Comparing Online Webcam- and Laboratory-Based Eye-Tracking for the Assessment of Infants’ Audio-Visual Synchrony Perception

• In Event: Poster Session 10
  In Poster Session: PS 10 Topic - Methods, History, Theory

Abstract

The current worldwide pandemic situation necessitated a change to online data collection methods for many psychological research labs. Previous research with adults showed that web cameras can be employed just as effectively as eye-tracking cameras to track eye movements on decisional tasks (Bott et al., 2017). Collecting eye-tracking data online entails higher variance and lower sampling rate, but no significant difference regarding spatial accuracy in comparison to in-lab recordings (Semmelmann & Weigelt, 2018). Research with infants demonstrated that looking time paradigms can be successfully applied online using webcam-based video recording (Semmelmann et al., 2017; Tran et al., 2017). However, the use of webcam-based eye-tracking and its comparison with in-lab eye-tracking is currently lacking in infancy research. Thus, the goal of the current study is to examine whether webcam-based eye-tracking is a feasible method to assess infants’ audio-visual synchrony perception. We will assess 4-month-old infants (N=60) in a preferential looking paradigm: one group will be tested online using webcam-based eye-tacking and video recording via the Labvanced platform (Finger et al., 2016), while the other group is currently assessed in a laboratory setting using an EyeLink 1000 Plus eye-tracker (SR Research) and a camera (Sony Action Cam). Our paradigm aims to reveal if infants can detect temporal synchrony more easily between simple audio-visual stimuli in contrast with complex stimuli. During the procedure, infants are presented with two side-by-side videos, one of which is synchronous (S) while the other is asynchronous (AS) with a musical stimulus (a children’s song). In the simple condition, infants see an infant being bounced to a simple version of the song. In the complex condition, infants see a person dancing to a complex version of the song. Each condition comprises of 6, 20-s trials. Difference scores (DS) between relative looking times to the S and AS stimulus will be calculated across trials for each condition (i.e., positive score = preference for S displays; negative score = preference for AS displays). Webcam- and in-lab based eye-tracking data will be compared in terms of data quality, viewing behaviour and experimental effects. First, we hypothesise that data quality will be similar between the two groups. To test this, we will compare video data for completeness, frame rate, resolution, and brightness (Semmelmann et al., 2017); and eye-tracking data for the frequency of calibration and technical issues between groups. Second, we expect that viewing behaviour of infants will be independent from the method used. To evaluate this, we will compare infants’ attention retention (ratio of on- and off-screen viewing coded from videos) and eye-tracking dwell times across trials between groups. Third, we anticipate that experimental effects will manifest in higher absolute DS for the simple compared to the complex condition irrespective of the method used. Thus, we will compare the two groups on DS differences between conditions. Our study will evaluate the feasibility of online infant eye-tracking. Our findings will contribute to the development of online eye-tracking paradigms that can widen participation and increase data reproducibility in infancy research.

Authors

• Anna Bánki, University of Vienna
  Non-Presenting Author

• Martina de Eccher, University of Vienna
  Non-Presenting Author

• Stefanie Hoehl, University of Vienna
  Non-Presenting Author

• Gabriela Markova, University of Vienna
  Presenting Author