Child-Led Play Supports Adults’ Exploratory Causal Inferences

• In Event: Parent-Child Interactions During Digital and Physical Playful Learning

Abstract

Prior research has demonstrated that children are better than adults at learning certain unconventional causal forms, suggesting that hypothesis search in early development is more flexible and eclectic than adults (Lucas et al., 2014; Gopnik et al., 2017). Specifically, young children are more likely to consider unconventional conjunctive (“AND”) causal relationships (i.e., multiple causes jointly produce an effect) than adults (Lucas et al., 2014; Wente et al., 2017). Do children draw flexible causal inferences in their explorations, and does it promote more creative and flexible thinking in causal learning for observing adults? In two experiments, we examined whether observing children’s exploratory causal explorations promote adults’ flexible causal learning.
In Experiment 1, 72 parent-child dyads (children aged 4-6 years) played an interactive tablet game involving a machine and objects. During training, a parent-child dyad saw an experimenter demonstrate events that were statistically best explained by the unusual conjunctive relationship. That is, the machine was activated by some combinations of objects but not by each of the objects individually. The experimenter then introduced a new set of objects and asked the parent-child pair to explore the objects independently (Solo), or jointly through interactions led by the child (Child-Led) or by the parent (Parent-Led). Next, the experimenter asked the parent and child to individually generate the effect using the objects (i.e., “Which of these objects would you use to turn on the machine?). To measure parents’ and children’s openness to consider the unfamiliar causal form, we coded whether two or more objects were used during the exploration trials and the final intervention. Overall, children outperformed parents in generalizing the unusual conjunctive rule regardless of condition: Children were more likely than adults to explore objects jointly, b=0.56, p=.021 and use multiple objects to activate the machine in the test, b=1.82, p<.001. Compared to children, parents used more unique combinations of objects, t(72.9)=2.36, p=.021. Critically, whether parents led or observed influenced their reasoning. The likelihood of using multiple objects to activate the machine was higher for parents in the Child-Led group than those in the Parent-Led group, b=1.44, p=.022.
In Experiment 2, we investigated whether “child” evidence is sufficient to support success in adults’ flexible causal reasoning. A group of 96 adults saw an adult model perform interventions, but we yoked the evidence to the specific interventions attempted by either children or parents of Experiment 1. Participants who observed child-yoked interventions performed by the adult model provided more correct conjunctive responses than participants who observed parent-yoked interventions, b=0.91, p=.033. Thus, child-yoked interventions performed by the adult model similarly improved adults’ flexibility in causal form inferences.
Consistent with previous research, children were better than adults at learning the unusual conjunctive relationship, suggesting that children make more flexible causal inferences than adults. Our research revealed that children’s broad, exploratory hypotheses were also reflected in their explorations. Furthermore, our findings suggest that observing evidence generated by children may help adults to be more flexible and open-minded in their causal learning.

Authors

• Koeun Choi, Virginia Polytechnic Institute and State University
  Presenting Author

• Milagros Grados
  Non-Presenting Author

• Elizabeth Bonawitz, Rutgers University - Newark
  Non-Presenting Author