The Emergence of Imitation From 9 to 12 Months in Younger Siblings of Children with Autistic Spectrum Disorders: Preliminary Findings From a Longitudinal Study

Background: Imitation deficits are well documented in autistic spectrum disorders (ASD; Rogers & Williams, 2006). Indeed, imitation deficits are among the predictors of ASD by as early as 12 months of age (Zwaigenbaum et al., 2005) and accuracy of early imitation is reduced in infants at risk for ASD (Young et al., 2011). However, little is known about the atypical, versus typical emergence of imitation over time. Few studies examined qualitative differences in pre-imitative behavior (approximations to imitation, Kaye & Marcus, 1981; Nichols, 2005). Moreover, no infant sibling study has examined emerging imitation during the critical developmental period between 9 to 12 months.   

Objectives: (1) To replicate a hierarchy of precursors of imitation in low-risk (LR) controls. (2) To extend the findings to a high-risk (HR) infant sibling sample, compared with LR controls. (3) To examine imitation differences in diagnostic groups (HR/LR) from 9 to 12 months.   

Methods: HR and LR participants were evaluated at 9 and 12 months of age using the imitation task from the Autism Observation Scale for Infants (AOSI; Bryson et al., 2008). The AOSI is a semi-structured play schedule that measures early signs of ASD. The examiner presented the infant with 3 actions (oral-facial movements or actions with objects) and 1 to 3 trials per action, depending on the infant’s successful performance. Video records of AOSI administrations were coded using Noldus Observer software and a novel detailed coding scheme. Imitation was scored in 3 ways: (1) total imitation, (2) best score across actions, and (3) approximations to imitation (predictable hierarchical patterns of behavioral responses to the model; e.g., touching model’s hands after model claps).   

Results: We compared infants’ imitation scores (i.e., total imitation, best score and approximations to imitation) using 2 (age) by 3 (diagnosis) mixed repeated measure ANOVAs. Infants were grouped according to 36-month outcome as: (1) siblings with ASD (ASD siblings); (2) siblings without ASD (non-ASD siblings); and (3) low-risk controls (LR). Preliminary results from the first 31 infants coded (n’s = 10 ASD siblings, 10 non-ASD siblings, 11 LR) replicated a hierarchy of ‘approximations to imitation’ in the LR group. Further, ASD siblings demonstrated fewer self-directed approximations than the LR controls. As expected, imitation performance increased with age [F (1, 28) = 8.79, p = .006]. A main effect of diagnostic group [F (2, 28) = 3.45, p = .046] revealed that ASD siblings imitated less frequently than did the non-ASD siblings and LR. There was no group by age interaction.    

Conclusions: The study provides initial evidence supporting the use of a hierarchy of approximations to imitation as an approach to studying the atypical emergence of imitation skills. Preliminary findings suggest HR-ASD siblings differ from non-ASD siblings and LR controls both in frequency of fully imitative acts and in quality/level of approximations to imitation. These differences are evident by age 9 months and remain at 12 months. The results may have implications for understanding psychological mechanisms underlying the emergence of imitation deficits, and for early detection and intervention in ASD.