Dear experts,
sorry for another parametric modulation question, but here it is:
my trial is the following:
expectation period (reward cue: 3.75s) - target (0.75s) - outcome period
(win/loss: 1.25s).
2x2 cue conditions and a (small) behavioral interaction between both
factors for reaction times (RTs).
i want to know whether the reaction time effect is reflected in activity
of "motivating" brain areas,
who should show activity throughout the expectation period.
what i did:
put all 4 cues in one regressor (0th) and included RTs as 1st order
regressor in 1st level analysis. one-sample t-test on 2nd level - bad
results :(
so i'm looking for errors i made and these are my questions:
a) is it wrong to model the cue period with a duration of 3.75 (not a
stick) in a parametric modulation model?
everytime i read about it, they take stick functions .... but i got
nice results with mini-blocks in an earlier study.
b) before entering RTs, i mean corrected them (which is done by spm5
anyway, which i learned later).
as i'm looking for areas showing greater activation with decreased RTs,
and since i thought the hrf is scaled (modulated) by RT values in its
height,
shouldn't i also invert values (meaning that e.g. -11,64s deviation from
mean RT should be entered as +11,64s)?
or will the [-1] contrast test for that opposite relationship on 2nd
level (i thought it would only test for the absence of linear modulation
...)?
c) in a discussion earlier in the list, Rik mentioned the possibility of
additionally modelling categorial responses within
such a model as +1 and -1 values in a second parametric regressor. what
kind of order should i assign to that regressor - linear?
thanks very much for your help
Markus
--
Dipl.-Psych. Markus Staudinger
Social Cognitive Affective Neuroscience Group
Department of Psychiatry
Division of Medical Psychology
University of Bonn
Sigmund-Freud-Strasse 25
D- 53105 Bonn
Germany
Tel: +49 (0)228 287 19704
Fax: +49 (0)228 287 19125
E-mail: [log in to unmask]
|