Causal blankets: Theory and algorithmic framework

@article{Rosas2020CausalBT,
  title={Causal blankets: Theory and algorithmic framework},
  author={Fernando E. Rosas and Pedro A. M. Mediano and Martin Biehl and Shamil A. Chandaria and Daniel Polani},
  journal={ArXiv},
  year={2020},
  volume={abs/2008.12568}
}
We introduce a novel framework to identify perception-action loops (PALOs) directly from data based on the principles of computational mechanics. Our approach is based on the notion of causal blanket, which captures sensory and active variables as dynamical sufficient statistics -- i.e. as the "differences that make a difference." Moreover, our theory provides a broadly applicable procedure to construct PALOs that requires neither a steady-state nor Markovian dynamics. Using our theory, we show… 

Figures from this paper

The Emperor's New Markov Blankets.
TLDR
This paper identifies a persistent confusion in the literature between the formal use of Markov blankets as an epistemic tool for Bayesian inference, and their novel metaphysical use in the free energy framework to demarcate the physical boundary between an agent and its environment.
Examining the Continuity between Life and Mind: Is There a Continuity between Autopoietic Intentionality and Representationality?
A weak version of the life-mind continuity thesis entails that every living system also has a basic mind (with a non-representational form of intentionality). The strong version entails that the same
Interpreting Dynamical Systems as Bayesian Reasoners
TLDR
This work focuses on the case in which a system can be interpreted as performing either Bayesian filtering or Bayesian inference, and provides formal definitions of what it means for such an interpretation to exist, using techniques from category theory.
The Markov blanket trick: On the scope of the free energy principle and active inference.
TLDR
It is argued that FEP is not the general principle it is claimed to be, and that active inference is notThe all-encompassing process theory it is purported to be either.
The neural correlates of consciousness under the free energy principle: From computational correlates to computational explanation
How can the free energy principle contribute to research on neural correlates of consciousness, and to the scientific study of consciousness more generally? Under the free energy principle, neural

References

SHOWING 1-10 OF 34 REFERENCES
Computational Mechanics: Pattern and Prediction, Structure and Simplicity
TLDR
It is shown that the causal-state representation—an ∈-machine—is the minimal one consistent with accurate prediction, and several results are established on ∉-machine optimality and uniqueness and on how∈-machines compare to alternative representations.
Parcels and particles: Markov blankets in the brain
TLDR
A treatment of effective connectivity that speaks to the emergence of intrinsic brain networks and critical dynamics is described, predicated on the notion of Markov blankets that play a fundamental role in the self-organization of far from equilibrium systems.
Markov blankets, information geometry and stochastic thermodynamics
TLDR
There is a natural Bayesian mechanics for any system that possesses a Markov blanket, which means that there is an explicit link between the inference performed by internal states and their energetics—as characterized by their stochastic thermodynamics.
The Umwelt of an embodied agent—a measure-theoretic definition
TLDR
It is shown that the global dynamics of the world can be simplified without changing the internal process, and this simplification can serve as a minimal world model that the system must have in order to be consistent with theinternal process.
Information and closure in systems theory
The notion of closure plays a prominent role in systems theory where it is used to identify or define the system in distinction from its environment and to explain the autonomy of the system. Here,
Representations of Space and Time in the Maximization of Information Flow in the Perception-Action Loop
TLDR
The evolved implicit representations of space and time are analyzed in an information-theoretic manner, which paves the way toward a principled and general understanding of the mechanisms guiding the evolution of sensors in nature and provides insights into the design of mechanisms for artificial sensor evolution.
Information Theory of Decisions and Actions
TLDR
Using a graphical model, a recursive Bellman optimality equation for information measures is derived, in analogy to reinforcement learning; from this, new algorithms for calculating the optimal trade-off between the value-to-go and the required information- to-go are obtained, unifying the ideas behind the Bellman and the Blahut–Arimoto iterations.
Beyond integrated information: A taxonomy of information dynamics phenomena
TLDR
Integrated Information Decomposition is obtained, which paves the way for more detailed analyses of interdependencies in multivariate time series, and sheds light on collective modes of information dynamics that have not been reported before.
Autonomy: An information theoretic perspective
TLDR
Analysis reveals some subtle facets of the concept of autonomy, in particular with respect to the seemingly innocent system-environment distinction the authors took for granted, and raises the issue of the attribution of control, i.e. the responsibility for observed effects.
Action and perception for spatiotemporal patterns
  • Martin Biehl, D. Polani
  • Computer Science, Mathematics
    ECAL
  • 2017
TLDR
This work presents definitions of actions and perceptions within multivariate Markov chains based on entitysets, and shows that the definition of actions leads to non-heteronomy and that of perceptions specialize to the usual concept of perception in the perception...
...
1
2
3
4
...