Prior Preferences in Active Inference Agents: Soft, Hard, and Goal Shaping
Neural Computation · 05.11.2026 · 1-29
DOI: 10.1162/neco.a.1605 ↗Kurz gefasst
Fachartikel in der Zeitschrift Neural Computation, online erschienen am 05.11.2026. Verfasst von Filippo Torresan und 2 weiteren Autor:innen.
Abstract (Originalsprache)
Active inference proposes expected free energy as an objective for planning and decision making to adequately balance exploitative and exploratory drives in learning agents. The exploitative drive, or what an agent wants to achieve, is formalized as the Kullback-Leibler divergence between a variational probability distribution, updated at each inference step, and a preference probability distribution that indicates what states or observations are more likely for the agent, hence determining the agent’s goal in a certain environment. In the literature, the questions of how the preference distribution should be specified and of how a certain specification affects inference and learning in an active inference agent have been given hardly any attention. In this work, we consider four possible ways of defining the preference distribution, either providing the agents with hard or soft goals and either involving or not goal shaping (i.e., intermediate goals). We compare the performances of four agents, each given one of the possible preference distributions, in a grid world navigation task. Our results show that goal shaping enables the best performance overall (i.e., it promotes exploitation) while sacrificing learning about the environment’s transition dynamics (i.e., it hampers exploration).
Zitieren
APA
Torresan, F., Kanai, R., & Baltieri, M. (2026). Prior Preferences in Active Inference Agents: Soft, Hard, and Goal Shaping. Neural Computation, 1-29. https://doi.org/10.1162/neco.a.1605
Vancouver
Torresan F, Kanai R, Baltieri M. Prior Preferences in Active Inference Agents: Soft, Hard, and Goal Shaping. Neural Computation. 2026:1-29. doi:10.1162/neco.a.1605
BibTeX
@article{Torresan2026,
author = {Torresan, Filippo and Kanai, Ryota and Baltieri, Manuel},
title = {Prior Preferences in Active Inference Agents: Soft, Hard, and Goal Shaping},
journal = {Neural Computation},
year = {2026},
pages = {1--29},
doi = {10.1162/neco.a.1605}
}
EUVIMED