I'm trying to use AFNI tools to run psychophysiological interaction analyses, and I have a question about how to carry out one particular step in that process. I'm trying to deconvolve a hemodynamic response function from a time series to arrive at what amounts to an item-response function for stimuli onsets. The easiest way for me to illustrate what I'm trying to do may be to make reference to the deconvolution example in the latest 3dDeconvolve documentation, example 1.4.2.1. In this example there's a simulated time series created by convolving an IRF with and HRF:

The IRF: F(t) = [0 0 1 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0]

The HRF: H(t) = [0 5 10 5 2]

The resulting simulated time series (with a constant and slope added):

z(t) = [100 101 102 108 114 110 108 107 108 114 120 116 114 113 114 120 126 122 120 119]

Example 1.4.2.1 shows how to deconvolve F(t) from z(t) to get H(t), the original hemodynamic response function. My question: is there any way using AFNI tools to take z(t) and H(t) and recreate the original IRF, F(T)? If so, what's the command line syntax to do it?

Thanks,

John Herrington