19448
Taking a Dimensional Measurement Approach to Test the Imitation Impairment Hypothesis in Autism
Despite more than 60 years of research which has consistently demonstrated an imitation impairment in autism spectrum disorders in many studies, the significance of imitation in social cognitive development remains poorly understood. It remains debated as to whether impairment in autism stems from an imitation-specific mechanism or is just an epiphenomenon of impaired social motivation, memory, attention or motor skills. Furthermore, studies of the imitation impairment have focussed on non-social or symbolic actions rather than social communication behaviours.
Objectives:
- To develop a robust experimental design for the measurement of imitation ability, that can control for other non-specific factors.
- To develop a method for measuring facial imitation ability in autism.
- To use these methods to examine imitation ability in autism.
Methods:
We recorded the movement of a pen on a computer tablet screen to provide accurate measures of movement parameters involved in drawing regular geometric shapes drawn at different sizes and speeds. By comparing the movement parameters of a model with those drawn by participants attempting to imitate these actions, measures of imitation fidelity could be provided. In a “ghost” control condition, relying on end-point re-enactment rather than imitation, the models actions were used to generate a dot which moved on the screen. Therefore, the 2 conditions were closely matched for demands placed on memory, attention and motor skills but one involved imitating an action, whilst the other required reproducing the movement of a dot. The experiment was conducted with 2 successive groups of young people, mainly adolescents with autism spectrum disorder, and age, sex and IQ matched controls.
In a second experimental design, designed to measure facial imitation ability, 2 sets of morphed facial expressions were generated. In each, 3 archetypal emotional expressions were blended in systematically varied amounts, such that each facial stimulus differed from the others to variable degrees, from being closely similar to markedly different. Imitation ability was measured by the accuracy with which blind raters could match an attempt at imitation to the correct model.
Results:
In the first experiment, error measures were generated for path-length (shape size) and movement speed (how fast the shape was drawn). For shape-size (but not speed), the error was significantly greater for both imitation and control conditions, but a significant interaction demonstrated a greater impairment specific to imitation (F(2, 70) = 8.36, p = .001, ηp2 = .19). In a second study with a different group of adolescents, principal component s analysis was used to extract an overall index of imitation ability which showed greater error in the ASD group compared to controls showed a group difference (t(36) = 2.42, p = .021). The ASD group also scored significantly higher on Facial Imitation Error (t(41) = 3.55, p = .001).
Conclusions:
There are deficits of imitation in autism above and beyond those associated with more general impairments that are evident in control conditions. Hence, hypotheses that problems with imitation might be explained through inadequate attention or motivation, or impairments of motor control are not supported by these studies.