Hi Steve,
yes, I spotted the problem meanwhile. On the first level, there were two
subjects with colinear regressors, rendering some of the contrasts invalid.
As I run all 37 subjects in batch mode by calling feat on the shell, this
evaded my attention quite some time. But as you pointed out, the
filtered_func_data had empty data points for those two subjects in the
respective cope directories.
Thanks for your help!
Cornelius
|