Hi FSL list,

When using a parametric modulator, is there a simple way to pull out an estimate of the weighting in the GLM?

I was initially using a 3 column EV in a feedback learning experiment. One EV is for a picture presentation (the cue), one EV is for positive feedback (post the picture), and one EV is for negative feedback (again, post the picture). 

My third columns for the feedback EVs are prediction error values derived from a computational model. I would like to examine the association with prediction error and BOLD response. Examining the FEAT contrast for positive feedback (or negative feedback) using a 3 column EV seems to combine both feedback and prediction error. In my brief experiences with afni and their amplitude modulation routine, you could pull out coefficient for the mean for the condition and a coefficient for the modulator. In FEAT, should I be using additional EVs? or ?

Any info or advice is greatly appreciated! Thanks much!


Jamie L. Hanson
Waisman Laboratory for Brain Imaging & Behavior | Child Emotion Research Lab
University of Wisconsin - Madison
1500 Highland Avenue
Madison, WI 53705
Email: [log in to unmask]