-musician there is a self-fulfilling prophecy that they won’t be able to (this is also related to the concept of self-efficacy - someone who believes they cannot do a task are more likely to fail at that task). But how to actually change this identity without also threatening it?

One really powerful thing teachers can do is treat students like dancers, like equals of the teachers. I think this justifies delving into PBL more than anything else. By setting problems that are problems that dancers solve, by showing the value of the solutions students come up with (possibly building on them, or explaining the grounds on which we reject them), teachers signal to their students: “We are part of the same community, dancing is just expressing ourselves in movement, we all have things to express, we all have the same number of bones and muscles to create movement with, we might be further along the path than you but we are on the same path”. By treating students as capable problem solvers (which many of them already are in real life), we allow them to express themselves as capable problem solvers and bring out the best in them.

The idea of giving “problems that dancers solve” is an important one. It’s the idea of creating authentic problems - problems that have meaning to students, that motivate them because they are problems they face on the dance floor and that motivate them because they are problems that belong to the identity they are trying to acquire.

Last, giving students problems to solve as a group creates a community. People who know each other, people who hang out, people who have battled through problems together and emerged, triumphant. These are people who will come out dancing for each other and people who will be eager to join the vast community of like-minded dancers (unlike beginner classes that feel like they are going to go dancing and will be terrible at it and that nobody will want to dance with them).

Dance and the education system

Before writing this, I re-read some of the literature I wanted to refer to (and the related wikipedia articles). I was surprised at how negatively teaching strategies based on constructivist ideas of learning appeared to be treated. Apart from the debate being a bit of an empty one (just as no instructionist researchers think they can just “give” knowledge to students, no constructivist researchers think that students can just learn, without guidance), there are at least two specifics of social dance that I think argue for more constructivist ideas being integrated into dance teaching.

Part of the reason that constructivist ideas have been promoted is that they are intended to improve deep, conceptual learning. This kind of improvement is not the kind that can be measured by most tests within the education system - at best, they are expected to perform only just as well as instructionist ideas on such tests. The end goal is to improve education in the long term and our ability to measure that improvement is deeply impaired by the difficulty in defining “better education” and our fundamental inability to change all the forces at work within an education system. In social dance, however, there is no education system to deal with, there are no tests - the kind of teaching that is designed to produce factory workers who are good at taking tests serves no purpose.

The other aspect of social dance is that students are, hopefully, social dancing. They are putting in a lot of non-reflective practice, internalising a lot of bad habits and misconceptions. This has implications for what it means for a problem to be authentic (learning to lead 4 8-counts of choreography is not authentic - learning to steal 4 8-counts of choreography is slightly more authentic - learning to improvise 4 8-counts of choreography is an actual problem), and for the absolute necessity of drawing on prior knowledge and creating cognitive conflict for the teaching to have any impact. What is more, students need the tools to reflect on their social dance experiences, transforming them into learning opportunities rather than practice opportunities, where they can take the things they do or try to do and organise them as knowledge that is readily available for them, for problem solving in class and problem solving at the next social dance.

Note to myself: A lot of Keith Sawyer’s introduction reflects on research into how *children learn. How are adults different? They have more prior knowledge, they have stronger identities, their misconceptions are less reliant on intuition and experience and more on declarative knowledge…*

So what does a good problem look like for a dance class?

Putting all these ideas together, a good problem:

  • Draws on participants’ prior knowledge (so what do we do when we have differing levels of knowledge?)
  • Is careful about the way it primes for identity
  • Is authentic (it has meaning for the student, both in their current and future projected identities)
  • Positions learners as capable problem solvers and capable dancers
  • Creates cognitive conflict
  • Is scaffolded to reduce cognitive load (but not cognitive conflict, i.e. it does not spoonfeed new knowledge)
  • Fits in a curriculum where scaffolds are progressively reduced
  • Accepts many answers that can be built on to create canonical answers rather than be rejected
  • Is “wicked”: the problem is defined such as clarifying what the problem is is as much part of the finding the solution as actually finding the solution
  • Allows students to understand how to evaluate candidate solutions
  • Prepares students to understand why the canonical solutions make sense (they evaluate “better” than other solutions)
  • Allows students to understand how the evaluation is related to abstract concepts (and thus prepares them to re-use similar evaluation methods to evaluate solutions to new problems)

Actually creating problems that evaluate well for all these criteria is a tough one that I’ll save for another time. I’m convinced that we can use PBL just as well in beginner classes (where we draw on many years of being experts at walking and moving their own bodies, existing problem solving abilities, and attempt to address identity issues such as having two left feet) as in advanced classes (where we work on existing knowledge students have of dancing and try to amke it mesh it in new and wonderful ways, while confirming their identities as dancers). The most difficult are probably all the “intermediate” classes, where students have different levels of knowledge, do not understand that some of this knowledge needs weeding out, and have identity issues related to not wanting to give up the things they already know (because it would mean they don’t know so much, or that they have to reject the absolute knowledge given to them by their favorite teacher).

As parting words, I have a researcher colleague who is a computer scientist. He believes strongly in “time on task”, which is ultimately one of the most consistent predictors of learning: how much time do students spend actively engaged in a learning task? He likes all the learning theories, but ultimately, he likes to reduce every new flavor of the month teaching/learning theory to “do students spend more time on task”? The more we “teach” dance, the more we tend to stop and explain and the less our students are actually dancing. So if all the above has left you unconvinced, PBL can also be reduced to one of the many ways we can produce less teaching and more learning - which rather unelegantly brings me back to the title of this piece.

Thanks to Rory, Stefan, Lilith, Mike, Chloe, Tobias and Ruth, who participated in the skillshare and whose numerous reflections forced me to produce class notes that basically summarize most of the findings of the Learning Sciences.