Seeing the Error in my “Bayes”: A Quantified Degree of Belief Change Correlates with Children's Pupillary Surprise Responses Following Explicit Predictions

Abstract

Bayesian models allow us to investigate children’s belief revision alongside physiological states like “surprise”. Recent work finds that pupil dilation (or the “pupillary surprise response”) following expectancy-violations may be predictive of belief revision. How can probabilistic models inform interpretations of “surprise”? Shannon Information considers the likelihood of an observed event, given prior beliefs – suggesting stronger surprise occurs following unlikely events. In contrast, Kullback-Leibler divergence considers the “dissimilarity” between prior beliefs and updated beliefs following observations – with greater surprise indicating more change between belief states to accommodate information. To assess these accounts under different learning contexts, we use Bayesian models that compare these computational measures of “surprise” to contexts where children are asked to either predict or to evaluate the same evidence during a water displacement task. We find correlations between the computed Kullback-Leibler divergence and children’s pupillometry responses only when children actively make predictions, and no correlation between Shannon Information and pupillometry. This suggests that when children attend to their beliefs and make predictions, pupillary responses may signal the degree of divergence between a child’s current beliefs and updated, more accommodating beliefs.

Predicting Learning: Understanding the Role of Executive Functions in Children's Belief Revision Using Bayesian Models

“…Explicit predictions might increase the subjective value of an observed outcome, resulting in a stronger prediction error response when expectations are incorrect, subsequently increasing attention to the outcome (Brod et al, 2022). Predictions might help learners bring potential competing models to bear both prior to, and subsequent to, observed outcomes, enhancing surprise that is linked to belief updating (Colantonio et al, 2023; see also Rutar et al, 2023). Such accounts suggest that prediction may particularly support learners who do not already spontaneously engage in active cognition during learning or metacognitive engagement -by helping the learner focus on potential alternatives to currently held beliefs.…”

“…This is suggested to stem from its declarative nature -with higher cognitive effort and engagement needed for explicit predictions (Brod, 2021b). This conscious engagement with learning material and prior beliefs p29 may heighten the saliency of conflicting evidence (e.g., inducing a "surprise" response; Brod et al, 2018;Theobald & Brod, 2021;Colantonio et al, 2023) or increase the subjective value of the conflicting information (Brod & Breitwieser, 2019). This may then prompt a learner to better encode the new information when updating their views of the world.…”