Identifying Informative Environments for Cognition Parameter Inference via Bayesian Experimental Design

2026-07-30

AI Summary

This study addresses the challenge of automatically identifying experimental environments that are maximally informative for inferring cognitive parameters, thereby enhancing the efficiency and accuracy of Bayesian cognitive modeling. Framing environment design as a Bayesian experimental design (BED) problem, this work is the first to treat the environment itself as a design variable and introduces an amortized BED framework that substantially reduces computational costs while preserving inference performance. Integrating Bayesian inverse planning, Monte Carlo BED, and the Mouselab-MDP process-tracing paradigm, the proposed method accurately replicates the environment rankings produced by exact BED in Mouselab-MDP tasks. The results further demonstrate that no single environment is universally optimal across all inference objectives, revealing inherent trade-offs among different goals.

Abstract

Computational cognitive modeling seeks to infer latent cognitive mechanisms underlying observed behavior. Bayesian inverse planning provides a principled framework for such inference, but its success depends critically on the experimental environment. Existing approaches typically treat environments as fixed, leaving open the question of which cognitive experiments are most informative for cognition parameter inference. We formulate the design of cognitive planning experiments as a Bayesian Experimental Design (BED) problem, treating the experimental environment as the design variable. We establish an exact Monte Carlo BED benchmark and introduce an amortized Bayesian experimental design framework for efficient posterior inference and design evaluation. Experiments on the Mouselab-MDP process-tracing paradigm show that amortized BED closely matches the environment rankings of exact Monte Carlo BED while substantially reducing computational cost. We further show that no single environment is uniformly optimal across cognitive inference objectives, revealing trade-offs between expected information gain, posterior recoverability, and information efficiency. These results provide a principled framework for designing informative cognitive experiments for Bayesian parameter inference.

Problem

• Bayesian Experimental Design
• cognitive modeling
• parameter inference
• informative environments
• experimental design

Innovation

• Bayesian Experimental Design
• amortized inference
• cognitive modeling
• experimental environment optimization
• Mouselab-MDP