Toward a large-scale collaboration for infant online testing: Introducing ManyBabies-AtHome

• In Event: Updates from ManyBabies: Collaborative efforts to promote best practices and improve robustness in developmental science

Abstract

Laboratory studies of infant development frequently rely on the measurement of infant gaze behavior in paradigms like preferential looking, visual habituation, or looking-while-listening. Classically, such experiments are conducted under - at least per lab - highly standardized conditions in order to control factors such as distance between participants and screens, movement, lighting, and ambient noise.
At present, most psychology experiments use participants from a small section of the world’s population, and we do not know if they will generalize. Online testing holds great promise towards more ecologically valid, more highly powered and more representative experiments. First, infants are tested in a familiar environment and, in an asynchronous set-up, at the caregiver’s time of choice. Second, caregivers do not face the burden of a lab visit and associated travel, making it easier to access populations beyond those living in proximity of an infant lab. Third, this increased accessibility enables testing of more diverse populations, both within and beyond the researcher’s home country.
Recently, there have been several promising initiatives to move experiments online to allow caregivers to participate from their home environments (e.g., https://lookit.mit.edu/, Scott & Schulz, 2017). Despite tremendous advances in at-home testing, there are significant obstacles, especially when aiming to increase accessibility to a wide range of participants. For example, remote data collection of infant audio and video data poses ethical issues ranging from recruitment, consent and reimbursement, to privacy protection and data storage. Such issues are necessarily bound to institutional and national regulations. In addition, in order to be inclusive, any at-home solution should be able to accomodate a large variation in home environments, including lighting and type of digital device. This issue poses challenges ranging from standardized stimulus display to analysis of video data, especially when venturing into automatic gaze coding.
To address these challenges, we introduce ManyBabies-AtHome. This project aims to collaboratively address the challenges of infant online testing. We will use Lookit, a solution which provides a flexible and secure testing environment for online infant looking studies, but is to date optimized for use in the US and/or with English-speaking populations. ManyBabies-AtHome will advance current online-testing methods by (1) establishing and translating generally applicable solutions in procedure, documentation, standardization and analysis to make this testing method accessible and robust across a range of home environments across the world, (2) collecting and annotating a large dataset of infant online gaze data that can be exploited for the development of automatic coding approaches. Our first proof-of-concept study will be a visual preference paradigm (Figure 1), designed with the purpose of assessing the feasibility of our proposed workflow for asynchronous online testing (Figure 2), as well as the robustness of online looking methods across populations. We will follow up with studies incorporating synchronized audio and video stimuli. These studies will also yield a dataset that can be used for training and evaluating automatic methods for coding infant looking behaviour in the noisy home environment.

Group Authors

The ManyBabies-AtHome consortium

Authors

• Sho Tsuji, The University of Tokyo
  Presenting Author

• Christina Bergmann, Max Planck Institute for Psycholinguistics
  Non-Presenting Author

• Helen Buckler, University of Nottingham
  Non-Presenting Author

• Rhodri Cusack, Trinity College, Dublin
  Non-Presenting Author

• Lorijn Zaadnoordijk, Trinity College, Dublin
  Non-Presenting Author