Capacity Limits and Learning: Two Views of Altered Visuo-Motor Imitation in ASD

Background: Impairments in visuo-motor imitation have been recorded in ASD since the 1970's, and are believed by some to have a fundamental, causal role in the development of the disorder. What is less well understood is how these imitation deficits come about. Based on clinical experience, we have developed two hypotheses regarding the basis for these imitation deficits. The first centers around capacity limitations. Specifically, we have noted that children with ASD have a challenge imitating gestures when they are required to do two components simultaneously. On the other hand, we also suggest a framework for impaired gesture imitation in which children with ASD may have trouble building up representations of the novel gestures through impaired learning mechanisms.

Objectives: We tested these two hypotheses regarding the basis for altered imitation in ASD independently.

Methods: Study 1A: High-functioning children with ASD aged 8-12 years and age- and IQ-matched controls (n = 25+25) were asked to imitate 9 novel, meaningless gestures with their right hand in which movement elements (e.g., supinate the forearm, flex fingers 1 and 2) were all serial, and 7 gestures in which these elements were overlapping (simultaneous). Performance was recorded as correct or incorrect. Study 1B: 15 pairs of gestures were constructed in which there was a serial version and an simultaneous version. 7 children with ASD and 7 controls participated. Study 2A: 4 children with high-functioning ASD and 5 controls participated in a novel gesture learning task, in which they watched videos of a model performing a meaningless gestures. At a Go cue, they copied that gesture and were scored as performing it correctly or incorrectly. They repeated this cycle 5 more times with the same gesture. 12 gestures were tested. Study 2B: Similar to 2A, but with four repetitions of each gesture. 16 children with ASD and 17 controls participated.

Results: Study 1A: Main effects were seen for diagnosis (ASD had more errors than TD) and for gesture type (simultaneous gestures evoked more errors than serial). The central finding was the interaction effect: children with ASD had greater performance cost for simultaneous (over serial) gestures than controls. Partial eta squared: 0.21, p=0.001. This effect size stayed constant even when controlling for the total number of elements in each gesture. Study 1B: When the serial and simultaneous gestures were paired (i.e., the only difference was the relative timing of the two elements), there were no group differences in serial gesture error rates. The group x gesture-type interaction effect size was 0.42 (partial eta squared), p = 0.017. Study 2A: Children with ASD took an average of 4.0+/-2.3 trials to perform a gesture correctly; TD, 2.7+/-2.1 (p=0.013). Study 2B: Children with ASD took an average of 3.3+/-2.3 trials to perform a gesture correctly; TD, 2.3+/-1.7 (p=0.01).

Conclusions: There is independent evidence for both accounts of impaired imitation in ASD--a more restrictive capacity limitation, and impaired learning of the motor representation. Work using EEG to establish neural correlates is ongoing.