A Model-Free Reinforcement Learning Implementation of Decision Making Under Uncertainty by Sequential Sampling

AI Summary

Model-free reinforcement learning algorithm implementing sequential sampling with an implicit decision boundary, learning when to commit or continue sampling under cost.
Model reproduces canonical perceptual decision features: accuracy and reaction time depend on evidence strength, and speed-accuracy trade-off adapts to payoff regimes.
Unifies learning and decision making within one framework, explaining flexible, context-dependent changes in behaviour and suggesting mechanisms for adaptability.

Neural Comput. 2026 Jun 22:1-32. doi: 10.1162/NECO.a.1543. Online ahead of print.

ABSTRACT

Although evidence integration to the boundary model has successfully explained a wide range of behavioral and neural data in decision making under uncertainty, how animals learn and optimize the boundary remains unresolved. Here, we propose a model-free reinforcement learning algorithm for perceptual decisions under uncertainty that implements a sequential sampling process with an implicit decision boundary. Our model learns whether to commit to a decision given the available evidence or continue sampling information at a cost. We reproduced the canonical features of perceptual decision making such as dependence of accuracy and reaction time on evidence strength, modulation of speed-accuracy trade-off by payoff regime, and many others. By unifying learning and decision making within the same framework, this model can inspire a new look at the mechanisms of flexibility in context-dependent changes of behavior.

PMID:42330493 | DOI:10.1162/NECO.a.1543

Document this CPD