1.: Use "parametric modulation." You can read about it here on the listserv, or in some SPM manual. It goes back at least as far as SPM99.
2.: Really depends on your model of what the brain is doing; in that sense it's up to you, your understanding of the literature, etc. The one technical issue with exponentials is that the decay constant has to be fixed (in advance) in a linear regression---allowing it to be estimated would involve non-linearities.
|