Neuropsychology

Recovery and Plasticity Simulator

Four different things all look like getting better. Watch five simulated trajectories fan out, then take one trained score apart.

Simulated - an invented model with a visible seed, not data, not a prognosis and not a prediction about anybody

Learning objective

By the end you should be able to name the processes that all raise a score after injury, and say why a trained score cannot separate them.

About 30 minutes, in two experiments. Nothing you do here is saved or sent anywhere.

Numbers from a model, not from people

Every curve here is produced by an invented equation from a visible seed. None of it is data, none of it is a norm, and none of it forecasts what would happen to any real person. The model exists to make a logical point about what a score can and cannot distinguish.

Recovery is not guaranteed, is not proportional to effort, and does not continue indefinitely. Some of the simulated trajectories plateau and some drift downwards, because real ones do. Nothing here is a promise about outcome or a prognosis for anybody.

Experiment 1. Set an injury severity and a rehabilitation intensity, and look at five people rather than one curve.
Turn the measurement noise up and see how much of a "change" is not one.
Experiment 2. Take one person's trained score apart, and compare it with a task they have never practised.

This part needs JavaScript

The trajectories are computed and drawn in your browser, so this section cannot run with JavaScript switched off. The rest of this page is readable without it.

1 Five people, not one curve
2 What is doing the work

Need a hand?

Key terms

Spontaneous recovery: Improvement in the period after an injury that happens without treatment, as acute effects subside.
Diaschisis: Reduced functioning in undamaged tissue connected to the injury. Its resolution is one of the things that lifts a score.
Restitution and compensation: Getting the original way of doing the task back, and reaching the same outcome by a different route. Both raise a trained score.
Measurement noise: Variation in a score between occasions that is not a change in the person.
Transfer: Whether an improvement also shows up on a task that was never practised.

When you measure

Moves the marker on the chart and drives every number in the readout.

The injury and the rehabilitation

Noise, and the other four people

Measurement noise

How much the same person's score moves from one testing session to the next, for reasons that have nothing to do with recovery.

The five people

Five simulated people on the trained task

The numbers

The four processes, and what tells them apart

Each process, what it changes, its time course and the evidence that would identify it
Process	What changes	Time course	What would identify it

A fictional rehabilitation service reports that scores on its training task rise by an average of 22 points over six months, and concludes that the programme restores function.

What evidence would distinguish genuine functional recovery from an improved strategy for that task? A larger rise on the same training task Performance on an untrained task of the same ability, an alternate form of the trained test to remove practice, a group who did not receive the programme, and observation of how the task is now being done A follow-up scan showing reorganisation The person's own report that things feel easier

Restitution, substitution, compensation and environmental support produce curves a trained score cannot tell apart: the same number at twelve months can be reached four different ways. An untrained task is the cheapest thing that discriminates between them, which is why a recovery claim needs more than the measure the therapy trained.

Recovery and Plasticity Simulator

Learning objective

Numbers from a model, not from people

First, a judgement

The simulator

Experiment 1 - five people, not one curve

Five simulated people on the trained task

What the trained score is made of

Reading this

The trained task and a task never practised

The numbers

Challenge - what would settle it

What this demonstrates