Variation—practise under variety and interference#

Here is one of the most counter-intuitive findings in skill learning: the practice that feels smoothest is often the practice that teaches you least. If you want a skill to hold up later—in the exam, on the runway, under pressure—you usually have to make practice a bit harder now. Two ways to do that are to vary the task and to add interference. Let’s look at both.

Vary the task. Perform the skill with deliberate variations in context: change the starting point, the method you use to reach the objective, or the objective itself. Practising aircraft landings, you might come in too high or too low, with the engine at idle, or onto a short runway. Practising docking a yacht, you might do it under sail or under engine, with help or without.
Add interference. Perform the same task, but with elements that make it harder without changing how the task is done. For the landing or the docking, that might be rain, a low sun, or time pressure. The task is unchanged; you just have to do it while coping with something.

These are the hands-on, motor-skill version of a principle you may have already met as interleaving—see Interleaving . Researchers call it the contextual-interference effect: jumbled, less predictable practice tends to depress your performance during practice while improving retention and transfer afterwards (Schmidt & Bjork, 1992). It is one of the family of “desirable difficulties”—see Desirable difficulties —where in-session ease is a poor guide to lasting learning.

Why harder practice can help#

When you do the same landing over and over in calm conditions, you settle into a groove. After the first few, you are barely solving the problem any more—you are repeating a movement you already have warm and ready. It feels like mastery. The feeling is misleading.

Varied, interfered practice breaks that groove. Each repetition is a little different, so you cannot coast: you have to set up the solution afresh, choose the right response, and adjust. That extra effort is doing real work. The leading explanation is that you process the task more deeply and rebuild the skill from memory each time, rather than keeping it idling—which is also why it transfers better to the messy, unpredictable conditions of the real thing.

The classic laboratory demonstration is decades old: learners who practised several movement patterns in a random, mixed order performed worse during practice than those who practised them in neat blocks—yet did better on a later retention test (Shea & Morgan, 1979). Same movements, same number of trials; only the order changed.

How strong is the effect—really#

I’ll be straight with you: this is a real effect, but it is not a guarantee, and the older Memletics text oversold it. Three caveats are worth holding onto.

The size depends on what you’re practising. The closest large review—a meta-analysis of interleaving, the cognitive sibling of contextual interference—found an overall benefit of medium size, but with big variation from task to task: substantial in some domains, small or absent in others (Brunmair & Richter, 2019). Treat “varied practice helps” as a reliable direction, not a fixed payoff.
It helps most when the variants are confusable. Variation earns its keep when you have to tell similar cases apart and pick the right response. Mixing things that are unrelated gives you little to discriminate, and little benefit (Brunmair & Richter, 2019). Vary the kinds of landing you might actually face—not random unrelated drills.
Beginners may need a gentler dose. When a skill is brand new, too much variety or interference can swamp you and stall learning rather than deepen it. The evidence here is mixed, but a sensible reading is to build a basic, stable version of the skill first, then ramp up the variation and interference as you become competent.

Keep it in context, too. Use variations and interference drawn from situations that are actually likely during or after your training. Landing in real turbulence is the right kind of difficulty; reciting lines from Romeo and Juliet mid-approach is just noise. The difficulty has to be the kind your performance will later face.

What to expect when you try it#

Don’t be alarmed when your scores dip. Suppose landing accuracy is scored on how close to the runway threshold you touch down. While you practise varied and interfered scenarios, your scores will probably be lower than if you just did ordinary landings—that dip is the cost you are paying on purpose. The reasonable expectation is that, once you’ve done the work, you’ll hold up better than someone who did the same number of easy, identical landings, especially when conditions get awkward.

Notice the careful framing there: tends to, probably, better than easy repetition. That is the truth of it. Varied practice is one of the best-supported ideas in skill learning, and also one of the easiest to overstate—so use it confidently, but judge it on how you perform days later under real conditions, not on how smooth it felt today.

Takeaway: To make a skill last, train it the way you’ll use it—varied, occasionally awkward, never on autopilot. Expect to look worse in the session and better afterwards. Lean on confusable, in-context variations, ease beginners in gently, and measure success by later, real-world performance rather than in-the-moment smoothness.