Modeling changes in probabilistic reinforcement learning during adolescence.

Liyu Xia,Anne Gabrielle Eva Collins,Beth Baribault,Sarah L Master,Maria K Eckstein,Linda Wilbrecht,Ronald E Dahl,Stefano Palminteri

doi:10.1371/journal.pcbi.1008524

Liyu Xia, Anne Gabrielle Eva Collins + Show 6 more

Open Access

https://doi.org/10.1371/journal.pcbi.1008524

Copy DOI

Abstract

In the real world, many relationships between events are uncertain and probabilistic. Uncertainty is also likely to be a more common feature of daily experience for youth because they have less experience to draw from than adults. Some studies suggest probabilistic learning may be inefficient in youths compared to adults, while others suggest it may be more efficient in youths in mid adolescence. Here we used a probabilistic reinforcement learning task to test how youth age 8-17 (N = 187) and adults age 18-30 (N = 110) learn about stable probabilistic contingencies. Performance increased with age through early-twenties, then stabilized. Using hierarchical Bayesian methods to fit computational reinforcement learning models, we show that all participants’ performance was better explained by models in which negative outcomes had minimal to no impact on learning. The performance increase over age was driven by 1) an increase in learning rate (i.e. decrease in integration time scale); 2) a decrease in noisy/exploratory choices. In mid-adolescence age 13-15, salivary testosterone and learning rate were positively related. We discuss our findings in the context of other studies and hypotheses about adolescent brain development.

Highlights

In the everyday world, perfectly predictable outcomes are rare
While we found that performance increased during adolescence and stabilized in early adulthood in this stable probabilistic learning task, a probabilistic switching task in the same sample of participants [3] found a pronounced inverse U shape in overall performance, which peaked at age 13–15
We sought to examine the development of learning in a stable probabilistic environment using a large adolescent and young adult sample with continuous age in the 8–30 range

Summary

Introduction

Perfectly predictable outcomes are rare. Yet, we need to track important events and their relationships to other events and actions. Our ability to learn about these probabilistic relationships is crucial for our daily life and decision making. This challenge needs to be met by the developing brain, especially during adolescence [4,5,6,7,8,9]. One might assume that the brain gets better at this (and possibly all) forms of learning with brain maturation. Slower learning that integrates over a longer time scale may lead to more robust and stable performance in probabilistic environments. There may be periods where one form of learning is emphasized over the other. There may be inverted U shapes [3, 7, 12], that peak to support a sensitive period when specific information is available in the environment and/or when an organism needs to accomplish its transition to independence [13,14,15]

Results

Discussion

Conclusion