Predictive Coding in Autism Spectrum Disorders

Predictive coding posits that the brain continuously generates hierarchical models to anticipate sensory events and minimises the discrepancy between expectation and input, known as prediction error. In individuals with autism spectrum disorders, accumulating evidence suggests that atypical weighting of prior beliefs and sensory evidence underlies core features such as sensory hypersensitivity, insistence on sameness and social communication challenges. Neurophysiological studies reveal that autistic individuals may rely more heavily on incoming sensory signals and less on contextual priors, leading to an over-precise representation of current inputs and difficulty in adapting to changes in environmental statistics. Across perception, action and social cognition, this imbalance can manifest in reduced flexibility of expectation updating, heightened sensory volatility and altered hierarchical integration of information. Understanding these mechanisms offers a unifying framework for diverse manifestations of autism, linking perceptual anomalies to broader cognitive and behavioural profiles and opening avenues for targeted interventions aimed at recalibrating predictive processes.

Research from Nature Portfolio

Recent investigations have employed advanced neuroimaging and animal models to elucidate how predictive coding differs in autism. Functional MRI studies in autistic adults performing associative learning tasks have shown intact basic predictive abilities but enhanced coupling of high-level predictions with neural activity in frontal regions, alongside distinct patterns of prediction-error signalling in the anterior cingulate and striatum. Complementary work in a primate model of autism, induced by prenatal valproate exposure, has demonstrated persistent sensory hypersensitivity and unstable predictions across multiple brain hierarchies, revealing both underestimation and overestimation of environmental regularities and identifying spatio-spectro-temporal biomarkers of imprecise inference. Foundational psychophysical research in autistic children has further documented a reduced central-tendency effect in time-interval reproduction, consistent with diminished influence of temporal priors in perceptual judgements.

Predictive Coding in Autism Spectrum Disorders publication trend

The graph below shows the total number of articles in predictive coding in autism spectrum disorders across all publications each year (not limited to Nature Index journals).

Technical terms

Predictive coding: A theoretical framework in which the brain minimises prediction errors by updating internal models of sensory inputs.

Prediction error: The difference between expected and actual sensory information, used to adjust future predictions.

Priors: Pre-existing beliefs or expectations that influence the interpretation of sensory data.

Bayesian inference: A statistical approach describing how prior beliefs and new evidence are integrated to form perceptual or cognitive judgements.

Efference copy: An internal copy of motor commands that predicts sensory consequences of self-generated actions for perceptual stability.

References

1. A predictive coding perspective on autism spectrum disorders. Frontiers in Psychology (2013).
2. Neural correlates of hierarchical predictive processes in autistic adults. Nature Communications (2023).
3. Erroneous predictive coding across brain hierarchies in a non-human primate model of autism spectrum disorder. Communications Biology (2024).
4. Central tendency effects in time interval reproduction in autism. Scientific Reports (2016).
5. Visuo-motor updating in individuals with heightened autistic traits. eLife (2024).
6. Is she still angry? Intact learning but no updating of facial expressions priors in autism. Autism Research (2024).