Hello SPM'ers -
We ran an event-related fMRI analysis on the same (7) subjects on two
different tasks: a counting stroop task and a go-nogo task. During the
cStroop, subjects were instructed to push a button that coincided with the
number of words they saw on the screen (stimulus presented for 1s with .5s
stimulus interval). The go-no task instructions required the subjects to
press button 1 (the go task) after seeing a specific combination of pics,
and button 2 (the no-go task) for the remaning sets (stimulus presented for
1s with no interval btw slides). After standard pre-preocessing of the raw
data we ran a fixed-effect analysis on the data using the following setup:
TR = 2s for stroop; 2.3 for go-nogo
scans per session = 149 for the stroop; 86 for the go-nogo
# of conditions = 7 for each stroop; 4 for each go-no
stochastic design = no
SOA = variable
time in scans was entered for each trial
variable durations = no
parametric modulation = none
type = events modelled with 'hrf alone'
no interaction among the trials
zero specified regressors
remove global effects = scale
hi-pass filter = 32 (default)
low-pass filter = none
model intrinsic correlation = none
trial specific F-contrats = no
The resultant design matrix (for the stroop task) is attached.
We were surprised to note a lack of activation in the anterior cingulate
(AC) (even at ridiculously low thresholds) in the stroop task, when the
behavioral data for the same subjects was very high (p < .001). When we
loaded the resMS.img into the viewer we noted a large area of 'white' in
close proximity to the area of interest. This seemed to indicate that a
large degree of variance was located near the AC. After some discussion, we
intially thought this may simply be due to the small number of subjects
within the design.
Upon subesequent analysis of the go-nogo data we observed the same of degree
of variance localized with the same region as the cStroop task (see attached
.jpeg files for illustration). I have also attached an overlay of one of the
.con files for further clarification. Our concern and question (yeah there
actually is one here) is, given the proximity of the variance to our area of
interest, could this be causing/suppressing our notable lack of AC
activation? Additionally, what may be causing this large observed variance?
Thank you in advance...
Keith
--------------------
Keith Harenski
Emory University
Dept. of Psychiatry and Behavioral Sciences
404.727.2992
|