A Computational Model of Children's Learning and Use of Probabilities Across Different Ages

Zilong Wang,Thomas R. Shultz,Ardvan S. Nobandegani
2023-05-07
Abstract:Recent empirical work has shown that human children are adept at learning and reasoning with probabilities. Here, we model a recent experiment investigating the development of school-age children's non-symbolic probability reasoning ability using the Neural Probability Learner and Sampler (NPLS) system. We demonstrate that NPLS can accurately simulate children's probability judgments at different ages, tasks and difficulty levels to discriminate two probabilistic choices through accurate probability learning and sampling. An extension of NPLS using a skewed heuristic distribution can also model children's tendency to wrongly select the outcome with more favorable items but less likely to draw the favorable ones when the probabilistic choices are similar. We discuss the roles of two model parameters that can be adjusted to simulate the probability matching versus probability maximization phenomena in children, and why frequency biases children's probabilistic judgments.
Neurons and Cognition
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to simulate and explain the development process of preschool children's non - symbolic probability judgment at different age stages. Specifically, the researchers used the Neural Probability Learner and Sampler (NPLS) system to model the experiment of O'Grady and Xu (2020), which explored the ability development of school - age children when dealing with probability reasoning tasks. The NPLS system can accurately simulate the probability judgments of children of different ages, including their performance under different tasks and difficulty levels. In addition, by adjusting the model parameters, the NPLS can also simulate the phenomenon that children tend to wrongly choose the option with more favorable items but a lower probability of being drawn when facing similar probability choices. This helps to understand how children learn and use probability and how age growth affects their decision - making strategies. The core problem of the paper lies in understanding how children learn to deal with uncertainty and probability information during the development process, especially when facing tasks that require making decisions based on probability. The research not only focuses on whether children can correctly identify options with higher probability, but also explores whether they are affected by frequency bias, that is, in some cases, they are more inclined to make decisions based on frequency rather than probability. Through these simulations, the researchers hope to reveal the mechanisms of probability learning and use in children's cognitive development.