Early diagnosis of Autism Spectrum Disorder (ASD) in toddlerhood is becoming more common since the development of early autism screening practices. However, the false positive rates of widely used screening tools remain high leading to many children without ASD who are referred to specialists. Given that access to specialists is limited due to long wait-lists and the dearth of specialists in certain regions, there is a pressing need to develop easily accessible and broadly applicable direct measures that will further streamline screening and diagnosis for at risk toddlers. Considering evidence that typical children as young as three months old make social interpretations about ambiguous shapes based on their movements, and that older individuals with ASD demonstrate reduced social interpretations compared to controls on similar tasks, performance on direct, nonverbal measures of social understanding may be a promising way to reliably differentiate ASD from other delays in very young children. Therefore, the primary aim of this study is to examine the utility of a novel, easily accessible direct measure of social understanding in reliably identifying ASD in a sample of at risk toddlers with a broad range of verbal and cognitive abilities. Participants include toddlers (18-36 months) considered at risk for an ASD based on the Modified Checklist for Autism in Toddlers (20 ASD and 20 non-ASD). Children are presented with two nonverbal social understanding tasks featuring interacting shapes; a live puppet task based on a paradigm that has been used in previous infant studies, and a novel, easily accessible, computerized version of this task presented on a touch screen computer tablet. In both tasks, children view a series of brief scenes featuring a green hill with three colored shapes (circle, square, and triangle); a climber, a helping shape and a hindering shape. Social interpretations will be measured by shape preference, and longer looking time toward a surprising event (i.e., climber approaches hinderer) versus an expected event (i.e., climber approaches helper). It is hypothesized that toddlers without a diagnosis of ASD will demonstrate evidence of social understanding on both tasks, whereas children with ASD will demonstrate reduced evidence of social attribution. Additionally, it is hypothesized that performance will predict a clinical diagnosis of ASD.