I have an experiment that looks at 2 different states within the same run/scan. One of the processes consists of a single response (categorization) and is therefore being modeled as an Event-Related, 0 duration response. The other process of interest consists of a continuous state (encoding) whereby each trial within the run is user-defined, variable and locked to reaction time. For the encoding stage we have modeled this using a Variable Epoch Model using RT as the duration.
In case it is of importance: There is no overlap in these processes (each process is time independent) and I am currently modeling with a Canonical HRF with no time or dispersion derivatives.
I know it is physically possible to enter Epoch and ER data and run 1st Level single subject in SPM5 and examine each independently, however I am unclear whether contrasting or even comparing Epoch with ER data is recommended. For instance, if I ran a contrast of "Categorization (ER) > Encoding (Epoch)" what would be the limitations? Would you recommend I approach this data in a different manner all together?
Any guidance on this would be appreciated. Thanks.
-Josh
|