Tripling the practice changed nothing: when to stop drilling

In a training, the moment someone solves the exercise correctly, the room wants another one like it. I have run that extra block many times: six more problems on the same pattern, everyone leaving the room fluent. Doug Rohrer and Kelli Taylor tested that instinct across two experiments with 216 college students. The extra reps bought nothing a month later.

Nine problems scored like three.

In their 2006 study, students learned one procedure for permutations with repeated items, then worked either 3 or 9 practice problems in the same sitting. Everyone came back for a test 1 or 4 weeks later. The 6 extra problems moved scores at neither delay ¹.

The earlier experiment from Rohrer, Taylor, Harold Pashler, John Wixted and Nicholas Cepeda shows the shape of the decay. 218 college students learned geography facts in one experiment and word definitions in the other. The vocabulary group cycled through the pairs either 5 times or 10 times. At the 1-week test the overlearners recalled far more. The gap then shrank fast: in the geography experiment the difference held at 1 and 3 weeks and had vanished by 9 ².

The meta-analysis is friendlier because its tests came sooner.

James Driskell, Ruth Willis and Carolyn Copper pooled the overlearning literature in 1992 and reported a moderate effect on retention, d = 0.75, with the size depending on how much extra practice was given, the type of task, and how long the wait before testing ³. That looks like a win for drilling. No study in the pool tested anyone beyond 4 weeks. The warm verdict was measured inside the window where extra reps still hold.

The same reps, moved a week.

The other half of Rohrer and Taylor's 2006 paper takes those same practice problems and changes only their calendar. One group worked 10 problems in a single session. The other worked 5, waited a week, then worked 5 more. Identical problem count, identical total minutes of practice.

At 1 week the two groups were hard to tell apart. At 4 weeks the spaced group scored 64% against 32% for the massed group, d = 0.86 ¹. The calendar did the work, with the same 10 problems and the same total minutes.

Drilling pays when the skill has to survive the next one.

Kazuhisa Shibata, Yuka Sasaki, Takeo Watanabe and colleagues trained volunteers on a visual discrimination task and kept training them past the point where performance plateaued. The overtrained group held on to the first skill while learning a second one afterwards, and magnetic resonance spectroscopy showed a shift toward inhibitory, GABA-dominant processing in early visual areas ⁴. Extra reps do something spacing does not: they lock a skill against interference from whatever you teach next.

This was a laboratory perception task rather than a sales workflow, so I read it as a reason to pick one skill for the drilling budget. The other clean case for drilling is a deadline inside the week, which is about as long as overlearning holds ²³.

The protocol.

1. Stop at 3 correct cold recalls. Katherine Rawson and John Dunlosky ran 533 students through practice schedules that varied the initial criterion from 1 to 4 correct recalls and the number of later relearning sessions from 1 to 5, then tested 1 to 4 months out. Their recommendation: 3 correct recalls in the first session, followed by 3 widely spaced relearning sessions ⁵. The initial criterion mattered a lot on its own and much less once relearning was in place, so extra reps in the first session buy what the follow-ups already deliver.

2. Count the reps you were about to add, then move half of them. In a 2-hour session, that means stopping the pattern at 3 clean solves and putting the rest in a calendar invite 7 days out ¹.

3. Book the relearning sessions before you leave the room. 3 sessions, spread wide, 20 to 30 minutes each. In Rawson and Dunlosky's schedules the relearning sessions carried the long-term retention ⁵.

4. Make every rep a cold recall. Notes closed, screen clear of worked examples. A rep with the answer in view does not count toward the 3.

5. Give one skill the overlearning budget. Choose the one that has to keep working while the team learns the next tool, and drill that one past the plateau ⁴. Everything else gets the stopping rule.

6. Test at 4 weeks, cold, with 5 problems. Rohrer and Taylor's groups looked alike at 1 week and sat 32 points apart at 4 ¹. An end-of-session check sees none of that, and a smile sheet signed by a confident room sees even less.

The day gets thinner when you design it this way: fewer exercises on the pattern, more empty space in the afternoon. 3 clean solves, then the calendar invites, then a cold test at week 4 with 5 problems, no notes, and a written score next to the date.

Sources.

1. Rohrer, D., & Taylor, K. (2006). The effects of overlearning and distributed practise on the retention of mathematics knowledge. Applied Cognitive Psychology, 20(9), 1209-1224. doi.org/10.1002/acp.1266
2. Rohrer, D., Taylor, K., Pashler, H., Wixted, J. T., & Cepeda, N. J. (2005). The effect of overlearning on long-term retention. Applied Cognitive Psychology, 19(3), 361-374. doi.org/10.1002/acp.1083
3. Driskell, J. E., Willis, R. P., & Copper, C. (1992). Effect of overlearning on retention. Journal of Applied Psychology, 77(5), 615-622. doi.org/10.1037/0021-9010.77.5.615
4. Shibata, K., Sasaki, Y., Bang, J. W., Walsh, E. G., Machizawa, M. G., Tamaki, M., Chang, L.-H., & Watanabe, T. (2017). Overlearning hyperstabilizes a skill by rapidly making neurochemical processing inhibitory-dominant. Nature Neuroscience, 20(3), 470-475. doi.org/10.1038/nn.4490
5. Rawson, K. A., & Dunlosky, J. (2011). Optimizing schedules of retrieval practice for durable and efficient learning: How much is enough? Journal of Experimental Psychology: General, 140(3), 283-302. doi.org/10.1037/a0023956