A behaviour that was shaky at the end of a session is frequently better at the start of the next one, with no practice in between. The gap is doing something.
What consolidation involves
A newly learned behaviour is initially held in a fragile state, easily disrupted and easily overwritten by whatever is learned next.
Over the following hours it is progressively stabilised, a process that depends heavily on sleep and continues without any further exposure to the task.
The behaviour therefore continues to improve while the dog does nothing. Performance measured at the end of a session understates what was actually learned.
Why sleep after training matters
Dogs allowed to sleep after a learning session perform better later than dogs kept awake and occupied for the same period.
Patterns of brain activity recorded during sleep resemble those recorded during the original learning, which is consistent with the material being replayed and strengthened.
The practical version is unglamorous: a short training session followed by a nap is more productive than the same session followed by a walk full of new experiences.
Interference between sessions
New learning arriving immediately after a session competes with material that has not yet stabilised, and it can degrade it.
Training two similar behaviours back to back — two cues that look alike, or two versions of the same movement — is the most reliable way to produce confusion in both.
Separating unrelated tasks by a rest period, and separating similar tasks by considerably longer, avoids most of this.
Why short sessions outperform long ones
Sustained sessions produce diminishing returns as attention drops and frustration accumulates, and the later repetitions are often worse than the earlier ones.
Ending while the behaviour is still good means the last repetition consolidated is a correct one. Working past that point consolidates the deteriorating version.
Several sessions of a few minutes spread across a day generally produce faster progress than one long block, for the same total time spent.
Why the first repetition next time is the real test
Performance at the end of a session is inflated by the immediate context, the warmed-up state and the recent history of reward.
The first attempt after a gap, in a slightly different setting, reflects what actually consolidated. It is the honest measurement.
Judging progress on that repetition rather than the last one of the previous day tends to produce more realistic decisions about when to raise difficulty.