AbstractRecognition of subtle emotional facial expressions is challenging for some individuals with autism spectrum disorders (ASD). Training that targets recognition of low intensity emotional expressions may therefore be effective as an intervention to improve social-emotional skills. This paper reports the results of two randomised controlled experiments looking at the effect of a training methodology designed to increase the recognition of happy emotion in low intensity happy facial expressions. The first study implements this training with children with autism spectrum disorder (ASD, N = 14) and the second study implements this training with adults with ASD (N = 27). The training paradigm used images from a morph sequence that mixed a happy expression with a mixed-emotion ‘norm’ expression to create a sequence of varying intensity happy expressions. Participants were asked to say whether or not individual faces from the sequence were happy, to measure their happiness detection threshold. Participants that received active training were given biased feedback to shift their detection threshold, while participants that received control training were given feedback consistent with their baseline threshold. There was some statistical evidence that thresholds in the active training group shifted more than in the control group. This suggests training was successful in increasing the number of expressions that individuals identified as happy. However, there was no evidence that training increased facial expression recognition accuracy, as measured by the Reading the Mind in the Eyes task completed after training (Study 2).
Publisher Correction: Neural network modeling of altered facial expression recognition in autism spectrum disorders based on predictive processing framework
