Oxytocin normalizes the implicit processing of fearful faces in psychopathy: a randomized crossover study using fMRI

The study addresses whether oxytocin can modulate neural abnormalities in the implicit processing of fearful facial expressions in violent antisocial men with antisocial personality disorder (ASPD) with and without psychopathy. Violent offenders with ASPD, particularly those with psychopathy (ASPD + P), are responsible for a disproportionate amount of violent crime and show pronounced deficits in empathic processing, including recognizing and responding to others’ fear and distress. Prior work has shown abnormalities in reinforcement learning and emotion processing networks in antisocial populations, with empathic deficits especially marked in psychopathy. However, implicit neural responses to fearful faces in adult violent offenders stratified by psychopathy have been underexplored, and no prior work has tested pharmacological modulation of these neural differences. Guided by evidence that oxytocin influences fear-related processing in social brain regions, the authors hypothesized that (1) ASPD + P offenders would show reduced modulation by fearful expression intensity within amygdala, anterior insula, and anterior/mid-cingulate cortices compared with ASPD – P offenders and healthy non-offenders under placebo, and (2) intranasal oxytocin would attenuate these group differences.

Evidence indicates partially separable neural circuits for processing specific emotions, with amygdala, insula, and anterior/mid-cingulate implicated in fear. In youth, callous-unemotional traits (a precursor to ASPD + P) are linked to reduced amygdala responses to fearful faces, whereas low callous-unemotional traits (ASPD – P precursor) show increased amygdala responses relative to typical controls. In adults, limited studies stratifying violent offenders by ASPD ± psychopathy report reduced activation in face-processing and emotion/motivation regions (orbitofrontal and ventromedial prefrontal cortices) and altered insula responses during implicit viewing of fearful expressions in ASPD + P; one larger study lacked non-offender controls. Dimensional studies in violent offenders suggest psychopathic traits modulate neural responses to emotional faces, though not all samples were strictly ASPD or assessed for ASPD. Oxytocin, a neuropeptide central to social behavior with receptors in social brain regions (amygdala, cingulate), enhances explicit recognition of fearful faces and modulates activity in fear-processing regions in healthy adults; a small behavioral study in antisocial adults suggested acute oxytocin (24 IU) can improve fearful expression recognition, but neural mechanisms were untested.

Design: Double-blind, placebo-controlled, randomized crossover fMRI study. Participants self-administered 40 IU intranasal oxytocin (Syntocinon) or placebo in two sessions separated by 3–28 days, beginning the fMRI task 25–30 minutes post-dosing. Allocation used permuted blocks; participants and researchers were blinded. Trial registered (NCT05383300); CONSORT adhered to. Participants: 58 men aged 20–58 years (IQ > 70) enrolled between Sept 2017–Mar 2020; violent offenders met DSM-5 ASPD via SCID-5-PD and had convictions for violent crimes; healthy non-offenders (NO) recruited from the community. All underwent SCID-5-PD and PCL-R assessment (25-point psychopathy threshold for UK context) and provided access to criminal records. Exclusions: major mental disorder (bipolar I/II, major depression, psychotic disorders), neurological disorder, significant head injury, severe sensory impairment, MRI contraindications. Only participants completing both MRI sessions were included. Final analytic sample: 34 violent offenders (19 ASPD + P; 15 ASPD – P) and 24 NO. Groups differed in education and PCL-R scores; ASPD + P showed higher reactive/proactive aggression. Lifetime substance-use disorders were more frequent in offender groups; same-day active substance use differences were covaried in imaging analyses. Ethical approval obtained; written informed consent provided. Task: Morphed Faces task assessing implicit processing of fearful facial expressions at graded intensities (40%, 60%, 80%, 100%). Participants performed a gender discrimination task (left/right button press) during one ~10-minute run. Behavioral analysis: Three-group (NO, ASPD – P, ASPD + P) × two-condition (oxytocin, placebo) × four-intensity (40–100%) repeated-measures ANOVAs on accuracy and response latency (SPSS v25; alpha 0.05). MRI acquisition and preprocessing: Whole-brain BOLD fMRI on a 3T GE scanner. Primary outcome: parametric modulation regressor indexing BOLD responsivity by fearful expression intensity. Preprocessing, individual-level GLMs, and quality control as per Supplementary Information. Group-level MRI analysis: Three-group × two-condition 3dMVM (AFNI) assessed modulation by fear intensity. Whole-brain cluster correction via AFNI 3dClustSim with autocorrelation function (10,000 simulations); initial voxelwise P = 0.005. A priori ROIs: bilateral amygdala, anterior insula, and mid-cingulate cortex with small-volume corrections (TT_Daemon masks); corrected thresholds: k = 13 (anterior/mid-cingulate), k = 8 (anterior insula), k = 2 (amygdala) at initial P = 0.005. Active illicit substance use on scan day included as covariate.

Sample/task performance: Participants performed the gender task accurately (mean accuracy 96.7%, SD 1.0; mean RT 938.8 ms, SD 183.3). No significant main effects or interactions for group, condition, or intensity on accuracy or RT. Whole-brain modulation by fear intensity: Significant parametric modulation observed in bilateral middle occipital gyri extending into fusiform gyri and a separate left fusiform region (visual/face-processing areas). ROI/group effects for fearful expressions: - Overall group effect: Mid-cingulate cortex bilaterally showed significant group differences. • Right mid-cingulate gyrus (BA 24): F = 9.464, P = 0.0003. • Left mid-cingulate gyrus (BA 23/24): F = 9.135, P = 0.0003. - Placebo condition, ASPD + P vs ASPD – P: ASPD + P showed reduced modulation by fearful intensity. • Left mid-cingulate gyrus (BA 23/24): t = 3.95, P = 0.0002, k = 109. • Right mid-cingulate gyrus (BA 24): t = 4.115, P = 0.0001, k = 92. • Right anterior insula (BA 13): t = 3.045, P = 0.003, k = 6 (did not survive strict multiple-comparisons correction). - Oxytocin effect within ASPD + P (oxytocin > placebo): Increased modulation by fearful intensity. • Right mid-cingulate gyrus (BA 24): t = 4.033, P = 0.0001, k = 173. • Left mid-cingulate gyrus (BA 24): t = 4.074, P = 0.0001, k = 119. • Left anterior insula (BA 13): t = 3.498, P = 0.0009, k = 15. - Group × condition interaction: Left mid-cingulate gyrus (BA 23/24): t = 3.21, P = 0.0022, k = 16, indicating that oxytocin abolished placebo-condition group differences between ASPD + P and ASPD – P. - Amygdala: No significant main, between-group, or within-group effects in either condition. Interpretation: Under placebo, ASPD + P offenders showed blunted mid-cingulate (and trend-level right insula) modulation to increasing fearful intensity relative to ASPD – P. Oxytocin enhanced modulation in bilateral mid-cingulate and left anterior insula in ASPD + P, normalizing differences with ASPD – P.

Findings show that violent offenders with ASPD + P have reduced implicit neural sensitivity to fearful facial expression intensity within the mid-cingulate cortex and anterior insula relative to ASPD – P, aligning with models of impaired processing of others’ distress in psychopathy. Contrary to some expectations, amygdala effects were not observed, potentially reflecting sample size and task/design specifics. Crucially, intranasal oxytocin increased modulation by fearful intensity in the mid-cingulate bilaterally and left anterior insula among ASPD + P, eliminating baseline group differences with ASPD – P in these regions. This suggests oxytocin may increase the salience of fearful expressions and modulate neural circuits underpinning empathic processing. The results provide initial evidence that neurochemical intervention can transiently normalize core social-cognitive processing deficits associated with psychopathy, underscoring potential avenues for adjunctive treatments targeting empathic dysfunction in high-risk offenders.

The study demonstrates that violent antisocial men with psychopathy exhibit diminished modulation of mid-cingulate and insula responses during implicit processing of fearful faces relative to antisocial men without psychopathy. A single 40 IU dose of intranasal oxytocin enhanced modulation in these regions in the psychopathy group, abolishing between-group differences in the mid-cingulate cortex. These findings provide proof-of-concept that oxytocin can acutely modulate neural substrates of empathic processing deficits in psychopathy. Future work should include larger, multi-site samples, explicit behavioral endpoints (e.g., fear recognition, aggression measures), longitudinal dosing to assess durability, exploration of dose–response and timing, and mechanistic studies of network-level changes and individual differences (e.g., callous–unemotional traits, substance use).

- The right anterior insula reduction in ASPD + P versus ASPD – P under placebo did not survive strict multiple-comparison correction and requires replication. - Laterality inconsistency: baseline effect in right insula, oxytocin effect in left insula; hemispheric laterality of oxytocin effects remains debated. - Mid-cingulate clusters included apparent white matter due to spatial smoothing/registration; despite motion control, some noise (e.g., motion) may contribute. - No significant differences versus non-offender controls in ROIs, possibly due to limited power; similarly, no amygdala effects observed, which may reflect underpowered samples. - Group differences in active substance use on scan day existed, though modeled as covariates; residual confounding cannot be entirely excluded. - The task probed implicit processing only; no assessment of downstream behavioral or clinical outcomes, limiting conclusions on functional impact.