Psychopathic and autistic traits differentially influence the neural mechanisms of social cognition from communication signals

The study addresses how psychopathic and autistic traits differentially influence neural mechanisms of social cognition, particularly in response to communicative voice signals. Prior work shows individuals with psychopathic traits exhibit social cognition deficits and altered activity in social brain networks comprising mirror neuron (inferior frontal cortex, intraparietal sulcus), mentalizing (superior temporal cortex, temporo-parietal junction, dorsal medial frontal cortex), empathy (anterior insula, anterior cingulate cortex), and limbic (amygdala, medial temporal lobe, ventromedial/orbitofrontal cortex) subsystems. Heterogeneity across psychopathy subtypes—primary (low anxiety, manipulative, callous, instrumental) versus secondary (high anxiety, impulsivity, reward dependency)—may explain inconsistent neural findings. Most prior research relies on visual stimuli (e.g., faces), whereas voice carries essential nonverbal and speech communication cues, engaging temporal dynamics and potentially basal ganglia mechanisms. Fundamental discrimination of social (voices) versus non-social sounds is accomplished in auditory cortex (including temporal voice areas). Given reported overlaps between psychopathy and autism spectrum traits in cognitive and neural domains (e.g., empathy and limbic networks), the study tests whether psychopathic and autistic traits share or dissociate in neural processing of communicative voice signals. The exploratory hypothesis is that psychopathic and autistic traits will differentially correlate with activity across mirror, empathy, mentalizing, and limbic networks during voice processing.

Design: Exploratory fMRI study examining neural responses to communicative sounds, relating individual differences in psychopathic (primary, secondary) and autistic traits to brain activity. Participants: N=113 adults (47 male, 66 female), age 18–40 years (mean 25.59, SD 4.79). Normal hearing and normal/ corrected-to-normal vision; exclusion: hearing impairments, psychiatric or neurological disorders. Two left-handed. Ethical approval by Canton of Zurich ethics committee; informed consent obtained. Stimuli and task: 140 natural sounds (duration 500 ms), comprising 70 human voice sounds (speech and non-speech vocalizations) and 70 non-voice sounds (animal, artificial, natural). Sounds RMS-normalized and presented at 70 dB SPL. Five sound conditions (voice: speech, non-speech; non-voice: animal, nature, artificial) presented in pseudo-random order with 3–5 s ISI. Each sound played once; 10% of trials were immediate repetitions (14 repetitions total). One-back repetition detection task: press button with right index finger when a sound repeats from the previous trial. Behavioral analysis: Reaction times and detection accuracy computed for hits (repetition present and detected), false alarms (no repetition but button press), and miss trials (repetition present but not detected). One-way repeated-measures ANOVA tested condition differences (alpha p=0.05). Pearson correlations tested associations between performance (RT, accuracy) and trait scores (psychopathy, autism), FDR-corrected. MRI acquisition and preprocessing: 3T Philips Ingenia scanner. Preprocessing in SPM12: manual realignment to AC-PC, motion correction (realignment to mean EPI), slice timing correction; co-registration of T1 to mean EPI; segmentation to estimate normalization; normalization to MNI space (functional resampled to 2 mm isotropic voxels); 8 mm FWHM Gaussian smoothing. First-level analysis: GLM with five regressors for each sound condition plus one regressor for repetition trials; six motion parameters as nuisance regressors. Events modeled with stick functions at stimulus onset, convolved with canonical HRF. Group-level analyses: Random-effects factorial analyses. Contrasts: (1) voice > non-voice; (2) speech > non-voice; (3) non-speech > non-voice; (4) speech > non-speech; (5) non-speech > speech. ROI analyses on significant clusters from these contrasts. Whole-brain multiple regressions assessed associations between neural effects and trait scores. Trait measures: Psychopathy assessed via Levenson Self-Report Psychopathy Scale (LSRP; total, primary [16 items], secondary [10 items]) administered on 5-point Likert and converted to 4-point scaling for comparability. Autism traits assessed via Autism-Spectrum Quotient (AQ; 50 items). Additional measures: Big Five Inventory (extraversion, agreeableness, conscientiousness, neuroticism, openness), PANAS (positive, negative affect; affective balance), STAI-trait, BDI-IA. In multiple regression models predicting neural activity, additional personality measures and demographics (age, gender) were included as regressors of no interest. Statistical thresholds: Functional activations visualized at voxel p<0.005, cluster extent k>55, corresponding to cluster-corrected p<0.05 (for display). Correlations and regressions FDR-corrected where applicable. Conjunction analyses employed conjunction null hypothesis to test overlap between LSRP and AQ effects.

- Trait distributions and reliability: Broad variability across psychopathic and autistic traits with acceptable reliability (Cronbach’s alpha: LSRP_total 0.84; LSRP_prim 0.84; LSRP_sec 0.70; AQ 0.78; other trait scales 0.74–0.90). LSRP_total strongly correlated with LSRP_prim (r=0.909, p<0.001) and LSRP_sec (r=0.726, p<0.001); LSRP_prim and LSRP_sec intercorrelated (r=0.373, p<0.001). Only LSRP_sec correlated positively with AQ (r=0.356, p<0.001). - Broader personality associations: All LSRP scores and AQ negatively correlated with agreeableness (r<-0.349, p<0.001). Overlap between LSRP_prim and AQ in lower openness (r<-0.231, p<0.001). LSRP_sec and AQ both associated with lower extraversion (r<-0.243, p<0.001), higher neuroticism (r>0.153, p<0.001), and higher trait anxiety (r>0.384, p<0.001). LSRP_sec specifically related to lower affective balance (PANAS_ab r=-0.288, p<0.001) and higher depression (BDI r=0.266, p<0.001). Male gender associated with higher AQ (r=-0.278, p<0.001; coded such that negative correlation indicates higher AQ in males). - Behavioral performance: High overall accuracy for hit trials (mean 92.21%), significantly higher than false alarms (14.41%) and misses (7.79%) (F1,219=196.170, p<0.001, GG-corrected). Slower RTs for false alarms vs hits (F1,16=6.349, p=0.016). Hit rate negatively associated with psychopathic traits: LSRP_total (r=0.257, p=0.014) and especially LSRP_prim (r=-0.292, p=0.009) after controlling for age and gender, indicating more difficulty detecting repetitions with higher primary psychopathy. - Voice processing network: Voice > non-voice activated bilateral auditory cortex (peaks in higher-order AC Te3), frontal motor and ventral premotor regions, and infero-orbital frontal cortex. Speech and non-speech each engaged similar networks; non-speech additionally activated ACC and bilateral amygdala (CMA/BLA). Speech > non-speech emphasized higher-order AC and left vPM; non-speech > speech emphasized secondary AC (PTe, PPo) and right amygdala (CMA). - ROI-level trait associations: Across all five sound categories, higher LSRP_total and LSRP_prim correlated with greater activity in right planum temporale (r≈0.269 and 0.264, both p≈0.028 FDR), and LSRP_prim also with right mid STC (r=0.287, p=0.025). For voice > non-voice, higher LSRP linked to lower activity in bilateral infero-orbital FC (left with LSRP_sec r=-0.276, p=0.019; right with LSRP_prim r=-0.240, p=0.041), right CMA (LSRP_total r=-0.275, p=0.019; LSRP_prim r=-0.286, p=0.019), and ACC (LSRP_prim r=-0.279, p=0.033). Similar negative associations held for speech > non-voice; for non-speech > non-voice, negative relationships emerged in right motor cortex (r<-0.257, p<0.035). No significant AQ associations in these ROIs for voice > non-voice. - Whole-brain multiple regressions: Increasing LSRP_total predicted widespread hypoactivity for voice > non-voice across low- and higher-order AC (PPo, PTe, mST, aST), motor/premotor (MC, vPM), fronto-insular (ioFC, aINS), medial limbic (BLA, hippocampus), and basal ganglia (putamen, caudate) regions. Similar negative effects were observed with LSRP_prim; secondary psychopathy showed more restricted effects (notably in mirroring/mentalizing networks and sensory AC). AQ predicted fewer negative associations localized to bilateral posterior STC (PST), left vPM, and ACC; speech-specific analyses showed larger PST associations. Conjunction analyses including both LSRP and AQ revealed no significant overlaps, indicating largely differential neural influences. - Summary of subtype-specific neural alterations: Primary psychopathy exhibited broad hypoactivity across mirroring, mentalizing, empathy, limbic, and basal ganglia systems during voice processing. Secondary psychopathy showed hypoactivity primarily in mirroring and mentalizing systems with additional deficits at auditory sensory processing levels (ventral auditory stream/object identification). High autistic traits were associated with deviations in sensory cortices, more consistent with dorsal auditory stream (communicative context encoding), and overall intact core voice processing in ROI contrasts.

Findings demonstrate that psychopathic and autistic traits are linked to distinct neural signatures during processing of communicative voice signals. Behaviorally, higher primary psychopathy related to poorer repetition detection, suggesting attention or monitoring deficits even in a basic auditory task. Neurofunctionally, primary psychopathy showed widespread hypoactivity across social cognition networks (mirror neuron, mentalizing, empathy, limbic) and in basal ganglia nodes, implicating disrupted decoding of socio-affective content and temporal/communicative patterns in voice signals. Secondary psychopathy presented with mirroring and mentalizing hypoactivity and novel evidence of upstream auditory sensory processing deficits in voice-sensitive cortex, suggesting that antisocial behaviors may be compounded by impaired auditory object identification of social cues. In contrast, autistic traits showed fewer and more focal associations, primarily within sensory and dorsal-stream-related regions (posterior STC, vPM, ACC), indicating subtler social voice processing differences and relative preservation of core voice-selective responses in the assessed ROIs. The lack of conjunction between LSRP and AQ effects underscores differential neurocognitive pathways underlying superficially overlapping social difficulties. Overall, the results address the study’s hypothesis by delineating trait-specific modulation of social cognitive and affective brain networks during vocal communication processing.

The study provides evidence that psychopathic traits, especially primary psychopathy, are associated with broad neural hypoactivity across social cognition and basal ganglia systems during voice signal processing, while secondary psychopathy shows more circumscribed deficits in mirroring/mentalizing networks together with previously undescribed sensory-auditory processing impairments. Autistic traits exhibited more limited, sensory-focused deviations, largely distinct from psychopathy-related effects. These results highlight that social cognition impairments from vocal communication signals have differential neural underpinnings across psychopathic subtypes and autistic traits, emphasizing basal ganglia involvement in primary psychopathy and sensory processing deficits in secondary psychopathy.

- Exploratory study design without a priori power analysis or stopping rule; sample size chosen to enable parametric regressions comparable to prior work. - Community sample with self-report trait measures; lack of clinical/institutionalized cohorts limits generalizability to clinical psychopathy/ASD. - Potential cross-sample/cross-country issues in applying psychopathy cutoff scores; noted that a commonly used LSRP cutoff may overestimate clinically relevant psychopathy in this sample. - MRI analyses used standard smoothing and GLM modeling; results rely on correlation/regression with multiple trait covariates and FDR corrections; causal inferences cannot be made. - Focus on auditory/voice stimuli may not generalize to other social modalities (e.g., visual cues).