Examples before practice

I’ve been wrong before, and I’ll be wrong again, and that’s ok <warning: link is NSFW>. It’s like with science: if you change your mind, you weren’t lying before, you’ve learned more now. So I’ve been wrong about the emphasis between practice and examples. What I’ve learned is that, in general, practice isn’t the only area of importance, and the benefits of examples before practice.

So, as part of the Learning Development Accelerator‘s YOK (You Oughta Know) series, I got the chance to interview John Sweller. I’ve known John, I’m very honored to say, from my days at UNSW. I was aware of his reputation as a cog sci luminary, but he also turned out to be a really  nice person. He’s the originator of Cognitive Load Theory (CLT), and he was kind enough to agree to talk about it.

As background, he’s tapped into David Geary’s biologically primary and biologically secondary learning. The core idea is that some things we’ve evolved to learn, like speaking. Then there are things we’ve developed intellectually, like reading and writing, that aren’t natural. Instruction is to assist us to acquire the latter.  The latter typically has high ‘element interactivity’, whereby there are complex interrelationships to master. That is, it’s complex.

CLT posits that we have limited cognitive capacity, and overwhelming that capacity interferes with learning. The model talks about two types of load. The first is intrinsic load, that implied by the learning task. The second is extrinsic load, coming from additional factors in the particular situation. The premise is that learning complex things (biologically secondary) has such a high intrinsic load that we really need to focus on managing load so we can gradually acquire the entailed relationships.

There are a number of implications of CLT, but one is about the value of worked examples. An important element is showing the thinking  behind the steps. A second empirical result is that worked examples  are better than practice! At least, initially, for novices. Yet this upends one of my recommendations, which is generally that the most important thing we can do to improve our learning is focus on better practice. I still believe that, but now with the caveat after worked examples.  

Now, he didn’t tell us when that happens, e.g. when you switch from worked examples to practice. However, like the answer to how much spacing needed for the spaced practice effect, I suspect the answer is ‘it depends’. There’s the ‘expertise reversal’ effect that says as you gain experience, the value of worked examples falls and the value of practice raises. That point, I’d suggest, is dependent on the prior knowledge of the learners, the complexity of the material, the scope, and more.

I’m now recommending, particularly for new material, that improving the learning outcomes includes meaningful practice  after  quality worked examples. That’s my new, better, understanding. Make sense?

As an aside, I talked about CLT in my most recent book, on learning science, with a diagram. In it, I only included intrinsic and extrinsic, as those two seemed critical, yet the classic theory also includes  germane intrinsic load. One of the audience members asked him about that, and John opined that he probably needn’t have included germane. Vindication!