Stop judging learning by today's performance

How well you do during practice is a bad predictor of how much you will keep.

Why it works

Bjork’s central distinction is that performance (how you do right now) and learning (durable change) can move in opposite directions. Conditions that boost immediate performance, like massed practice, often produce little lasting learning, while conditions that depress immediate performance can produce more. Judging learning by in-session ease therefore systematically rewards the weaker methods.

How to do it

1. Do not treat smooth, error-free practice as proof you are learning.
2. Measure learning by performance after a delay, on a different occasion.
3. Expect the better methods to look worse in the moment and judge them later.

Evidence

The learning-versus-performance distinction is well established: numerous studies show manipulations that improve practice performance can reduce long-term retention and transfer, and vice versa. (rct)

The dissociation is robust in many paradigms, but performance and learning are not always opposed; the point is that current performance is an unreliable proxy, not that hard always beats easy.

Sources

• Soderstrom & Bjork (2015), "Learning versus Performance", Perspectives on Psychological Science

Common mistake

Choosing the study method that produces the best results during the session, which is exactly the method most likely to fade fastest afterward.