Scope:
This is a metaphor, not a scientific claim. I use “Markov blanket” as a boundary metaphor, not, for example as a claim about the brain or consciousness. The “blanket” facilitates understanding of complex systems by simplifying the relationships between variables.
In some contemporary theoretical biology and cognitive science (e.g., free-energy/active-inference work), “Markov blanket” is also used to describe how a system maintains a boundary with its environment via mediating states (a membrane-like interface).
Philosophical uses of the term in discussions of dissociation in psychology, or nonlocality in physics are noted as examples of wider metaphorical uptake, but they are not evidence for my argument so are not cited.
Source:
In its technical origin (Pearl), a Markov blanket [1] is the minimal set of variables that makes a target variable conditionally independent of the rest of a model; it formalises relevance and screens off noise.
My Use Case:
I borrow the term only to express a practical idea about the demarcation problem in science: boundaries should be selective and functional, not merely exclusionary.
In my related essay work, a “thin” boundary means open to candidate inputs; a “thick” boundary means strict about the testing channels that upgrade an input into scientific standing.
Citation:
[1] Pearl, Judea. 1988. Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference. San Mateo, CA: Morgan Kaufmann.