Brain Organization Underlying Superior Math Problem Solving Abilities in Children with Autism

Background: In addition to social, communicative and sensorimotor processing deficits, it has been reported that individuals with autism spectrum disorder (ASD) can exhibit remarkable strengths, such as the ability to rapidly absorb and precisely remember tremendous amounts of details, or a particular interest towards regular sets with repeatable rules (i.e. “hyper-systemizing”). A synergistic interplay of these cognitive processes seem to have a role in predisposing to special abilities of savant sort. Mathematics represents the most concrete instantiation of these processes, as it is built upon systematic axiomatic procedures, and therefore represents an ideal domain to experimentally measure the cognitive and neural bases of superior-like abilities ― and their heterogeneity ― in ASD. Critically, the cognitive and brain mechanisms that might support such interplay ― and thus foster proficiency in math ― remain elusive. An emerging theoretical account of ASD has proposed that superior abilities in analytical and procedural thinking are attributable to greater engagement of primary sensory visual areas.

Objectives: We investigated: (i) whether greater engagement of primary sensory visual areas or their patterns of neural activity could explain better math abilities in children with ASD; and (ii) whether functional connectivity between these higher-order visual areas and medial temporal lobe systems implicated in procedural memory processes could support better math abilities in ASD.

Methods: We tested a group of 7-12 year old children with ASD, and a group of age-, gender-, and IQ-matched neuro-typical peers. Cognitive measures included standardized tests of calculation and problem solving, as well as strategy assessments. Brain measures included task-based functional magnetic resonance imaging (fMRI) during math problem solving.

Results: Behaviorally, we show that, on average, children with ASD showed superior math problem solving skills and relied on sophisticated analytical strategies more frequently than neuro-typical peers. Yet, not all individuals with ASD displayed proficient skills in math problem solving. Neurobiologically, we report differences between children with ASD and their neuro-typical peers in patterns of activity related to arithmetic problem complexity in the ventral temporal-occipital, parietal, and temporal cortices that support visuo-spatial, numerical, and mnemonic processes. Activation levels as well as multivariate activation patterns in ventral-temporal occipital cortex predicted individual differences in math problem solving abilities in ASD, but not in neuro-typical children. Remarkably, task-based multivariate connectivity analyses revealed that the interactions between regions of the ventral-temporal cortex, and the medial temporal lobe significantly and uniquely modulated individual math abilities in ASD.

Conclusions: Our results suggest that superior math abilities in children with ASD are engendered through a unique pattern of brain organization characterized by differential recruitment of posterior regions of the ventral stream and their functional coupling with memory systems. This is consistent with the hypothesis that analytical thinking is attributable to different engagement of posterior visual areas in ASD. Critically, our data extend this model to incorporate mnemonic systems residing in the medial temporal lobe as a locus for enhanced procedural thinking in ASD. More generally, these circuits can provide candidate biomarkers for explaining individual differences of superior cognitive skills in ASD.