If anyone is curious, the psychologists discovered this long ago; it's related to spaced repetition, but in education & sports & psychology, this is known as blocking: you get better skill gain by not massing practice on a single skill, but by regularly rotating through multiple tasks. (So in a baseball study, the players would rotate through batting, throwing, catching etc, instead of spending a long time at each activity.)
This seems to be a different thing than just practicing different activities related to the same sport. These are two functionally similar activities which use slightly different muscle patterns to accomplish, so that the trainee is directly applying their knowledge of the first action to the second, but can’t just do the precise same motion.
> “If you make the altered task too different, people do not get the gain we observed during reconsolidation. The modification between sessions needs to be subtle”
A better baseball analogy might be practicing catching line drives then practicing catching popups then practicing catching one-bounce throws then practicing catching a lobbed egg.
Or maybe practicing batting and then practicing golfing and then practicing kendo swings.
If I'm reading the source article correctly (linked below by Perixoog), it sounds like change was so subtle that the participants themselves weren't even aware of it:
"Participants were unaware of this manipulation"
"We chose to manipulate sensorimotor variability so that participants were unaware of the fact that there was any change in the task"
So not only are the tasks similar, but they're so similar that participants weren't consciously aware that they were doing something different. Wacky
They also mention that "participants strengthened skill through the re-exploration of sensorimotor space."
I wonder how large an exploration of the sensorimotor space could be while generalizing/improving the original skill. (re batting and kendo). No idea, but interesting article either way
Which makes me wonder if you can apply this as an autodidact. If you know you have made subtle changes, do the results still hold. I'd guess the answer would be yes but it could be an interesting problem.
There also might be a delta (variation of the activity) that optimizes the learning progress. My guess for that would be...change too little and you're inefficient, change too much and you are inefficient, too.
This is all speculation, on my part (certainly not an expert in this area)
In the article they say that "Contextual variability can strengthen retention [15] and generalization of skills", but that they "chose to increase sensorimotor variability while maintaining constant the original learning context."
They also mention that "attributing errors to internal sources can strengthen learning [19, 20] and generalization [21] of motor behavior"
My guess for an autodidact would be how powerful internal attribution is compared to the change in context and exploration of the sensorimotor space. Of course, they're probably not discrete, and probably also impact one another.
Sounds like some really interesting possible implications, and possibilities for further research. Really makes me want to reread the article and go through some of the articles they cite
> “If you make the altered task too different, people do not get the gain we observed during reconsolidation. The modification between sessions needs to be subtle”
I saw that quote but I have no idea what the researcher is basing it on. As described, the study can only show that blocking/interleaving is superior to massed practice:
> The volunteers were split into three groups, and each spent 45 minutes practising this. Six hours later, one of the groups was asked to repeat the same training exercise again, while another group performed a slightly different version that required different squeezing force to move the cursor. The third group only completed the first training session, so they could act as a control.
That is, as described, there is no 'interleaved subtle-difference' vs 'interleaved gross-difference' comparison, because they only compared 'near-zero-practice' vs 'massed-practice' vs 'interleaved-practice'. They seem to be wildly extrapolating from there, in ignorance of a good deal of prior work.
> practicing batting and then practicing golfing and then practicing kendo swings
Actually, as the subject progresses toward being an "expert" their practice should be increasingly specific. Beginners will certainly benefit from the approach you mention initially, but professionals will not.
Regular successful achievement of a physical task (reliably hitting a ball with a bat) can be considered a walk through a problem space. Initially, any training that approximates the task will take you closer to the target in the space. However as you approach that target, the margin for error diminishes sharply. Consequently your training should become more specific.
We got best improvement on our catcher by getting close enough to force a rapid response, or get hit by the ball, and a series of varied throws. High left, low right, center fast, high slow...
Take a bucket of balls and just run through them quick.
This seems to align with the study. Same basic task, lots of minor, well in our case occasionally major too, variations. The kids would improve notably and significantly in a single session.
I think it has more to do with subtle variation, not repetition in time. I haven't read the whole paper but just from the intro I got that one shouldn't robotically overtrain doing the same things, but should change the rhythm, syncopation, etc. Neural network overfitting. The same phenomena exist in physiological athletic improvement.
Just a quick correction (because it is a very common mistake). I think you are referring to "spaced learning" or "spacing", not "spaced repetition".
Every time you reinforce your memory, the amount of time it takes to forget it lengthens. Or, more properly, the slope of the curve that describes the probability of forgetting it decreases. That curve is known as the "forgetting curve". In spaced repetition, you are trying to time the repetition so that you still have a low percentage chance of forgetting.
"Spaced learning" or "spacing" is almost the opposite effect. What it says is that inserting a space in learning such that you forget something makes the material stronger when you subsequently relearn it. Although I am not aware of research that actually links spacing to the forgetting curve, you can probably imagine that if you learn something once, the forgetting curve is fairly steep (i.e., the probability of you forgetting it grows fairly quickly with time). With spacing, if you wait until you forget something and then relearn it, the forgetting curve is much shallower the second time (i.e. the probability of you forgetting it grows fairly slowly with time). In fact, it is apparently much shallower than if you had done spaced repetition (i.e. timed a reinforcement so that that allowed you to remember the item rather than forgetting it). Forgetting an relearning is rather an annoying and difficult way of learning, but some research suggest that it is dramatically more effective. This has been called "desired difficulty".
Also, there is a slightly nuanced difference between "blocking" and "interleaving". In interleaving, you fairly rapidly change topics and then loop back to the first again. So, if you have to learn 5 tasks, (A, B, C, D and E) you could spend 10 minutes on each and then loop back to the beginning. If you did that 6 times, you would have spent 1 hour on each task. Blocking would be spending 1 hour on each task, in blocks and then going back to it the next day. These are both different from the approach of training A until you learned it, then moving on to B, etc, etc.
In schools you can talk about "blocked" schedules (i.e. you try to fit in every subject every day -- not to be confused with the opposite of cramming a single subject into 3 or 4 periods, annoyingly also called a "blocked schedule"). Inside each subject they often don't interleave the material, though. There are some textbooks available that try to interleave the material within a subject. I have also read papers of some experiments where they actually tried to interleave all of the subjects (i.e., in each period you would have 10 minutes of each subject) with some success.
Anyway, I hope that helps. I spent a huge amount of time reading about this a few years ago, but have neglected it for about 3 years, so I hope I have not made any mistakes myself ;-)
No, I'm referring to spaced repetition, which is also often called spaced learning; the forgetting curve is why it works, but it is not otherwise relevant. The question about learning is how to distribute the reviews over time: all at once, or gradually over time with possibly expanding intervals. Similarly, with interleaving vs massed practice, the question is how to split up a fixed number of activities: should activity A be practiced for the maximum allowable amount of time, then B, then C and so on (analogous to massing on a single item), or should it be broken up with a few instances of A, then a few of B, and so on in a cycle until you've gone through all activities for the maximum allowable time? It turns out that for the same number of reviews or actions or exercises or time, spaced produces better recall than massed, and blocked/interleaved produces better skill learning than massed as well.
