EUVIMEDEUVIMED entdecken
Crossref

Hierarchical Active Inference Using Successor Representations

Prashant Rangarajan, Rajesh P N Rao

Neural Computation · 21.08.2026 · 38 (9) · 1491-1544

DOI: 10.1162/neco.a.1545 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Neural Computation, online erschienen am 21.08.2026. Verfasst von Prashant Rangarajan und 1 weiteren Autor:innen. Beteiligte Einrichtung: Paul G. Allen Center for Computer Science and Engineering, University of Washington, Seattle 98195, USA prashr@cs.washington.edu. Die Arbeit nennt 55 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Active inference, a neurally inspired model for inferring actions based on the free energy principle (FEP), has been proposed as a unifying framework for understanding perception, action, and learning in the brain. Active inference has previously been used to model ecologically important tasks such as navigation and planning, but scaling it to solve complex large-scale problems in real-world environments has remained a challenge. Inspired by the existence of multiscale hierarchical representations in the brain, we propose a model for planning actions based on hierarchical active inference. Our approach combines a hierarchical model of the environment with successor representations for efficient planning. We present results demonstrating (1) how lower-level successor representations can be used to learn higher-level abstract states, (2) how planning based on active inference at the lower level can be used to bootstrap and learn higher-level abstract actions, and (3) how these learned higher-level abstract states and actions can facilitate efficient planning. We illustrate the performance of the approach on several planning and reinforcement learning (RL) problems, including a variant of the well-known four rooms task, a key-based navigation task, a partially observable planning problem, the mountain car problem, and PointMaze, a family of navigation tasks with continuous state and action spaces. Our results represent, to our knowledge, the first application of learned hierarchical state and action abstractions to active inference in FEP-based theories of brain function.

Autor:innen und Institutionen

  • Paul G. Allen Center for Computer Science and Engineering, University of Washington, Seattle 98195, USA prashr@cs.washington.edu.
  • Paul G. Allen Center for Computer Science and Engineering, University of Washington, Seattle 98195, USA rao@cs.washington.edu.

Zitieren

APA

Rangarajan, P., & Rao, R. P. N. (2026). Hierarchical Active Inference Using Successor Representations. Neural Computation, 38(9), 1491-1544. https://doi.org/10.1162/neco.a.1545

Vancouver

Rangarajan P, Rao RPN. Hierarchical Active Inference Using Successor Representations. Neural Computation. 2026;38(9):1491-1544. doi:10.1162/neco.a.1545

BibTeX
@article{Rangarajan2026,
  author = {Rangarajan, Prashant and Rao, Rajesh P N},
  title = {Hierarchical Active Inference Using Successor Representations},
  journal = {Neural Computation},
  year = {2026},
  volume = {38},
  number = {9},
  pages = {1491--1544},
  doi = {10.1162/neco.a.1545}
}
Kostenlos starten →