Central serotonin modulates neural responses to virtual violent actions in emotion regulation networks

Abstract

Disruptions in the cortico-limbic emotion regulation networks have been linked to depression, anxiety, impulsivity, and aggression. Altered transmission of the central nervous serotonin (5-HT) contributes to dysfunctions in the cognitive control of emotions. To date, studies relating to pharmaco-fMRI challenging of the 5-HT system have focused on emotion processing for facial expressions. We investigated effects of a single-dose selective 5-HT reuptake inhibitor (escitalopram) on emotion regulation during virtual violence. For this purpose, 38 male participants played a violent video game during fMRI scanning. The SSRI reduced neural responses to violent actions in right-hemispheric inferior frontal gyrus and medial prefrontal cortex encompassing the anterior cingulate cortex (ACC), but not to non-violent actions. Within the ACC, the drug effect differentiated areas with high inhibitory 5-HT1A receptor density (subgenual s25) from those with a lower density (pregenual p32, p24). This finding links functional responses during virtual violent actions with 5-HT neurotransmission in emotion regulation networks, underpinning the ecological validity of the 5-HT model in aggressive behavior. Available 5-HT receptor density data suggest that this SSRI effect is only observable when inhibitory and excitatory 5-HT receptors are balanced. The observed early functional changes may impact patient groups receiving SSRI treatment.

Keywords: Medial prefrontal cortex; Pharmaco-fMRI; SSRI; Serotonin; Virtual violence.

Fig. 1

Gameplay event types. On each of the two measurement days, participants played the racing game Carmageddon during four 10-min fMRI sessions. During the gameplay, participants could run over and kill virtual pedestrians with their car (event type ‘violent action’; 1st panel). When participants narrowly missed the pedestrian ‘attempted violent action’ was coded (2nd panel). During the ‘non-violent action’ a bonus-item was collected (3rd panel). ‘Non-intended actions’ were coded when the player crashed into objects or scenery in the game map (4th panel). Video and sound of the gameplay were recorded; time point and type of each event were coded; and a response model was generated for fMRI analyses

Fig. 2

Mean count of coded event types. On each measurement day, participants played four 10-min sessions of the racing game Carmageddon. Gameplay was video recorded and subsequently coded with four event types (compare Fig. 1). A total of 40,548 events were coded throughout the study. The counts of violent actions and non-violent actions were similar. Similar event counts under PLAC and SSRI conditions suggest a high robustness of the behavioral paradigm: violent action under PLAC: mean count ± sem = 120.8 ± 32.5, SSRI = 120.5 ± 32.3; attempted violent action: PLAC = 50.8 ± 28.3, SSRI = 50.5 ± 27.8; non-violent action: PLAC = 109.1 ± 21.6, SSRI = 108.5 ± 23.2; non-intended action: PLAC = 253.6 ± 57.9, SSRI: 253.3 ± 55.8). sem standard error of the mean, PLAC placebo

Fig. 3

Effect of event type on functional responses. a The factor event (four game actions: violent, attempted violent, non-violent, and non-intended) explained variance in brain activity of extended sensory, motor, and reward networks (see Table 1). b Violent actions yielded similar distributed activation patterns and deactivation in the prefrontal cortex compared to non-violent actions. Maps were thresholded according to a corrected p < 0.05 after Monte Carlo simulation (voxel-wise p < 0.005 and cluster-size > 8). Z values are indicated beneath each slice

Fig. 4

Effect of the SSRI challenge on functional responses. a The factor drug explained variance in the right inferior frontal gyrus (IFG). The SSRI challenge decreased responses to violent and to attempted violent actions but not to non-violent and to non-intended actions (see ROI analysis in bar graph). b The interaction between drug and event yielded significance in the right medial prefrontal gyrus (mPFC) and the right anterior cingulate cortex (ACC). In these two clusters, the SSRI challenge reduced functional activation to violent actions and attempted violent actions only (see bar graph inserts). Maps were thresholded according to a corrected p < 0.05 after Monte Carlo simulation (voxel-wise p < 0.005 and cluster size > 8). Post hoc t tests: **p < 0.01; *p < 0.05. ACC anterior cingulate cortex, IFG inferior frontal gyrus, mPFC medial prefrontal cortex, PLAC placebo

Fig. 5

Effect of the SSRI challenge on functional responses in ACC subregions. The pregenual ACC subregions p32 and p24, and the subgenual region s25 were defined anatomically based on their cytoarchitecture (Palomero-Gallagher et al. 2008); p24 and p32 exhibited average and s25 elevated 5-HT_1A receptor density (Palomero-Gallagher et al. 2009). Z values are indicated beneath each slice. (B) The SSRI challenge enhanced deactivation in the pregenual subregions p32 (left panel) and p24 (middle panel) to violent action and to attempted violent action but not to non-violent or to non-intended actions. In the subgenual s25, the SSRI yielded no significant effect (right panel). Post hoc t tests: **p < 0.01; *p < 0.05. PLAC placebo