Long-range temporal correlations in violence and psychosis: An EEG-based study

A subgroup of individuals with psychosis have increased violence risk. Long-range temporal correlations (LRTC) in electroencephalography (EEG) reflect neural dynamics, but their relevance to violence risk remains unclear.

Resting-state EEG was collected from four groups: healthy controls (HC, n = 134), individuals with psychosis and violence history (violent-PSY, n = 16), individuals with psychosis without such history (nonviolent-PSY, n = 38), and violent individuals without psychosis (violent non-PSY, n = 23). LRTC was quantified using detrended fluctuation analysis (DFA) of mean and peak amplitude dynamics in alpha and beta bands during eyes-closed (EC) and eyes-open (EO) conditions, yielding eight metrics: LRTC for mean and peak amplitudes of alpha_EC, beta_EC, alpha_EO, and beta_EO.

No significant group differences in LRTC were observed in eyes-open conditions (all p > 0.05). In contrast, eyes-close conditions showed significantly reduced LRTC in the nonviolent-PSY group relative to controls for mean alpha_EC, peak alpha_EC, and mean beta_EC (all p < 0.05; Cohen's d's = 0.49-0.59). No significant LRTC difference emerged between violent and nonviolent psychosis groups. Other group comparisons were non-significant. Lower alpha-band LRTC was associated with greater symptom severity, including higher total PANSS and depressive subscale scores (all p < 0.05; r's = -0.26 to -0.28).

LRTC was reduced in the nonviolent-PSY group compared to controls but not significantly altered in the violent-PSY group. Nor did it differentiate between psychosis with or without violence history. Although further studies are needed, these findings provide limited support for LRTC as a marker of violence propensity in psychosis.