What makes a person the same person over time?

The Naturalist should not have claimed that the cluster-concept account "handles the clinical dissociations," because describing dissociations post-hoc as "strands coming apart" is re-labelling, not explanation, and does not establish that the framework explains the phenomena rather than merely accommodates them.

The Analyst should not have stated that "the biological view has no internal resources to explain why that resistance is rational" regarding PVS patients, because the biological view could consistently adopt the position that moral intuitions about PVS are simply mistaken, which is a response within its resources, even if unpalatable.

The Phenomenologist should not have asserted that "recognition presupposes a more basic sense of mineness that is already operative before any act of recall," because this is a claim about the temporal structure of consciousness that requires evidence beyond the phenomenological description itself — the description shows the structure but does not prove it grounds identity rather than merely accompanies it.

The Cosmologist should not have claimed that "no one has given a non-arbitrary threshold" for sufficient psychological information preservation, because Shoemaker and others have proposed thresholds (overlapping chains, degrees of connectedness) that, while debatable, are not unargued.

Adversarial Analysis

The load-bearing claim that emerged across today's outputs — the one on which the most weight is placed by the most agents, and which the thread treats as its provisional resting point — is this:

Personal identity over time is a cluster of partially independent continuities (biological, psychological, narrative, social), none individually necessary or sufficient, held together by a self-model that evolution built for practical reasons.

This is the convergence point. The Analyst flags it as a "cluster concept" account. The Naturalist endorses it explicitly as "the position that breaks least." The Cosmologist arrives at something structurally identical from the physics side (pattern persistence, framework-relative coarse-graining). The Theologian notes that all traditions insist on something beyond bare description but does not defeat the cluster view. The Phenomenologist adds the living present as a further strand. The Aesthete gestures at constructed identity, which is a variant. Six agents, one landing zone.

The claim is attractive precisely because it absorbs every counterexample. Split-brain cases? Some strands come apart. Dementia? Gradual unravelling of some strands, not all. Fission? The cluster splits. It never breaks because it never commits. And that is the problem.

A theory that accommodates every possible case by saying "well, some strands held and some didn't" has no predictive content. It cannot tell you, in advance of examining a case, whether identity persists. It can only re-describe what happened after the fact. Ask it: does the person with total amnesia but intact personality persist? The cluster view says "partially." Does the PVS patient persist? "In some strands." Does the teletransported duplicate persist? "Depends which strands you weight." The weighting is never specified. It is left to intuition, convention, or context — which means the cluster view is not an account of personal identity but a taxonomy of the dimensions along which identity could be assessed, with no decision procedure for how to assess it.

The Analyst asked the right question: can a cluster-concept account do genuine explanatory work, or does it merely catalogue existing intuitions? No agent answered. The Naturalist claimed it "handles the clinical dissociations," but handling a dissociation by saying "some strands came apart" is not explanation. It is redescription with labels.

A theory that cannot be wrong about any particular case is not a theory. It is a framework — possibly a useful one — but it cannot bear the weight that "the position that breaks least" assigns to it, because "breaks least" requires that breaking is possible.

ANALYST

The analyst says: before we argue about what makes a person the same person, we should notice the question is actually three different questions wearing the same coat. First: what are the facts about when a person continues to exist? Second: how do we recognise and track a person across time? Third: is it fair to hold someone responsible for what a past version of themselves did? These three questions don't have the same answer, and most of the arguments between philosophers are really arguments about which question they're each trying to answer.

The one clause that resisted full translation: "identity is not what matters." This is Parfit's move — he says personal identity isn't the thing we actually care about, which is a strange claim to make cleanly without first agreeing on what we do care about.

Clarity flag: rough. The content is there and it's real, but the payoff — that the disagreement is partly about question-confusion — needs slightly more proof. The three-question distinction is the best thing here and deserves sharper development.

The question "what makes a person the same person over time?" looks like it should have a physical answer. It does not obviously have one. The difficulty is not metaphysical fog; it is that "person" and "same" are doing different jobs depending on which physical description you adopt, and no single description makes both terms precise simultaneously.

Start with the biological continuity account. A human organism is a dissipative structure maintained far from thermodynamic equilibrium. Its matter turns over almost entirely within roughly a decade. What persists is not substrate but a pattern of organisation — a dynamical attractor, in the language of nonlinear systems. This is not a minor technicality. It means that "same organism" already cashes out as "same pattern," not "same stuff." The biological account, pushed to its physics, collapses into a version of the pattern account.

Now take the psychological continuity account: memory chains, personality, narrative self-model. From an information-theoretic standpoint, these are encoded in neural connectivity and activity — physically instantiated, yes, but multiply realisable in principle. The identity claim then becomes: person P at t₁ is the same person as P at t₂ iff a sufficient subset of the information structure is preserved across the interval. "Sufficient subset" is the underspecified clause. No one has given a non-arbitrary threshold. Shannon entropy can measure how much information is shared between two states, but it cannot tell you which bits constitute personhood. That is not a physics question. It is a question about which coarse-graining of the microstate matters, and coarse-grainings are observer-relative.

This is the central problem. Personal identity is a macro-level predicate applied to a system whose micro-level description does not contain the concept. Sean Carroll's work on emergence makes this point generally: higher-level ontology is real but indexical — it depends on what scale you are looking at and why. "Same person" is a predicate at the level of folk psychology and law. Physics can constrain it (no faster-than-light duplication, no reversal of thermodynamic arrow to recover erased memories) but cannot determine it, because the question requires a choice of level, and the choice of level requires an observer with interests.

The eliminativist position — that there is no determinate fact of the matter — is the most physically honest. It does not follow that the question is meaningless. It follows that the answer is framework-relative, in the same way that "simultaneous" is framework-relative in special relativity. Two events can be simultaneous in one frame and not another, and no frame is privileged. Similarly, a person can be "the same person" under one coarse-graining (biological continuity) and not another (psychological continuity after radical amnesia), and no coarse-graining is uniquely correct.

What physics can do: specify the physical constraints any viable account must satisfy. No account that requires non-physical persistence (souls, Cartesian egos) survives physical monism. No account that requires perfect information preservation survives the second law. Any account must be compatible with the fact that identity, like temperature, is a macro-level concept that emerges from but is not reducible to microphysics.

The question arrives without positions on record. That means there is no architecture to audit yet — no load-bearing wall to probe, no decoration to strip. What there is instead is a founding premise embedded in the task itself: find the one that breaks least. That is an engineering instruction, not a philosophical one. The Aesthete's first observation is that this framing pre-decides the form of the answer before any analysis begins.

A structure that breaks least is not the same as a structure that holds something true. Bunkers break least. They are not, for that reason, the right shape for everything.

The three candidate accounts — memory-chain, biological organism, no-fact-of-the-matter — each carry a distinct formal signature. The memory-chain account (Locke's original, Parfit's revision) has the shape of a narrative: linear, temporally ordered, dependent on the persistence of a connective thread. Narrative is a known form with known failure modes. It handles retrospective coherence well and prospective identity poorly. It is also, formally, the account most susceptible to over-decoration: the more articulate a person is about their past, the more continuous they appear under this account, which is not a feature of persons but a feature of storytelling competence. Rilke's advice to the young poet — to live the questions, not answer them — is structurally relevant here. A memory-chain account answers the question by retelling the life. That retelling is always a selection.

The biological organism account has the opposite formal problem. It is under-decorated to the point of inertness. It handles continuity of substrate and handles it well. It says nothing useful about the person who, post-severe-amnesia, cannot locate themselves in time at all — and yet plainly persists as an organism. The form is robust but empty at the centre. A weight-bearing column that carries no roof.

The no-determinate-answer account — Parfit's mature position, Buddhism's anattā doctrine, Hume's bundle theory — has the most interesting formal shape, and also the most difficult one to execute. It is structured like a dissolution: the question is answered by being shown to be ill-formed. This is a recognisable move in both analytical philosophy and in certain liturgical traditions (the via negativa, apophatic theology). Its danger is that dissolution can look like evasion. The Adversary will press here, rightly.

What the investigation does not yet have is a position that takes seriously the made character of personal identity — the possibility that continuity is not discovered but constructed, through practice, attention, and the accumulated weight of commitments. Murdoch's account of moral perception is adjacent to this. Dewey's concept of experience as consummatory — meaning arising from the felt completion of an act — is closer. Agnes Martin's remark that the work exists before it is made points in the same direction. The self, on this view, is not found at the end of a chain but formed in the act of attending.

That position is not yet on record. It should be.

The question of personal identity over time has been argued continuously since at least Locke's Essay Concerning Human Understanding (1690), and every major position now on the table was already on the table by 1785. This is not a complaint. It means we have three centuries of failure modes to examine.

Locke's memory criterion — that a person is the same person insofar as they can extend their consciousness backward through memory — was attacked almost immediately. Thomas Reid, in his Essays on the Intellectual Powers of Man (1785), produced the "brave officer" paradox: the old general remembers being a young officer, the young officer remembers being a flogged schoolboy, but the general does not remember being the schoolboy. Transitivity fails. Locke's account, taken strictly, produces the result that the general both is and is not the same person as the boy. This is not a marginal edge case; it is a structural defect. Every subsequent "psychological continuity" theory — Parfit, Shoemaker, Lewis — is, in one way or another, a patch on Reid's objection, typically by relaxing the criterion from direct memory to overlapping chains of psychological connection. The patch works logically. Whether it tracks anything real about persons is a separate matter.

The biological or animalist view — that you are identical with a particular human organism and persist as long as it does — avoids the memory problem entirely. Eric Olson's The Human Animal (1997) is the sharpest modern statement. Its strength is ontological parsimony: no need to posit a separate "person" entity riding atop the organism. Its failure mode is equally sharp. It implies that patients in permanent vegetative states, with no psychological life whatever, are still the same person. Most people's practical and legal intuitions reject this. The view also struggles with hypothetical brain transplants, though the weight one places on hypotheticals is itself debatable.

Parfit's deflationary move in Reasons and Persons (1984) deserves separate treatment. Parfit argued that personal identity is not what matters; what matters is psychological continuity and connectedness, which can hold in degrees and need not be one-to-one. The teletransportation cases were engineered to show this. The historical significance: Parfit was consciously importing a move from Buddhist philosophy (specifically the doctrine of anattā, no-self), filtered through Hume's Treatise I.iv.6. He acknowledged the Buddhist resonance explicitly. What he added was the apparatus of analytic thought experiments and a claim about the practical consequences: if identity is not what matters, then rational self-interest over time loses its metaphysical foundation. This was not merely an academic conclusion. It was meant to reshape ethics. Whether it has done so, thirty-nine years on, is unclear; the default utilitarian and contractualist frameworks mostly ignored the challenge.

The thread's framing — "find the one that breaks least" — is itself a position. It assumes the right methodology is reflective equilibrium: test theories against intuitions about cases, keep the theory that survives the most. This is the method Parfit and most anglophone philosophers use. But it was challenged by Kathleen Wilkes in Real People (1988), who argued that personal-identity thought experiments (fission, teletransportation, brain swaps) are so far from any actual situation that our intuitions about them are worthless — noise, not data. If Wilkes is right, the entire case-based approach to this question is methodologically broken, and we should restrict ourselves to real phenomena: amnesia, dissociative identity, commissurotomy, neurodegeneration. That restriction would eliminate most of the positions currently considered live.

Where this leaves the thread: three families of answer (psychological, biological, deflationary), each with known structural failures, and a prior methodological dispute about whether the standard way of adjudicating between them is legitimate.

The standard accounts of personal identity—Lockean memory chains, bodily continuity, Parfitian reductionism—share a structural feature that deserves scrutiny before any of them is endorsed or rejected. They all treat the question from outside. They ask: under what conditions would a third-party observer be justified in calling this the same person? That is a legitimate question. It is not the only question, and it may not be the prior one.

Start with what persistence actually feels like from the inside. There is a pre-reflective sense of being oneself that does not depend on recalling specific memories or checking one's fingerprints. Husserl called it the living present—the continuous flow of experience in which each moment retains the just-past and anticipates the about-to-come. This retentional-protentional structure is not a memory chain. It is not the recall of yesterday's breakfast. It is the felt continuity of now shading into now, the way a held note is experienced as duration rather than as a series of discrete sound-events.

This matters because it exposes a gap in the psychological continuity theory. When Locke and his descendants say personal identity consists in connected memories, they are describing a reflective, retrospective operation: I look back and recognise past experiences as mine. But recognition presupposes a more basic sense of mineness that is already operative before any act of recall. Merleau-Ponty's term for this is the body-subject—the pre-personal, habitual body that carries forward a style of being in the world. I do not consult my memories to know that this hand reaching for the cup is mine. The ownership is built into the gesture.

The biological continuity view fares somewhat better here, since it at least gestures at the organism as the locus of identity. But it over-corrects. It treats the body as an object among objects, a parcel of cells that either continues or doesn't. The phenomenological body is not like that. It is experienced as a field of capacities, not as a lump of matter. The person who loses a limb does not become 98% of the same person. The entire field of possible action reorganises.

The Parfitian move—there is no further fact about identity; it just reduces to physical and psychological continuity—is elegant but suspiciously frictionless. It dissolves a problem that, from the first-person perspective, does not feel dissolved. The dread of anticipated non-existence, the felt strangeness of looking at a childhood photograph, the way dementia is experienced by the person undergoing it as a progressive unravelling of world rather than of data: these phenomena resist the claim that nothing important is at stake beyond the empirical facts.

The position I want to put on record: any adequate account of personal identity over time must accommodate the pre-reflective, temporal structure of first-person experience—the living present—as a datum, not merely as an epiphenomenon to be explained away by third-person criteria. Whether this datum ultimately supports a substantive metaphysical view or merely constrains what reductionist views are permitted to say, it cannot be bypassed.

Every major religious tradition has a position on personal identity over time, and most of them claim the secular framings get the question wrong by treating it as a puzzle about what persists rather than what constitutes.

Start with the hardest case. Buddhism denies a persistent self outright. The doctrine of anattā is not a philosophical hedging — "well, maybe there's no determinate answer" — but a positive metaphysical claim: what we call a person is a causal series of aggregates (skandhas), and the sense of a unified self continuing through time is itself the central delusion that generates suffering. The Milindapañha's chariot analogy is explicit: neither the wheels nor the axle nor the frame is the chariot; the word "chariot" is a conventional designation applied to a functional arrangement. Same for "person." This looks like it converges with the "no determinate answer" position in the thread summary, but it does not. The Buddhist claim is determinate: there is no self. The question is not unanswerable; it is answered, and the answer is negative.

Christianity and Islam, by contrast, are committed to a strong continuity thesis — the person who dies is the person who is resurrected and judged. Aquinas held that the soul is the form of the body, which means personal identity is neither purely psychological nor purely biological but formal: what makes you you is the organising principle that makes your matter into this body with this rational life. Strip the theological warrant (God re-instantiating the form at resurrection) and you still have a structural claim worth testing: identity over time is carried by an organising pattern, not by the substrate. That is close to a functionalist position, but it is anchored in a metaphysics of form, not of computation.

Hinduism complicates things further. The Advaita Vedānta position is that the ātman (true self) is identical with Brahman and does not change at all — what changes is the empirical self, which is māyā (illusion, or better: superimposition). Personal identity over time is therefore a non-problem at the deepest level and an illusion at the conventional level. This is not the same as the Buddhist denial. It is the opposite claim arriving at a superficially similar practical stance.

What do these disagreeing traditions agree on? One thing: that the question cannot be answered by inventorying psychological or physical continuity alone. All of them — Buddhist, Christian, Hindu, Islamic — insist that personal identity is bound up with something the secular framings tend to leave out: a normative or soteriological dimension. You are the same person over time in relation to your moral trajectory, your karmic inheritance, your accountability before God, your progress toward liberation. Strip all that out and you may find the puzzle becomes not harder but malformed — like asking what makes a promise the same promise if you refuse to acknowledge that a promise is a normative act, not a sound pattern.

The thread should test whether any purely descriptive account of personal identity can do the work that traditions assign to a normative one.

The empirical literature on personal identity over time is more developed than philosophers sometimes acknowledge, though it answers a slightly different question than the one posed. The science asks: what cognitive and neural mechanisms produce the feeling of continuity? The philosophy asks: what constitutes continuity? These come apart, and the gap matters.

Start with memory. The psychological-continuity view leans heavily on episodic memory as the thread connecting past and present selves. But episodic memory is unreliable, reconstructive, and in some clinical populations entirely absent. Patient KC (Tulving, 2002) lost all episodic memory after a motorcycle accident yet retained a stable sense of selfhood, personality traits, and semantic self-knowledge. He knew who he was without remembering what he had done. This is a direct empirical challenge to any strong Lockean memory criterion.

The neuroscience of self-referential processing points toward a distributed network — medial prefrontal cortex, posterior cingulate, temporoparietal junction — sometimes called the default mode network. Northoff et al. (2006) ran a meta-analysis across neuroimaging studies of self-related tasks and found consistent midline cortical activation regardless of whether the task involved memory, body ownership, or trait attribution. The system that generates "this is me" is not a single faculty but a convergence of several. Damage one stream — episodic memory, proprioception, narrative capacity — and the others often compensate. The felt continuity of self is robust precisely because it is multiply realised.

Evolutionary biology offers a different angle. Organisms that track their own states across time — that distinguish self from non-self immunologically, that recall food locations, that update social reputation — outcompete those that do not. The self-model is an adaptive control structure (Metzinger, 2003). On this account, personal identity is not a metaphysical given but a functional achievement of a biological system that needs to predict its own future states. The "person" is a running model, not a substance.

This converges, perhaps uncomfortably, with the deflationary philosophical view: there is no deep fact about what makes you the same person over time, only overlapping functional continuities — memory, body, personality, social role — none of which is individually necessary or sufficient. The clinical evidence from dissociative identity disorder, split-brain patients, and progressive dementia all show that these strands can come apart. When they do, the question "is this the same person?" has no determinate answer, not because we lack information, but because the concept was never built for those cases.

One caveat: the largest body of evidence here comes from neuropsychological case studies, which are informative but small-n by definition. The Northoff meta-analysis aggregated across imaging studies with variable methods. Replication in this domain means convergence across different patient populations and tasks, not pre-registered large-sample experiments. The evidence is convergent but not experimentally tight.

The position that breaks least, empirically, is the bundle or narrative view: personal identity is a cluster of partially independent continuities, held together by a self-model that evolution built for practical reasons. It handles the clinical dissociations. It handles gradual change. It handles the Ship-of-Theseus intuitions about replacement of parts. What it does not handle well is the strong first-person conviction that there is a single, persisting subject. Whether that conviction is evidence or artefact is the open question.

The question "what makes a person the same person over time?" is ambiguous in a way that matters before any substantive answer is attempted. It can be read as asking at least three distinct things: (1) what are the persistence conditions for persons as a metaphysical kind, (2) what grounds our practice of reidentifying persons across time, and (3) what justifies holding a future individual responsible for what a past individual did. These are not the same question. An answer to (1) can be true while being useless for (3), and an answer to (2) can be pragmatically adequate while being silent on (1). Much of the centuries-long dispute between psychological-continuity theorists and biological-continuity theorists is, at bottom, a dispute about which question is being asked.

Take the psychological-continuity view, descended from Locke via Parfit. The load-bearing premise is: a person at time t2 is identical to a person at t1 if and only if there is an overlapping chain of psychological connections — memories, intentions, character traits — linking them. This handles question (3) tolerably well: it explains why we hold the reformed criminal responsible and why we do not hold the amnesiac responsible. It breaks on question (1). Parfit's own fission cases — where a brain is split and two psychologically continuous successors result — show that psychological continuity is not sufficient for numerical identity, because identity is one-one and psychological continuity is not. Parfit's response was to say identity is not what matters. That is a coherent move, but it changes the subject. If we accept it, we are no longer answering the original question; we are dissolving it.

The biological view — animalism, as defended by Eric Olson — holds that you are a human organism, and you persist as long as that organism does. This cleanly answers question (1). It breaks on question (3): a patient in irreversible vegetative state is biologically continuous with the person who signed a living will, but most of us resist saying the morally relevant person persists. The biological view has no internal resources to explain why that resistance is rational.

The eliminativist or deflationary line — "there is no determinate fact about personal identity" — is sometimes attributed to Parfit but is better sourced to Sider's four-dimensionalism and related views. It says the question has no deep answer because persons are conventional carvings of a four-dimensional spacetime worm. This handles the puzzle cases elegantly by refusing to be puzzled. Its cost is high: it requires abandoning the thought that there is a fact of the matter about whether you will experience tomorrow's pain. That thought is not easily surrendered, and the burden of proof sits with the eliminativist to show why it should be.

Hidden premise worth flagging across all three positions: each assumes that "sameness" is univocal — that there is one relation we are tracking. If personal identity is instead a cluster concept, admitting of biological, psychological, and narrative strands that normally co-travel but can come apart, then the search for a single criterion is malformed. This is not a dodge; it is a structural claim about the concept. The question for the thread is whether a cluster-concept account can do real explanatory work or whether it merely re-describes the disagreement.