Dear AFNI minds,

A recent paper in Neuron (40:1251-1257, J Jensen et al) used an event-related design with general linear modeling to assess and detect anticipatory activation in ventral striatum in response to visual cues predicting electric shock after a 5 s delay(CS+) contrasted with cues predicting no shock (CS-). This would be a fairly straightforward AUC linear contrast matrix with AFNI, e.g. "0 0 0 0 1 -1" or whatever.

However, in a second experiment, they sought to pinpoint whether this activation was more a function of cue stimulus onset versus offset by modeling (the HRF) to the onset time and to the offset time for/within each of the two cue types, i.e.

(CS+ onset - CS+ offset) - (CS- onset - CS- offset) p. 1252, right

How, in the conext of the methods of AFNI deconvolution 1.4.4.6 AUC differences (page 47), would one come up with a matrix file to do the above higher-order contrast? My task includes two events *within each trial type* that I would like to similarly contrast with each other before contrasting that contrast with the analagous contrast in another trial type, here:

(RESPONSEyellow - CUEyellow) - (RESPONSEgreen - CUEgreen)

My modeling currently generates the idealized HRF waveform for each of the two events (cue and response) in each trial.

Jim B