A study published in Biological Psychiatry: Cognitive Neuroscience and Neuroimaging reveals that the wide range of sensory experiences reported by autistic individuals maps onto distinct patterns of information exchange in the brain. Researchers analysed 146 autistic children and adolescents aged three to twenty-one and found that specific functional connections between cortical and subcortical regions can predict a child’s sensory profile. The work was led by Western University researchers Matthew Kolisnyk and senior author Ryan Stevenson.
Differences in sensory processing are a defining feature for up to 87 percent of autistic people, with sights, sounds, textures, smells or tastes perceived at unusually high or low intensities. Some individuals seek out particular sensory input, while others go to great lengths to avoid everyday stimuli, and these variations can influence cognitive development and daily functioning.
In a prior large-scale study, scientists identified five distinct sensory phenotypes among autistic youth. The first phenotype is “sensory adaptive,” describing children who show minimal sensory differences from neurotypical peers. The second is “generalized sensory differences,” involving severe sensitivities across nearly all sensory domains.
The remaining three phenotypes describe more specialised sensory responses: one group is primarily sensitive to taste and smell, another features children who are generally underresponsive yet actively seek out intense sensations, and the final group is characterised by movement difficulties alongside low physical energy. Until now, it was unknown how these five behavioural profiles correspond to the brain’s internal wiring.
To explore this, Kolisnyk, Stevenson and their team examined whether these sensory profiles arise from unique patterns of brain activity. The researchers aimed to determine if different sensory behaviours align with distinct functional connectivity configurations within the brain.
Participants’ caregivers completed a standard clinical questionnaire detailing daily responses to various sensory stimuli. Based on these scores, each child was classified into one of the five sensory phenotypes. Resting-state functional magnetic resonance imaging was then conducted to observe brain activity while each participant rested quietly, enabling the mapping of functional connectivity through synchronous activation across regions.
The brain scans were divided into 200 cortical areas and 32 subcortical regions. Graph theory provided a mathematical framework to quantify how information moves across this network, with metrics describing the strength of connections and the efficiency of information transfer between regions.
Among the measures, researchers calculated betweenness centrality to identify hub regions that facilitate long-distance communication and the clustering coefficient to gauge how interconnected a region’s immediate neighbours are. These calculations yielded hundreds of data points for each participant, describing their unique brain network.
The team then fed these graph theory metrics into a machine learning model to test whether connectivity data could reliably identify each child’s sensory phenotype. The model was evaluated across all ten possible pairings of the five groups.
The algorithm successfully distinguished seven of the ten phenotype pairs based on brain connectivity alone, performing best when separating the most divergent profiles. The clearest separation occurred between the sensory adaptive group and the generalized sensory differences group.
Specific brain networks emerged as robust predictors of sensory profiles. Children in the sensory adaptive group showed stronger global connections in the prefrontal cortex, associated with decision making and executive function, and in the orbitofrontal cortex, which integrates sensory information with emotional rewards.
By contrast, those with generalized sensory differences exhibited stronger local connections within the somatomotor network, which handles touch and voluntary movement. Their primary sensory processing regions appeared highly interconnected locally but demonstrated weaker communication with the prefrontal cortex, suggesting potential differences in how higher-order brain regions modulate sensory input.
Distinct connectivity signatures also characterised the more specialised profiles. The isolated taste and smell sensitivity group showed altered hub activity in the posterior parietal cortex, a region linked to spatial reasoning and attention. Variations in broad attentional networks and the default mode network also helped the model separate several phenotypes.
Subcortical structures proved particularly informative for distinguishing certain groups. Regions such as the thalamus and the nucleus accumbens were highlighted, with notable differences in subcortical connectivity for the generalized sensory differences group and the taste and smell sensitivity group.
The study notes several caveats. The machine learning model failed to produce statistically significant distinctions for three phenotype pairs, including the underresponsive and movement-difficulties groups, indicating possible shared neural mechanisms or the need for a larger sample to detect subtler brain differences.
Relying on caregiver reports to classify sensory behaviours introduces subjectivity that may not perfectly reflect participants’ internal experiences. The study did not control for overall autism symptom severity, intelligence quotient, or co-occurring conditions, all of which could influence brain connectivity patterns. Additionally, head movement during scans can affect functional connectivity measurements, despite motion exclusion and correction procedures. Future work should test the predictive models on independent samples to verify their generalisability.
The study, “Decoding the Neural Basis of Sensory Phenotypes in Autism,” lists authors including Matthew Kolisnyk, Kathleen Lyons, Eun Jung Choi, Marlee M. Vandewouw, Bobby Stojanoski, Evdokia Anagnostou, Azadeh Kushki, Rob Nicolson, Elizabeth Kelley, Stelios Georgiades, Jason Lerch, Jennifer Crosbie, Russell Schachar, Muhammad Ayub, Jessica Jones, Paul Arnold, Xudong Liu, and Ryan Stevenson.
