In my last post I outlined my concerns with the idea of ‘thinking hard’ being a good proxy for learning. Briefly, thinking hard about a problem appears to be an inefficient way to alter long-term memory structures. This means that it’s perfectly possible to struggle with a difficult exercise, successfully complete it, and still not have learned how to repeat the process independently. The problem is that ‘thinking hard’ exhausts limited working memory reserves.
In fact – as Daisy Christodoulou states in Making Good Progress? – the evidence on ‘overlearning’ seems to suggest that repeating a task to the point where almost no thought is required in its completion is a better way to change long-term memory:
Overlearning refers to the continued practice on a task after some criterion of mastery on that task has been achieved. A pianist, for example, might continue to practice a piece despite already being able to perform it. (Soderstrom & Bjork, 2013)
Way back in 1929, WCF Krueger showed that overlearning can represent considerable advantages in retention. Subjects were divided into two groups and given a list of words to learn. The first group practised learning the words to the point of mastery – i.e. to the point where they would reliably score 100% on a test – whereas the second group overlearned the list, studying for twice as long as the mastery group. While both groups’ performance was equally good immediately after the period of study, the overlearning group showed much greater retention on a long-term test. In a later study, Krueger showed that long-term performance increases in line with the amount of overlearning during practice: more overlearning equals better retention.
This same effect has been shown to hold true for meaningful prose (Gilbert 1957) as well as motor skills (Adams & Reynolds 1954 and Stelmach 1969). The importance of continuing practice beyond the point where performance no longer seems to improve is well established. Building on this early research, James Driskell and colleagues confirmed the beneficial effects of overlearning across a range of domains and drew the following conclusions:
- Although overlearning is beneficial for both cognitive and physical tasks, there was significantly greater benefit in cognitive tasks.
- Overlearning material by both 100% and 150% produced ‘strong to moderate’ improvements.
- The benefits of overlearning appeared to decrease over time, so that with a delay of 19 days between overlearning and testing resulted in performance falling by as much as 50%. However, if further overlearning were to take place approximately 3 weeks after the first session, improvements were maintained.
This last finding is consistent with what we know about spaced instruction: what we learn is prone to a decay in retrieval strength without topping up on what we have learned.
All of this needs to be filtered through the research on developing expertise. In his wonderfully useful book Peak, K Anders Ericsson, the pre-eminent researcher in the field, has written about the need for both ‘purpose practice’ – practice with the express aim of improving performance – and developing effective ‘mental representations’ of what success looks like.
At first glance the guidance on practice seems to contradict these findings about overlearning. Ericsson is clear that the gains from further practice start to disappear when we enter the ‘autonomous stage’ of skill development. Practice only seems to confer improvements to performance when we are in the ‘cognitive stage’ which forces us to concentrate on what we’re doing and be mindful of our performance. However, even though overlearning – practising beyond mastery to the point of automaticity – results in performance plateauing, my suggestion would be that future practice should attempt to overlearn beyond the point of our current ability. If we continue to raise the bar, overlearning should continue to be beneficial.
The idea of having a clear idea of what expert performance is like is crucial. If we practice without knowing what good looks like, we have no internal frame of reference to provide us with feedback on whether we’re making progress. This means that we’re dependent on someone else for this feedback. I’ve argued before that giving feedback can often have a detrimental effect on long-term learning and this is why. The benefits of overlearning may in part be due to creating an increasingly familiar sense of what effortless performance should feel like. If we only practice up to the point of mastery we may not have much experience of what mastery is like, but practising beyond this point could give us that sense of what it must be like to be an expert.
The implications of all of this is that rather than continually raising the bar and expecting students to contend with ever more complex challenges, perhaps we should allow considerably more time for consolidation before moving on to more difficult material. For instance, instead of giving young children increasingly demanding reading books every time they finish one, perhaps we should get them to continue to practise reading their book beyond the point at which they’re fluent, allowing them to enjoy the experience of reading without effort. As well as the potential benefits to retention, I think this approach is also likely to improve students’ motivation and enjoyment.