¥ Can have activations with multiple phases that are not always
in the same time
relationship to each other
H e.g., solving a visually presented
puzzle:
² subject
sees puzzle
² subject cogitates a while
² subject responds with solution
¥ With fixed-shape regression, this isnÕt too hard, since we
can treat each phase as a
separate stimulus class and lay down a separate fixed-shape HRF
for each type of stimulus time
H But fixed-shape regression is probably
not reasonable for this problem, since (at
least) the ÒcogitationÓ phase isnÕt fixed among the trials
¥ Deconvolution assumes that the HRF (although unknown) is the same among all trials
in the same class