Probing the Probabilistic Reasoning and Belief Change Abilities of Preschool-Age Children

Thumbnail Image
Hilton, Brooke
Psychology , Development , Children , Preschool , Cognitive Development , Developmental Psychology , Belief Change
Research on cognitive development has challenged the assumption that children are passive participants in their own learning. Far from ‘passive sponges’, children are active participants in their learning process who select sources of evidence, actively weigh the strength and validity of the evidence they glean, and use those determinations to build and dynamically revise their beliefs. Past literature has generally investigated the belief change abilities of children using deterministic, experimenter-generated evidence in the lab (e.g., Bonawitz et al., 2019; Kimura & Gopnik, 2019; Macris & Sobel, 2017). Little is known about how children might use self-selected, probabilistic evidence to revise their beliefs over time. The current study investigated the ability of 3- to 5-year-old children to revise their beliefs on a novel ‘fishing’ game. Thirty-two children were induced with a belief about which of two possible ponds was most likely to result in catching a fish versus seaweed. After belief induction, children then continued to fish from the ponds in one of two conditions. In the “belief-consistent” condition, the pond that contained fish during the belief-induction phase continued to be advantageous for catching fish. In the “belief-inconsistent” condition, the opposite was true. Importantly, in both conditions, the advantageous ponds only resulted in catching fish 75% of the time, and thus was probabilistic. Results demonstrated that children in both conditions of the study used the self-generated, probabilistic evidence to either maintain or revise their beliefs. Children also showed sensitivity to the probability values of the two ponds, and showed evidence of probability matching behaviour when making their pond selections. Exploratory modelling analyses illuminated the potential impact of time on children’s pond selections, revealing that children in the inconsistent condition of the study experienced a period of lability in their pond selections before stabilizing on a revised belief. These findings are interpreted with respect to theory-theory approaches to cognitive development that emphasize children’s self-directed learning from experience. Future work in this area may expand on this work by investigating neural mechanisms of belief revision and the impact of reward-learning on children’s pond selections during the testing phase of the experiment.
External DOI