Dear Caroline,
in most cases, one should model all known experimental conditions. During model specification and estimation, SPM doesn't make any distinction between conditions "of interest" or conditions "of no interest". You express your interest by specifying appropriate contrasts for the estimated model. Exceptions to this rule are baseline conditions ("rest" or a continuous higher level baseline) and highly correlated conditions (e.g. conditions that always follow each other after a small, fixed time). Baseline conditions are implicitly modeled, and the inclusion of highly correlated conditions needs to be decided depending on the focus of your experiment.
Hope this helps,
Volkmar
|