Dear SPM experts,
I am planning to run a 2x2 factorial fMRI experiment. Factor A (2
levels) is type of stimulus; factor B (2 levels) is context.
Due to technical issues, I can only acquire about 130 volumes
consecutively as long as I do not modify my acquisition parameters (TR
3secs; N. Slices 32 ; Slice Thickness 3.5mm/0.5mm gap). In other words I
am forced to split my experiment into multiple runs (6).
While one factor (say factor A) can easily be manipulated within runs,
the manipulation on the other factor (factor B) requires participants to
switch from one context to another which in principle could be quite
confusing and difficult for our participants. I was wondering whether
manipulating factor B across (and not within) runs would be a suitable
strategy to adopt.
I have two main questions:
1) Can I present only one level of factor B in RUN 1, 3, and 5 and the
other level in RUN 2, 4, and 6?
By adopting this strategy participants could easily set them selves in
advance in a specific context instead of having to switch continuously
within runs. However, this will force me to contrast trials which are
relatively far apart in time posing some high-pass filtering issues.
What would you do?
2) If this strategy were appropriate, should I construct my runs (e.g.
order and onsets of my stimuli) so that RUN1 is an exact replication of
RUN2, RUN3 of RUN4 and RUN5 of RUN6 (please note that RUN1, 3, and 4
would still differ)? Would this increase correlations between my
regressors so that a good estimation would be prevented?
I am asking this as I am worried that differences in design efficiency
due to the way stimuli will be presented within runs may render any
manipulation of factor B difficult to interpret.
For instance, suppose that I create six completely different runs. The
haemodynamic responses evoked by the stimuli presented in RUN1 could
differ from those evoked by the very same stimuli presented in RUN2.
While this difference could be interpreted on the basis of the different
perspective (context1 in RUN1 and context2 in RUN2) assumed by
participants in the two RUNS, one might argue that such difference may
be due to a different efficiency between the two RUNS.
My apologies for the silly questions and many thanks in advance to
anyone has a spare minute to comment on this.
Any help/suggestion would be very much appreciated.
Best Regards,
Andrea
|