Objectives: To assess atypical processing of facial and vocal affect in the autistic spectrum, using a recently validated set of ecological stimuli comprised of dynamic visual and auditory (non-verbal) vocal clips of emotional expression (Belin et al., 2008; Simon et al., 2007).
Methods: Twelve autistic, 10 nonautistic, and 11 Asperger syndrome participants matched for chronological age, full-scale IQ, gender, and handedness were asked to distinguish between facial and vocal affect stimuli representing disgust and fear as quickly and accurately as possible. Participants were presented with stimuli expressing fear or disgust (produced by trained actors) either visually (dynamic facial expression), audibly (non-linguistic vocal clip), or bimodally. For bimodal presentations, the same emotion was presented in both modalities for congruent conditions, while different emotions were presented in the two modalities for incongruent conditions. Signal-to-noise ratio scores for both visual and auditory signals resulting in an 85% discriminability (fear vs disgust) level of performance were also collected using the QUEST adaptive staircase method. In a second part of the experiment, the same procedure was repeated for each participant using visual and auditory information whose saliency was adjusted for (i.e., equally salient) based on thresholds previously obtained using the QUEST adaptive staircase.
Results: Results demonstrate a tendency for less efficient processing of emotional expressions in all conditions for the autistic group, but not the Asperger group, when compared to controls. Moreover, autistic participants necessitated an increased signal-to noise ratio to discriminate at an 85% correct level between fear and disgust. For incongruent bi-modal conditions, only autistic and Asperger participants preferentially categorized the affective expression using visual cues, suggesting a strong visual dominance for the processing of dynamic emotion information.
Conclusions: Our results suggest a reduced ability to categorize dynamic emotion expressions in autism, whether via visual or auditory channels. We are presently assessing whether the nonlinear probabilistic summation for the congruent bi-modal condition will differ in autistic spectrum and control groups by looking for a violation of the “race model” (an indicator of the level of neural integration of different sensory-emotional information) before and after equating the saliency of auditory and visual information for each participant. This is an important methodological consideration since we will be able to dissociate whether a putative impairment in multisensory emotion processing is contingent on less efficient processing in either of the perceptual channels.