Hi Govinda,
Another thing to think about is the fact that your regressors implicitly
assume that the shape of the BOLD response remains constant as a function of
RT. I think this is exactly opposite of the assumption you want to make in
order to estimate the activity per unit time. For very short durations of
neural activity, the shape of the predicted BOLD response is very similar to
the impulse response. However, as the durations get longer, the predicted
response significantly deviates from the impulse response. In other words,
your design matrix introduces a non-linearity into your estimate of the
response. The effect this will have on your interpretation of the data will
depend on the mean and shape of your RT distribution. I discuss this in
more detail in Neuroimage. 2008 Nov 15;43(3):509-20.
>> 2) I want to create a plot with % signal change in Y axis and
>> response duration in X-axis. I think this will require different EVs
>> for different response duration. Is there any other way of doing this?
If you want to do this in the GLM framework, then you should probably use
the FIR approach and split your trials into different groups corresponding
to different trial durations. This will avoid the non-linearity problem
above. Alternately, you can use Matlab to plot the mean BOLD response for
each response duration.
cheers,
jack
|