Hierarchical Neural Variability Reveals Adaptive and Maladaptive Mechanisms of Non-Suicidal Self-Injury

Why It Matters

Neural variability emerges as a transdiagnostic biomarker that distinguishes NSSI and forecasts treatment response, opening avenues for personalized interventions.

Key Takeaways

• NSSI patients exhibit heightened connectivity‑level and topology‑level neural variability.
• Connectivity variability predicts better emotional/attentional function and NSSI reduction.
• Topology variability associates with impulsivity and worse treatment outcomes.
• Variability‑based classifiers reach AUC 0.66, comparable to static FC models.
• Global signal regression eliminates variability differences, highlighting global brain dynamics.

Pulse Analysis

Neural variability—the moment‑to‑moment fluctuation of brain activity—has emerged as a fundamental substrate of cognitive flexibility. While stable activation patterns support routine processing, dynamic changes enable the brain to reconfigure networks in response to shifting internal states or external demands. Disturbances in this variability have been linked to a range of neurodevelopmental and psychiatric disorders, from autism to schizophrenia, where rigid thinking hampers adaptive behavior. In adolescents, non‑suicidal self‑injury (NSSI) often co‑occurs with emotional dysregulation and impaired set‑shifting, suggesting that altered neural variability may underlie the disorder’s core deficits.

The recent multi‑site fMRI investigation examined 160 patients with NSSI and 50 psychiatric controls, quantifying variability at two hierarchical levels: connection‑wise fluctuations in functional coupling and topology‑wise changes in whole‑brain network architecture. Across both levels, the NSSI group showed significantly higher variability, yet the clinical meaning diverged. Greater connectivity‑level variability correlated with higher scores on attentional awareness and lower alexithymia, and it forecasted larger reductions in self‑injury after a 13‑week psychotherapy program. Conversely, elevated topology‑level variability aligned with higher impulsivity scores and predicted poorer therapeutic response, highlighting a maladaptive signature.

These findings position neural variability as a promising transdiagnostic biomarker for NSSI. Unlike static functional connectivity, variability measures captured both adaptive and maladaptive dynamics and added prognostic value beyond traditional metrics, achieving a modest but reliable classification performance (AUC ≈0.66). The sensitivity of the signal to global‑scale brain activity—evidenced by its disappearance after global signal regression—suggests that large‑scale physiological processes may drive the observed patterns. Future work should explore task‑based paradigms, longitudinal monitoring, and integration with genetic or behavioral data to refine personalized treatment strategies and potentially intervene on the neural mechanisms that sustain self‑injurious behavior.