Isolating Visual- and Proprioception-Based Generalization of Motor Learning in Autism Spectrum Disorder

Background:  Individuals with autism spectrum disorders (ASD) show impaired acquisition of social communication skills, a defining feature of ASD; in addition, other more basic skills are also impacted. Development of new skills requires integrating information from multiple sensory modalities. This input is then used to form internal models of action that can be accessed when both performing skilled movements, as well as understanding those actions when they are performed by others. Learning skilled gestures is particularly reliant on integration of visual and proprioceptive feedback and may provide a critical developmental building block for social communication.

Objectives: We used a modified serial reaction time task (SRTT) to decompose proprioceptive and visual generalization components and examine whether the patterns of skill learning differ in ASD participants as compared to healthy controls (HC). 

Methods: Data were analyzed from 18 ASD participants (16 male, 38.72±18.362 years) and 11 healthy controls (6 male, 36.36±17.42 years) while performing a modified SRTT. Visual cues appeared in one of four positions, corresponding to one of four buttons on a response pad. To isolate different motor learning components, two separate sessions of the task were used for each participant. During both sessions participants trained using the dominant right hand. A test generalization period using the left hand followed where either the visual pattern of the sequence was kept constant (with the finger pattern mirrored) or the finger (proprioceptive) sequence was kept constant (with the visual pattern  mirrored). Diagnostic differences in generalization during visual- and proprioceptive-based sessions were assessed using a repeated measures two-way ANOVA on learning scores.  

Results: Analysis revealed a significant effect of generalization domain on reaction time learning scores (p=0.025) with greater generalization in the visual domain as compared to the proprioceptive domain. There was no main effect of diagnosis (p=0.284). However, there was a significant interaction between generalization domain and diagnosis on learning score (p=0.003). During the visual component, healthy controls showed greater improvement than ASD participants (p=0.010). Single sample t-tests revealed that only the control group showed significant visual generalization (HC: p=0.001; ASD: p=0.799). The two groups did not significantly differ in proprioceptive generalization (p=0.483), with neither group showing any significant learning effect (HC: p=0.637; ASD: p=0.618).

Conclusions: Findings revealed significant group differences when visual feedback was required for generalization of the learned motor sequence: healthy controls showed robust visual generalization whereas ASD participants demonstrated little generalization. In contrast, no group differences were observed when proprioceptive feedback was required for generalization, with both groups showing limited generalization. The findings suggest that when learning a motor sequence, individuals with ASD tend to rely less on visual feedback than do healthy controls. Visuomotor representations underlie imitative learning and action understanding and are thereby thought to be crucial to development of social skills and social cognition. It follows that anomalous patterns of procedural learning, with a tendency to discount visual feedback, may be an important contributor in the development of the core social communication deficits that characterize ASD.