Taxonomic Categorization in Children with High-Functioning Autism Spectrum Disorder

Background: Taxonomic categories can be formed at different levels of inclusiveness. Basic-level categories contain the most within-class similarity of local details (e.g. cats vs. dogs), while superordinate level categories contain less similarity (e.g. animals vs. vehicles); finally, animate-inanimate level categories have the least within-class similarity (e.g. animates: people and animals vs. inanimates: vehicles and furniture). Few studies have examined the formation of these categories in children with autism.

Objectives: to examine taxonomic categorization in typically developing (TD) children and children with high-functioning autism (HF-ASD) across varying levels of category inclusiveness. The weak central coherence account would suggest superior performance on categorization tasks requiring attention to local details (e.g. basic-level categories), as well as difficulty on tasks requiring the ability to extract broader, more conceptual concepts (e.g. animate-inanimate categories).

Methods: Categorization at three levels of inclusiveness was tested using the matching-to-sample and object sorting procedures. In the matching-to-sample procedure, a sample and two comparison pictures were presented and children were instructed to touch the picture that is “the same kind of thing.”  A proportion of correct responses was computed for each category level. In the object sorting procedure two sample objects were placed in transparent plastic bowls and children categorized a set of new objects by placing each in the bowl with “the same kind of thing.” On this task, taxonomic categorization yielded a maximum score of 4.

Participants:  22 children with HF-ASD were matched with 21 TD children on non-verbal mental age (Differential Abilities Scale, Second Edition), M = 6.73 years, p = .97 and chronological age, (M= 7.15 years), p = .34. All diagnoses for children with ASD were confirmed using the Autism Diagnostic Observation Schedule (ADOS).

Results: Children with HF-ASD performed similarly to TD controls across categorization levels on the matching-to-sample task. However, both groups performed more poorly on animate-inanimate trials, compared to the superordinate and basic levels. On the object sorting task, the two groups performed differently across levels, whereby children with HF-ASD and TD controls performed similarly on basic and superordinate level categorization, but children with HF-ASD tended to perform better on animate-inanimate trials (M ASD = 2.86, M TD = 1.91), p = .07. This group effect was due to TD children creating thematic associations when forming object groups (e.g. putting people on the furniture, or inside the vehicles). 

Conclusions: No evidence for enhanced basic-level categorization or autism specific deficits in animate-inanimate categorization was found. Thus, the profile of categorization abilities demonstrated by children with HF-ASD did not fit the expected profile based on the weak central coherence hypothesis. Interestingly, children with HF-ASD maintained a taxonomic categorization strategy on the object sorting task, while TD children more flexibly adopted both taxonomic and thematic categorization strategies, reflecting the typical deficit in playful pretense observed in ASD children.