Computational and theoretical approaches to Autism: From neural mechanisms to behavior

Marking World Autism Awareness Day 2026, this Research Topic aims to bring together computational, theoretical, and data-driven studies that advance a mechanistic understanding of autism across multiple levels of analysis, from synaptic and circuit-level processes to large-scale brain dynamics, cognition, and behavior.

Autism Spectrum Disorder (ASD) is a complex neurodevelopmental condition characterized by profound heterogeneity in sensory processing, social communication, and cognitive profiles. Despite significant advancements in identifying genetic and molecular markers, a major challenge persists: bridging the gap between microscopic neural dysfunctions and the macroscopic behavioral phenotypes observed clinically. Traditional descriptive methods often struggle to capture the non-linear, multi-scale interactions within the brain that drive these diverse experiences, and computational neuroscience has emerged as a transformative discipline in this landscape, providing the formal mathematical frameworks required to integrate data across levels of analysis. By utilizing dynamical systems theory, predictive processing models, and advanced network analysis, researchers can move beyond mere correlation toward a truly mechanistic understanding of how altered neural dynamics manifest as autistic traits and behaviors.

The primary goal of this Research Topic is to address the challenge of biological and behavioral heterogeneity in autism by fostering a synthesis of computational and theoretical perspectives. While research has identified various neural signatures associated with autism, the field lacks the predictive power to relate these physiological markers to specific developmental trajectories or individualized clinical outcomes.

To achieve a more unified understanding, this collection aims to consolidate research that uses quantitative models to bridge the gap between brain and behavior. We seek to advance the "computational psychiatry" of autism, where models can explain how individual differences in neural architecture lead to specific cognitive and sensory styles. By integrating expertise in machine learning, nonlinear dynamics, and multimodal data analysis, we aim to identify robust biomarkers and stratification strategies. Ultimately, this Research Topic seeks to demonstrate that theoretical frameworks are essential instruments for translating complex neurobiological data into meaningful, personalized insights for the autistic community.

This Research Topic invites contributions that employ computational, theoretical, and data-driven methodologies to investigate the mechanistic underpinnings of autism. We welcome articles that connect neural processes with behaviorally and clinically relevant outcomes. We particularly encourage work that addresses the individual diversity within the spectrum and studies that utilize longitudinal data to capture developmental changes.

Specific themes for contributors include:

- Synaptic and circuit-level mechanisms: Investigating E/I balance and local neural dynamics.
- Nonlinear dynamics: Applying state transitions and bifurcation-based approaches to brain activity.
- Predictive processing: Computational frameworks for understanding sensory and social processing.
- Linking brain and behavior: Models that map neural function to cognitive and social phenotypes.
- Network organization: Analyzing large-scale brain connectivity and topology.
- Machine learning: Novel approaches for patient stratification, biomarker discovery, and clinical prediction.
- Developmental modeling: Longitudinal computational studies of brain maturation.
- Multimodal integration: Combining imaging, electrophysiology, and behavioral datasets through quantitative modeling.