Dear SPMers,
I am analyzing a block design visual search experiment with three
conditions. The blocks associated with a condition are randomly
distributed across the run. Reaction times were obviously varied and
I am interesting in comparing activity between trials where there were
short reaction times versus longer reaction times within and between
conditions. Trials will be be considered "short" if the reaction
times were below the median and "long" when above the median.
Question 1:
How do I enter the trial length as a regressor? Does separate
regressors from the onset and for the offset of the trial? Do I just
have a regressor of reaction times? Also if I do any of the above,
does defining a condition change, or do I just enter the length of the
entire block of that condition as normal?
Question 2:
This question is similar to the Question 1. For 5 seconds before each
block, the subject sees verbal instructions. If I wanted to look at
this, do I need a separate regressor for the onset and offset times?
Question 3:
Within each block, half of the trials have no search target. Should
treat these trials as "long" trials (Question 1) or does there need to
be a separate regressor specifically for no target trials? The
assumption is that the subject is actively searching throughout the
entire trials as opposed to stopping searching when the target is
found
Question 4:
I am uninterested in activity associated with the pressing of the
response button. Given that the button press is about 100 ms, do I
need to account for this as a nuisance variable? If so, how does one
enter a this under "Covariates" when this is not a covariate of
interest?
Question 5:
I will also be entering % correct as a regressor. How is this
technically different to doing a PPI analysis, or does the PPI call up
the .mat file generated when running the STATs analysis..
Apologies for all the questions. Thanks so much in advance for any help.
Michelle
|