Untapped Mathematical Learning Capacity in Children with Autism Spectrum Disorders

Background:

While mathematics is often cited as an area of preserved abilities for individuals with autism spectrum disorders (ASD), emerging research suggests this domain is not a universal strength for children on the spectrum. However, a child’s current math achievement reflects, in part, their educational history and thus may not capture their full learning potential. Assessing mathematical learning in children of varying achievement levels has the potential to inform the design of educational interventions and impact employment opportunities for individuals with ASD.

Objectives:

We sought to investigate learning trajectories in an arithmetic training task in children with ASD and relate them to standardized measures of mathematical achievement. We then assess whether children with ASD show the same relationship between learning rates and math ability as their typically developing (TD) peers.

Methods:

21 children with autism ages 8-11 and 19 age- and IQ- matched TD children participated in a five session one-on-one math training program focused on memorizing 14 double- plus single-digit arithmetic problems (47+9=56). Each session involved 14 exposures to each problem in a variety of physical and computerized settings. In a physical flash card task, children completed three rounds, where they verbally solved all 14 training problems, twice. After the first round, children were encouraged to “beat their time” on two subsequent rounds. To identify distinct learning profiles within each group, we performed a latent class growth modeling (LCGM) analysis using the average daily performance of each participant. Mathematical skills were assessed prior to training using the Broad Math composite of Woodcock Johnson Tests of Achievement- III.

Results:

In the ASD group LCGM identified three classes: two with slow initial starting reaction times (11.7 vs. 10.6 sec per problem), and a third, much faster class (3.9 sec). Between the two initially slow groups, one had a much faster learning rate (-1.6 vs. -1.2). Notably, this group had poorer math skills (87 on Broad Math) than the slower learning group (103). This pattern of larger learning gains for lower math skills was also captured by a significant correlation between math achievement and learning slope in the ASD group as a whole (r =.62, p=.003). In the TD group, LCGM also identified three classes, but they had more similar initial performance (4.4, 6.4, 6.0) and learning rates (-.4, -.9, -.6), and notably learning rates did not correlate with math ability (r=.43, p>.05).

Conclusions:

Emerging research suggests that math skills are not a universal strength in children with ASD. However, it remains to be determined if children with poorer math achievement could excel if presented with more optimal learning environments. In the current study we used one-on-one training, and a variety of activities to engage children in a math fact learning task. While all children improved on the task, only among children with ASD did we find larger learning gains for children with the lowest achievement levels. Together, these results suggest that children with ASD who struggle in math may have untapped mathematical learning potential.