Main content
Coupling intrinsic and extrinsic rewards to account for context dependence in reinforcement learning
Date created: | Last Updated:
: DOI | ARK
Creating DOI. Please wait...
Category: Project
Description: When selecting among multiple alternatives, the subjective values of available options and their neural correlates tend to scale as a function of the range of available options in the choice set. This phenomenon, which we call context-sensitive valuation, has been documented even in tasks where option values must be learned through repeated exposure to outcomes and their associated probabilities. The present study investigates a novel, "intrinsically enhanced" model as an alternative to range adaptation (Bavard and Palminteri, 2022), and aims to compare the explanatory power of each model for context-sensitive valuation in reinforcement learning through an experimental setup purposefully designed to disentangle them with different qualitative predictions.