The lexicon — lexicon

RENDERED FROM STRUCTURED DATA — this page is not a markdown file. Source: lexicon/terms/blanket-mind-body-world.json in uni-cookbook at 575fc93d9d31. The sha256 recorded for this page is the digest of that JSON, not of this text. The renderer is generators/derive_docs.cjs; it copies fields and adds no facts. Every value below is a field of the source, verbatim, including the ones that read NOT RUN and PENDING.

The five-register lexicon file (Sanskrit / Latin / English / Hindi / Spanish) for the blanket / mind / body / world concept cluster. It renders the 14 concepts of the markov-blanket domain of lexicon/CONCEPTS.json and NOTHING ELSE. Per rule 18.14 of the operator's charter ('Do not claim perfect translation'), NO term in this file is claimed to be a correct or perfect rendering. Each register carries a status (CITED | PENDING-CITATION | COINED | NOT-ATTEMPTED), and every entry carries a back_translation_test that has NOT BEEN RUN. The definitions, carried verbatim from CONCEPTS.json, are the translatable object and the test object; the words are candidates. Where a register's word is likely to mislead, that is recorded in ambiguity_notes rather than smoothed over -- those notes, and not the headwords, are the useful content of this file.

authored 2026-07-15 · reviewer: NONE - PENDING EXPERT REVIEW · counts: {"entries":14,"registers_per_entry":5,"back_translation_tests_specified":14,"back_translation_tests_RUN":0,"back_translation_tests_PASSED":0,"expert_reviews_completed":0,"sanskrit_headwords_retrieved_from_monier_williams":17,"latin_headwords_retrieved_from_lewis_and_short":11,"hindi_headwords_retrieved_from_mcgregor":6,"spanish_headwords_retrieved":0,"registers_marked_CITED":1,"count_correction_note":"The Sanskrit and Latin retrieval counts were first written as 13 and 10 by hand, and both were wrong; a recount against the retrieval log gave 17 and 11. The erroneous figures are recorded here rather than quietly overwritten, on the same principle CONCEPTS.json applies to its own hand-count correction.","retrieval_log":{"note":"Every headword this file retrieved from an authority on 2026-07-15, listed so the counts above are auditable rather than asserted. A headword appearing here was READ; a headword absent from here is NOT cited anywhere in this file.","monier_williams_slp1_keys":["avasTA","antar","bahis","vyavaDAna","saMyoga","AvaraRa","indriya","upADi","manas","sApekza","sKalana","paraspara","nirapekza","tala","garBa","praveSa","nirgama"],"lewis_and_short_entries":["velamen","status2","superficies","mutuus","informatio","interior","exter","lapsus2","sensus2","nidus","activus"],"mcgregor_headwords":["avasthā","āntarik","bāhya","kambal","satah","paraspar"],"rae_headwords":[]}}

Falsifiers for this file

• A headword marked CITED is not in the named authority, or is not there in the quoted sense. That entry is a pseudopropheta and must be struck to PENDING-CITATION immediately. This is the worst defect this file could contain.
• A quoted citation string does not match what the authority's page actually says at the given retrieval URL.
• A register marked COINED lacks a morphological derivation, or its derivation invokes a component whose status is not stated in component_provenance.
• Any register anywhere in this file is described as correct, perfect, verified, or proven. Charter rule 18.14 forbids it and no test has been run that could support it.
• A back_translation_test is reported as passed while back_translation_tests_RUN remains 0.
• An entry renders μ with a word for mind/soul/consciousness, η with a word for cosmos/earth, or b with a word for anatomy -- the three failures the rendering_architecture exists to prevent.
• MIND_BODY is rendered as a compound noun in any register, thereby naming a third substance and reopening dualism.
• SENSORY_STATES and ACTIVE_STATES are rendered with one word, or with words that do not distinguish the direction of influence -- the sparsity condition is then unexpressible.
• PEARL_BLANKET and FRISTON_BLANKET are rendered with the same term with no distinguishing name, which performs the BLANKET_SLIDE in the lexicon itself.
• BLANKET_SLIDE's entry is filed, cited or merged as a TRUE-store observation. Its store is META and crossing the stores is the cardinal sin.
• This file is cited as evidence that any system has a Markov blanket, that I(mu;eta|b) has been estimated for anything, or that any biological hierarchy is nested. It says none of those things and names all three as NOT-MEASURED.
• The Spanish register is promoted to CITED without someone actually reading dle.rae.es.

MARKOV_BLANKET —

Confidence: low. vyavadhāna and velamen are both CITED and both carry a concealment sense the concept forbids; they are the best of a bad field rather than good renderings. The English metaphor is defective at the root and no register inspected here repairs it. · Reviewer: NONE - PENDING EXPERT REVIEW

(no definition in this file; see CONCEPTS)

register	form	literal gloss	status
sanskrit	मार्कोव-व्यवधानम् (mārkova-vyavadhānam)	the Markov intervening-screen; that which is placed between, named for Markov	COINED
latin	velamen Markovianum	the Markovian covering / veil	COINED
english	Markov blanket	The set b = {s,a} of sensory and active states whose conditioning renders internal and external states independent, so that p(mu,eta|b) = p(mu|b)p(eta|b), equivalently I(mu;eta|b) = 0.	—
hindi	मार्कोव व्यवधान ()	the Markov intervening-screen	PENDING-CITATION
spanish	manta de Markov	the Markov blanket (calque of the English)	PENDING-CITATION

Ambiguity notes

• vyavadhāna carries 'obstruction, hiding from view' alongside 'intervening'. The blanket does NOT hide the world from the mind and does not obstruct anything; it MEDIATES. A reader taking the concealment sense has been handed the Friston reading (a real thing being hidden behind a real surface) at no evidential cost. This is the reason the term is admitted with its definition attached and never alone.
• vyavadhāna also carries 'cessation, termination' and, in grammar, the very narrow 'intervention of a syllable or letter'. Neither is meant.
• velamen is a CLOTH word - a veil or garment. It therefore renders the English metaphor faithfully, which is precisely the risk: it reproduces 'blanket' rather than repairing it. Per the source concept's own note, this register is rendering a METAPHOR, not a mechanism, and that fact is recorded here rather than smoothed over.
• 'Blanket' in English is already a bad metaphor for b: a blanket on a bed covers by lying on top and does its work by hiding. b does its work by conditioning. Every register inherits this defect from the English and none of them repairs it. The definition, not the word, is the object.
• The name 'Markov' attaches to a construct Markov did not define in this form; the term is eponymous convention, and no register should be read as a claim about Markov's own work.

Forbidden mistranslations

register	form	why it is wrong
sanskrit		Monier-Williams gives āvaraṇa as 'covering, hiding, concealing'; 'the act of covering, concealing, hiding'; 'an obstruction'; and, in the Jain sense, 'mental blindness'. Rendering the blanket āvaraṇa asserts that b CONCEALS η from μ. It does not. Worse: in Advaita Vedānta āvaraṇa names māyā's veiling power, which hides a reality that IS there. That single word therefore smuggles in (a) a real external world, (b) the claim that it is hidden, and (c) a Vedānta metaphysics - i.e. it asserts the FRISTON_BLANKET reading by vocabulary alone, in a corpus that explicitly declines it. This is the worst available rendering and it is also the most natural one.
sanskrit		Tempting, because Vedānta uses upādhi for a limiting adjunct and b does limit. But Monier-Williams gives its primary sense as 'that which is put in the place of another thing, a substitute, substitution', and it also carries 'appearance, phantom' and 'deception'. b is not a substitute for anything and is not an illusion; in the Advaita usage the adjunct is precisely what is to be SEEN THROUGH, which inverts the concept - b is the only thing there is to see.
hindi		McGregor gives कंबल kambal, m., as 'a blanket' - the textile, the thing on a bed. This is the literal calque a translator or a machine reaches for first, and it renders a conditional-independence structure as bedding. It is the single most plausible failure in this register.
latin		Both are 'covering' words reached for by the same cloth instinct as velamen but WITHOUT even velamen's advantage of being the conventional calque. This file did not retrieve either headword from Lewis & Short, so the glosses here are the author's recollection and are NOT cited; they are listed as traps to check, not as facts. Falsifier: retrieve both from Perseus before anyone uses them.
any		b is a SET OF STATES, not a thing. Every noun in every register here is a metaphor for a set. The admission rule is that the term never travels without p(μ,η|b) = p(μ|b)p(η|b) beside it.

Back-translation test — status: NOT-RUN. No reviewer has been engaged. This test is specified, not passed.

MIND —

Confidence: medium for the Sanskrit and Latin as STRUCTURES (both built from CITED components, both deliberately free of substance-nouns); low that any reader steeped in the philosophical registers will hear them as flatly as they are meant. The avasthā-traya resonance is a live, unresolved risk. · Reviewer: NONE - PENDING EXPERT REVIEW

(no definition in this file; see CONCEPTS)

register	form	literal gloss	status
sanskrit	अन्तरवस्थाः (antaravasthāḥ)	the states that are within; the interior states	COINED
latin	status interiores	the inner states / the interior conditions	COINED
english	mind (internal states, μ)	The internal states mu inside a blanket, which have no term in external states and whose only evidence about the world is the sensory states.	—
hindi	आंतरिक अवस्थाएँ ()	internal states	COINED
spanish	estados internos	internal states	PENDING-CITATION

Ambiguity notes

• THIS IS THE MOST DANGEROUS ENTRY IN THE FILE, and the danger is not that a register will render μ badly - it is that every register will render μ BEAUTIFULLY, with a word for mind that has two thousand years of philosophy behind it, and thereby assert what the corpus explicitly disclaims. Nothing in the partition names an experiencer. The formalism admits partitions for a Newton's cradle and a burning fuse.
• The whole design of this file follows from that: μ is rendered STRUCTURALLY (antar + avasthā / status + interior), off position-and-state words, precisely so that no register's vocabulary can quietly install a subject. The renderings are deliberately flat. Flatness is the feature.
• antar is an indeclinable, not a substance-noun, and avasthā is a condition, not a bearer of conditions. The Sanskrit therefore says 'the states that are within' and says nothing about whose states they are or whether it is like anything to be them. That silence is the accurate rendering of the mathematics.
• Residual freight even so: avasthā in McGregor carries 'stage (of life, progress); age', and in Indian philosophical usage the avasthā-s are canonically the states of the SELF (waking, dream, deep sleep, turīya). A Sanskritist reading antaravasthāḥ may hear the avasthā-traya doctrine and supply an ātman to have the states. The word does not say it; the tradition may hear it. Recorded, not resolved.
• Latin status is comparatively clean, but its section I is corporeal - 'posture, position, attitude, carriage'. status interiores could be misread as 'inner posture' (a bearing, a mood). The Trop. sense II ('condition, state, situation') is the one meant.

Forbidden mistranslations

register	form	why it is wrong
sanskrit		Monier-Williams gives manas as 'mind (in its widest sense as applied to all the mental powers), intellect, intelligence, understanding, perception, sense, conscience, will', and philosophically as 'the internal organ or antaḥkaraṇa of perception and cognition, the faculty or instrument through which thoughts enter or by which objects of sense affect the soul'. Every one of those is a capacity of a cognizer. μ is a set of states. Rendering μ as manas asserts that there is a soul being affected, which is exactly the claim the corpus refuses to make about UNI or about any system.
sanskrit		Imports Yoga/Buddhist psychology wholesale - citta is the mind-stream that is to be stilled or purified, a thing with a career and a soteriology. μ has neither. NOT retrieved from Monier-Williams by this file; the gloss is the author's and is NOT cited. Falsifier: retrieve citta before anyone relies on this reasoning.
sanskrit		The worst of the three. ātman is the self/soul, and Monier-Williams's OWN entry for manas distinguishes manas from ātman and calls manas perishable and belonging to the body - i.e. the lexicon itself treats ātman as the eternal subject. Rendering μ as ātman asserts a persisting experiencer behind the internal states. The partition contains no such thing. NOT retrieved directly by this file; the reasoning rests on the manas entry, which WAS retrieved and which does state the distinction.
latin		Same defect as manas, in Latin: all three name a cognizing or animating principle rather than a set of states, and anima additionally imports the soul. NOT retrieved from Lewis & Short by this file - listed as traps to check, not as cited facts.
hindi		The ordinary Hindi word for mind/heart, and the reflex of manas; it carries the full folk-psychological load (mind, heart, inclination, desire). NOT retrieved from McGregor by this file. Falsifier: query McGregor for man before relying on this.
any		Banned outright. None of these is in the definition. μ is defined by two negative structural facts - no term in η, and evidence only via s - and by nothing else. A register that adds an experiencer has not translated the concept; it has replaced it.

Back-translation test — status: NOT-RUN. No reviewer has been engaged. This test is specified, not passed.

BODY —

Confidence: medium. superficies is the one CITED single-word rendering here and it is a good fit for 'the surface that carries the dependencies'. The Sanskrit compound is weaker: tala's 'base/bottom' sense pulls against the intended reading and only the vyavadhāna element holds it in place. · Reviewer: NONE - PENDING EXPERT REVIEW

(no definition in this file; see CONCEPTS)

register	form	literal gloss	status
sanskrit	व्यवधानतलम् (vyavadhāna-talam)	the surface of the intervening screen; the intervening surface	COINED
latin	superficies	the surface; the upper side of a thing	CITED
english	body (the blanket, b)	The blanket b = {s,a} itself: the surface that carries every dependency between mind and world, and that thereby determines what is inferable at all.	—
hindi	व्यवधान-सतह ()	the intervening surface	COINED
spanish	superficie	surface	PENDING-CITATION

Ambiguity notes

• b is NOT the anatomical body except by modeling choice. The source concept says so outright. Every register here therefore refuses the anatomy words and takes the SURFACE words instead - which is why the Latin is superficies and not corpus.
• The cost of that choice: 'surface' in every register suggests a thin outer skin of a solid object, and b is not the outside of anything. b is a set of states that happens to factorize the dependency. A surface word is a metaphor exactly as much as a blanket word is; it is merely a less misleading one, because it carries mediation rather than concealment.
• The concept's real content is a claim most readers will miss and no register conveys: change the surface and you change the set of possible minds behind it, because you have changed the only evidence there will ever be. Embodiment is not decoration. No word in any register says this; only the definition does.
• tala in Monier-Williams also carries 'the lower part, base', 'the palm of the hand', 'the sole of the foot', and it is the name of a hell. The 'base/bottom' sense is a live misreading: tala alone would suggest that b is the BOTTOM of something rather than its boundary. The compound with vyavadhāna is what pins the intended sense, and it does so weakly.
• superficies carries a strong pejorative shadow in the modern European languages descended from it - superficial, superficialidad, meaning shallow or trivial. b is the opposite of trivial: it is the only thing that carries anything. A Spanish or English reader may hear 'superficial' where 'surface' is meant.

Forbidden mistranslations

register	form	why it is wrong
sanskrit		Both are the anatomical/physical body - flesh, frame, corpse. b is a set of states, not tissue, and the source concept explicitly says b is NOT the anatomical body except by modeling choice. śarīra additionally imports the body/soul contrast (the śarīra is what the ātman wears and sheds), which reinstalls exactly the dualism this partition dissolves. NEITHER headword was retrieved from Monier-Williams by this file; the glosses are the author's and are NOT cited. Falsifier: retrieve both before relying on this.
latin		The direct calque of the English label 'body', and therefore the first thing a translator reaches for - which is precisely why it is banned. corpus is a physical body; b is the blanket. Choosing corpus turns a claim about conditional independence into a claim about anatomy, and quietly answers the modeling question (where does the organism end?) that the corpus insists must be measured, not assumed. NOT retrieved from Lewis & Short by this file; listed as a trap, not a cited fact.
hindi		Same anatomy defect as the Sanskrit. NOT retrieved from McGregor by this file. Falsifier: query McGregor before relying on this.
spanish		Same defect as corpus, from which it descends. PENDING-CITATION like the whole Spanish register.
any		All four are better than 'body' and all four are still wrong in the same direction: each names a physical partition of a physical thing. b is a partition of a STATE SPACE. 'Boundary' is the most dangerous of them, because in this literature 'boundary of the organism' is the FRISTON_BLANKET claim - a rendering that calls b a boundary has performed the BLANKET_SLIDE inside a single word.

Back-translation test — status: NOT-RUN. No reviewer has been engaged. This test is specified, not passed.

MIND_BODY —

Confidence: low. The grammatical requirement (relation, not substance) is met in every register by construction, and that much is solid. But the Sanskrit head saṃyoga buys its correct relation-type with an incorrect substance-ontology and an explicit 'direct material contact' gloss, and the Latin head coniunctio is not yet cited at all. This entry is the least settled in the file and is the one most in need of an expert. · Reviewer: NONE - PENDING EXPERT REVIEW

(no definition in this file; see CONCEPTS)

register	form	literal gloss	status
sanskrit	अन्तरवस्था-व्यवधान-संयोगः (antaravasthā-vyavadhāna-saṃyogaḥ)	the conjunction of the interior states with the intervening screen	COINED
latin	coniunctio statuum interiorum cum superficie	the joining of the inner states with the surface	PENDING-CITATION
english	mind.body (the interface)	The coupling mu to b -- a pair of one-way doors, sensory states the world's only route in and active states the mind's only route out -- which is not a fifth thing but the interface itself.	—
hindi	आंतरिक अवस्थाओं का सतह से संयोग ()	the conjunction of the internal states with the surface	COINED
spanish	el acoplamiento de los estados internos con la superficie	the coupling of the internal states with the surface	PENDING-CITATION

Ambiguity notes

• This is the operator's fourth term and the load-bearing one, and the whole difficulty is grammatical rather than lexical: mind.body is a RELATION, not a thing. M12 leaves it implicit. Every register in this entry is therefore built as a PHRASE (genitive/postposition), never as a compound noun, because a compound noun in any of these languages would name a third substance - and naming a third substance is how you reopen dualism, which the partition does not contain.
• That is why the Latin is a five-word phrase rather than an elegant single word. The elegance would be the defect. Any reviewer tempted to compress 'coniunctio statuum interiorum cum superficie' into something like 'menticorpus' or 'animocorporeum' has, in the act of compressing, asserted the thing this entry exists to deny.
• saṃyoga was chosen because Nyāya-Vaiśeṣika treats it as a RELATION (a guṇa inhering in substances), not as a substance - which is precisely right. But the freight is heavy and cuts both ways: to treat saṃyoga as a guṇa is to accept an ontology in which qualities inhere in SUBSTANCES, i.e. it presupposes that μ and b are substances that can bear qualities. They are sets of states. The register buys the correct relation-type at the price of an incorrect substance-metaphysics. This is exactly the doctrinal-freight problem the charter names and it is NOT resolved here.
• Monier-Williams's philosophical gloss of saṃyoga is 'direct MATERIAL contact'. The μ-to-b coupling is not material and involves no contact. This is a genuine mismatch, not a quibble.
• saṃyoga also carries 'carnal contact, sexual union' and 'matrimonial connection'. A naive reader of antaravasthā-vyavadhāna-saṃyogaḥ is unlikely to reach for these, but a naive reader of saṃyoga alone might, and the term must not travel alone.
• The 'one-way doors' image in the definition is doing real work and no register carries it: the asymmetry (s is the world's ONLY route in, a is the mind's ONLY route out) is what makes this an interface rather than a mixture. Only the definition says it.

Forbidden mistranslations

register	form	why it is wrong
any		The central forbidden move of this entry, and the most natural one, since the English label 'mind.body' LOOKS like a compound. A compound noun names a thing. mind.body is not a fifth thing; it is the interface itself. Rendering it as a compound noun asserts a third substance between mind and world, which is Cartesian dualism's exact failure mode - and it does so while appearing to be a faithful, even literal, translation. A translator will feel most correct at the moment of being most wrong.
spanish		Doubly wrong: it is a compound noun (see above), AND it is the settled Spanish name for the mind-body PROBLEM in philosophy of mind. Using it imports a two-millennium debate about how mental substance relates to physical substance into a concept that posits neither. PENDING-CITATION like the whole Spanish register.
sanskrit		The Vaiśeṣika relation of INHERENCE - the intimate, inseparable relation of a quality to its substrate or a whole to its parts. Superficially attractive as a 'tighter' relation than saṃyoga, but wrong: inherence is precisely the relation of a thing to what it cannot exist apart from, and μ is not inherent in b. NOT retrieved from Monier-Williams by this file; the gloss is the author's and is NOT cited. Falsifier: retrieve samavāya before anyone uses or rejects it on this reasoning.
latin		Both name a merging. The interface is the opposite of a merging: it is what keeps μ and η apart while letting them depend on each other. NOT retrieved from Lewis & Short; listed as traps to check, not cited facts.

Back-translation test — status: NOT-RUN. No reviewer has been engaged. This test is specified, not passed.

WORLD —

Confidence: medium. DOWNGRADED from 'medium-high, the highest in this file' by the back-translation adversary 2026-07-15. What survives the downgrade: both Sanskrit components and both Latin components are CITED and were independently re-retrieved and re-confirmed; the construction is exactly parallel to MIND (which is the symmetry the mathematics demands); and the deliberate avoidance of every world-word does give this rendering real immunity to the SOVEREIGN HOMONYM that the English label cannot achieve - that claim is unaffected. What forced the downgrade: a second, independent doctrinal-freight risk the entry had not recorded (the bāhya-artha / external-object realism dispute - see ambiguity_notes), which is not a homonym problem but an ontological one, and which is worse than the 'apparent/for show' sense that was previously logged as the only residual risk. Immunity to one hazard was being read as immunity generally. · Reviewer: NONE - PENDING EXPERT REVIEW

(no definition in this file; see CONCEPTS)

register	form	literal gloss	status
sanskrit	बहिरवस्थाः (bahiravasthāḥ)	the states that are outside; the exterior states	COINED
latin	status exteriores	the outer states / the exterior conditions	COINED
english	world (external states, η)	The external states eta outside the blanket, which a mind reaches exactly never and which enter only as a hypothesis that explains the blanket.	—
hindi	बाह्य अवस्थाएँ ()	external states	COINED
spanish	estados externos	external states	PENDING-CITATION

Ambiguity notes

• SOVEREIGN HOMONYM, flagged by the source concept itself: η is NOT 'world' as Earth or biosphere (the corpus's WORLD_CONTEXT entries - GAIA, LABORATORY, TOY_WORLD - are a different registered domain), and NOT 'World C', the corpus's count-based reader. Three different things wear the English word 'world' in this corpus. The registers here translate only the third-person-plural set of states outside a blanket.
• This is why every register renders η as 'the exterior STATES' and never with a world-word: status exteriores and bahiravasthāḥ cannot be confused with the planet, and that immunity is the point.
• The hardest content to carry is the epistemics, and no register carries it: a mind reaches η exactly NEVER. It touches only its own blanket. The generative model is not a model of η but of how η is expected to show up in b. Nothing in 'the exterior states' warns a reader that the exterior states are unreachable in principle; the definition must travel with the term.
• McGregor's sense 3 for bāhya is 'outward, apparent (as the external world)' - the dictionary itself supplies 'the external world' as the idiomatic use. Convenient, and a hazard: it is the sense that pulls bāhya toward the cosmological reading this entry forbids.
• McGregor's sense 4 for bāhya is 'intended for show (conduct)' - i.e. superficial, insincere, for appearances. η is not for show; it is the one thing that is never shown. A Hindi reader may hear exactly the wrong connotation.
• DOCTRINAL FREIGHT NOT PREVIOUSLY RECORDED, added by the back-translation adversary 2026-07-15 and left UNRESOLVED. The freight this entry logged for the outside-words is COSMOLOGICAL (bāhya pulling toward 'the external world' as a place). There is a second and sharper risk it did not log, and it is an EPISTEMOLOGICAL one: in Indian philosophy the outside-word is the standing term of art in the dispute over BĀHYA-ARTHA, the 'external object' - whether objects outside cognition exist at all, contested between the realists and the Yogācāra idealists. That is a debate about whether there IS an η, which is precisely the question the FRISTON_BLANKET entry says this corpus declines to adjudicate. A Sanskritist or a Hindi reader trained in darśana may therefore hear bahir-/bāhya- as TAKING A SIDE - asserting external realism - in the same way āvaraṇa was rejected for asserting a concealed reality by vocabulary alone (see MARKOV_BLANKET forbidden_mistranslations). The two risks are different and the second is worse: the cosmological one makes η a place, the epistemological one makes η's EXISTENCE true by word choice.
• STATUS OF THE NOTE ABOVE: UNCITED, and honestly so. The adversary retrieved Monier-Williams s.v. bāhya (SLP1 key bAhya, https://www.sanskrit-lexicon.uni-koeln.de/scans/MWScan/2020/web/webtc/getword.php?key=bAhya , 2026-07-15) expressly to warrant it, and MW DOES NOT WARRANT IT: the only bāhya+artha collocation MW records is Pāṇinian and linguistic ('external to, i.e. not resulting from, the sounds or letters forming a word'), not epistemological. MW's bāhya does carry 'not belonging to the family or country, strange, foreign' and 'excluded from caste or the community, an out-caste', neither of which is the freight named above. So the bāhya-artha risk is real in the philosophical literature and is NOT in this file's admission authority, and it is recorded here as an uncited adversary observation rather than dressed in a citation it does not have. NOTE ALSO: bāhya is NOT in this file's retrieval_log and is NOT cited anywhere in this file - the Sanskrit register for WORLD is built on bahis, and the Hindi bāhya is cited to McGregor, not to MW. Falsifier: put the bāhya-artha question to the same expert who runs this entry's back-translation test; if a competent reader hears bahiravasthāḥ as asserting that the exterior states exist, the rendering has taken the side this corpus declines and must be reworked.
• Latin exterior carries no such shadow but is marked 'rare but class.' by Lewis & Short, so the phrase status exteriores will read as slightly unusual Latin. Unusual is acceptable; misleading is not.

Forbidden mistranslations

register	form	why it is wrong
sanskrit		loka is world-as-realm - a place, a plane of existence, a population, and in the plural the cosmological loka-s. It answers 'which world?' η is not a place and has no inhabitants. Rendering η as loka asserts a cosmos where the concept posits only a set of states, and it collides head-on with the sovereign homonym the source concept warns about. NOT retrieved from Monier-Williams by this file; the gloss is the author's and is NOT cited. Falsifier: retrieve loka before relying on this.
sanskrit		jagat is literally 'the moving/living', the world as the totality of moving things - and in Vedāntic usage it is the term of art for the phenomenal world whose reality is at issue. Both are wrong: η need not move, and η's reality is not what this concept adjudicates. NOT retrieved from Monier-Williams by this file; the gloss is the author's and is NOT cited.
latin		The direct calque of the label 'world' and therefore the first reach. mundus is the cosmos, the universe, the ordered whole (and, separately, 'clean' and 'women's ornaments'). η is a set of states outside a particular blanket - it is relative to that blanket, and one system's η contains other systems. There is no single mundus. NOT retrieved from Lewis & Short; listed as a trap, not a cited fact.
hindi		All three are world-as-cosmos-or-society; saṃsār additionally imports the cycle of rebirth, which is about as much metaphysics as a single word can carry. NOT retrieved from McGregor by this file. Falsifier: query McGregor before relying on this.
spanish		'el mundo' repeats the mundus defect. 'entorno'/'ambiente' (environment, surroundings) are subtler and worse in one respect: they are the standard renderings in the active-inference literature, they sound technical, and they smuggle in the idea that η SURROUNDS the system spatially. η is defined by conditional independence, not by geometry. PENDING-CITATION like the whole Spanish register.
any		'Reality' asserts what FRISTON_BLANKET asserts and what this corpus declines. 'The real world' is worse - it implies the mind's side is unreal. μ and η have the same ontological standing in the partition; only their position differs.

Back-translation test — status: NOT-RUN. No reviewer has been engaged. This test is specified, not passed.

SENSORY_STATES —

Confidence: medium for the Sanskrit (both components CITED, and the route-not-faculty strategy genuinely evades the indriya trap); LOW for the Latin, whose adjective sensorius is not cited at all and which keeps the faculty word the Sanskrit was careful to drop. The two registers should not be treated as equally warranted. · Reviewer: NONE - PENDING EXPERT REVIEW

(no definition in this file; see CONCEPTS)

register	form	literal gloss	status
sanskrit	प्रवेशावस्थाः (praveśāvasthāḥ)	the entry states; the states that are the way in	COINED
latin	status sensorii	the sensory states	PENDING-CITATION
english	sensory states	The blanket states s influenced by external but not internal states, forming the world's only route into the mind.	—
hindi	प्रवेश अवस्थाएँ ()	entry states	COINED
spanish	estados sensoriales	sensory states	PENDING-CITATION

Ambiguity notes

• The English label says 'sensory', which invites every register to reach for a sense-faculty word. The DEFINITION says something quite different and much thinner: s are the blanket states influenced by external but not internal states. That is a sparsity condition on a graph. It says nothing about sensing, and it would hold of a thermometer's mercury.
• The Sanskrit therefore deliberately renders the ROUTE (praveśa, 'entrance, way in') rather than the FACULTY, and buys real safety by doing so - praveśāvasthāḥ cannot be heard as a sense-organ. The Latin does NOT yet do this: status sensorii keeps the faculty word, and its adjective is not even cited. The two registers are not equally safe and this file does not pretend they are.
• SCOPE NOTE carried from the source concept, and it is a real one: the directional sparsity (external influences s; internal does not) is a requirement ADDED on top of Pearl's condition by the free-energy literature. It is NOT part of Pearl's original construct. A register that renders s as 'the senses' obscures the fact that this asymmetry is a stipulation with an author and a date, not a mathematical necessity.
• praveśa's own freight is comic but real: Monier-Williams also gives it 'income, revenue, tax, or toll', 'employment, use, utilisation', 'manner, method', 'entrance on the stage', 'the entrance of the sun into a sign of the zodiac', and 'the syringe of an injection pipe'. None is likely to be reached for in a technical compound with avasthā, but praveśa ALONE is badly overloaded and must not travel alone.
• praveśa also carries 'intrusion' and 'interfering with another's business, obtrusiveness'. s is not an intrusion; η's influence on s is the ordinary condition of there being a blanket at all.

Forbidden mistranslations

register	form	why it is wrong
sanskrit		The obvious, idiomatic, and completely wrong choice. Monier-Williams gives indriya as 'faculty of sense, sense, organ of sense' - an ORGAN or POWER belonging to a cognizer, not a state. Its Vedāntic sense is worse: the entry records fourteen indriya-s comprising five sense organs, five motor organs, PLUS manas, buddhi, ahaṃkāra and citta, each governed by a presiding deity. Rendering s as indriya therefore (a) makes s an organ rather than a state, (b) installs the very cognizer that MIND forbids, and (c) imports a fourteen-fold psychology with deities attached. It also collapses SENSORY_STATES and ACTIVE_STATES into one category, since the indriya-s comprise both jn̄ānendriya (sense) and karmendriya (action) - destroying the exact distinction μ/η sparsity depends on.
sanskrit		Independently of the philosophy: Monier-Williams's PRIMARY senses for indriya are 'power, force, the quality which belongs especially to the mighty Indra', 'exhibition of power, powerful act', 'bodily power', 'virile power', and 'semen virile'. The word is built on the god Indra. A translator using it for 's' is not reaching for a neutral technical term; they are reaching for a word whose core sense is divine potency.
latin		Lewis & Short gives sensus, us, m. as 'the faculty or power of perceiving; perception, feeling, sensation, sense', and its whole section II is mental - 'feeling, sentiment, emotion, affection; sense, understanding, capacity; humor, inclination, disposition, frame of mind', plus 'opinion, thought, view' and sensus communis. It is a faculty-and-feeling word from end to end. s is neither a faculty nor a feeling. Using sensus alone for s would assert both a perceiver and something it is like to perceive.
any		'Perception' and 'observation' both name an ACT performed by a subject; s is a set of states, and the act-reading reinstalls the subject. 'Input' is the mirror error: it is engineering-neutral but implies an intended recipient and a designed channel, and it invites the reading that s is data delivered to μ. Nothing delivers anything. s is simply the part of b that η influences and μ does not.
any		The single structural failure to avoid in this pair of entries. s and a are distinguished ONLY by the direction of influence. A register with one word covering both (indriya is exactly this) cannot express the sparsity condition and therefore cannot express the blanket.

Back-translation test — status: NOT-RUN. No reviewer has been engaged. This test is specified, not passed.

ACTIVE_STATES —

Confidence: medium for the Sanskrit (both components CITED; the exit-route strategy avoids both the karman trap and the busyness reading, and mirrors SENSORY_STATES as the mathematics requires). LOW for the Latin despite activus being CITED - the citation is real and the attested sense ('practical, opposed to contemplative') actively fights the concept. A CITED headword with the wrong sense is not a better rendering than an honest coinage, and this entry is the clearest case in the file of that distinction mattering. · Reviewer: NONE - PENDING EXPERT REVIEW

(no definition in this file; see CONCEPTS)

register	form	literal gloss	status
sanskrit	निर्गमावस्थाः (nirgamāvasthāḥ)	the exit states; the states that are the way out	COINED
latin	status activi	the active states	COINED
english	active states	The blanket states a influenced by internal but not external states, forming the mind's only route out into the world.	—
hindi	निर्गम अवस्थाएँ ()	exit states	COINED
spanish	estados activos	active states	PENDING-CITATION

Ambiguity notes

• The source concept's warning is blunt and every register violates it by default: 'Active' means ACTED-UPON-BY-THE-INTERIOR, not energetic or busy. a is defined by what influences it (μ does, η does not). The English word 'active' says the opposite - it says a is doing something. It is a passive participle's job dressed in an active adjective.
• The Sanskrit sidesteps this entirely by rendering the ROUTE (nirgama, 'exit, way out') instead of the activity, exactly as praveśāvasthāḥ does for s. nirgamāvasthāḥ makes no claim that anything is busy. This is the strongest reason to prefer the Sanskrit construction over the Latin one here.
• Latin activus is CITED but its attested sense is the WRONG one, and wrong in an especially expensive way: Lewis & Short gives it as 'active, practical', of philosophy, OPPOSED TO CONTEMPLATIVE. That opposition is precisely the action/thought dualism the blanket partition dissolves - a is not what the system does INSTEAD of thinking; a is the only channel μ has. A Latinist reading status activi against Seneca hears 'the practical states, as against the contemplative ones', which reintroduces the split this corpus is trying to describe away. The citation is real and the fit is bad; both facts are recorded.
• Latin activus's other attested sense is grammatical - verba activa, transitive verbs. A reader may hear 'the transitive states'. Harmless but noisy.
• nirgama's freight cuts one real way: Monier-Williams gives 'disappearing, vanishing, cessation, end' alongside 'exit, outlet'. a does not vanish and does not cease. A reader taking that sense would read nirgamāvasthāḥ as 'the states in which things stop', which is not merely different but nearly opposite - a is where μ reaches the world, not where anything ends.
• nirgama also carries 'export-place (of goods)' (Manu viii, 401) and 'escaping from'. The export sense is an oddly apt accident - a is where the interior's influence leaves - but it is an accident and carries a commercial frame that is not meant.

Forbidden mistranslations

register	form	why it is wrong
sanskrit		karman is the single most doctrinally loaded word available in this register: act, deed, work, rite - and the moral residue of action that determines rebirth. a has no moral valence and no consequences beyond the next state. karmendriya ('organ of action') adds the indriya defect on top (an organ, not a state; a cognizer implied; see SENSORY_STATES). NEITHER headword was retrieved from Monier-Williams by this file; the glosses are the author's and are NOT cited. Falsifier: retrieve both before relying on this reasoning - but the doctrinal risk is severe enough that karman should be treated as forbidden pending that check, not after it.
sanskrit		'Action, doing, performance' - cleaner than karman but still names an ACT, and a is not an act. NOT retrieved from Monier-Williams by this file; the gloss is the author's and is NOT cited.
latin		Both name an act or a doing, and both are nouns for events; a is a set of states, not an event. Additionally actus imports the Aristotelian-scholastic actus/potentia pair, which is a whole metaphysics of act and potency. NOT retrieved from Lewis & Short; listed as traps, not cited facts.
any		'Output' and 'effector' import a designed device with a purpose, and 'motor' imports muscles - the source concept's whole point is that a is defined by an influence asymmetry, which is equally true of a bacterium's flagellum, a thermostat's relay and nothing at all. 'Motor states' additionally re-anatomises b, which BODY forbids.
any		Directly contradicts the source concept's own note. a can be perfectly still. A rock's surface has active states in this sense if anything inside it influences them.

Back-translation test — status: NOT-RUN. No reviewer has been engaged. This test is specified, not passed.

CONDITIONAL_INDEPENDENCE —

Confidence: medium for the Sanskrit coinage, which is the most defensible construction in this file on structural grounds (it makes conditionality morphologically inseparable from independence) while carrying nirapekṣa's 'indifferent/accidentally' freight, which is bad. LOW for Latin, Hindi and Spanish, none of which has a single retrieved headword in this entry. This entry has the weakest citation base in the file and should not be leaned on. · Reviewer: NONE - PENDING EXPERT REVIEW

(no definition in this file; see CONCEPTS)

register	form	literal gloss	status
sanskrit	सापेक्षनिरपेक्षता (sāpekṣa-nirapekṣatā)	dependent-independence; independence that is itself relative to something	COINED
latin	independentia condicionalis	conditional independence	PENDING-CITATION
english	conditional independence	The relation mu perp eta given b, which is the whole content of a blanket: given the blanket, internal and external states carry no further information about each other.	—
hindi	सापेक्ष निरपेक्षता ()	conditional independence	PENDING-CITATION
spanish	independencia condicional	conditional independence	PENDING-CITATION

Ambiguity notes

• The source concept is emphatic and the emphasis is the content: NOT 'mostly' and NOT 'usually' - by definition, or the partition is not a blanket. Every register must carry an exact factorization claim, not a tendency. No natural-language word for 'independent' in any of these registers carries exactness; they all admit degrees. This is a defect of all five registers equally and the math is the only fence.
• The deep trap, and the reason this entry is hard: 'independence' in ordinary language is a property of ONE thing (a person is independent). Here it is a three-place relation - μ is independent of η GIVEN b - and it is FALSE without the third place. Drop 'given b' and the claim becomes not merely weaker but wrong: μ and η are emphatically NOT independent unconditionally; that is exactly why there is a blanket. A register that renders this with a one-place word for independence has stated the opposite of the concept.
• The Sanskrit coinage is designed against exactly that failure: by yoking sāpekṣa ('dependent, having regard to') to nirapekṣa ('independent, without regard to') - the two morphological opposites of a single root - it makes the conditionality structurally inseparable from the independence. You cannot drop the third place without dropping half the word. This is the one place in the file where a coinage does something no attested term does.
• nirapekṣa's freight is severe and pulls the wrong way: Monier-Williams's other senses are 'desireless, careless, indifferent, disinterested', and the adverb nirapekṣam is 'regardlessly, ACCIDENTALLY'. That makes independence an ATTITUDE (not caring about something) rather than a probabilistic factorization, and the 'accidentally' sense is close to a direct contradiction: conditional independence is not an accident but a structural fact. A reader with the Mānava-dharmaśāstra or Kāvya usage in mind will hear an ascetic's detachment where a factorization is meant.
• sāpekṣa's attested sense 'having regard or respect to' is also personal - it is about someone showing consideration. Neither component is a mathematical term and neither pretends to be.

Forbidden mistranslations

register	form	why it is wrong
any		The cardinal error of this entry and the most likely one. μ and η are NOT independent; they are independent GIVEN b. A bare independence word asserts unconditional independence, which is false and which, if true, would mean there was nothing to infer and no reason for a blanket to exist. This is not a shade of meaning - it is the negation of the concept.
sanskrit		'Self-dependence, autonomy, freedom' - a one-place property of an agent, and in Śaiva and grammatical usage a term of art for AGENCY (the kāraka svatantraḥ kartā, 'the agent is the independent one'). It therefore commits both available errors at once: it drops the conditioning term, and it installs an agent. NOT retrieved from Monier-Williams by this file; the glosses are the author's and are NOT cited. Falsifier: retrieve svatantra before relying on this.
latin		Both are the classical idiom for independence and both are POLITICAL or LEGAL - freedom from another's power, being one's own master. Statistical factorization has nothing to do with either. Listed as traps; NOT retrieved from Lewis & Short.
any		'Uncorrelated' is a strictly weaker condition than independence (zero correlation does not imply independence except in special cases) and substituting it silently weakens the claim to something the blanket does not require and which would not suffice. 'Orthogonal' is the same error dressed in geometry. 'Unrelated' is simply false: μ and η are richly related THROUGH b.
any		Technically near-synonymous and therefore tempting, but it is a screening METAPHOR, and in a corpus where the blanket is already fighting a covering/concealment reading (see MARKOV_BLANKET, āvaraṇa), importing a second screen-word compounds the error. It also imports a specific literature on common causes that this concept does not invoke.

Back-translation test — status: NOT-RUN. No reviewer has been engaged. This test is specified, not passed.

CONDITIONAL_MUTUAL_INFORMATION —

Confidence: low, and low in a way no amount of further work in these registers would fix. Two components are CITED (paraspara, mutuus) and one CITED component is actively misleading (informatio). There is no attested term for Shannon information in Sanskrit or Latin because the concept postdates both by two millennia; the charter predicted exactly this outcome and this entry records it rather than pretending otherwise. The mathematics is the only warranted content here. · Reviewer: NONE - PENDING EXPERT REVIEW

(no definition in this file; see CONCEPTS)

register	form	literal gloss	status
sanskrit	सापेक्ष-परस्पर-सूचना (sāpekṣa-paraspara-sūcanā)	conditional mutual indication	COINED
latin	informatio mutua condicionalis	conditional mutual information -- but see ambiguity_notes: informatio does NOT mean information	COINED
english	conditional mutual information	The quantity I(mu;eta|b), in nats, which is the operable test of whether a candidate boundary is a real blanket, since I greater than zero beyond estimator noise means a route from world to mind bypasses the body and the partition is drawn in the wrong place.	—
hindi	सापेक्ष परस्पर सूचना ()	conditional mutual information	COINED
spanish	información mutua condicional	conditional mutual information	PENDING-CITATION

Ambiguity notes

• THE HONEST OUTCOME THE CHARTER PREDICTS, arriving exactly where it was predicted: there is NO attested Sanskrit term and NO attested Latin term for information in Shannon's sense. None. The quantity was defined in 1948 and neither language has a word for it. Every register in this entry is therefore a coinage or a calque, and the ONLY thing in this entry with any warrant is the mathematics: I(μ;η|b), measured in NATS. The unit is the fence. No register may be read as supplying a meaning the math does not already carry.
• Latin informatio is the trap of this entry, and it is a beautiful one because the word is CITED, classical, and Ciceronian - and means something else entirely. Lewis & Short: 'an outline, sketch, first draft' and, tropologically, 'an idea, conception'. Cicero uses it at N.D. 1.36.100 for a PRECONCEIVED NOTION in someone's mind. Cicero's informatio is a mental conception in a person; Shannon's information is a measure over a joint distribution and requires no mind at all. Using informatio therefore imports a cognizer into the one entry whose whole job is to be a mind-free measurement. The word looks like the right answer and is an anachronism of two thousand years.
• mutuus's primary attested sense is 'borrowed, lent', with the substantive mutuum meaning 'a loan'; the 'reciprocal, mutual' sense is marked Transf. (transferred) by Lewis & Short, and its gloss there opens 'in return, in exchange'. informatio mutua could therefore be construed as 'a borrowed conception' - a loan of an idea. This is not a serious reading risk in context but it is what the lexicon actually says, and it is recorded rather than quietly dropped. ADVERSARY CORRECTION 2026-07-15: this note previously said the sense is 'marked Trop. by Lewis & Short'. It is marked Transf. The note's SUBSTANCE survives the correction unchanged - 'borrowed, lent' is still the literal sense and 'mutual' is still the transferred one - but the label was wrong and a note that names what a lexicon marks must name it correctly.
• Hindi sūcanā is the everyday modern word for 'information' as in a notice, an announcement, something someone tells you. That is the folk sense, not the measure-theoretic one, and it is arguably the worst kind of false friend: it will feel entirely natural to a Hindi reader and will convey nothing of I(μ;η|b). Because sūcanā's Hindi sense is a communication BETWEEN PERSONS, it also reinstalls the cognizer.
• THE SCOPE FACT THAT MATTERS MOST, carried from the source concept and softened by nobody: this criterion exists, is clean, and is overwhelmingly UNEXERCISED. I(μ;η|b) is NOT-MEASURED for essentially all real biological systems and for every UNI boundary. When any text says 'system X has a Markov blanket', the default assumption should be that I was never estimated. Naming this quantity elegantly in five registers does not measure it in one.
• The nats are load-bearing and no register carries them. I > 0 beyond ESTIMATOR NOISE is the operable test; the noise floor is part of the claim; a term without a unit and an error bar is not a measurement.

Forbidden mistranslations

register	form	why it is wrong
sanskrit		The natural reach for 'information' and catastrophic. jn̄āna is KNOWLEDGE - something a knower has; vijn̄āna is discriminative cognition and, in Buddhist usage, one of the five skandhas and a term of art for consciousness itself. I(μ;η|b) is a number in nats computed from a joint distribution; it is not knowledge, nobody has it, and it applies to a rock. Rendering information as jn̄āna installs a knower in the one place a knower is most damaging, and vijn̄āna additionally imports the whole Yogācāra apparatus. NEITHER headword was retrieved from Monier-Williams by this file; the glosses are the author's and are NOT cited - but the doctrinal risk is severe enough that both should be treated as forbidden pending that check, not after it.
latin		All three name knowledge or acquaintance held by a knower - same defect as jn̄āna, in Latin. NOT retrieved from Lewis & Short; listed as traps, not cited facts.
any		I(μ;η) and I(μ;η|b) are different quantities and the difference is the entire concept. Unconditional mutual information between mind and world is large and its being large is unremarkable. It is the CONDITIONAL quantity that tests the boundary. Dropping the condition does not weaken the claim; it tests nothing.
any		Not a lexical error but an evidential one, and it is the error this concept exists to prevent. A term of art that reads as established fact ('the conditional mutual information of the system') licenses the reader to assume an estimate exists. For essentially every system anyone has named, none does.
any		'Shared' and 'common' imply a substance possessed jointly by two parties; mutual information is not a stuff that μ and η both hold. 'Overlap' is a geometric metaphor and suggests the two sets intersect, which they do not - they are disjoint by construction.

Back-translation test — status: NOT-RUN. No reviewer has been engaged. This test is specified, not passed.

PEARL_BLANKET —

Confidence: low. The head nouns are CITED but the concept's entire content is an evidential status (free, model-relative, claiming nothing) that no noun in any register can carry. The registers name the object correctly and convey nothing that matters about it. · Reviewer: NONE - PENDING EXPERT REVIEW

(no definition in this file; see CONCEPTS)

register	form	literal gloss	status
sanskrit	पर्ल-व्यवधानम् (parla-vyavadhānam)	the Pearl intervening-screen	COINED
latin	velamen Pearlianum	the Pearlian covering	COINED
english	Pearl blanket	A blanket as Pearl coined it: a formal statistical construct over a graph you drew, doing conditional-independence bookkeeping relative to a model, and thereby cheap, well-defined and unobjectionable.	—
hindi	पर्ल व्यवधान ()	the Pearl intervening-screen	PENDING-CITATION
spanish	manta de Pearl	the Pearl blanket	PENDING-CITATION

Ambiguity notes

• The load-bearing word in this entry is not a noun but an adjective the registers cannot carry: CHEAP. A Pearl blanket is free BECAUSE IT CLAIMS NOTHING ABOUT THE WORLD. It is conditional-independence bookkeeping relative to a graph YOU ALREADY DREW. The graph is a modeling artefact; the blanket over it inherits exactly that status and no more.
• No register conveys 'relative to a model'. Every register produces a noun that sounds like a thing that exists. This is the entry where the reader is most likely to be misled by good translation, because the term is unobjectionable and the reader will assume its referent is out there. The referent is in the model.
• 'Pearl' is a surname (Judea Pearl) and in every register it will look like the English common noun for the gem - and in Devanagari transliteration it is worse, since पर्ल carries no capitalisation cue at all that would mark it as a name. Recorded as a real, unfixed hazard of the transliteration.
• This entry and FRISTON_BLANKET are deliberately given the SAME head noun in each register (vyavadhāna / velamen), differing only in the attached name. That is correct and it is also the danger: the two concepts differ in EVIDENTIAL COST, not in structure, and identical heads make the costly one look as free as the free one. The distinction lives entirely in the definitions. See BLANKET_SLIDE, which is the name for what happens when it does not.

Forbidden mistranslations

register	form	why it is wrong
any		The whole content of 'Pearl blanket' is that it is a formal construct over a graph the modeller drew. A rendering that suggests finding, revealing or locating a blanket has converted a free construct into an expensive claim - i.e. has performed the BLANKET_SLIDE.
any		A real and stupid failure mode that machine translation reaches for immediately, and which the Devanagari transliteration cannot signal against. 'The pearl blanket' is not a thing. The name is a person.
any		The two are distinguished only by the name attached. Dropping the name, or rendering both as simply 'the blanket', collapses a free bookkeeping construct into a metaphysical claim about where an organism ends - which is precisely the field's characteristic error and has its own concept_id here.

Back-translation test — status: NOT-RUN. No reviewer has been engaged. This test is specified, not passed.

FRISTON_BLANKET —

Confidence: low, for the same reason as PEARL_BLANKET: the head noun is CITED and the concept's content is an evidential cost that no noun carries. Additionally the Sanskrit and Hindi transliterations of the eponym are orthographically unstable (no native /f/), which this file has not resolved. · Reviewer: NONE - PENDING EXPERT REVIEW

(no definition in this file; see CONCEPTS)

register	form	literal gloss	status
sanskrit	फ्रिस्टन-व्यवधानम् (phrisuṭana-vyavadhānam)	the Friston intervening-screen	COINED
latin	velamen Fristonianum	the Fristonian covering	COINED
english	Friston blanket	A blanket read as the metaphysical boundary of a real thing -- the claim that some system out there genuinely has an inside, a surface and an outside -- which requires premises beyond the formalism's mathematical success.	—
hindi	फ्रिस्टन व्यवधान ()	the Friston intervening-screen	PENDING-CITATION
spanish	manta de Friston	the Friston blanket	PENDING-CITATION

Ambiguity notes

• The load-bearing word here is EXPENSIVE, and as with PEARL_BLANKET no register carries it. A Friston blanket is an EMPIRICAL CLAIM: that some system out there genuinely has an inside, a surface and an outside. It requires I(μ;η|b) ≈ 0 in the actual system - a measurement which, per CONDITIONAL_MUTUAL_INFORMATION, has essentially never been made.
• STATED PLAINLY BECAUSE IT MUST BE: this corpus does not assert that anything has a Friston blanket, and UNI is NOT claimed to have one. This entry names a position; it does not take it. A reader who finds these five renderings elegant should notice that elegance is not evidence and that no register here is a warrant for the claim it names.
• The dispute is LIVE. The corpus takes no side beyond insisting the slide be named (see BLANKET_SLIDE). Nothing in this lexicon should be read as adjudicating whether Friston blankets exist; a lexicon cannot settle that and this one does not try.
• The eponym is doubly awkward in the non-Latin registers. Devanagari has no native /f/: the nukta letter फ़ is the standard device and is routinely written as plain फ, so फ्रिस्टन may be read with an aspirated /ph/. The transliteration is unstable and this file does not pretend to have fixed it.
• As with PEARL_BLANKET, the head noun is deliberately identical to its sibling's. The two concepts are structurally the same object and differ ONLY in evidential cost. Identical heads are therefore correct AND are the mechanism by which the cheap term and the expensive one get confused. The confusion has its own name and its own entry.

Forbidden mistranslations

register	form	why it is wrong
any		This concept NAMES a metaphysical reading; it does not endorse it. A register that renders 'Friston blanket' as, in effect, 'the true boundary of the organism' has converted the naming of a contested position into the assertion of it - inside the entry whose whole purpose is to keep that assertion visible and attributable.
sanskrit		Doubly forbidden, and interestingly so. āvaraṇa is forbidden for MARKOV_BLANKET because its concealment sense (Monier-Williams: 'covering, hiding, concealing'; Jain 'mental blindness') smuggles in a hidden reality. Here that smuggled content happens to BE the Friston claim - so āvaraṇa would look almost right. It remains forbidden, because a register must not make a contested empirical claim TRUE BY VOCABULARY. If āvaraṇa were admitted here, the Friston reading would become unfalsifiable in Sanskrit: the word would assert what the measurement is supposed to test.
any		Same defect as PEARL_BLANKET. Dropping the eponym collapses the expensive claim into the free one - the BLANKET_SLIDE in a single editorial decision.
any		All three assert the conclusion. The third additionally installs a self, which MIND forbids for independent reasons.

Back-translation test — status: NOT-RUN. No reviewer has been engaged. This test is specified, not passed.

BLANKET_SLIDE —

Confidence: medium-high for the Sanskrit and Latin - now the highest in the file outright, WORLD having been downgraded to medium by the back-translation adversary 2026-07-15 (this entry was NOT downgraded; the cross-reference is corrected here only so it does not go stale) - and for an unusual reason: both head nouns are CITED and both independently carry the SAME slip/error doubling the English metaphor depends on, so the translation preserves the figure rather than approximating it. The unresolved risk is skhalana's 'displacement of a garment' sense, which is a live pun in a compound with a covering-word. · Reviewer: NONE - PENDING EXPERT REVIEW

(no definition in this file; see CONCEPTS)

register	form	literal gloss	status
sanskrit	व्यवधानस्खलनम् (vyavadhāna-skhalanam)	the blanket-stumble; the slip concerning the intervening screen	COINED
latin	lapsus velaminis	the slip of the covering; the blanket's slide	COINED
english	the Pearl-to-Friston slide	The field's characteristic error of defining a Pearl blanket, which is free, and then quietly concluding one has discovered where the organism ends, which is expensive and requires a measurement nobody made.	—
hindi	व्यवधान-स्खलन ()	the blanket-slip	PENDING-CITATION
spanish	el desliz de Pearl a Friston	the slip from Pearl to Friston	PENDING-CITATION

Ambiguity notes

• STORE = META. This is the only META-store concept in the file and the routing is load-bearing: BLANKET_SLIDE names a defect in COMMENTARY ABOUT signals, not a signal. It is not a fact about any system. Under SIGNUM SIGNUM MANET the signal and the commentary about it travel separately, and this entry is commentary - it must never be filed, cited or merged as though it were an observation about a blanket.
• Both the Sanskrit and the Latin get a genuine gift here, and it is the only such gift in the file: skhalana and lapsus BOTH carry the physical sense (stumbling, sliding) AND the intellectual sense (mistake, error, fault) in their own attested entries. The English 'slide' does exactly the same double duty. Three registers independently make an error into a slip, so the metaphor survives translation intact rather than being reconstructed. This is a real, checkable convergence, not a poetic one.
• What no register carries is the ASYMMETRY that makes the slide a slide: it runs one way, cheap-to-expensive, because you get the Pearl blanket for free and then find yourself holding a Friston blanket you never paid for. Nothing in skhalana, lapsus or 'slide' says which way the slope runs. Only the definition does.
• skhalana's freight is real and awkward: Monier-Williams also gives 'discharge, emission (of semen)' and 'displacement (of a garment)'. In compound with vyavadhāna - itself a covering-word - vyavadhāna-skhalanam could be read, quite literally and by the lexicon's own senses, as 'the slipping of the garment'. Given that the concept is ABOUT a blanket, this is a live pun and not a remote one. Recorded, unresolved.
• Correct mathematics does not license the metaphysical reading. That sentence is the concept and no register asserts it; the terms name the error without explaining why it is one.
• The corpus takes NO SIDE in the underlying dispute beyond insisting the slide be named. A register that made the slide sound like a proven fallacy would overstate the corpus's position as badly as one that made it sound harmless.

Forbidden mistranslations

register	form	why it is wrong
any		Overstates the corpus's position. The corpus insists the slide be NAMED; it does not adjudicate the dispute, which is live. 'Fallacy', 'error proven', 'refuted inference' all take a side this file has no warrant to take.
any		Understates it in the opposite direction. The slide is the difference between a claim that costs nothing and a claim that requires a measurement nobody made. That is an evidential jump, not a wording preference.
sanskrit		bhrama is error-as-delusion (and in Vedānta, error about what is real - which is a metaphysical thesis, not a methodological slip); pramāda is negligence or carelessness, which moralises the mistake and implies the field is being sloppy rather than making a specific, principled, expensive inference. NEITHER was retrieved from Monier-Williams by this file; the glosses are the author's and are NOT cited.
latin		Both assert that the move IS wrong. lapsus is preferred precisely because it names a slipping without adjudicating - which matches the corpus's actual position. NOT retrieved from Lewis & Short; listed as traps, not cited facts.
any		Not a lexical error but a store violation, and the cardinal sin: this is META. Merging it with TRUE-store material about actual blankets is exactly the crossing this corpus forbids.

Back-translation test — status: NOT-RUN. No reviewer has been engaged. This test is specified, not passed.

INTEROCEPTION —

Confidence: low. The Latin is an honest, openly-labelled modern coinage (nothing is pretended). The Sanskrit is built from three CITED components but shares the English term's central defect - it names where the signal comes from rather than which side of the partition it sits on - and the author predicts every register fails the load-bearing question. This is an entry where the definition does all the work and the vocabulary actively misleads. · Reviewer: NONE - PENDING EXPERT REVIEW

(no definition in this file; see CONCEPTS)

register	form	literal gloss	status
sanskrit	अन्तःप्रवेशावस्थाः (antaḥpraveśāvasthāḥ)	the inward entry-states; the states that are the way in from within	COINED
latin	interoceptio	interoception	COINED
english	interoception	Sensory channels whose external states are other parts of the same organism, forming a blanket nested inside a blanket, and rendered in M12's engineering clause as hardware signals.	—
hindi	अंतःप्रवेश अवस्थाएँ ()	inward entry-states	COINED
spanish	interocepción	interoception	PENDING-CITATION

Ambiguity notes

• The fact that makes this concept hard, and that every register fumbles: FROM THE BRAIN'S POSITION, THE VISCERA ARE η. Interoception is not the interior sensing itself; it is a blanket nested inside a blanket, whose external states happen to be other parts of the same organism. The 'inter-' in the English says 'within the body' and therefore points at the WRONG side of the partition - it names the anatomical location of the source, not the topological role, and the topological role is that the source is OUTSIDE the inner blanket. Every register inherits this misdirection from the English.
• The Sanskrit coinage antaḥpraveśāvasthāḥ shares the defect knowingly: antar- says 'inner', which locates rather than types. A register that got this right would have to say something like 'the entry-states whose exterior is the same organism', which no compound in any of these languages will carry.
• M12 renders interoception in its engineering clause as HARDWARE SIGNALS, and that gloss is deliberately deflationary. Nothing here licenses reading interoception as felt bodily experience, as affect, or as the substrate of emotion - a large literature does read it that way, and this corpus does not.
• SCOPE VIOLATION ON RECORD, carried verbatim from the source concept because it is the most useful thing in it: '80% of the vagus is afferent' is a RAT ABDOMINAL measurement in near-universal circulation as a fact about the human vagus, where the cervical figure is ~73-74%. A number quoted outside its scope is a defect EVEN WHEN THE NUMBER IS RIGHT. This is a lexicon entry and cannot fix that, but any register rendering 'interoception' inherits a term whose most-cited quantitative fact is misattributed by scope.
• praveśa's overload (income, tax, zodiac, injection-syringe, obtrusiveness) applies here as much as in SENSORY_STATES. The syringe sense is an especially unfortunate accident in an entry about viscera.

Forbidden mistranslations

register	form	why it is wrong
any		The exact inversion of the concept. From the brain's position the viscera are EXTERNAL states. A rendering that makes interoception 'the inside sensing itself' has dissolved the nested blanket it is supposed to name, and with it the only structure the concept has.
sanskrit		Disastrously attractive: it is literally 'the inner instrument', it is a standing term of art, and Monier-Williams's OWN manas entry defines manas as 'the internal organ or antaḥkaraṇa of perception and cognition'. So the lexicon itself ties antaḥkaraṇa to the mind-as-organ. Using it for interoception would install the whole Sāṃkhya/Vedānta inner-instrument apparatus (manas, buddhi, ahaṃkāra, citta) inside an entry that is supposed to name hardware signals from the gut. The evidence for this warning is the retrieved manas entry, which does carry that definition.
any		All four import phenomenal experience. M12's engineering clause renders interoception as hardware signals; none of these is a hardware signal, and all of them install the experiencer that MIND forbids.
latin		Doubly wrong: sensus is a faculty-and-feeling word (Lewis & Short: 'the faculty or power of perceiving', and a whole mental section II) so it installs a perceiver; and 'internus' repeats the anatomical-location error, asserting that the sensed thing is inside, when its topological role is to be outside the inner blanket. sensus IS cited; internus was not retrieved.
any		Not a lexical error but the scope violation named above, recorded here because the term drags it along. The figure is rat abdominal; the human cervical figure is ~73-74%.

Back-translation test — status: NOT-RUN. No reviewer has been engaged. This test is specified, not passed.

BLANKET_NESTING —

Confidence: low for the Sanskrit: garbha is CITED and its dominant sense is the womb, so the coinage is admitted provisionally and against its own lexicon entry's centre of gravity; the author's stated prediction is that it fails its own back-translation test. Medium for the Latin, which carries the recursion in syntax rather than vocabulary and refuses the bird metaphor the English is stuck with. · Reviewer: NONE - PENDING EXPERT REVIEW

(no definition in this file; see CONCEPTS)

register	form	literal gloss	status
sanskrit	व्यवधानगर्भता (vyavadhāna-garbhatā)	the containedness of screen within screen; screen-within-screen-ness	COINED
latin	velamina in velaminibus	coverings within coverings	COINED
english	blanket nesting	The recursion in which one level's blanket becomes part of the next level's internal states, running organelle to cell to tissue to organ to organism to colony to ecosystem.	—
hindi	व्यवधान के भीतर व्यवधान ()	screens within screens	PENDING-CITATION
spanish	mantas dentro de mantas	blankets within blankets	PENDING-CITATION

Ambiguity notes

• ATTESTATION HONESTY, carried from the source concept and applying to this file's own English: the corpus writes this as 'blankets of blankets' / 'Nesting'. The hyphenated term 'blanket-nesting' appears in 0 FILES of the corpus. The English headword being translated here is therefore itself a lexicon-internal coinage, and the four other registers are translations of a term the corpus does not actually use. Recorded because it would be easy to present this entry as though the English were settled.
• THE DISTINCTION THIS ENTRY EXISTS TO PROTECT: that blankets CAN compose this way is a MATHEMATICAL FACT. That any particular biological hierarchy IS so composed is a SEPARATE EMPIRICAL CLAIM. And the number of nested levels in any named organism is NOT-MEASURED. The concept is a CONSTRUCTION, not an observation. The organelle-cell-tissue-organ-organism-colony-ecosystem ladder is an illustration of a possibility, not a report of a finding - and it reads exactly like a report of a finding, in all five registers.
• garbha is the contested choice in this file and the argument runs both ways. FOR: Monier-Williams attests 'the inside, middle, interior of anything' and 'any interior chamber, adytum or sanctuary of a temple' - the garbhagṛha of a temple is literally a chamber nested inside a larger structure, which is the right image. AGAINST: the entry's FIRST sense is 'the womb', and it also carries 'a foetus or embryo, child, brood or offspring (of birds)' and 'a woman's courses'. The reproductive senses dominate the entry. A reader will hear gestation, and nesting is composition, not reproduction: the inner blanket is not the outer blanket's offspring and is not developing into anything.
• The bird sense of garbha ('brood or offspring OF BIRDS') collides with the English metaphor in a way that looks helpful and is not: English 'nesting' is already a bird metaphor, and a reader may take the convergence as confirmation. It is a coincidence of two unrelated metaphors, and taking it as confirmation would be exactly the kind of false comfort this file is built to refuse.
• The Latin refuses a nesting-noun entirely and uses recursive syntax instead. This is the file's only register that carries recursion by GRAMMAR rather than by LEXICON, and it is the more honest solution for a concept whose content is a structural relation.
• What no register carries: the DIRECTION of the recursion (one level's BLANKET becomes the next level's INTERNAL STATES - a b becomes a μ, which is a type change, not merely a containment). 'Screens within screens' says containment and says nothing about the type change. The definition does.

Forbidden mistranslations

register	form	why it is wrong
latin		The direct calque of English 'nesting' and firmly refused. Lewis & Short gives nidus as 'a nest' (Cic. de Or. 2.6.23; Verg. G. 4.307), with transferred senses 'young birds in a nest', 'a litter of pigs', 'a receptacle or case for books', 'a dwelling, residence, home'. Every one is a CONTAINER OR A HOME - a physical thing that holds another physical thing, with a bird in it. Blanket nesting is a recursive type relation in which one level's blanket becomes the next level's internal states. nidus imports birds, dwellings and litters and conveys no recursion at all. That the English word 'nesting' is itself this metaphor is the defect being repaired here, not a licence to reproduce it in Latin.
sanskrit		The Sanskrit bird's-nest word - same defect as nidus. NOT retrieved from Monier-Williams by this file; the gloss is the author's and is NOT cited.
any		The garbha risk made explicit, and the reason this entry's Sanskrit is only provisionally admitted. Nesting is COMPOSITION. The inner blanket is not born from the outer one, is not growing, and is not becoming anything.
any		Not lexical but evidential, and the most consequential error available here. That blankets CAN nest is mathematics; that a given organism's levels DO is an unmade measurement. The number of nested levels in any named organism is NOT-MEASURED.
any		All three are theory-laden imports. 'Hierarchy' adds control and rank, which the relation does not contain; 'levels of organisation' imports a whole biological research programme; 'emergence' asserts a contested thesis the concept does not make. The relation is composition, full stop.

Back-translation test — status: NOT-RUN. No reviewer has been engaged. This test is specified, not passed.