University of Hertfordshire

From the same journal

By the same authors

Standard

Expanding the Active Inference Landscape: More Intrinsic Motivations in the Perception-Action Loop. / Biehl, Martin; Guckelsberger, Christian; Salge, Christoph; Smith, Simon; Polani, Daniel.

In: Frontiers in Neurorobotics, Vol. 12, No. AUG, 45, 30.08.2018, p. 45.

Research output: Contribution to journalArticle

Harvard

APA

Vancouver

Author

Bibtex

@article{d0e86745b493420f8646d5c199acc26e,
title = "Expanding the Active Inference Landscape: More Intrinsic Motivations in the Perception-Action Loop",
abstract = "Active inference is an ambitious theory that treats perception, inference, and action selection of autonomous agents under the heading of a single principle. It suggests biologically plausible explanations for many cognitive phenomena, including consciousness. In active inference, action selection is driven by an objective function that evaluates possible future actions with respect to current, inferred beliefs about the world. Active inference at its core is independent from extrinsic rewards, resulting in a high level of robustness across e.g., different environments or agent morphologies. In the literature, paradigms that share this independence have been summarized under the notion of intrinsic motivations. In general and in contrast to active inference, these models of motivation come without a commitment to particular inference and action selection mechanisms. In this article, we study if the inference and action selection machinery of active inference can also be used by alternatives to the originally included intrinsic motivation. The perception-action loop explicitly relates inference and action selection to the environment and agent memory, and is consequently used as foundation for our analysis. We reconstruct the active inference approach, locate the original formulation within, and show how alternative intrinsic motivations can be used while keeping many of the original features intact. Furthermore, we illustrate the connection to universal reinforcement learning by means of our formalism. Active inference research may profit from comparisons of the dynamics induced by alternative intrinsic motivations. Research on intrinsic motivations may profit from an additional way to implement intrinsically motivated agents that also share the biological plausibility of active inference.",
keywords = "Active inference, Empowerment, Free energy principle, Intrinsic motivation, Perception-action loop, Predictive information, Universal reinforcement learning, Variational inference",
author = "Martin Biehl and Christian Guckelsberger and Christoph Salge and Simon Smith and Daniel Polani",
year = "2018",
month = aug,
day = "30",
doi = "10.3389/fnbot.2018.00045",
language = "English",
volume = "12",
pages = "45",
journal = "Frontiers in Neurorobotics",
issn = "1662-5218",
publisher = "Frontiers Media S.A.",
number = "AUG",

}

RIS

TY - JOUR

T1 - Expanding the Active Inference Landscape: More Intrinsic Motivations in the Perception-Action Loop

AU - Biehl, Martin

AU - Guckelsberger, Christian

AU - Salge, Christoph

AU - Smith, Simon

AU - Polani, Daniel

PY - 2018/8/30

Y1 - 2018/8/30

N2 - Active inference is an ambitious theory that treats perception, inference, and action selection of autonomous agents under the heading of a single principle. It suggests biologically plausible explanations for many cognitive phenomena, including consciousness. In active inference, action selection is driven by an objective function that evaluates possible future actions with respect to current, inferred beliefs about the world. Active inference at its core is independent from extrinsic rewards, resulting in a high level of robustness across e.g., different environments or agent morphologies. In the literature, paradigms that share this independence have been summarized under the notion of intrinsic motivations. In general and in contrast to active inference, these models of motivation come without a commitment to particular inference and action selection mechanisms. In this article, we study if the inference and action selection machinery of active inference can also be used by alternatives to the originally included intrinsic motivation. The perception-action loop explicitly relates inference and action selection to the environment and agent memory, and is consequently used as foundation for our analysis. We reconstruct the active inference approach, locate the original formulation within, and show how alternative intrinsic motivations can be used while keeping many of the original features intact. Furthermore, we illustrate the connection to universal reinforcement learning by means of our formalism. Active inference research may profit from comparisons of the dynamics induced by alternative intrinsic motivations. Research on intrinsic motivations may profit from an additional way to implement intrinsically motivated agents that also share the biological plausibility of active inference.

AB - Active inference is an ambitious theory that treats perception, inference, and action selection of autonomous agents under the heading of a single principle. It suggests biologically plausible explanations for many cognitive phenomena, including consciousness. In active inference, action selection is driven by an objective function that evaluates possible future actions with respect to current, inferred beliefs about the world. Active inference at its core is independent from extrinsic rewards, resulting in a high level of robustness across e.g., different environments or agent morphologies. In the literature, paradigms that share this independence have been summarized under the notion of intrinsic motivations. In general and in contrast to active inference, these models of motivation come without a commitment to particular inference and action selection mechanisms. In this article, we study if the inference and action selection machinery of active inference can also be used by alternatives to the originally included intrinsic motivation. The perception-action loop explicitly relates inference and action selection to the environment and agent memory, and is consequently used as foundation for our analysis. We reconstruct the active inference approach, locate the original formulation within, and show how alternative intrinsic motivations can be used while keeping many of the original features intact. Furthermore, we illustrate the connection to universal reinforcement learning by means of our formalism. Active inference research may profit from comparisons of the dynamics induced by alternative intrinsic motivations. Research on intrinsic motivations may profit from an additional way to implement intrinsically motivated agents that also share the biological plausibility of active inference.

KW - Active inference

KW - Empowerment

KW - Free energy principle

KW - Intrinsic motivation

KW - Perception-action loop

KW - Predictive information

KW - Universal reinforcement learning

KW - Variational inference

UR - http://www.scopus.com/inward/record.url?scp=85053140841&partnerID=8YFLogxK

U2 - 10.3389/fnbot.2018.00045

DO - 10.3389/fnbot.2018.00045

M3 - Article

VL - 12

SP - 45

JO - Frontiers in Neurorobotics

JF - Frontiers in Neurorobotics

SN - 1662-5218

IS - AUG

M1 - 45

ER -