# A neuro-computational framework for modeling the development of cross-sensory interactions in Autism: from mechanistic understanding to targeted intervention

## TL;DR

The paper introduces a computational model explaining how sensory interactions develop in children and why this process is delayed in autism.

## Contribution

A novel neurocomputational model is proposed to explain atypical sensory development in autism and suggest potential interventions.

## Key findings

- Inhibitory cross-modal projections become excitatory during development, enabling the shift from sensory competition to facilitation.
- Delayed sensory transition in autism may result from reduced neural plasticity or limited multisensory experience.
- Multisensory rehabilitation could improve perceptual and social abilities in children with autism.

## Abstract

Recent studies have shown that multisensory processing in children shifts from a competitive to a facilitative state as development progresses, and this transition appears delayed in children with autism spectrum diagnosis (ASD). The neural mechanisms underlying this developmental change, and its alteration in ASD, remain largely unknown. To address this gap, we investigated how sensory modalities interact in the developing brain using a biologically plausible neurocomputational model governed by Hebbian learning rules. We also explored the neural substrates that may underlie atypical multisensory development in ASD. Our results suggest that inhibitory cross-modal projections gradually become excitatory during development, mediating the observed shift from competition to facilitation in typical children. Furthermore, our simulations show that the delayed transition in ASD may stem either from reduced neural plasticity or diminished multisensory experience. Our model informs the potential benefits that a multisensory rehabilitation strategy may have on the development of perceptual abilities of ASD children, and possibly on the core social symptoms characterizing Autism as well. By linking computational modelling with behavioural findings, our work provides a framework for understanding atypical sensory development. We anticipate that this model could inform future neurophysiological studies and guide the design of multisensory-based therapeutic interventions for ASD.