One of the markers of autism spectrum disorders (ASD) in infants and toddlers is impaired selective attention to faces1-6 (Preliminary Study, PS#1). This impairment diminishes their ability to learn from and interact adaptively with others in real-world environments.7 Attentional selection in the social domain relies, in part, on one?s ability to encode reward values of people and store these values in long-term memory as stable values.8- 10 The ?stable? values (henceforth, ?values?) are learned over the course of repeated learning opportunities, and once acquired, they are signaled rapidly, preferentially directing gaze to encoded faces of importance (high- value, HV) based on their hedonic or informative properties in the past.11 Automatic responses based on values stored in long-term memory are essential for survival when decisions have to be made rapidly (e.g., mother versus stranger). Learning about values is subserved by the reward learning system in the brain involving basal ganglia (BG) circuitry.11 This circuitry is implicated in the pathophysiology of ASD12, 13 and extant evidence suggests that individuals with ASD exhibit specific impairments in learning the reward value of social stimuli such as faces.14 (PS#2) Based on this evidence, we propose that limited attention to faces in toddlers with elevated autism symptoms (ASD+) is, in part, driven by impaired value learning in the social domain, affecting their ability to rapidly and preferentially select HV faces and ignore low-value (LV) faces in the complex real-world environment. Consequently, they exhibit diminished spontaneous attention to faces in general, and when they look at faces, they may distribute their limited attentional resources between high- (e.g., mother or therapist) and low-value (stranger) individuals in a trial-and-error fashion. We further hypothesize that reinforcing attention of children with ASD+ toward specific faces through social value training (SVT) will increase their attention to these faces in real-world environments. Here we propose to test a novel SVT in the ASD+ group at 18 months (n=48). In the proposed pilot study, ASD+ toddlers will undergo SVT using a gaze-contingent eye-tracking paradigm. SVT will be administered over a two-day period and the training effects will be assessed by changes in visual attention to HV faces as compared to LV faces between baseline, post-baseline (day 3), and a long-term follow-up (1 month) using two tasks: a laboratory selective attention (LSA) task and real-world selective attention (RWSA) task. We will also evaluate acceptability and feasibility of the value training and contribution of sex, nonverbal developmental level, and severity of autism symptoms to response to the training. Taking advantage of the overlapping cohorts between Projects 1 (neonatal imaging) and Project 5, we will examine if characteristics of resting-state functional connectivity at birth predict response to training in the ASD+ toddlers. To our best knowledge, this is the first study to directly target a marker for ASD by engaging the reward learning system and to evaluate functional integrity of neonatal connectome as a potential predictor of response to training at 18 months.
