Thanks Gang,

I think I'm not understanding something very fundamental about 3ddeconvolve here-- does it require a model function to be able to deconvolve? Is it actually deconvolving, or just fitting or extrapolating?... What sort of method is it using to actually "deconvolve"?

From my understanding of deconvolution, we have a y(t), which is our time series, and a x(t) which is our stimulus (the ones and zeroes). After deconvolving the two, we should get a h(t) which is our hemodynamic response function (imperfect due to noise, of course), which should be length of the length of y minus the length of x.

If 3dDeconvolve truly does deconvolution, why does it require a pre-set model HRF? Just to test 3dDeconvolve, I inputted the time series for a voxel as a 1D input in 3dDeconvolve, threw in my stimulus times with a model gamma function, and it required me to pre-set the parameters for the gamma function.

If I'm pre-setting parameters for a model function, is 3dDeconvolve actually performing deconvolution at all?


-Alice