NOTE-151: Register — The Depth of the Body

Contribution

The paper names a second dimension of the body besides spatial breadth, which it calls depth: the level of abstraction, or microphysical commitment, of the objects that make it up. It defines depth for descriptions (degree of microphysical commitment) and extends it to objects through real definitions (§3). It also defines high-level neutrality for events. On that basis it states two moral principles restricting bodily rights and interests to events that are not high-level neutral (§5). It applies them to BCIs whose computation runs on third-party or cloud hardware (§6): the implemented software object is a body part and the server is not. That reconciles the user’s unqualified bodily ownership with the operator’s property rights, and it yields a new duty of backup provision for server operators.

Key insight

Two overlapping objects can sit in the same place at different depths. One is the heart, or the implemented BCI process. The other is the exact parcel of matter now composing the heart, or the server now running the process. Only the higher-level one is a part of the moral body. So whether an intervention touches “your body” depends on whether it changes high-level properties, not on whether it touches matter that is spatially inside you. In the paper’s metaphor (§2.5), there is a “floor” in “the mereological constitution of the body”, and “what happens below this floor does not really concern the body”.

Assumptions

Premises and authorities, since this is a philosophy paper.

• BCIs can be body parts for moral purposes. This is assumed (§2.1: “I will assume as much in this paper”), motivated by Harbisson’s antenna, Aas & Wasserman 2016 and Aas 2021.
• No part of a person may be owned by another entity (Quigley & Ayihongbe 2018; Bublitz 2024), treated as a “compelling principle” (§§1, 2.2, 2.3).
• Intransience requirement (§2.4): a body part must figure “in a lasting or irreplaceable way” in bodily processes (cf. Aas 2021; de Vignemont 2011; Heersmink 2015). It applies to organic parts too: individual molecules fail it.
• Bodily rights are unconditional, not qualified (§2.3): the paper rejects the qualified-rights strategy rather than refining it.
• Real definitions (Rosen 2015) fix an object’s identity and persistence conditions. An object’s depth equals the depth of the “primary constituents” of its real definition (§3, n. 3).
• Distinct coinciding objects: the heart and its matter, or the software and the server, may be “different objects, similar to how a statue may be a distinct object from the clay” (Thomson 1998; Paul 2010). This is hedged (“we might even say”, §2.5).
• Software objects exist as “a persisting, implemented computational process” (Marr 1982; Chalmers 2019), higher-level than the hardware (§2.5, §6).
• Microphysical vs macrophysical is the working construal of depth. Determinacy and fundamentality are “nearby versions that sometimes diverge” (§3, n. 2–4).
• The high-level threshold is context-sensitive, like “tall” (§3).

Key results

What the paper argues, by section.

• §2.3 (negative): Qualified bodily rights fail. A harm condition is too narrow: an unconsented pelvic exam under anaesthesia is a violation without harm (Friesen 2018). An interests or wishes condition is too broad: an interest in sole use of a server does not generate a right to it. Autonomy and disrespect conditions presuppose the answer to whether the object is a body part. Any qualification still leaves a third party owning part of a person.
• §3 Df. high-level neutrality: “an event e is high-level neutral with respect to a state or object just in case e does not change the high-level properties of that state or object”. It is gradable: an event is neutral “down to a certain minimum level”.
• §3 depth of objects: “objects that are too deep are not genuine body parts”. A water molecule (real definition in terms of atoms) is deeper than a heart (“the organ that circulates blood in an organism”).
• §4: Subjectivist and interest-based accounts of the body place it “above a certain metaphysical floor”. Functionalist and extended-mind accounts face a role-versus-realizer question that depth answers. The natures of hylomorphic objects (“x is an object constituted by p1…pn arranged F-wise”) require the pᵢ to be “somewhat high-level and not essentially maximally determinate” if they are to allow variable embodiment.
• §5 principles: Bodily interests are high-level and Bodily rights are high-level: both “concern events that are not high-level neutral with respect to their body”, and high-level neutral events “generally” do not bear on them or infringe them.
• §5 MRI objection: An unconsented MRI is wrong for reasons other than its high-level-neutral physical effect: disregard of autonomy over medical tests, moving the patient without consent, and extracting information that is not high-level neutral.
• §6 BCIs: The software object (defined by algorithms) is a body part. The server (defined by transistors and circuits) is not. Interference that leaves the software unaffected is not interference in the body, so maintenance and replacement are permissible. Interference that changes high-level properties is “generally impermissible”. Server operators acquire duties about risk, for example backup servers against fire or blackout, which the paper calls “an unprecedented species of moral obligation”, and it says new legal requirements are warranted. Transferring a process between servers is not necessarily high-level neutral, because it changes location. That issue is left open.

Claims

id	claim	strength	support
C1	Some objects composing a body (water molecules, the momentary matter of the heart) are not parts of the moral body, though they are parts in a “run-of-the-mill” sense	informal argument	§1, §2.4–2.5, §4. Intuitions (gravitational and electromagnetic effects, glucose metabolism) plus the intransience requirement
C2	The depth of an object is the depth (microphysical commitment) of the primary constituents of its real definition	stipulative definition	§3, n. 3. Rests on Rosen 2015. Alternatives by determinacy or fundamentality are acknowledged
C3	Objects that are too deep are not genuine body parts	informal argument	§3. Real definitions fix persistence, so deep objects are transient and fail the intransience requirement (§2.4)
C4	Qualified bodily rights cannot resolve the ownership/transience tension	informal argument	§2.3. Counterexamples to harm and interest conditions, and a circularity argument against autonomy and disrespect conditions
C5	Subjectivist, interest-based, functionalist and hylomorphic theories of the body accommodate or support depth	informal argument, uneven	§4. Strongest for subjectivist and Aas 2021 accounts. Biological views are said to be “not well positioned” for moral parthood. Hylomorphism is treated “briefly”
C6	Bodily interests and rights concern only events that are not high-level neutral	informal argument from cases	§5. Metal detector, phone radiation, MRI objection. The paper says the principles do not depend on the ontology of C2–C3
C7	A BCI user’s implemented software is a body part, and the third-party server that runs it is not	informal argument, conditional	§2.5, §6. Depends on the assumption that BCIs are body parts (§2.1) and on software being a distinct higher-level object
C8	Depth lets users keep unqualified bodily ownership while operators may maintain or replace servers without consent	informal argument	§6
C9	Server operators acquire novel duties (e.g. backup servers), and new legal requirements are warranted	assertion with brief argument	§6 (Almond; Zimmerman 2006). No analysis of the content or limits of these duties
C10	Moving a BCI process between servers is not necessarily high-level neutral, and its ethics is unsettled	stated open question	§6
C11	Whether brain tissue needs a lower “floor” (context sensitivity) is unsettled	stated open question	§5. Three options given (lower threshold, univocal notion, mental-integrity right per Douglas 2020)

Concepts

• breadth: the spatial extension of the body. Its parts occupy spatial regions.
• depth: “roughly … the level of abstraction of its parts” (abstract). Formally, for a description, the degree to which it is “committal about the microphysical properties”. For an object, the depth of the primary constituents of its real definition (§3).
• moral body: “the body as it figures in questions of bodily ownership, integrity, and autonomy” (§1).
• body part (moral sense): an object of suitable depth that meets the intransience requirement. The paper’s mereological claim concerns this relation, which it separates from ordinary parthood (§4 parenthesis).
• high-level neutrality: see Key results. It is gradable and context-sensitive.
• intransience requirement: a body part must figure lastingly or irreplaceably in bodily processes (§2.4).
• problem of ownership / problem of transience: the two obstacles to BCI embodiment (§2.2). §2.5 once calls the second the problem of “temporality”.
• software object: “a persisting, implemented computational process”, higher-level than the server that implements it (§2.5).

Connections

What mereological thesis it defends, and about which relation. The thesis is about parthood, specifically a restricted, morally loaded body-part relation, with depth as the ordering that restricts it. It is not a thesis about depth as a relation in its own right. The paper explicitly concedes a “run-of-the-mill conception of parthood” on which water molecules are parts of the body (§4). So it posits two relations: ordinary parthood, and moral-body parthood, which excludes everything below a context-sensitive “floor”. Depth itself is a gradable ordering (“deeper than”) between objects, often spatially coincident ones, fixed by their real definitions. The supporting ontology is coincidence (statue/clay) plus hylomorphic variable embodiment (Fine 1999; Johnston 2002). The paper does not address classical mereological principles. One consequence goes unremarked: the heart is a moral body part and a molecule is an ordinary part of the heart, yet the molecule is not a moral body part. So moral-body parthood cannot be the transitive closure of ordinary parthood. This is my observation, not the paper’s. It bears directly on LIT-124, whose summary stresses that beyond parthood being a partial order every further principle is contested. Register’s relation is at best a restriction of parthood, not a rival to it.

What the ethical conclusion rests on. The two principles (§5) are the operative premises, and the paper says they do not strictly depend on the real-definition ontology. The ontology is offered as explanation and justification, not as a premise. The BCI conclusion (§6) rests on: (i) the assumption that BCIs are body parts (§2.1); (ii) the no-ownership-of-persons principle; (iii) the intransience requirement; (iv) the rejection of qualified rights (§2.3); (v) the claim that software is a distinct, higher-level object; and (vi) case intuitions (metal detectors, water molecules, the pencil-sharpening case against Carter & Palermos 2016).

Links in this record.

• LIT-124 (SEP Mereology): see above. Register’s floor restricts a parthood relation rather than proposing a composition principle.
• LIT-142 (Meincke on pregnancy): a contrasting route to a “which things are parts” question. Meincke rejects both the Containment and Parthood views through a process ontology. Register keeps a substance-and-coincidence picture and restricts parthood by level. Both make body boundaries depend on something other than spatial containment.
• LIT-097 (The Extended Mind): Register’s extended-mind reading of the body (Carter & Palermos 2016) inherits the Clark–Chalmers framing, and his intransience requirement is the analogue of the extended mind’s reliability and availability conditions (he cites Heersmink 2015).
• LIT-147 (Pernu on neuroprosthetics): both turn on multiple realisation in BCIs. Pernu locates it at a cortical, physical variable. Register locates the body part at the software level above many possible servers.
• LIT-136 (Floridi, informational self): Floridi’s corporeal “membrane” and Register’s software body parts both let information-level objects count as constituents of a person.
• LIT-130 (Personal Identity and Ethics): a comparable metaphysics-to-ethics move for persons that Register makes for body parts. §5’s disclaimer (the ethics does not strictly depend on the ontology) bears on LIT-130’s question of priority between metaphysics and ethics.
• LIT-111 (Birch): the dossier’s pairing, that conversations are spread across data centres, is an analogy about location in hardware. Register does not discuss AI systems.
• LIT-049, LIT-167 and LIT-168 concern individuating biological wholes by level or organisation. They are adjacent, but Register cites none of their authors except Pradeu (2012).

Bearing on the record

It carries nothing for ML practice, and there is no Anthology document it should inform. The software/hardware depth distinction is suggestive for questions about where an AI system or a user’s cognitive tool is “located”, but the paper makes no claim about AI systems themselves, only about BCI software as part of a human body. Within nucleation it is a clean data point for the mereology/ethics interface: a proposal that a morally loaded restriction of parthood is fixed by level of abstraction, with a stated disclaimer that the ethics can stand without the ontology.

Limitations

• The central premise that BCIs are body parts is assumed, not argued (§2.1). The paper resolves a tension conditional on it.
• The abstract says depth is “crucial” to the moral significance of BCIs. The body shows that it gives a unified explanation, but concedes that the principles do not depend on the ontology (§5). So the ontology’s necessity is claimed more strongly in the headline than it is supported.
• The distinct-coinciding-objects step is hedged (“we might even say”, §2.5), and the known puzzles of material constitution are not addressed.
• The high-level threshold is deliberately vague and context-sensitive (§3). No criterion is offered for fixing it in a given case, which the paper admits for brain tissue (§5).
• Only one high-level property of software, location, is flagged as non-neutral under transfer (§6). There is no account of which software changes are neutral (n. 7 notes that some small changes are).
• The MRI reply moves the wrong to other features of the act. That preserves the principle, but it also means the principle alone does not classify the act.
• The novel duties (backups, legal requirements) are asserted with little analysis.
• The citations are inconsistent (see corrections). This does not affect the argument.

Open questions

• Is a context-sensitive floor needed (brain vs skin), or does a univocal notion plus mental-integrity rights (Douglas 2020) suffice? The paper says “a more specific and elaborate investigation is required” (§5). A worked brain-tissue case would settle it.
• Do people have a right to know or control where their BCI software runs, and what is the moral weight of moving it? This is left open in §6.
• How does moral-body parthood interact with ordinary parthood (the transitivity failure noted above), and can it be stated as a mereological relation at all? This is my question; the paper does not raise it.
• Does the view extend to software that is not integrated into a body, for example an extended cognitive tool that fails the intransience requirement? The pencil case suggests not, but a cloud service used intermittently falls between the cases.

Corrections to the seeded skim

• The dossier’s access line takes OpenAlex’s 2025-08-21 as “possibly an earlier online-first date, unverified”. The article page’s publication-details block gives 2025-08-21 as “Accepted on” and 2025-12-12 as the publication date. So 2025-08-21 is the acceptance date, not an online-first date.
• The dossier’s §1–2 bullet stops at the two problems (ownership, transience) and so misses where the argument actually turns. §2.3 argues that qualified bodily rights (conditioned on harm, interests, autonomy or disrespect) cannot resolve the tension, and that they would still violate the no-ownership-of-persons principle. §2.4 states the intransience requirement (“for something to count as a body part, it must figure in a lasting or irreplaceable way in someone’s bodily processes”). §2.5 introduces depth, the heart/blood-circulator/parcel example and the “metaphysical floor” in the body’s “mereological constitution”. The software/server resolution is first given in §2.5, not in §6.
• The dossier says the resolution “dissolves the ownership conflict”. The paper says (§6) that depth separates server ownership from bodily ownership but does not make every intervention on the server permissible. It also says that moving a BCI process between servers “is not necessarily high-level neutral” (it changes location), so backups mitigate the risk without removing it. The conflict is narrowed, not dissolved.
• The dossier’s §7 gloss says the view “fits existing (subjectivist, interest-based, functionalist, hylomorphist) theories”. §4 is more qualified than that. Biological functionalist views “are not well positioned to account for the morally relevant notion of body parthood” and only use depth implicitly. The hylomorphism link is “briefly” made. “Support” is strongest for the subjectivist and interest-based (Aas 2021) accounts.
• The dossier and NOTE-151 imply the ethics rests on the real-definition metaphysics (“backed by … real definitions à la Fine and Rosen”). The paper says (§5) the two principles “don’t depend on the ontological claim that body parts themselves have partly abstract, high-level real definitions”; that claim only offers “a simple and unified explanation”. The real-definition machinery is Rosen (2015). Fine (1999) enters only through variable embodiment and hylomorphism (n. 2, §4).
• Minor, and in the paper rather than the dossier: several in-text citations do not match the reference list. de Vignemont 2011 is cited in §2.2 and §4 but only de Vignemont 2024 is listed. The text has Almond 2004, the list 2005. The text has Brumberg et al. 2019, the list 2018. The text has “Pradeau 2012”, the list Pradeu.