Derek Parfit (1971), “Personal Identity”, The Philosophical Review 80(1), January 1971, pp. 3–27 — DOI-10.2307/2184309
On the citation. The brief’s details are all right: author, title,
journal, year, volume 80, issue 1, pages 3–27. Crossref gives only the first
page and the month; the JSTOR cover page gives the range. published: is
the issue month, since no source gives a day.
Key takeaways
- Two targets. The belief that in every describable case “the question about identity must have an answer”, and the belief that “unless the question about identity has an answer, we cannot answer certain important questions” (survival, memory, responsibility) (pp. 3–4).
- Division breaks the tie between identity and what matters. If each half of my brain is housed in a new body, “I do not survive” is absurd (“How could a double success be a failure?”), and “I survive as one” is arbitrary. “The relation of the original person to each of the resulting people contains all that interests us—all that matters—in any ordinary case of survival” (p. 10).
- Identity is one-one and all-or-nothing; what matters need not be either. Division shows it need not be one-one; fusion and gradual change show “most of the relations which matter in survival are, in fact, relations of degree” (p. 11).
- A criterion survives the objection. “X and Y are the same person if they are psychologically continuous and there is no person who is contemporary with either and psychologically continuous with the other” (p. 13), offered as sufficient, not necessary.
- The relations can be described without identity. Q-memory and q-intention are memory and intention without the condition that the experience or action was one’s own (pp. 14–17).
- Connectedness versus continuity. Connectedness (direct relations) is not transitive and comes in degrees; continuity (overlapping chains) is transitive (p. 20). Speaking of “past selves” and “future selves” by degree of connectedness, “there is no underlying person who we both are” (p. 25).
- The principle of self-interest has no force of its own (p. 26). Its apparent force comes from caring what happens to one, or from impartiality; “all the parts of his future are equally parts of his future” is “a truth too superficial to bear the weight of the argument”.
Standing in the record
Filed on 2026-10-02 at the owner’s request, to fill out the record’s coverage of personal identity. NOTE-378 is the reading, and it placed the work: Active. It is the paper the record’s identity-and-ethics survey means by “Parfit’s fission case”, and the source of the claim the record leans on when it asks whether an AI interlocutor persists. Read it for the division argument of §I and the q-memory of §III. Sections IV–V are a proposal about how we could talk, offered as such. The self-interest conclusion of §VI is drawn in two pages.
No anthology topic holds it, it carries no instruction for machine-learning practice, and the anthology does not hold it.
Where it meets the record, set out in the NOTE:
- The survey (LIT-130, NOTE-166) reports the fission argument as the route by which “what matters” is detached from identity (its C1), and gives the psychological criterion as unique psychological continuity. Both are here, in their first form; the survey cites the later works.
- Birch’s persisting-interlocutor argument (LIT-111, NOTE-183) denies that a chatbot “friend” persists because conversation history is not psychological continuity. The deciding condition is this paper’s clause (2) of q-memory: a q-memory must depend on the experience “in the same way (whatever that is)” as a memory does. A transcript handed to a successor does not. Birch argues from Parfit’s later “Relation R”; the condition is already here.
- Goldstein and Lederman (LIT-206, NOTE-138) make the per-conversation instance, not the model, the bearer of beliefs and desires. In this paper’s terms, instances drawn from one model are a tree of descendant selves with a shared ancestral self and no connectedness between branches. The paper’s point that each branch is “as good as survival” while none is the original is the pattern their instance view needs, and does not argue for. The pairing is mine.
- Dennett’s “Where Am I?” (LIT-405, NOTE-338). Yorick and Hubert in step are Parfit’s divided mind before reunion; their drift is division without reunion, and Dennett’s “two persons” is Parfit’s verdict that past the borderline “it would become intolerable to deny that they were different people” (p. 7).
- The narrative self (LIT-442, NOTE-364). Dennett’s two selves in one body when the story coheres around two points (its C5) and Parfit’s successive selves within one life are both selves individuated more finely than persons. Parfit’s are fixed by degree of connectedness, Dennett’s by the coherence of an interpretation.
- Frankfurt and Velleman (LIT-293, LIT-292, NOTE-268). NOTE-268 records Velleman reading Frankfurt’s late view as Parfitian in counting the persistence of valued traits toward survival. That strand is in this paper as “lasting q-characteristic” among the direct relations (p. 20), though the weighting by value comes later.
The other works filed beside it on 2026-10-02 frame it. Williams’s “The Self and the Future” (1970) is its footnote 11, the case that the answer must be all-or-nothing and carries a “risk”. Lewis’s “Survival and Identity” accepts its account of what matters and denies its conclusion that identity is not it. Reasons and Persons (1984), Part III, is its long version. Locke’s chapter is where Parfit says division is “implicit”. Schechtman’s narrative account and the animalist answer, being filed in parallel by another agent, are the two main replies to its psychological criterion. Shanahan’s work on role-play and the multiplicity of simulacra, filed by a third agent, is where the paper’s branching selves meet language models.