Assessing test-retest reliability of rule learning measures in infants

The ability to learn and generalize abstract algebraic rules (Marcus et al., 1999) has often been demonstrated in infants and is currently assessed in the large-scale multi-lab ManyBabies3 (MB3) study (Visser et al., accepted pending data collection). To study individual differences in infant development, it is essential to know how stably a certain ability can be measured over multiple measurement time points. Research on the reliability of infant looking time measures to date has been sparse and has found little or no evidence of test-retest reliability. Marimon & Höhle (2022) found relatively stable individual looking times across 3 test sessions with the same infants, but no correlations across sessions with regard to the manipulations of interest. Crista et al. (2016) reported good test-retest reliability for some but not all experiments in a larger battery of speech perception tasks with infants. Results from a previous ManyBabies study testing infant preference for infant-directed speech (Shreiner et al., 2023) found low test-retest reliability. Therefore, the aim of this study is to assess the test-retest reliability of looking-time measures of abstract rule learning in the context of the MB3 research project. (Note on contributions: Author list will be finalized after data collection at the time of finalizing the manuscript to be submitted. Authors of the final manuscript are expected to contribute to data collection and/or to provide substantive input in multiple stages of project preparation, data analysis, writing, and preparation of the final manuscript.)