The Attunement of Visual Salience From 2 until 24 Months in TD and ASD Infants

Background: Throughout development, infants filter environmental information by specifically attending to stimuli they perceive to be salient, thereby restricting the information they process and learn from. While previous studies of social engagement have identified social vulnerabilities in infants with autism spectrum disorders (ASD) by assessing global looking patterns, these methods fail to capture dynamic, context-driven changes in infants’ interest and attention. By dynamically assessing group agreement in visual scanning, we can define and quantify group tendencies in attention as well as individual deviation from group norms. Previous research suggests that dynamic measures of attention, as well as the timing of deviations in attention, may be useful in the early identification and characterization of ASD. 

Objectives: This study aims to map developmental trajectories of visual attention in typically developing infants and infants with ASD to 1) identify physical and social stimuli that are salient to infants at various developmental milestones and 2) investigate the developmental process by which children with ASD construct alternative schemas of salience during infancy. 

Methods: Data were collected prospectively and longitudinally from infants at high- or low-risk for ASD (infant sibling study design), and conventional diagnostic evaluations at twenty-four months defined 26 typically developing children (TD), and 13 children with confirmed ASD diagnosis. Eye-tracking data were collected during viewing of naturalistic movie scenes. Allocation of visual resources was quantified by kernel density analysis at each moment in time in TD children to create a continuously changing map of normative salience in relation to movie-content. This process was repeated to create differential landscapes of salience for infants with ASD.

Results: By 24 months of age, typical patterns of attention become stable and constitute a developmental baseline, providing a landmark for comparison in typically developing infants. In relation to this baseline, TD infants exhibit the following developmental milestones (in order of first emergence): dyadic attention to eyes (4 months), dyadic attention to mouths (5 months), triadic attention to faces (12 months), and triadic attention to body and gestures (18 months). During this period, TD infants gradually diminish their attention to physically salient stimuli in favor of more socially relevant content. In contrast, children with ASD begin to exhibit unique patterns of attention as early as 4 months of age, and individual measures of deviance from typical viewing patterns increase over time for children with ASD.

Conclusions: This research demonstrates that viewing patterns of TD infants serve evolving purposes related to their developmental goals. Additionally, the timing of attentional milestones suggests that triadic interactions become important later in development, and attention to faces is important earlier in their development than attention to body and gestures. This research also suggests that infants with ASD employ alternative attentional strategies during the first months of life that fail to capture certain experiences important for learning, resulting in deviant developmental trajectories. These measures of the dynamic allocation of visual resources may be useful in the early identification of risk for ASD as well as early treatment of the disorder.