Emotion Recognition Trajectories in Autism Spectrum Disorders: Effects of Internalizing and Externalizing Comorbidities

Background:  Summer months are crucial for youth with autism spectrum disorders (ASD; Cross, 2013).  Schools often mandate services in the summer to avoid “significant regression” of core deficits (Etschiedt, 2002).  Facial emotion recognition (FER) is a deficit domain in ASD that worsens over time (Lozier et al., 2014), and may be prone to such regression.  FER varies by intensity (e.g., Mazefsky, 2002) and stimulus type (child vs. adult; Nowicki & Mitchell, 1998), revealing information about FER dynamics.  Internalizing and externalizing symptoms are common in ASD (e.g., Noordhof et al., 2014), and negatively correlate with FER in non-ASD individuals (e.g., Rodemaker, 2000; Schmitt, 2000).  Longitudinally (in non-ASD samples), internalizing symptoms predict greater deficits in FER over time (e.g., Székely, 2014); externalizing symptoms show similar effects over time, though results are inconsistent (e.g., Bowen & Dixon, 2010; Székely, 2014).  Thus, internalizing and externalizing symptoms may affect trajectories of FER in ASD during summer months.  

Objectives:

1) To use longitudinal, objective measures to examine whether FER deficits in ASD worsen during the summer months.  We hypothesized that FER deficits would increase over time.

2) To examine the effects of internalizing and externalizing symptoms on FER trajectories.  We hypothesized that these symptoms would predict greater FER regression, with internalizing evincing a larger effect.

3) To explore if obtained patterns vary by stimulus type (intensity & age of faces).  

Methods:

Seventeen adolescents (ages: 10-17, M_age = 14.31, SD_age= 1.58) with ASD completed the computer-based Diagnostic Analysis of Nonverbal Accuracy-2 (DANVA-2; Nowicki, 2004) to assess FER every 3 weeks over an 18-week summer period.  Parents completed the Child Behavior Checklist (CBCL; Achenbach, 1991) as a measure of internalizing and externalizing symptoms at baseline.  Hierarchical Linear Growth Modeling was used to examine longitudinal change across DANVA-2 faces, with externalizing and internalizing symptoms as predictors.  Significant results were explored in terms of variation in intensity (high vs. low) and stimulus type (adult vs. child).

Results:

FER errors on the DANVA-2 decreased over time (β₁₀ = -.32, p < .01).  Greater internalizing symptoms predicted fewer errors (β₁₂ = -.04, p < .01); greater externalizing symptoms predicted more errors (β₁₃ = .04, p <.01; Figures 1 & 2).  When probing FER stimuli, the internalizing effect was found for low intensity (β₁₂ = -.03, p < .01), child (β₁₂ = -.03, p < .01) and adult (β₁₂ = -.01, p = .03) faces.  The externalizing effect was found for low intensity (β₁₃ = .02, p < .01) and adult (β₁₃ = .03, p< .01) faces.

Conclusions:

Contrary to our hypotheses, youth with ASD improved in FER over the summer, and this effect was especially great among higher internalizing symptoms; as expected, externalizing attenuated this trajectory.  The internalizing effect was strongest among subtle faces, while the externalizing effect was strongest among subtle and child faces.  This suggests that ASD youth with internalizing symptoms may become especially attuned to subtler faces over the summer, while those with externalizing symptoms may become especially insensitive to subtle faces and those of their peers.