Dear all,

In my dataset, there are a few runs in which not all types of events occurred. In order to make the scripting of the first-level analysis a little easier, I still included those runs, but added a single row of " 0    0    0" to the 3-column event file for those types of events. At a later stage of the analysis, I selected only the good ones.

When modelling the HRF with a double-gamma function, this procedure worked just fine. However, currently I am trying to use FIR basis functions instead of double-gamma to model the HRF, while keeping everything else the same, and now the models crashes right at the beginning.

I investigated a little and the script fails at the command "feat_model design" with the error message:

Warning: at least one EV is (close to) a linear combination of the others. You probably need to alter your design.
(Design matrix is rank deficient - ratio of min:max eigenvalues in SVD of matrix is 6.10232e-17)
 Contrasts involving these combinations will be set to zero.

F-test 1 isn't valid - each included contrast cannot be a linear combination of the others.

Is there a way that I can avoid the error and make feat model the run regardless? As only a subset of events were missing, I would prefer to not throw away the run and use the events that do exist in the higher-level analysis.

Thanks for the help!

Eduard