My experimental design for each trial is as follows:

Stimulus 1 (Predicts feedback to be delivered by Stim2)
Fixation1 (waiting for feedback given by Stim2)
Stimulus 2 (delivers feedback)
Fixation 2

Lengths of both Fixaion 1 and 2 are jittered. Since Fixation 1 is a time period where participants are waiting to receive feedback delivered by S2, it doesn't seem like this should be included as part of baseline. However, since Fixation1 always immediately follows S1 this is causing a problem for the deconvolution analysis.

S1 and S2 last 1 timepoint
Fix1 and Fix 2 last 0 to 3 or 4 timepoints

Since the anticipation fixation (Fix1) lasts for multiple timepoints is there a special way this should be set up in the stim_file (right now I have a 1 for every time point where the events occur and 0 where they don't).

When I run 3dDeconvolve with Fix1 as a regressor I get collinearity warnings. I assume this is because there are no baseline periods between S1, Fix1, and S2.

Any help or suggestions on how to solve these problems is very much appreciated.