Detecting Conflict-Related Changes in the ACC

1

• Detecting Conflict-Related Changes in the ACC
• Judy Savitskaya1, Jack Grinband1,3, Tor Wager2,
  Vincent P. Ferrera3, Joy Hirsch1,3
• Program for Imaging and Cognitive Sciences
  (PICS), 2. Dept. of Psychology, 3. Dept. of
  Neurobiology Behavior, Columbia University

682.16 RR89
Effect of Response Time
Imaging Scanning was performed on a 1.5T GE
Twinspeed Scanner. Sixteen subjects were scanned
for 5 runs (8 min each). TR 2.0 s, 25 slices,
4.0 mm slice thickness, FOV 200, 64 x 64.
Effect of Cognitive Control
Introduction Functional imaging studies have
suggested that the anterior cingulate cortex
(ACC) is an integral component of the
decision-making system. Using the Stroop task, it
has been shown that BOLD activity in the ACC is
increased for high conflict (incongruent
stimuli), compared to low conflict (congruent
stimuli), decisions. This result has been
interpreted to mean that the ACC is involved in
conflict detection and/or resolution and that the
ACC may represent a conflict-related module in
the brain. Recent work (Grinband et al, 2008)
has shown that stimulus processing duration can
modulate the BOLD response independent of
condition-type. In some decision-making tasks,
response time (RT) may be correlated to the
condition type. This creates ambiguity in brain
activation results neural activity can be due to
differences in condition or differences in RT.
We used a model-free (non-GLM) analysis of fMRI
data. determine whether ACC activity stems from
the processing of cognitive conflict or is
related to conflict-independent effects of
response time.
Low Control Trials (preceded by
congruent) Incongruent RTs Congruent RTs
Incongruent RTs gt Congruent RTs (All Trials)
Incongruent gt Congruent
Incongruent
Incongruent
Congruent
Congruent
BOLD signal change (x 10-2)
BOLD signal change (x 10-2)
A
Time (s)
Time (s)
Incongruent RTs Congruent RTs (Trials within
100 ms of median)
High Control Trials (preceded by
incongruent) Incongruent RTs Congruent RTs
Stroop Task
Figure 3. A standard GLM analysis was performed.
A contrast of incongruent trials gt congruent
trials produced robust activity in medial
prefrontal cortex (mPFC). This activity was used
to create a mask to perform event-triggered
averaging (Figs 4 5).
Incongruent
Congruent
BOLD signal change (x 10-2)
BOLD signal change (x 10-2)
Incongruent
Long Trials gt Short Trials
Congruent
B
Time (s)
Time (s)
Figure 6. To determine the effect of cognitive
control on activity in mPFC, congruent and
incongruent trials with equal RTs were sorted by
previous trial type. When the previous trial was
congruent (low control), mPFC showed more
activity on incongruent trials. When the
previous trial was incongruent, the result was
reversed.
Incongruent RT lt Congruent RT (Short incongruent
Long congruent trials)
Figure 1. Subjects performed a modified version
of the Stroop task. Four colors (red, green,
blue, yellow) were used. To eliminate repetition
priming effects, color combinations were
randomized such that the same color was not
presented on two consecutive trials.
Incongruent

• Discussion
• Differences in RT between conditions can obscure
  differences in neural activity due to condition
  type.
• The response in the mPFC does not depend on the
  presence of response conflict. Instead, it is
  determined by mean RT of each condition and the
  level of cognitive control.
• After eliminating differences in RT, mPFC
  activity was determined by the level of cognitive
  control.
• GLM analyses should be validated by model-free
  analyses.

RT Performance
BOLD signal change (x 10-2)
Congruent
Figure 4. To simulate the effects of RT on a
standard GLM contrast, we compared long duration
trials vs short duration trials on a data set in
which there were no differences in condition
type. A simple visual stimulus (flashing
checkerboard) was presented to subjects for
variable durations (mean 840 ms, gamma
distributed). There were no differences in
condition type, only differences in stimulus
duration. A contrast of long trials (gt 840 ms)
vs short trials (lt 840 ms) produced a result in
visual cortex similar to the activation map in
mPFC for the incongruent vs congruent contrast.
C
Time (s)
Figure 5. A model free analysis. Hemodynamic
responses were measured in the mPFC for congruent
and incongruent trials of varying RTs. Fig. 5A
shows the event-triggered average for all the
trials this result is consistent with the GLM
result in Fig 3. Fig. 5B shows HDRs for trials
with equal RTs (i.e. RTs within a 200ms window).
Fig. 5C compares fast incongruent trials (gt mean
RT) with slow congruent trials (lt mean RT). The
activity in mPFC is completely determined by RT
rather than the presence of response conflict.
Figure 2. Incongruent trials have longer mean RTs
than congruent trials.
Grinband, Wager, Lindquist, Ferrera, Hirsch
(2008) Detection of time-varying signals in
event-related fMRI designs. Neuroimage 43,
509-520.
For further information contact Jack Grinband,
jg2269_at_columbia.edu