Š76Š
One Solution: Multiple HRFs
 By treating each phase of the activation that is temporally uncoupled from the other phases as a separate stimulus class, then can compute a separate HRF for each
H Using basis function deconvolution, since timing will be arbitrary
 This will only work if the fluctuations in timing are great enough so that the different HRFs donÕt always overlap in the same way
H Otherwise, it becomes impossible to tell the tail end of HRF #1 from the start of HRF #2, if they are always locked together in the same temporal relationship