I am doing an experiment with two brief (~100ms) stimulus conditions I will call Pulse and Prepulse&Pulse. Besides these two stimuli which are part of an event-related design (100 events of each trial type over four ten minute runs) and whose lag is 6TR, I wish to include in this model two additional "conditions" as regressors which are EMGPulse and EMGprepulse&pulse (i.e. the intensity of the eyeblink response to the two stimuli for the 200 events). These regressors represent behavioral EMG data which were collected at the same time as the fMRI in the scanner. We wish to include this data into the model somehow to "relate" this data with the fMRI data and find any correlations and to improve the original model. The question is how do we do this?
Since the EMG peaks last less than a TR, I think we are allright in using only the amplitudes of the peaks at the stimulus onsets in the current .1D files for the "new" .1D files of the EMG regressors. Also the amplitudes I think should be normalized on the same scale for both EMGPulse and EMGPrepulsepluspulse. I believe to use waver is not necessary here but I maybe wrong. Even though the Pulse and PPI conditions have 7
coefficients each, I think one can add the two EMG regressors without any lag. This maybe equivalent to adding motion regressors to the model. The question remains is this the "right" way of adding the EMG regressors?
What we would ultimately like to identify are 1) the brain areas where the pulse F stat is significantly increased by the addition of information about the intensity of the eyeblink response (EMGPulse). The hope is this will correspond to the brain areas that are specifically associated with producing eyeblink rather than any other aspect of the sensory-motor activity; and 2) the brain areas which are specifically associated with inhibition of the eyeblink in the PPI condition (is this equal to (prepulse&pulse- EMGPrepulseplus pulse) - (pulse - EMGpulse). If not
how would you make a "new" Pulse-PPI GLT - which accounts for brain areas that are significant for this difference?

Thanks,
Linda Heidinger
Research Associate
University of Chicago Hospitals