Hi David,
Just to elaborate on what Donald said, there is a followup paper with the same authors (http://www.ncbi.nlm.nih.gov/pubmed/22659443) that has a slightly different version of Least Squares- Separate, where the trial of interest gets its own regressor, the other trials of that condition get their own regressor, and each of the rest of the conditions gets its own regressor as well. In the original paper, there are just two regressors in the model- one for the trial of interest and one for all of the trials of all of the other conditions.
There is also this paper by Cisler (http://www.ncbi.nlm.nih.gov/pubmed/24055504) that compares gPPI and beta series functional connectivity for different experimental designs, using both simulated and real data. They found better results with beta series for event-related designs and better results with gPPI for block designs.
- Taylor
|