Hello André,
in case you are not interested in the contrasts "No-Go" (e.g. [1]) and "Go" (e.g. [0 1]) itself, simply build the contrasts "NoGo > Go" (e.g. [1 -1]) on single-subject level and enter the corresponding con images into second-level analysis (e.g. t-test).
When subsequent trials are only a few seconds appart, some studies used regressors for no-go trials, commision errors and misses, but no regressor for go trials. You would then enter the con images corresponding to [1] into second-level analysis. As the go trials are typically much more frequent, this regressor would highly correlate with beta zero (~ implicit baseline) when trials are packed densly. Basically, the go regressor would be "on" all the time. So omitting a go regressor might be an option. But this really depends on the design, so check the relevant papers :-)
Best,
Gabor
|