Performance on a Novel Kinect Emotional Choice Game Correlates with Broader Autism Phenotype Characteristics in Typically Developing Adults

Background: While video game overuse has caused difficulties in the lives of some individuals with autism spectrum disorders (ASD) (Mazurek et al., 2011), video gameplay may provide a useful way to measure social cognition. We designed a video game that physically engages participants in a fun, unstructured task that implicitly measures emotion processing, using the Microsoft Kinect platform. Difficulties in emotion processing and other autistic-like traits have been shown to vary beyond populations with ASD diagnoses, with ASD representing extrema (Lundstrom et al., 2012).

Objectives: To examine relationships among preferences for emotional information expressed through whole-body, volitional actions, the ability to recognize emotional information, and traits associated with the broader autism phenotype.

Methods: TD adults (N = 19; 5 females; age M = 22.7 years, SD = 6.1 years; FSIQ M = 111, SD= 21.85) played a video game on the Kinect platform using the custom-built Yale Interactive Kinect Environment Software (YIKES). Live video images of each participant were integrated in the game environment, which was projected on to a wall. The participants were asked to move left and right to position their images to catch emotional (fearful, happy, and neutral) faces “falling” from the top of the screen. A total of 280 face images were presented,divided over eight rounds of gameplay. Outcome variables were ratios of number of emotional (fearful or happy) faces, to total number of faces caught, in each of round. Given the large number of variables associated with gameplay dynamics and high levels of correlations among variables, we used principal component analysis (PCA) for dimensionality reduction. We examined correlations (Pearson’s) between the identified principal components (PCs) and performance on the Developmental Neuropsychological Assessment (NEPSY-II) Affect Recognition subtask, the Broader Autism Phenotype Questionnaire (BAPQ), and IQ (measured using the Wechsler Abbreviated Scale of Intelligence; WASI).

Results:  PCA revealed 3 factors accounting for 83% of all variance: principal component 1 (PC1: 51% of variance) represents diminished selection of negative (fear) content and increased selection of positive (happy) content; PC2 (21%) represents increased selection of any emotional content; and PC3 (11%) represents avoidance of negative content and selection of positive content specifically in mood-negative contexts. PC1 showed no correlations with any examined phenotypic characteristic. PC2 was associated with lower BAPQ scores (r = -.47, p < .05) and with higher Affect Recognition scores (r =.47, p < .05). PC3 correlated with performance IQ (r = .46, p< .05).

Conclusions:  Performance on a simple, emotional-choice, whole-body-movement video game can provide information about phenotypic characteristics of typical adults. Correlations between PC2 and BAPQ or AR scores suggest that preferential orienting towards emotional information is related to the ability to identify emotions in pictures of faces, and to fewer broader autism features in TD adults. These results provide evidence that video games may be used as an appealing platform for understanding cognitive ability, emotional processing, and social capabilities. Future directions include extension to children and to individuals with ASD, and exploration of therapeutic applications.