regressor colinearity when build 3DD model

AFNI Message Board
1

Hello Gang!

I have two regressors of interested across trials, one is feedback, can be right(-1) or wrong (1), the other is prediction error, when it's right feedback it's zero, when it's wrong feedback its a continuous variable that ranges between 0 and 1, which varies across wrong trials.

I am interested in two effects, one is generic feedback, i.e. the contrast between right feedback and wrong feedback; the 2nd is prediction error. If i create the modulator separately for generic feedback (-1,1) and prediction error (0 for right trials and positive values range 0-1 for wrong trials). Those two modulators can be highly correlated. But for the effect of prediction error, i clearly do not want the component of generic error effect in it. What should I do regarding this?

Thanks!

all the best,
mengxing

2

Mengxing,

Could you provide details about the experimental design? Specifically:

  • How many trial conditions are there?
  • How many trials per condition?
  • What are the trial durations?
  • How are the trials arranged?

A visual illustration of the experimental design would also be helpful.

Gang Chen