Empirical Inference

Bounded rationality, abstraction and hierarchical decision-making: an information-theoretic optimality principle

2015

Article

ei


Abstraction and hierarchical information-processing are hallmarks of human and animal intelligence underlying the unrivaled flexibility of behavior in biological systems. Achieving such a flexibility in artificial systems is challenging, even with more and more computational power. Here we investigate the hypothesis that abstraction and hierarchical information-processing might in fact be the consequence of limitations in information-processing power. In particular, we study an information-theoretic framework of bounded rational decision-making that trades off utility maximization against information-processing costs. We apply the basic principle of this framework to perception-action systems with multiple information-processing nodes and derive bounded optimal solutions. We show how the formation of abstractions and decision-making hierarchies depends on information-processing costs. We illustrate the theoretical ideas with example simulations and conclude by formalizing a mathematically unifying optimization principle that could potentially be extended to more complex systems.

Author(s): Genewein, T and Leibfried, F and Grau-Moya, J and Braun, DA
Journal: Frontiers in Robotics and AI
Volume: 2
Number (issue): 27
Pages: 1-24
Year: 2015
Month: October

Department(s): Empirical Inference
Bibtex Type: Article (article)

DOI: 10.3389/frobt.2015.00027

BibTex

@article{GeneweinLGB2015,
  title = {Bounded rationality, abstraction and hierarchical decision-making: an information-theoretic optimality principle},
  author = {Genewein, T and Leibfried, F and Grau-Moya, J and Braun, DA},
  journal = {Frontiers in Robotics and AI},
  volume = {2},
  number = {27},
  pages = {1-24},
  month = oct,
  year = {2015},
  doi = {10.3389/frobt.2015.00027},
  month_numeric = {10}
}