Minecraft Working Memory Task: Considering Content in the Working Memory Abilities in School-Age, Higher-Functioning Children with Autism Spectrum Disorders

Background: Higher-functioning children with ASD (HFASD) often experience working memory (WM) problems (Geurts, de Vries, van den Bergh, 2014). These difficulties can be similar to those observed in children with ADHD (Geurts et al., 2004). Incorporating specific interests—such as video game content—into WM tasks may affect WM outcomes in both clinical groups but for possibly different reasons.

Objectives: Our study examined WM abilities in 9–18-year-old school-age children with HFASD compared to children with ADHD or typical development (TD) across two types of memory tasks—two standardized WM tasks and an experimenter-constructed task based off the video game Minecraft. This study examined the hypothesis that children with HFASD will perform better than controls on a memory task incorporating items of interest compared to performance on content-irrelevant WM measures. An alternative hypothesis is that children with HFASD demonstrate difficulties on a content-specific task due to syndrome-specific difficulties in WM.

Methods: The participants were 71 children with HFASD, 31 children with ADHD, and 38 children with TD. ASD symptoms were confirmed with the ADOS–2; ADHD symptoms were confirmed with the Conners-3. IQ was assessed with the WASI-II. WM abilities were assessed with Story and Verbal WM assessments of the Wide Range Assessment of Memory and Learning-2. The Minecraft WM task required participants to watch a 2-minute clip of Minecraft gameplay and to retell the video to an experimenter; participants were scored across a list of 35 possible items. Minecraft familiarity was assessed by self-report on a questionnaire.

Results: A MANCOVA controlling for IQ revealed a Diagnostic Group effect on WM measures, F(4,270)=3.84, Wilks' Λ=.90, p=.005, partial η²=.05. Univariate effects were observed for Story Memory, F(2,136)=5.61, p=.005, partial η²=.08; pair-wise comparisons with Sidak corrections indicated that the HFASD group performed significantly lower than the TD group (p=.003). No univariate effects were observed for Verbal WM. Almost all participants across groups reported being familiar and knowing about Minecraft (97%, 100%, and 100% for HFASD, ADHD, and TD, respectively). Within the HFASD group, VIQ (r=.43, p<.002), ADOS-2 Total score (r=-.41, p<.002), and Story Memory (r=.55, p<.002) were correlated with the Minecraft Recall score; these significant correlations remained after controlling for age. Within the ADHD group, only age (r=.44, p=.02) and Story Memory (r=.58, p=.001) were significant. An ANCOVA controlling for age revealed a Diagnostic Group effect on Minecraft recall score, F(2,136)=5.53, p=.005, partial η²=.08; the HFASD group performed significantly lower than the TD group (p=.003) but did not differ from the ADHD group.

Conclusions: This study demonstrated that children with HFASD display syndrome-specific WM difficulties regardless of the content of the WM task. The HFASD group demonstrated significantly lower performance on both the Story Memory scale and the Minecraft Recall task; Minecraft Recall performance in HFASD was related to verbal IQ, story memory, and ASD symptoms, while the performance in ADHD was related to story memory and age. Our study suggests that the memory difficulties observed in HFASD may be due to syndrome-specific impairments regardless of the task content.