A Test of Three Practice Schedules, and the One That Works Best (Especially for Older Learners)

It’s often said that kids learn faster than adults. And as I’ve reached the age at which I used to think adults were considered old¹, I’ve certainly felt increasingly like one of those well-used sponges that no longer absorbs quite as much water as it once did. =(

Of course, there is research which suggests that this may not actually be as true as it feels – but we do experience various physical and cognitive changes over the course of our lifetime. Like in motor performance, working memory, and selective attention, for instance.

So what does this mean for older folks still seeking to learn new things and continue to improve? Does it mean we need to practice differently as we get older? And if so, what would that even look like?

A coordination/timing task

A pair of researchers (Beik & Fazeli, 2021) recruited sixty adults between the ages of 60-75, to test the effect of different practice strategies on their learning and motivation.

The task was a pretty simple one. Basically, they were placed in front of a computer screen and given a giant, square, 1.5ft number pad. The computer screen would show them which numbers to press, and what tempo to press them at, and then it’d be the participant’s turn to give it a try. So basically, it was like a super simple version of Dance Dance Revolution , but for the right hand, rather than the legs.

Three groups

To keep things simple, the pattern of numbers that participants were asked to press was always the same. The only thing that changed was the tempo, or the speed at which they had to press the number keys.

A third of the participants were assigned to the “blocked” practice group, and asked to practice the first timing pattern 54 times (let’s call it tempo A). Then they were asked to practice tempo B 54 times, and then tempo C 54 times. For a total of 162 practice repetitions.

Another third of the participants were assigned to the “random” practice group. Unlike the blocked group, which had a nice, predictable, orderly sort of structure, theirs was all jumbled up. For them, the computer kept switching up the order of the tasks, where they would never see the same tempo more than twice in a row. So their practice looked something more like tempo A, tempo C, tempo B, then B, C, A, C, C, A, B, A, B, A, C, etc.

The final third of the participants were assigned to the “learner-adapted” group. Their practice was a combination of blocked, “serial,” and random practice, where the type of practice they did, was dependent on how well they were doing and how many mistakes they were making.

So they would start with a little bit of blocked practice, to get comfortable with each task. And if they were doing pretty well, they would switch to serial practice. And if they made too many mistakes, no biggie, they’d go back to a bit of blocked practice, before giving serial practice another try. 

But if they did well with serial practice, then they’d advance to random practice. And if they struggled with random, again, no big deal, they’d go back for some more serial practice, but with the goal being to eventually get to a high level of performance even with random practice.

The idea behind this type of adaptive practice was to try to keep the learner at a “sweet spot” of task difficulty, where optimal learning could take place. Where they weren’t getting bored because the task was too easy, but they also weren’t feeling overwhelmed because the constant switching between tempos made the task too difficult.

Wait! What’s serial practice?

Ok, but quick sidebar – what’s serial practice?

Ah yes! Serial practice is somewhere in between blocked and random practice. It’s like random practice, where you keep rotating between different passages during the same practice session, but instead of rotating between passages in a random order, you would rotate between them in a specific order.

So in the case of this study, instead of one rep of tempo A, then C, B, B, C, A, C, C, A, B, A, B, A, C, etc. all randomized, it would look something like AAAAA, BBBBB, CCCCC, AAAAA, BBBBB, CCCCC, etc.

24 hours later…

Then, to see how effectively their practice “stuck,” participants were asked to return to the lab 24 hours later, to be tested on the skills that they learned the day before.

And which type of practice led to the best learning?

The results

During practice…

During the practice session, the blocked group did a lot better and their timing was more spot-on, than either the learner-adapted or random groups (average error for blocked=41.55ms, learner-adapted=60.74ms, random=72.38ms – where lower scores, means better performance).

But as you can probably guess, it was a different story the next day, when some forgetting had set in, there was no real warm-up, and it was time to put their skills to the test.

The retention test…

When it came to retention of the skills they learned, i.e. how effectively practice stuck, the learner-adapted group performed the best – with an average error score of 47.78ms, vs. 62.61ms for random, and a whopping 102.60ms for the blocked group.

So why does learner-adapted practice seem to work so much better? At least for older adults?

Why?

Well, to keep this post from getting too long, I’m leaving out some details of the study that are actually pretty relevant and meaningful. Like how the researchers also looked at the effect of these strategies on motivation. And whether there was a difference in these strategies when it came to developing more flexible skills that were adaptable to new tempo patterns that they hadn’t practiced.²

But the most important thing I’m glossing over might be how the researchers looked at the differences between these strategies when learning similar skills, vs. dissimilar skills. Or in other words, when the task was easier vs. when the task was harder.

And the gist, is that there does seem to be something to this notion of a “challenge point,” where learning is best. Where if a task is too easy, and we’re getting it right every single time, then we’re not really learning much. But if a task is too difficult, and every repetition is a mess, then here too, we’re probably not learning much.

The idea is that one type of practice isn’t inherently “better” than another, but each is a tool that we can use to adjust the challenge point of our practice up or down. And keep our practice in that learning “sweet spot,” where things are challenging enough, that we’re making a few mistakes here or there, but not so many that we’re just reinforcing bad habits and getting discouraged. Kind of like adjusting the metronome to make things easier or more difficult for us.

So what are we to do with all of this?

Caveats

Well, before we get into that, a couple caveats. The main thing to keep in mind is that participants were asked to learn a pretty simple task. One that emphasized timing, rather than on the sort of small, intricate motor movements or coordination of multiple fingers/sides of the body that might be involved in playing an instrument. But still, conceptually, this seems to fit with other research in this area, so presumably it’d still apply to more complex skills as well.

The other thing to keep in mind, is that participants were between 60-75 years of age, so it could be that the results might be a little different if participants had been 6 or 16 years of age. But here too, there are other studies which have looked at this sort of thing with younger adults, with similar results, so it’s likely that the same principles would apply to younger learners as well.

The learner-adapted practice algorithm

So how can we apply these findings to practice?

Well, let’s take a look at a flowchart from the paper, which shows exactly what the learner-adapted practice looked like:

In terms of interpreting what this means, it might help to know that practice repetitions were organized in groups of 9 repetitions, and “good enough to advance” to the next level was a “batting average” of 66% or better.

So everyone started off with 6 blocks of 9 practice repetitions on tempo A (54 total practice reps). Then, they moved on to tempo B for 6 blocks of 9 reps, and then tempo C for 6 blocks of 9 reps.

If they were able to get to a level where they were getting “perfect”³ scores on at least 6 out of 9 repetitions in each block, then they would advance to Stage 2. If, by the end of Stage 2, they were only getting 5/9 repetitions or fewer correct, then they stayed in Stage 1 to solidify their skills a bit more.

But if they were able to get at least 6/9 reps correct, they advanced to Stage 2. And if by the end of Stage 2, they were able to get 6 or more perfect reps out of every block of 9 tries, they moved on to Stage 3. But if not, then they took a step back and went to Stage 1.5 to work on their skills a bit more, until they could achieve the necessarily level of performance to move on to Stage 2, and then, eventually, Stage 3.

Takeaways?

So could you just take this formula and apply it to your or your students’ practice exactly as is?

Well, sort of, I guess. But I think that might miss the point a bit.

Even though I think we’d all love it if there were just some neat and tidy formula that we could apply to every piece, passage, or excerpt, for me, the main finding of the study is that for learning to be maxed out, it’s important to find the right balance between the difficulty of a task, and our current skill level.

So when you’re just getting a new piece in your fingers, and you’re months away from a performance, sure, being within 5% of what you want an excerpt to sound like, and getting it right 2/3 times might be totally good enough to promote yourself to serial practice, and then random practice.

But at some point, when you’re closer to a performance, maybe a batting average of 2/3 tries isn’t good enough. Especially if you often get it wrong on the first try, and right only on the 2^nd and 3^rd try.

And at some point, allowing yourself wiggle room of +/- 5% whether it’s intonation, rhythm, or anything else, might be way too generous as well.

For me, the main takeaway is not the algorithm, per se, but the larger idea that it’s important for us to be mindful of our learning sweet spot when practicing. And to monitor our practice, so we’re neither making too many mistakes, nor too few.

And to use practice strategies like blocked, serial, random, or learner-adapted, to increase the difficulty of the task, either by allowing ourselves more consecutive repetitions to work something out, and get it sounding right more consistently. Or to interleave our practice repetitions a bit more, so we are upping the challenge, and pushing ourselves to practice getting things right with fewer warmup repetitions to feel things out and make those tiny adjustments that we won’t get when we’re on stage…

And where does deliberate practice fit into all of this?

Ah, good question! Today’s study was about how to structure practice repetitions.

But if you started to wonder what we’re supposed to do between or within the repetitions, then yes, this is where deliberate practice and self-regulated learning factor into the equation, and help guide the smaller moment-to-moment details of our practice.

And speaking of deliberate practice, if practicing has always felt like taking 2 steps forward, 1 step back. Or if you consistently put in the time, but feel like the results just aren’t there. Or if you’re familiar with the idea of deliberate practice, but have had difficulty figuring out how to actually put this into practice, there’s a course for that! If you’re looking for that sort of thing, the new Practice That Sticks course might be just the thing that could help put a few more pieces together:

Practice That Sticks – a quick and to-the-point course on smart, effective, practice

Reference

Beik, M., & Fazeli, D. (2021). The effect of learner-adapted practice schedule and task similarity on motivation and motor learning in older adults. Psychology of Sport and Exercise, 54, 101911. https://doi.org/10.1016/j.psychsport.2021.101911

1. I’ve since added a decade or two to that number, and reserve the right to revise it further, of course…
2. The short of it is that learner-adapted was best for motivation, and both learner-adapted and random were better than blocked for developing flexible skills.
3. Anything within 5% of the target tempo was considered correct. If their tempo was more than 5% faster or slower than the target tempo though, then it was considered an error.