Free-energy minimization in joint agent-environment systems: A niche construction perspective

Jelle Bruineberg (Corresponding Author), Erik Rietveld, Thomas Parr, Leendert van Maanen, Karl J. Friston

Research output: Contribution to journalArticleAcademicpeer-review

3 Citations (Scopus)
90 Downloads (Pure)

Abstract

The free-energy principle is an attempt to explain the structure of the agent and its brain, starting from the fact that an agent exists (Friston and Stephan, 2007; Friston et al., 2010). More specifically, it can be regarded as a systematic attempt to understand the ‘fit’ between an embodied agent and its niche, where the quantity of free-energy is a measure for the ‘misfit’ or disattunement (Bruineberg and Rietveld, 2014) between agent and environment. This paper offers a proof-of-principle simulation of niche construction under the free-energy principle. Agent-centered treatments have so far failed to address situations where environments change alongside agents, often due to the action of agents themselves. The key point of this paper is that the minimum of free-energy is not at a point in which the agent is maximally adapted to the statistics of a static environment, but can better be conceptualized an attracting manifold within the joint agent-environment state-space as a whole, which the system tends toward through mutual interaction. We will provide a general introduction to active inference and the free-energy principle. Using Markov Decision Processes (MDPs), we then describe a canonical generative model and the ensuing update equations that minimize free-energy. We then apply these equations to simulations of foraging in an environment; in which an agent learns the most efficient path to a pre-specified location. In some of those simulations, unbeknownst to the agent, the ‘desire paths’ emerge as a function of the activity of the agent (i.e. niche construction occurs). We will show how, depending on the relative inertia of the environment and agent, the joint agent-environment system moves to different attracting sets of jointly minimized free-energy.

Original languageEnglish
Pages (from-to)161-178
Number of pages18
JournalJournal of Theoretical Biology
Volume455
DOIs
Publication statusPublished - 14 Oct 2018

Fingerprint

Niche
Energy Minimization
Free energy
Free Energy
niches
Joints
energy
Markov Chains
statistics
foraging
brain
Canonical Model
Path
Simulation
Generative Models
Foraging
Markov Decision Process
Brain
Inertia
State Space

Keywords

  • Active inference
  • Adaptive environments
  • Agent-environment complementarity
  • Desire paths
  • Free energy principle
  • Markov decision processes
  • Niche construction

Cite this

Bruineberg, Jelle ; Rietveld, Erik ; Parr, Thomas ; van Maanen, Leendert ; Friston, Karl J. / Free-energy minimization in joint agent-environment systems : A niche construction perspective. In: Journal of Theoretical Biology. 2018 ; Vol. 455. pp. 161-178.
@article{4da6b96933c84ffaad88ed07050e08ad,
title = "Free-energy minimization in joint agent-environment systems: A niche construction perspective",
abstract = "The free-energy principle is an attempt to explain the structure of the agent and its brain, starting from the fact that an agent exists (Friston and Stephan, 2007; Friston et al., 2010). More specifically, it can be regarded as a systematic attempt to understand the ‘fit’ between an embodied agent and its niche, where the quantity of free-energy is a measure for the ‘misfit’ or disattunement (Bruineberg and Rietveld, 2014) between agent and environment. This paper offers a proof-of-principle simulation of niche construction under the free-energy principle. Agent-centered treatments have so far failed to address situations where environments change alongside agents, often due to the action of agents themselves. The key point of this paper is that the minimum of free-energy is not at a point in which the agent is maximally adapted to the statistics of a static environment, but can better be conceptualized an attracting manifold within the joint agent-environment state-space as a whole, which the system tends toward through mutual interaction. We will provide a general introduction to active inference and the free-energy principle. Using Markov Decision Processes (MDPs), we then describe a canonical generative model and the ensuing update equations that minimize free-energy. We then apply these equations to simulations of foraging in an environment; in which an agent learns the most efficient path to a pre-specified location. In some of those simulations, unbeknownst to the agent, the ‘desire paths’ emerge as a function of the activity of the agent (i.e. niche construction occurs). We will show how, depending on the relative inertia of the environment and agent, the joint agent-environment system moves to different attracting sets of jointly minimized free-energy.",
keywords = "Active inference, Adaptive environments, Agent-environment complementarity, Desire paths, Free energy principle, Markov decision processes, Niche construction",
author = "Jelle Bruineberg and Erik Rietveld and Thomas Parr and {van Maanen}, Leendert and Friston, {Karl J.}",
year = "2018",
month = "10",
day = "14",
doi = "10.1016/j.jtbi.2018.07.002",
language = "English",
volume = "455",
pages = "161--178",
journal = "Journal of Theoretical Biology",
issn = "0022-5193",
publisher = "Academic Press Inc.",

}

Free-energy minimization in joint agent-environment systems : A niche construction perspective. / Bruineberg, Jelle (Corresponding Author); Rietveld, Erik; Parr, Thomas; van Maanen, Leendert; Friston, Karl J.

In: Journal of Theoretical Biology, Vol. 455, 14.10.2018, p. 161-178.

Research output: Contribution to journalArticleAcademicpeer-review

TY - JOUR

T1 - Free-energy minimization in joint agent-environment systems

T2 - A niche construction perspective

AU - Bruineberg, Jelle

AU - Rietveld, Erik

AU - Parr, Thomas

AU - van Maanen, Leendert

AU - Friston, Karl J.

PY - 2018/10/14

Y1 - 2018/10/14

N2 - The free-energy principle is an attempt to explain the structure of the agent and its brain, starting from the fact that an agent exists (Friston and Stephan, 2007; Friston et al., 2010). More specifically, it can be regarded as a systematic attempt to understand the ‘fit’ between an embodied agent and its niche, where the quantity of free-energy is a measure for the ‘misfit’ or disattunement (Bruineberg and Rietveld, 2014) between agent and environment. This paper offers a proof-of-principle simulation of niche construction under the free-energy principle. Agent-centered treatments have so far failed to address situations where environments change alongside agents, often due to the action of agents themselves. The key point of this paper is that the minimum of free-energy is not at a point in which the agent is maximally adapted to the statistics of a static environment, but can better be conceptualized an attracting manifold within the joint agent-environment state-space as a whole, which the system tends toward through mutual interaction. We will provide a general introduction to active inference and the free-energy principle. Using Markov Decision Processes (MDPs), we then describe a canonical generative model and the ensuing update equations that minimize free-energy. We then apply these equations to simulations of foraging in an environment; in which an agent learns the most efficient path to a pre-specified location. In some of those simulations, unbeknownst to the agent, the ‘desire paths’ emerge as a function of the activity of the agent (i.e. niche construction occurs). We will show how, depending on the relative inertia of the environment and agent, the joint agent-environment system moves to different attracting sets of jointly minimized free-energy.

AB - The free-energy principle is an attempt to explain the structure of the agent and its brain, starting from the fact that an agent exists (Friston and Stephan, 2007; Friston et al., 2010). More specifically, it can be regarded as a systematic attempt to understand the ‘fit’ between an embodied agent and its niche, where the quantity of free-energy is a measure for the ‘misfit’ or disattunement (Bruineberg and Rietveld, 2014) between agent and environment. This paper offers a proof-of-principle simulation of niche construction under the free-energy principle. Agent-centered treatments have so far failed to address situations where environments change alongside agents, often due to the action of agents themselves. The key point of this paper is that the minimum of free-energy is not at a point in which the agent is maximally adapted to the statistics of a static environment, but can better be conceptualized an attracting manifold within the joint agent-environment state-space as a whole, which the system tends toward through mutual interaction. We will provide a general introduction to active inference and the free-energy principle. Using Markov Decision Processes (MDPs), we then describe a canonical generative model and the ensuing update equations that minimize free-energy. We then apply these equations to simulations of foraging in an environment; in which an agent learns the most efficient path to a pre-specified location. In some of those simulations, unbeknownst to the agent, the ‘desire paths’ emerge as a function of the activity of the agent (i.e. niche construction occurs). We will show how, depending on the relative inertia of the environment and agent, the joint agent-environment system moves to different attracting sets of jointly minimized free-energy.

KW - Active inference

KW - Adaptive environments

KW - Agent-environment complementarity

KW - Desire paths

KW - Free energy principle

KW - Markov decision processes

KW - Niche construction

U2 - 10.1016/j.jtbi.2018.07.002

DO - 10.1016/j.jtbi.2018.07.002

M3 - Article

VL - 455

SP - 161

EP - 178

JO - Journal of Theoretical Biology

JF - Journal of Theoretical Biology

SN - 0022-5193

ER -