Neuropsychology · Simplified
What Is the Score Actually Made Of?
The score went up. That is four different things, and only one of them means the damage has mended.
Two years of recovery
One rising line, four contributions
A fictional patient is tested every few months for two years on a task they practise in rehabilitation. Their score climbs. Change how much each of the four things contributes and watch the same total arrive by different routes.
Key terms
- Restitution
- The damaged function itself working better. This is the only one that would show up on a task nobody practised.
- Relearning
- Getting better at this particular task through doing it, without the underlying function changing.
- Compensation
- Achieving the same result a different way, such as using a notebook instead of remembering.
- Practice effect
- Improvement that comes from having taken the test before.
The upper line is the task they practise, built up from the four bands beneath it. The lower dashed line is a task of the same kind that they never practised, which can only rise by the restitution band. The gap between the two lines is everything that did not generalise.
What this shows
Improvement is not one thing
Key idea: Two patients can show the same rising line and be in completely different positions. If most of the rise is restitution, the underlying function has improved and it will show up anywhere that function is needed. If most of it is relearning, compensation or practice, the person has got better at this task and the gain stops at its edges. Nothing about the trained score distinguishes them, which is why an untrained task of the same kind is worth more than another session on the trained one. None of this makes compensation second best: a person who manages their week with a notebook has solved their problem. It matters because it changes what you predict, what you offer next, and what you tell a family who are asking whether the recovery will keep going.
The four contributions are drawn as separate bands that add up, and they are not separable like that in reality: relearning and restitution interact, compensation can itself drive neural change, and practice effects are not constant across the two years. The time courses here are illustrative shapes rather than measured ones, and real recovery is noisier, less monotonic and much more variable between people. The untrained task is also an idealisation: a genuinely untrained task sharing the same underlying function is difficult to find, and any second task carries some practice effect of its own once it has been given twice. Nothing here describes a real patient or predicts any real recovery.
The longer version adds five simulated people to compare, staged experiments on each contribution separately, and a treatment of what a plateau does and does not mean. It is at Recovery and Plasticity Simulator in the main collection.