Modeling Boundedly Rational Agents with Latent Inference Budgets

TL;DR

This paper introduces a latent inference budget model (L-IBM) that explicitly captures computational constraints of agents, enabling better modeling of suboptimal decision-making across diverse tasks and correlating with agent skill and difficulty.

Contribution

The paper presents the L-IBM, a novel approach that models agents' computational constraints explicitly using a latent variable, improving the understanding of suboptimal decision-making.

Findings

L-IBMs outperform Boltzmann models in various tasks.

Inferred inference budgets correlate with skill and difficulty.

L-IBMs are efficient and meaningful to compute.

Abstract

We study the problem of modeling a population of agents pursuing unknown goals subject to unknown computational constraints. In standard models of bounded rationality, sub-optimal decision-making is simulated by adding homoscedastic noise to optimal decisions rather than explicitly simulating constrained inference. In this work, we introduce a latent inference budget model (L-IBM) that models agents' computational constraints explicitly, via a latent variable (inferred jointly with a model of agents' goals) that controls the runtime of an iterative inference algorithm. L-IBMs make it possible to learn agent models using data from diverse populations of suboptimal actors. In three modeling tasks -- inferring navigation goals from routes, inferring communicative intents from human utterances, and predicting next moves in human chess games -- we show that L-IBMs match or outperform Boltzmann models of decision-making under uncertainty. Inferred inference budgets are themselves meaningful, efficient to compute, and correlated with measures of player skill, partner skill and task difficulty.