An asynchronous, automated workflow for looking time experiments with infants

The study of infant gaze has long been a key tool for understanding the developing mind. However, labor-intensive data collection and processing limit the speed at which this understanding can be advanced. Here, we demonstrate a fully asynchronous, automated workflow for conducting Violation-of-Expectation (VoE) experiments. We first replicate four classic VoE experiments in a synchronous online setting, and show that VoE can generate highly replicable effects through remote testing. We then confirm the accuracy of a state-of-the- art gaze annotation software, iCatcher+ in a new setting. Third, we train parents to control the experiment flow based on the infant's gaze. Combining all three innovations, we then conduct an asynchronous automated infant-contingent VoE experiment. The automatic workflow successfully replicates a classic VoE effect: infants look longer at inefficient actions than efficient ones. We compare the resulting effect size and statistical power to the same study run in-lab and synchronously via Zoom. The automated workflow significantly reduces the marginal cost and time per participant, enabling larger sample sizes. By enhancing the reproducibility and robustness of findings relying on infant looking, this workflow could help support a cumulative science of infant cognition. Tools to implement the workflow are openly available.