Neuropsychology · Simplified
What One Score Hides
Two people with the same executive score can have nothing in common.
The model
Four capacities, three tasks, one number
No task measures one thing. Each of these three draws on the four capacities in different proportions, so moving a capacity moves some tasks and not others.
Key terms
- Inhibition
- Stopping a response that is ready to go but wrong here.
- Switching
- Dropping one rule and taking up another.
- Updating
- Keeping something in mind and revising it as things change.
- Planning
- Working out a sequence of moves before starting.
The upper bars are the three task scores. The lower bars are the capacities you set. A task score is a weighted mixture of the capacities, so no task score is a measure of any one of them.
What this shows
A composite is an average of things that differ
Key idea: Every executive task is impure. It needs several capacities at once, plus everything else a task needs, so a low score says the person struggled without saying with what. A composite made by averaging several such tasks is worse rather than better for that purpose: averaging is exactly the operation that removes the differences between them. What identifies the problem is the pattern across tasks that load differently, which means the tasks have to be chosen so their profiles of demand differ. Two people with the same composite and opposite profiles need different things done for them, and the composite is the number least able to tell you that.
The weights here are invented and the model is a caricature: real tasks do not decompose into four tidy capacities with fixed weights, the capacities themselves are inferred from performance rather than observed, and how separable they are is an active argument rather than a settled fact. Motivation, fatigue, speed and comprehension move every one of these scores together in ways nothing here represents. Treat the model as a way of thinking about impurity, not as a description of any person or any test. Nothing here is an assessment instrument.
The longer version adds two more capacities, a fourth task, a prediction stage on which processes each task requires, and the weights revealed after the prediction. It is at Executive Function Task Laboratory in the main collection.