Optimizing adult guidance to facilitate children’s learning: Recent advances in developmental and computational cognitive sciences

Teaching and learning in everyday life are fundamentally social. When children observe demonstrations from adults, for example, what they learn from those demonstrations often depends on their inferences about the intentions and knowledge states of the adults, which in turn depend on the adults’ choice of pedagogical methods. This chapter summarizes the literature on the effects of different pedagogical methods on children’s learning and discusses insights from empirical investigations and computational models of pedagogical reasoning. Specifically, the Bayesian model of pedagogical reasoning has formalized teaching and learning in pedagogical settings where teachers intentionally choose examples to guide learners (Shafto et al., Cogn Psychol 71:55–89, 2014). Applying this model to research with children, studies have shown that presenting the same piece of information in different ways may lead children to learn differently. For example, whereas direct instructions are efficient in transmitting information, they may restrict children’s own exploration and discovery as the model predicted (Bonawitz et al., Cognition 120:322–330, 2011). On the other hand, reframing these instructions into “pedagogical questions” may facilitate both information transmission and further learning at the same time (Yu et al., Dev Sci 21, 2018). Building upon these studies, it has been suggested that computational modeling and data science tools could be used to study how adult guidance can be optimized in terms of timing and form, to facilitate children’s learning during everyday activities (Yu et al., Front Psychol, 9[1152], 2018). Such tools may have implications for both formal and informal education in Singapore and may open up new areas for future research.