Variation, motor learning, and educational psychology

Training Q&A [Moderated]
1

This is for Dr. Baraki, following up on a comment in the “Understanding Hypertrophy” thread.

In the motor learning and educational psychology literature, practicing a main skill plus close variations together is often more effective for skill acquisition than just practicing the main skill in isolation.

Most of the the studies have the following form: they take a few closely related skills and compare practicing them separately (in blocks) to practicing them in rotation (interleaved). Almost universally, they find interleaved learning is superior. From badminton serves, to putting in golf, to practicing clarinet, to med students reading EKGs, the interleaved group gets superior results with the same amount of practice. The effect replicates very well, applies to basically any area of learning (both motor skills and more cognitive skills, like the EKG example), and is usually large enough to be practically significant.

Here are some fun examples. You can find over a hundred more, if you look.

Badminton (old, but historically important, I think):
http://www.tandfonline.com/doi/abs/1…nalCode=urqe20

Music (full text available):

Training med students:
https://link.springer.com/article/10…A1022687404380

Here’s a meta-analysis of 139 studies:
http://journals.sagepub.com/doi/abs/…s.99.1.116-126

The theoretical justification for this is the “contextual interference” that comes from practicing different skills in quick succession. On the most deflationary reading, the explanation is just that switching tasks frequently makes you maintain focus better. This is consistent with the finding that in the short term the mixed group performs worse (because they’re confused by the task switching) while in the long term the mixed practice group has better skill retention. (This is sort of analogous to how spaced repetition versus massed learning works in educational psychology, if you’re familiar with that. Given an equivalent amount of practice, massed learning is better for “peaking” short term memory, while spacing out the study of a subject is better for long term retention.)

Some further possible explanations are discussed here:
https://www.sciencedirect.com/scienc…6794579090005X

In any case, it works, it works well, and it works in every domain of learning that’s been studied, as far as I can tell.

There’s also some research on blocked practice of variant skills versus constant practice of a single main skill. I am less familiar with this but I think there is still some edge to the blocked practice group.

We then have a rough hierarchy of the efficacy of various skill learning methods:

  1. Mixed practice (main and variant skills practiced in an interleaved manner)
  2. Block practice (main and variant skills practiced in separate blocks)
  3. Constant practice of a main skill

I propose that this literature shows you an easy way to make your programming of variant lifts more effective. It’s still somewhat unclear to me why introducing variant lifts into a program “works” and why people like RTS find them so effective (compared to constant practice of S/B/D for powerlifting purposes). I think this reasoning suggests it “works” in part because of the motor learning benefits associated to going from (3) to (2) on that hierarchy. Maybe you don’t buy that explanation, and that’s fine and not really relevant to my main point. However, if you believe variants should be in your programming, you need to pick between (1) and (2). Currently your templates use approach (2). For example, paused benching on one day, and TnG benching on another. Instead, I think you should do both on the same day.

For example, a reasonable workout might be 5x3 at 80% on TnG bench. Instead I think you should be doing something like:
3 at 80% TnG
1 at 80% 2 count pause,

and repeat this pairing 4 times. It’s the same amount of “work” but the motor learning benefits should be greater. And since strength is largely a skill, I’d guess it would make you stronger.

I realize this is a pretty crazy idea, but here are three reasons you should have faith in it. First, in the single study I can find where this approach has been examined in the context of strength training, it (edit: something like it, using multiple variations on the same day) worked:

Second, while there is very little strength specific evidence for this proposal, the effect is highly generalizable. Mixed practice seems superior for every skill where it’s been studied, across a very large number of disciplines. This strongly suggest mixed practice would be effective here, too.

Third, you routinely make programming decisions based in part on the terrible, small sample, noisy, often contradictory exercise science literature. If you take that stuff seriously and let it inform your training, you should definitely take this seriously, since it replicates like gangbusters and the evidence base is as good as psychology is capable of producing.

I’m very interested in your thoughts on this.

Thanks,
Patrick

2 Likes
2

Hi Patrick,

Thanks for the helpful explanation and overview of this topic, rather than just sending me a bunch of papers to read :slight_smile:

This is interesting stuff. We will need to digest it a bit longer.

1 Like
3

Very interesting and well articulated consideration.

Just to be sure (as I’m unfamiliar with the term interleaving in this context), in your example:

For example, a reasonable workout might be 5x3 at 80% on TnG bench. Instead I think you should be doing something like:
3 at 80% TnG
1 at 80% 2 count pause,

and repeat this pairing 4 times. It’s the same amount of “work” but the motor learning benefits should be greater. And since strength is largely a skill, I’d guess it would make you stronger.

This would be 3 reps TnG and 1 rep 2 count pause all at the same weight as a single “set,” and then repeated for a total of 4 “sets?”

I.E. “interleaving” as suggested in option “1” would be to do the variants in the same set as the main skill lift. Or perhaps the mechanics of how to do “interleaving” optimally still need to be worked out shrug

Also, I’m curious as to your background. Med student/practitioner? Coach? Competitive athlete?

4

Thanks for posting this and your hypertrophy questions Patrick, it is stimulating some very interesting conversations. I’ve always been interested in the science of learning and have some experience in using these practice models in different contexts including golf, olympic lifting and most recently learning how to type at a half way descent speed.

What I have learned as a general rule is that rate of learning correlates to the degree to which the practice environment (random, blocked or constant) stimulates an active problem solving process of the participant. The more active problem solving the environment requires of the participant the more they end up learning as long as they can accomplish these processes (the learning threshold is not so high as to prevent any success if the outcome is objectionable).

With golf practicing at the driving range this meant taking no more then 2 shots aiming at a specific target with the same club before switching to another club. I would use the feedback of the ball flight of the first shot and make a minor adjustments on the next shot at the same target before switching clubs and repeating the process.

With Olympic weightlifting I would alternate between doing a set of snatches and a set of clean and jerks. I would hit a near max effort double on a snatch and then the next set switch to a near max effort clean and jerk. When I went back for the next set I had to remember why I might of missed a rep on the previous set and come up with a planned adjustment to succeed on my next attempt. I did this for around 2 months and it would be pretty difficult for me to find the data, but I believed it was working. The problem was the strategy was impractical (too much plate changing).

With the idea your presenting for interleaving I’m not sure it would work the same way. In both practice scenarios above I as the participant was using feedback to make technical adjustments, or changes to my motor plan on a following attempt. The practice strategies I was implementing forced a greater cognitive demand on the execution of a skill.

I’m not saying the above implementation would not work, and I am all for experimenting with new ideas to get stronger (this is basically what I’ve been doing my entire professional life), but I think you would need to create an environment that forces the lifter to go through an active problem solving process. If I am doing a complex that I know the exercise order and have warmed up and grooved what I’m doing before my work sets I think its closer to a slightly different form of constant practice on the practice continuum.

5

Thanks for your reply, Nick. I tried my proposal today and it appeared to work exactly as you suggest it might.

Specifically, I did sets of TnG bench press at about 80% alternated with 2 count paused singles, also at 80%. The TnG sets after the first paused single were actually faster than the first. Instead of using my sternum as a trampoline, I actually had to “think” about what I was doing at the bottom and while pushing off my chest (e.g. getting the bar moving toward my face). “Think” is perhaps not the right word here – my higher cognitive functions, to the extent that they exist, shut off while benching – but there was definitely some sort of motor pattern error detection and correction going on. I usually spend only a split second in the bottom few inches of the movement, so it was a chance to learn how to perform that part efficiently. I think this might be another way to say what Mike T says when he suggests, for example, paused benching and paused squatting to people “weak” out of the bottom of those lifts. He calls it strengthening a weak range of motion, I call it optimizing the performance of a part of the lift you usually blow through without thinking.

Greg Nuckols makes a very similar point in his article “Double-Paused Deadlifts” for T-Nation: “The ‘magic’ behind double-paused deadlifts is they quickly clean up deadlifting form, and a lifter with better technique is a stronger lifter.”

Except, with both Mike T and Nuckols, the variant lift is done on the a separate day from the main lift. Imagine how much more effective that technical feedback would be when the variant lift and the main lift are the done on the same day, in an alternating fashion…

This is a good point. I think what I suggested is actually called “serial” practice in the literature, and as you mention this is shown to be less effective than randomized practice. But the fix is easy: randomize the exercise with a coin flip immediately before the set. I’d guess a single 2 count paused rep and a double or triple at 80% TnG have about the same “recovery cost,” so it shouldn’t be a problem. So do 5-6 sets with the exercise determined randomly. Now we’re getting really nuts, but hey, the science says what is says.