Dear SPM list,
I didn't hear back about my earlier question, so I'd like to try to
clarify what I'm asking. I have an event-related design with 4 types of
events. I'd like to do a 1 1 -1 -1 contrast on the predictors for these
event types. The trouble is that my events are close enough together in
time that I think there is nonlinearity in the BOLD response. That means
that my regressors, which are based on a linear, additive response, won't
be good models for the neural response.
How can I estimate this contrast, but take nonlinear effects into account?
I thought one way to do this would be to use the Volterra series
expansion, but I'm not sure how this would work. Is it right that the
diagonals of the H matrix (which are the squared betas from the linear
component?) model 'saturation' effects, and that coefficients of
interactions between trials are modeled by the off-diagonal elements of
the H matrix? If so, when creating contrasts could I do something like
entering a 1 for the linear effect and a 1 for the 'saturation' effect for
one trial type, vs. -1 for the linear and -1 for the saturation effect of
another trial type?
Alternatively, is there a way to use the coefficients from the Volterra
analysis, estimate a 'saturation' factor, and rebuild a linear model
taking into account saturation effects? I could do this if I knew where
to find the estimates - but would it be a valid procedure?
Any help on this issue would be very much appreciated, as I am confused
about this!
Thanks,
Tor
_____________________________
Tor Wager Department of Psychology University of Michigan
Cognition and Perception Area
525 East University
Ann Arbor, MI 48109-1109
Office: 734-936-1295
Home: 734-995-8975
Email: [log in to unmask]
_____________________________
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|