Semantic and Visuospatial Analogical Reasoning in Autistic Children

Background: Studies have demonstrated intact fluid reasoning abilities, such as relation manipulation and logical solution inference, in children with an autism spectrum disorder (ASD) when assessed with analogical visuospatial reasoning problems (Dawson et al., 2007). Similar performance between ASD and typically developing (TD) children has also been found when semantic analogical problems are presented with pictures (Morsanyi & Holyoak, 2010), but literature supports preferential use of visuospatial reasoning strategies in ASD individuals (Stevenson & Gernsbacher, 2013). Only a few studies used equivalent formats of semantic and visuospatial analogical problems to compare the performance of ASD and TD children on both types of problems. Furthermore, few studies assessed how analogical reasoning develops in autism, but little evidences suggest slower reasoning development in ASD individuals (Green et al., 2014).

Objectives: (1) To assess performance in semantic and visuospatial analogical reasoning in ASD and TD children. (2) To document reasoning development in autism. 

Methods: 31 ASD and 42 TD children matched on age (6-13 years; M=10.00, SD=1.11) and on Raven’s Progressives Matrices’ percentiles (M=56.65, SD=24.66) completed 240 reasoning problems on a computerized task. Problems consisted of 2x2 matrices with last entry to be filled with one of the three responses options presented. Problems were composed of pictures and varied in content (semantic vs visuospatial) and complexity (0-, 1-, 2-relations to be manipulated to correctly solve the problem). Accuracy and reaction time (RT) were recorded. Mixed ANOVAs were conducted with Content and Problem Complexity as within-subject factors and the Group and Age (6-9, 10-13 years) as between-subject factors.

Results: In both ASD and TD groups, visuospatial content, increased complexity and younger age decreased accuracy and increased RT (p<.05). Moreover, for the 1-relation visuospatial problems, older children of both ASD and TD groups were more accurate than younger children (p<.05), while there was no significant difference between younger and older children for the semantic problems of the same level of complexity (1-relation). Overall, TD group (M=.89, SD=.08) was significantly more accurate than ASD group (M=.83, SD=.15), though the difference was relatively small. In terms of RT, ASD group (M=5485.84, SD=2113.11) was faster than TD group (M=6649.41, SD=1913.09) in the most difficult condition, 2-relations visuospatial problems (p<.05). There was a trend for ASD group being faster than ASD group in other conditions, but it did not reach significance.

Conclusions: Both ASD and TD groups showed a similar pattern of performance: increased accuracy with age, decreased accuracy with relational complexity and better accuracy for semantic than for visuospatial problems. In both groups, accuracy was lower for visuospatial reasoning problems than for semantic ones. While TD children were overall more accurate, ASD children were faster. Finally, considering language deficits and visuospatial reasoning strategies in autism, the present semantic over visuospatial advantage in ASD group suggests that the way problems are presented influences their performance. Pictured problems and matrix format seem to suit well their reasoning strategies.