Eye Tracking in Between: Gaze Behaviors on Structural Trial Components in Toddlers with ASD

Background:

Individuals with ASD look at social information atypically in multiple contexts and across their lifespan. These atypicalities have been quantified through eye tracking, typically through repeated presentation of experimental stimuli. Trials in these paradigms often begin with (1) a central fixation (CF) simultaneously displayed with salient sound to orient participants to a consistent visual starting point, (2) followed by the stimulus of interest, (3) followed by an interstimulus interval (ISI; e.g. blank screen) to disrupt visual processing. Little is known about the behaviors of individuals with ASD during these structural components of experimental trials.

Objectives:

To examine eye tracking data metrics collected from CFs and blank screen ISIs across several years of studies on young children with ASD in order to gain insights into attentional processes associated with developmental heterogeneity. More broadly, to discuss strategies for large-scale aggregation of eye tracking data, techniques to enable long-term consistency in data acquisition, and unique opportunities for studying early typical and atypical human development.

Methods:

Data were collected from toddlers (N=554) 1 to 3 years of age with confirmatory diagnoses of ASD (N=208, age=23.3months), developmental delay (DD, N=59, 23.5m), or typically development with or without a sibling with ASD (TDHR, N=57, 24.1m, and TDLR, N=230, 24.2m, for High Risk and Low Risk for ASD, respectively). CFs were analyzed for valid data retention (%Valid) and looking time at the CF (%CF) in 200ms bins up to 1s of total presentation time. Blank ISIs were examined for blinks and fixation counts as defined by a 1 degree 80ms distance dispersion algorithm. Linear mixed model analyses on over 2.5 million sifted rows of data with group x age x bin + Mullen nonverbal developmental quotient (NVDQ) fixed and intercept with bin slopes by subject nested in protocol random effects for CFs and group x age + NVDQ with subject intercept random effects nested within protocol for ISIs.

Results:

Toddlers with ASD looked longer at screens with CFs (%Valid) than DD (p<.05, d=.39), TDHR (p<.01, d=.31), and TDLR (p<.01, d=.20) toddlers, though they showed less looking with age compared to other groups (all p<.01). Similar results were found for %CF for age-related changes (p<.01), though only ASD-TDHR contrasts were significant (p<.01). For blank ISIs, toddlers with ASD also showed more Valid data and fewer blinks than TDHR (p<.001,p<.01) and TDLR (p<.001,p<.001) but not DD (p=.11,p=.37) groups, and more fixations than the TDHR group (p<.001). Higher NVDQ was associated with more %Valid, fewer blinks, and more fixations.

Conclusions:

Toddlers with ASD show increased looking towards scenes with salient non-social perceptual cues, as evident in CFs, but also during periods of visual inactivity. They also show greater exploratory activity during these blank trials as evident by more fixations, and may show decreased disengagement from the computer monitor in general, as evidenced by decreased blink rates and. These results contrast to opposing effects observed for social events (Shultz, Klin, & Jones, 2011; Chawarska, Macari, & Shic, 2012). Applications of explored approaches will be discussed.