Guido,
If I understand it correctly, SPM's PSTH function creates a new basis set
in which each time point after the stimulus is modelled as a separate regressor, and the graph plotted by
PSTH is the betas associated with each of these regressors (so, for
example, if on the 3rd timepoint following each
stimulus there is a lot of activation, this will show up in the betas and
become a 'bump' at dp=3 in your PSTH).
I think it's an important question how we model activity which precedes
the stimulus in functional imaging experiments. One way is obviously to
model a pre-stimulus cue (if you have one), but then it is difficult to
dissociate activity
evoked by the cue from that anticipating the forthcoming stimulus. Another
possibility might be to model the immediate pre-stimulus period with a new
set of regressors alongside your standard stimulus-evoked regressors.
However, you would have to think carefully about which basis set to
choose. For example, I don't know if pre-stimulus activity would
necessarily take the form of the canonical HRF. I don't know of any
studies which have explicitly tried to do this, and would be very
interested to hear from anyone who does -
best
Chris
Christopher Summerfield
[log in to unmask]
On Mon, 13 Mar 2006, <Guido> <Hesselmann> wrote:
> Dear SPMers,
>
> I would like to see the PSTH for event-related responses at NEGATIVE
> time points, ie before the onset of the event. How I can convince the
> spm_graph function of SPM5 to do that?
>
> Any help would be highly appreciated.
>
> Guido Hesselmann
>
|