Social Visual Engagement in the First Six Months of Life: The Role of Contingency

Background: Typically-developing babies engage preferentially with social aspects of the environment from the first days of life. Examples include both their ability to distinguish adults looking at them from those who are not, as well as their preferential fixation, from at least 3 months of age, to the eyes of other people. An important goal of current research in autism should be to capitalize on these and other early-emerging mechanisms of sociability in order to trace the earliest detectable deviations from normative development. This is a key step in identifying autism at the earliest possible time point. With this goal in mind, recent research in our group has made use of videotaped actresses portraying infant-directed approaches. The videos were created to maximize naturalistic conditions while preserving experimental control. While the utility of these stimuli has been shown, particularly in work with 2-year-olds diagnosed with autism, these stimuli may not be optimal for maximizing between-group differences precisely because they still are not truly natural interactions: videotaped caregivers lack the contingency inherent in real-life, infant-parent interaction. More recently, we developed a testing laboratory that enables live interaction between mother and child, with bi-directional eye-tracking and direct line-of-sight gaze for both parties. In the current proof-of-concept study, we compare visual scanning for typically-developing 1- to 6-month-old infants in two conditions: watching videotaped actresses (Condition 1) and live interaction with mothers (Condition 2). We disambiguate the factors impacting on differences between the two conditions (identity of adult [stranger/mother] versus presence of contingency [videotaped/live]) by adding a third condition: a pre-recorded, and hence, non-contingent, video of the infant’s mother (Condition 3).

Objectives: This experiment is intended to test the hypothesis that visual scanning behavior in typically-developing children will evidence discrimination between contingent and non-contingent interactions within the first six months of life.

Methods: Eye-tracking data were collected during each of the three conditions (pre-recorded, live/contingent, and non-contingent). During collection of eye-tracking data, simultaneous video recordings captured the field-of-view of each participant (thus baby’s view showed mom, while mom’s view showed baby). Field-of-view recordings were coded into four regions (eyes, mouth, body and object) using fully-automated face-tracking and pattern-matching software. The eye-tracking data were then analyzed for time spent fixating on each of the four regions-of-interest.

Results: Preliminary results suggest that typically-developing infants, during contingent interactions with mothers (Condition 2), significantly increase their fixation on eyes as compared with their responses to pre-recorded clips of actresses (Condition 1). Additionally, initial findings show that these children increase their fixation on mouths when viewing non-contingent, pre-recorded clips of their own mothers (Condition 3). These results support our attempts to eliminate a potential confound in the data (lack of contingency, as well as recognition of identity vs. recognition of contingency).

Conclusions: Preliminary results suggest that live interaction increases eye-fixation in typically-developing babies. This experimental paradigm is likely to potentiate between-group differences relative to infants at-risk for autism, thus increasing its utility in the detection of early deviations from the course of normative social visual engagement.