¥ Instead of inputting a signal model time series (e.g., created with waver and stimulus timing), you input the stimulus timing
directly
H Format: a text file with 0s and 1s, 0 at TR-grid times with no stimulus, 1
at time with stimulus
¥ Must specify the maximum lag (in units of TR) that we expect HRF to last
after each stimulus
H This requires you to make a judgment
about the activation Ñ brief or
long?
¥ 3dDeconvolve returns estimated
values for each bq for each
stimulus class
H Usually then use a GLT to test the HRF
(or pieces of it) for
significance