Self-organisation · Article

Nested Markov blankets and hierarchical self-organisation

A Markov blanket is a set of states that separates internal states from external states in a statistical sense.³ In the free energy literature, blankets are said to nest. The blanket of an animal encloses those of its organs, cells and nuclei.¹ Authors argue that this recursion lets a whole system minimise variational free energy across scales.⁷ Palacios and colleagues tested whether such nesting emerges in simulation.⁹

Earliest held: 2008, Friston K
Most discussed in: On Markov blankets and hierarchical…, 2020
In the library: 197 passages in 20 works
Rewritten: 2026-10-23

The blanket concept

Friston, Levin, Sengupta and Pezzulo (2015) describe a Markov blanket as a set of states that separates two other sets. The term comes from Bayesian networks.³ They give the cell surface as an example, with intracellular states inside and extracellular states outside.³ External states can then be known to the internal states only indirectly, through the blanket. Bruineberg, Kiverstein and Rietveld (2018) summarise Friston's 2013 claim that such blankets are all pervasive in natural systems dominated by short-range interactions.⁸ On that view a system divides into internal and external states, shielded from each other by a blanket.⁸

Blankets within blankets

In Life as we know it (2013), Friston observed that a system can have many partitions and many blankets. He gave the example of an animal whose blanket encloses those of its organs, cells and nuclei.¹ Allen and Friston (2018) drew a related conclusion about organisms. They argued that an organism is defined by a near-infinite regress of interacting blankets, not by one blanket.⁴ Ramstead, Badcock and Friston (2018) went further. Because every ergodic system is said to possess a blanket, they proposed describing the universe as blankets of blankets, all the way up and all the way down.⁵

Construction across scales

Ramstead and colleagues (2018) described how to build the next level. States at one level are partitioned into internal subsets and their blankets. States for the level above are then formed as mixtures of blanket states.⁶ They called these mixtures slow modes, known as order parameters in synergetics.⁶ Kirchhoff and colleagues (2018) added a claim about causation. Keeping blankets at every scale lets dynamics at one scale enslave the blankets below, so that the whole organisation minimises variational free energy.⁷

Which blankets, and a test

Friston (2013) asked whether anything marks the blankets of living systems as special, given that blankets are so numerous. He offered a speculative answer. Perhaps minimum entropy blankets characterise biological systems.² Palacios, Razi, Parr, Kirchhoff and Friston (2020) noted that no numerical proof of principle existed. They aimed to show that blankets of blankets can emerge from gradient flows on variational free energy, under an appropriate generative model.⁹ They framed their cell-like simulations as an analogy, not as a detailed account of cell communication.⁹

SourcesEach quotation was checked word for word against the passage it opens.

the Markov blanket of an animal encloses the Markov blankets of its organs, which enclose Markov blankets of cells, which enclose Markov blankets of nuclei and so on.Friston K, 2013 · Life as we know it · open at passage 46
One might conjecture that minimum entropy Markov blankets characterize biological systems.Friston K, 2013 · Life as we know it · open at passage 47
the surface of a cell may constitute a Markov blanket separating intracellular (internal) and extracellular (external) states [43,51].Friston K, Levin M, Sengupta B, Pezzulo G, 2015 · Knowing one's place: a free-energy approach to pattern regulation · open at passage 11
any organism will be defined not by a singular blanket, but instead by a near-infinite regress of causally interacting Markov blankets within Markov blanketsAllen M, Friston KJ, 2018 · From cognitivism to autopoiesis: towards a computational framework for the… · open at passage 42
we should be able to describe the universe in terms of Markov blankets of Markov blankets—and Markov blankets all the way up, and all the way downRamstead MJD, Badcock PB, Friston KJ, 2018 · Answering Schrödinger's question: A free-energy formulation · open at passage 17
The mixtures can be regarded as slow (unstable) modes that are referred to as order parameters in synergetics [18].Ramstead MJD, Badcock PB, Friston KJ, 2018 · Answering Schrödinger's question: A free-energy formulation · open at passage 18
the conservation of Markov blankets (of Markov blankets) at every hierarchical scale enables the dynamics of the states at one scale to enslave the (states of) Markov blankets at the scale belowKirchhoff M, Parr T, Palacios E, Friston K, Kiverstein J, 2018 · The Markov blankets of life: autonomy, active inference and the free energy… · open at passage 53
Markov blankets are mainly used in the context of Bayesian graphical models, but as Friston (2013a) shows, they are all pervasive in natural systems dominated by short-range interactions.Bruineberg J, Kiverstein J, Rietveld E, 2018 · The anticipating brain is not a scientist: the free-energy principle from an… · open at passage 53
hierarchal compositions of Markov blankets of Markov blankets can emerge from gradient flows on variational free energy, under an appropriate generative model.Palacios ER, Razi A, Parr T, Kirchhoff M, Friston K, 2020 · On Markov blankets and hierarchical self-organisation · open at passage 4

Linked ideas

Autopoiesis and the Markov blanketrelated to

Treats the blanket as the boundary that lets a self-producing system persist, the same idea nested blankets extend across scales.

Free energy principle and active inferencepart of / contains

Nested blankets are one consequence drawn from the free energy principle, whose gradient flows drive the hierarchical organisation described here.

Hierarchical dynamic models and variational Bayesian inferencerelated to

Hierarchical generative models supply the formal structure that the numerical blanket-of-blankets simulations rely on.

Downward causation and coarse-grainingparallels

Higher-scale blankets constraining lower-scale ones echoes the question of how collective states act on their components.

Scale of the selfrelated to

Blankets at cell, organ and organism scales bear on where one draws the boundary of a self.

Quantum reference frames and the membrane as boundaryrelated to

Another treatment of the Markov blanket as the boundary that defines an observer's view of the world.

Predictive processing and interoceptive inferencerelated to

Passages treat the Markov blanket as the boundary enabling free-energy minimisation (A's inference framework); Fields & Levin describe multi-layer blankets from cells expanding into internal processes, hinting at nesting, but hierarchy isn't developed.

Markov blanket and sensory-active state partitionpart of / contains

Develops the recursive construction in which blankets at one level are built from mixtures of blanket states below.

Synergetics, slaving principle and separation of timescalesrelated to

Mixtures of blanket states at one scale act as order parameters for the scale above, with slower dynamics at each level.

Neutral monism, dual-aspect monism and the mind-body problemrelated to

Friston et al. propose 'Markovian monism': one kind of thing, with any system that has a Markov blanket relevant to mind, offered as a repair for Cartesian duality. Nesting isn't mentioned.

Where it is discussedPassages matching blankets of blankets, Markov blanket, hierarchical self-organisation, variational free energy minimisation

2020	On Markov blankets and hierarchical self-organisation · Palacios ER, Razi A, Parr T, Kirchhoff…	46
2018	The Markov blankets of life: autonomy, active inference and the free energy… · Kirchhoff M, Parr T, Palacios E…	41
2013	Life as we know it · Friston K	22
2018	Answering Schrödinger's question: A free-energy formulation · Ramstead MJD, Badcock PB, Friston KJ	19
2022	Active inference, morphogenesis, and computational psychiatry · Pio-Lopez L, Kuchling F, Tung A…	11
2018	From cognitivism to autopoiesis: towards a computational framework for the… · Allen M, Friston KJ	8
2015	Knowing one's place: a free-energy approach to pattern regulation · Friston K, Levin M, Sengupta B, Pezzulo…	7
2019	Somatic multicellularity as a satisficing solution to the prediction-error… · Fields C, Levin M	7
2018	The anticipating brain is not a scientist: the free-energy principle from an… · Bruineberg J, Kiverstein J, Rietveld E	7
2022	Competency in Navigating Arbitrary Spaces as an Invariant for Analyzing… · Fields C, Levin M	6