The Effects of Robot-Child Interactions on Imitation and Praxis Performance of Typically Developing (TD) Children and Children with Autism Spectrum Disorders (ASDs) Between 4-10 Years of Age

Background:  

While social communication delays are the primary impairments of Autism Spectrum Disorders (ASDs), motor impairments are quite frequent. Motor impairments in ASDs include incoordination and dyspraxia, namely difficulties in performing complex actions like skilled gestures. Traditional autism interventions address social communication impairments; they are intensive (30-40 hours/week) and socially demanding. However, motor impairments are rarely addressed in these contexts. In contrast, an interaction game with a robot could be a highly motivating context and a simpler social interaction for children with ASDs. We are currently developing a novel embodied social intervention involving robot-adult-child interactions during which children copy various complex actions performed by the robot based on karate or dance themes.

Objectives: To examine the effects of a novel robot-child intervention program on praxis and imitation performance in typically developing children and children with ASDs/ADHD.

Methods: Fifteen typically developing children and four children with ASDs/ADHD were examined. Each child received eight, 30-minute sessions of robot-adult-child interaction training across four weeks. The postural praxis subtest of Sensory Integration and Praxis Testing (SIPT) is a normed and standardized measure of gross-motor praxis and was used to examine generalized changes in praxis. We also examined task-specific improvements in imitation within the training context by offering five novel actions during a pretest and a posttest, before and after the training period. Dependent variables for SIPT included errors in modulation, directness, use of body part, as well as spatial aspects of movement. Percent imitation error for each robot action was also measured. Pre and posttest measures were obtained for each variable.

Results: In terms of generalized SIPT performance, Wilcoxon non-parametric tests showed an effect of training with a decrease in spatial (p<0.05), body part (p<0.05) and total (p<0.1) errors in the posttest compared to the pretest. In terms of task-specific performance during robot imitation, a two-way repeated measures ANOVA showed a significant effect of training with lower errors in the posttest as compared to the pretest (p<0.05, effect size=0.362). Case reports on two high-functioning children with ASDs and one high-functioning child with ADHD reveal improvements in both SIPT as well as task-specific robot imitation. Moreover, one child with low-functioning ASD dramatically improved his performance in the SIPT as well as task-specific, robot-imitation.

Conclusions: Our first study on robot-child interactions was a proof of concept study and helped develop a structured training paradigm and a sound testing protocol. Our results suggest moderate improvements in the posttest scores of TD children and the majority of the children with ASDs/ADHD.  The moderate improvements may be attributed to a short-training period (i.e., only 8 sessions of training). Our future research proposal is to conduct a randomized controlled trial in children with ASDs who receive robot-child interaction training in addition to the traditional interventions for an extended period while being compared to a comparison group. Thus, our preliminary data suggest that the robot-adult-child interaction context appears to be a promising tool for facilitating imitation and praxis in children with ASDs.