Brain Activity Underlying Fluid Reasoning in School-Aged Autistic Children

Background: Autistic children perform as well as typically developing (TD) children at visual analogy tests measuring fluid reasoning (Green et al., 2014; 2016; Morsanyi & Holyoak, 2010; Sahyoun et al., 2009). Previous fMRI studies in adults revealed stronger posterior brain activations paired with weaker prefrontal and parietal activations in autistic relative to non-autistic participants during fluid reasoning (Soulières et al., 2009). However, the development of brain mechanisms underlying fluid reasoning has not yet been investigated in young autistic children.

Objectives: To investigate the cerebral activity patterns during a fluid reasoning task in autistic and TD children aged between 8 and 14 years old.

Methods: Sixteen (1 female) autistic and 14 (2 females) TD children (data collection in progress) equivalent in age (M=12.39, SD=1.81), Raven’s Progressive Matrices raw scores (M=40.23, SD=8.88) and percentiles (M=53.97, SD=30.14), and laterality were scanned in a 3T MRI scanner while completing a computerized reasoning task containing 168 problems. Problems were 4x4 pictorial matrices with the last entry to be filled with one of 3 response choices. The problems varied in content (either semantic or visuospatial content) and in complexity, implemented as the number of relations to jointly consider in order to solve the problem (0, 1 or 2 relations). SPM12 was used for preprocessing and whole-brain analyses (k=20 voxels, FWE corrected p=.05) of the fMRI data.

Results: Autistic and TD children did not differ in accuracy (p=.65) and response times (p=.93) at the reasoning task. There was a Content X Complexity interaction (p=.02), the 2-relations visuospatial condition being the hardest in both groups. Accuracy also correlated with age in both groups (r=.378). At the cerebral level, a large reasoning network was revealed in both groups, involving mainly occipital, temporal and parietal regions. Autistic children presented stronger activations in the pulvinar, left superior temporal and inferior frontal gyri than TD children, who showed stronger cingular and cerebellar activations. For the semantic problems, there was stronger pulvinar activity in the autistic group. For the visuospatial problems, the precuneus was more active in the TD group. In terms of complexity, there was no significant between-group differences for the 2-relations versus easier problems. In both groups, performance at most complex visuospatial problems was positively correlated with cerebral activity in the occipital lobe and inferior frontal gyrus.

Conclusions: In line with previous findings (Santarnecchi et al., 2017), fluid reasoning was supported by a network of frontal, temporal, parietal and occipital activations in both of our groups. However, even with similar fluid reasoning skills, autistic and TD children presented distinct cerebral activity patterns. In the semantic problems, autistic children seemed to rely on regions involved in visual cognition, such as the pulvinar (Bourne & Morrone, 2017), more so than TD children. During the visuospatial problems, the TD children may have relied more extensively than autistic children on the precuneus (involved in motor and/or visual imagery), a finding we previously reported in adults (Soulières et al, 2009). Thus, cerebral activity would differ between autistic and TD children according to the content of reasoning.