Conflict processing networks: A directional analysis of stimulus-response compatibilities using MEG

Abstract

The suppression of distracting information in order to focus on an actual cognitive goal is a key feature of executive functions. The use of brain imaging methods to investigate the underlying neurobiological brain activations that occur during conflict processing have demonstrated a strong involvement of the fronto-parietal attention network (FPAN). Surprisingly, the directional interconnections, their time courses and activations at different frequency bands remain to be elucidated, and thus, this constitutes the focus of this study. The shared information flow between brain areas of the FPAN is provided for frequency bands ranging from the theta to the lower gamma band (4-40 Hz). We employed an adaptation of the Simon task utilizing Magnetoencephalography (MEG). Granger causality was applied to investigate interconnections between the active brain regions, as well as their directionality. Following stimulus onset, the middle frontal precentral cortex and superior parietal cortex were significantly activated during conflict processing in a time window of between 300 to 600ms. Important differences in causality were found across frequency bands between processing of conflicting stimuli in the left as compared to the right visual hemifield. The exchange of information from and to the FPAN was most prominent in the beta band. Moreover, the anterior cingulate cortex and the anterior insula represented key areas for conflict monitoring, either by receiving input from other areas of the FPAN or by generating output themselves. This indicates that the salience network is at least partly involved in processing conflict information. The present study provides detailed insights into the underlying neural mechanisms of the FPAN, especially regarding its temporal characteristics and directional interconnections.

Fig 1. Stimuli design.

During congruent conditions (LL, RR) the arrow was placed at the same side of the screen as where the arrow was pointing to, while during incongruent conditions (LR, RL), the placement of the arrow was opposite to its pointing direction. In each of the four different conditions (LL, LR, RR, RL) the first character denoted the direction of the arrow, while the second character indicated the ipsi- or contralateral side of the key-pressing finger. Subjects were asked to indicate the direction of the arrow utilising a left or right key press, respectively. The stimulus onset asynchrony (SOA) and stimulus duration are indicated at the bottom.

Fig 2. Mean reaction times.

(RT, ms) during congruent (LL, RR) and incongruent (LR, RL) conditions. The error bars indicate the standard deviation. The asterisk marks a significant difference between the RTs of congruent versus incongruent conditions (p < 0.01).

Fig 3. Map of regions of interest (ROIs) identified by a spatio-temporal cluster permutation test.

Bilaterally activated regions were the MFC (caudal middle frontal cortex, 1), PoC (postcentral cortex, 2), PrC (precentral cortex, 3), SPC (superior parietal cortex, 4), SMC (supramarginal cortex, 5), AI (anterior insular cortex, 6), Tem (middle and inferior temporal regions, 7), Prec (precuneus and superior parietal regions, 8), Para (paracentral region, 9), ACC (anterior cingulate cortex, 10), and MVC (medial visual cortices, 11). In the right hemisphere, the IPC (inferior parietal cortex, 12) and in the left hemisphere the SFC (superior frontal cortex, 13) were significantly activated.

Fig 4. Time courses of significant bilateral activity.

Group averaged time courses from single trail source analysis showing activity from the MFC (caudal middle frontal cortex), PrC (precentral cortex) and SPC (superior parietal cortex) for the left (lh) and right (rh) hemisphere and for each of the significant conditions of the conflict processing task (LL, RR, LR, RL) are presented. SD indicates the standard deviation; AU: arbitrary unit.

Fig 5. Causal information flow during conflict processing.

Significant causal interactions (p < 0.01) are shown for both incongruent tasks, LL vs LR (a) and RR vs RL (a) at four different frequency bands (4–30 Hz). At each left and right part of the circle, ROI labels from the left and right hemispheres are shown respectively (cf. Table 3). The lines indicate significant connections, while arrows indicate the direction of information flow. The location of significant causal connections is mapped onto the surface of the brain for each frequency band. The same colour coding is used in the circles and brain plots.