Developing a Novel Visual Search Task Using Personalized Interests

Poster Presentation
Thursday, May 2, 2019: 5:30 PM-7:00 PM
Room: 710 (Palais des congres de Montreal)
B. Silver, E. Clark-Whitney, E. Barnes, M. M. Conte, J. D. Victor and R. M. Jones, Weill Cornell Medicine, New York, NY
Background: Circumscribed interests are a common feature of Autism Spectrum Disorder (ASD), and can significantly interfere with daily functioning. Previous work in our lab has demonstrated that children with ASD have decreased cognitive flexibility to cues that represent their interests compared to Typically Developing (TD) children (Bos et al., under review), suggesting cognitive interference. One hypothesis as to why interests have greater affective salience in ASD is that individuals with ASD may show increased visual processing abilities for interests, similar to how TD individuals show enhanced processing skills for faces.

Objectives: Building upon prior work demonstrating that experts on a specific topic perform better on visual search tasks related to that topic, we created a novel visual attention task personalized to each individual’s circumscribed interest. The goal was to determine whether low-level visual features were processed differently for interests in ASD vs. TD. We hypothesized that participants with ASD would perform better than TD participants due to the increased visual processing abilities that ASD individuals have for interests.

Methods: 29 individuals (5-30 years of age, mean = 15.38; 14 ASD) completed a novel visual search task personalized for his or her circumscribed interest. In the lab, participants were required to locate an unscrambled image in a group of scrambled distractor images as quickly as possible, presented on an iPad. Each trial had either 3, 15, or 35 distractor images, with a total of 108 trials. There were 3 conditions: pictures of faces, pictures of houses, and pictures of the participant’s interest. Accuracy and reaction times were calculated for the three array sizes and by each condition. Caregivers completed the RBS-R as well as a novel questionnaire asking about the intensity and prior history of circumscribed interests. Analyses were conducted using Matlab.

Results: The effect of the array size on performance, calculated as the slope of reaction time over number of array elements, was significantly lower for the face condition as compared to the house condition and the interest condition for both ASD (p = .001, p < .001) and TD (p < .001, p = .030). The effect of the array size on performance was significantly correlated with age for the house condition (R2 = .600, p = .018) and the interest condition (R2 = .639, p = .010) in the TD group, and for the interest condition (R2 = .812, p < .001) in the ASD group. There was no effect of array size for the face condition in either group. In addition, there were no significant differences in any condition between the ASD and TD groups.

Conclusions: Our current task did not demonstrate significant difference in performance between ASD and TD individuals, suggesting that the difference in affective salience occurs at a different, higher level of visual processing than our current low-level visual processing task required. In the future, we will further probe at what level of the visual stream differences emerge between ASD and TD individuals for processing cues that represent one’s interests.