I've been playing around with creating impulse response functions using the TENT response model. I have a question about the following warning that I have seen. I'm not sure what the consequences are
...
*+ WARNING: -stim_times 1 .. TENT(-3,18,22) has inter-knot TR=1 but -iresp output TR=3
...
I believe this stems from the fact that I have a 3 s TR (data sampling) and am trying to model the response at a higher resolution (1 s). Although I get the warning, 3dDeconvolve still produces reasonable response functions. What is the real issue here? Am I overfitting the data? Are the statistics associated with each TENT coefficient over-estimating significance? Perhaps because the same raw data is contributing to so many of the basis function estimates. Are the coefficient estimates corrupted by this oversampling? And finally, if I really only care about the shape of the response (e.g., to pull out the "peak" over some range), is this warning a problem?
Thanks,
Ryan