NOTE-996: Nichols & Bruno — Intuitions about personal identity

Contribution

The paper turns Williams’s (1970) claim that personal-identity intuitions depend on how a case is told into survey data. It shows that the divergence Williams described can be produced in ordinary participants, that his own explanation of it (first against third person) does not account for it, and that abstract questions and a forced reflective choice come down on the psychological side. It also states a method: when intuitions conflict, find which are robust across frames and weigh those more.

Key insight

The conflict among personal-identity intuitions is a framing effect, and framing effects can be studied instead of only lamented. Which frame to trust is not fixed by the frame type (abstract is better for the Linda problem, concrete for grammaticality), so the authors use convergence and reflective preference as the tie-breaker. Everything they conclude about persistence itself is conditional on intuitions being the right evidence.

Assumptions

• Intuitions bear on the theory. Stated as a condition, not defended: “Insofar as our philosophical theory of personal identity should be based on our intuitions” (abstract, p. 293); “We have not attempted to defend the role of intuitions” (p. 306).
• More robust means more weight. “If intuitions are supposed to guide theory building, then, ceteris paribus, intuitions that are more robust ought to be given more weight” (p. 307).
• A liberal “psychological theory”. Any view on which “persistence or continuity of distinctive psychological factors is necessary for persistence of self” counts (pp. 298–299; n. 12 notes the usual usage is necessary and sufficient). Only necessity is tested; sufficiency was dropped because pilot participants found Williams’s Lockean case “very difficult … to comprehend” (p. 298; n. 11).
• “Self” and “person” are used interchangeably. “Persistence of self” and “personal identity” name one question throughout. A person is “a more-or-less autonomous agent that thinks, acts on the basis of reasons, and is subject of perceptual experiences” (p. 293).
• The population. Introductory philosophy students at the University of Arizona, before personal identity was taught, in class (pp. 300, 304, 306; n. 18). The authors call it “homogenous on several important factors, including age, culture, and socioeconomic status” (p. 307).
• Yes/no and free-response measures. No scales in the Pain and abstract studies; a 0–9 scale in the Blok replications (nn. 14–15).

Key results

• The two frames (§1, pp. 294–295). The authors label Williams’s third-person body exchange “the Lockean frame” and his first-person torture series “the Pain frame”. Williams’s two explanations of the divergence are set out (§3, pp. 297–298): the Lockean case is “the product of the will of the experimenter” (Williams p. 179), and the mentalistic verdict came from a third-person argument, the bodily one from a first-person one.
• Lockean side, replicating Blok, Newman & Rips (2005) (§4, pp. 299–300). Jim’s brain is placed in a stock body and his original body destroyed; in one version the memories are intact, in the other the scan finds “no memories”. Third person: agreement that it is “still Jim” is higher with memories, t(28) = 3.17, p < 0.01; means 5.45 against 3.24 on 0–9 (n. 14). First person, the names replaced by “I”: t(42) = 3.33, reported p = 0.01; means 5.95 against 4.09 (n. 15). No first- against third-person difference, t(70) = 0.69 (memory) and 1.09 (no memory). The design was within-subject, both versions to each participant (n. 16). The authors note their means are “generally closer to midline” than Blok et al.’s.
• Against the experimenter’s-will explanation (p. 300). In the Blok case “all that is specified is that the memories are erased. There is no hint that the memories are recreated somewhere else”, and the psychological answer still comes.
• Pain frame, first person (§5, pp. 300–301). “Helpful doctors” treat a lethal brain infection with a procedure that “will permanently eliminate your distinctive mental states (including your thoughts, memories, and personality traits)”; days later they “administer a series of painful shots”. Four yes/no items, the first a comprehension screen, the fourth the target: “When the doctors administer the series of shots, you will feel the pain.” 68% agree, χ²(1, N = 50) = 6.48, p = 0.01; among those who passed the screen and gave adequate explanations, 75%, χ²(1, N = 28) = 7.000, p = 0.0082. One adequate explanation from an agreeing participant: “the doctors stated that you will lose your thoughts, memories, and personal thoughts, not your sense of feeling” (p. 301).
• Pain frame, third person (pp. 302–303). The same with “Jerry”: 72% agree, χ²(1, N = 54) = 10.67, p < 0.01; comprehending subgroup 72%, χ²(1, N = 36) = 7.111. No difference from the first person, χ²(1, N = 64) = 0.062, p = 0.80 (Fig. 1). So the anti-psychological response “is not restricted, as Williams suggested, to the first-person rendering” (p. 302).
• The authors’ diagnosis (§6, pp. 303–304). The demand is in the Pain frame: “if I am not going to feel it then who is? Similarly for the third-person version, if Jerry isn’t going to feel the pain, then who?” This is offered as a hypothesis, not tested directly. Note 20 adds Emily Esch’s alternative, that pain’s vividness drives it and “a less vivid mental state, like itch, might generate different responses”.
• Abstract question (§6, pp. 304–305). Free response to “what is required for some person in the future to be the same person as you?”, with a request for plain words added after pilots “often triggered religious or spiritual responses” (n. 21). Over 70% named psychological factors, χ²(1, N = 53) = 9.98, p < 0.01 (by my recomputation, 38 of 53). To the statement “In order for some person in the future to be you, that person doesn’t need to have any of your memories”, more than 80% disagreed, χ²(1, N = 54) = 21.41, p < 0.001 (44 of 54 by recomputation). Against the Pain frame, χ²(1, N = 88) = 25.018, p < 0.001 (Fig. 2). The cell counts behind that N are not given, and I could not reconstruct them from the reported samples.
• Abstract is not always better (pp. 305–306). The Linda conjunction fallacy is a case where the abstract answer is right; grammaticality judgements (“Before his band broke up, Milo went to college”) a case where the concrete one is.
• Reflective choice (§7, p. 306). Each participant got both the Pain case and the abstract question, order counterbalanced, then: “it wouldn’t really be consistent to say both that you would feel the pain of the shots and also that in order for a person to be you, that person must have some of your memories. In light of this, which one are you more inclined to agree with?” 64% chose memory, χ²(1, N = 45) = 3.76, p = 0.053, reported as “greater than what one would predict by chance alone” (Fig. 3). A “sizable minority” of 36% did not (n. 23). In a comparable condition on free will, Nichols and Knobe (2007) found a roughly 50/50 split (n. 24).
• Conclusion (§8, pp. 306–308). Psychological intuitions “are resilient across significant changes in the cases” and converge with abstract judgement; Williams’s anti-psychological intuitions “don’t seem to stand to scrutiny when examined in a broader and aggregative context”. “We think all of this points in favor of resolving Williams’ intuitive conflict in favor of a psychological view of personal identity (assuming that intuitions should inform our theories of personal identity).” And: “Obviously we don’t think that in general the way to solve philosophical problems is to ask undergraduates.”

Claims

id	claim	strength	support
C1	In a brain-transplant vignette, participants judge the person persists when memories are kept and not when they are erased	moderate	Blok et al. replicated, t(28) = 3.17 and t(42) = 3.33, within-subject, means near the scale midpoint
C2	Most participants say a person whose distinctive mental states are eliminated will feel later pain	moderate	68% (N = 50), 72% (N = 54); 75% and 72% among comprehenders
C3	Neither result depends on first- against third-person telling	moderate as a null	t(70) = 0.69, 1.09; χ²(1, N = 64) = 0.062; small samples, no equivalence test
C4	Williams’s experimenter’s-will explanation of the Lockean verdict fails	weak to moderate	the Blok case has no recipient of the memories, yet memory still matters; his explanation concerned a full exchange, which was not tested
C5	Abstract questions elicit psychological criteria	moderate	over 70% free response (N = 53); over 80% reject persistence without memories (N = 54)
C6	Under reflection, people prefer the psychological judgement	weak	64% of 45, p = 0.053, with a forced choice whose wording names memory and frames the conflict
C7	The Pain frame’s verdict comes from a thought experimenter demand	weak	the authors’ hypothesis; no manipulation of it; alternatives (vividness of pain, n. 20; imaginative flexibility of “I”, n. 19) are not excluded
C8	If intuitions should guide the theory, the more robust psychological intuitions favour a psychological criterion (necessity only)	conditional, as argued	§8; the antecedent is not defended

Concepts

• Lockean frame / Pain frame — the authors’ names for Williams’s third-person body exchange and his first-person torture series.
• Psychological theory (liberal) — any view on which continuity of distinctive psychology is necessary for persistence (n. 12).
• Thought experimenter demand — pressure a case’s construction puts on the answer, by analogy with experimenter demand (p. 303).
• Robustness — persistence of an intuition across changes of frame, used as its weight in theory-building.
• Reflective equilibrium (folk) — the authors’ label for asking people to choose between two of their own conflicting judgements.

Connections

The paper sits in the program of Nichols and Knobe (2007) on free will, using framing (Tversky and Kahneman 1981) and the abstract/concrete contrast (Sinnott-Armstrong 2008). Its Lockean data come from Blok, Newman and Rips (2005) and Rips, Blok and Newman (2006). Bartels and Rips, and Bartels and Urminsky, are cited for behavioural evidence that expected psychological change lowers willingness to wait for a later reward (n. 17). Nichols (2008), “Imagination and the I”, had explained the Pain verdict by the flexibility of “I” in imagination and had assumed the first-person telling drove it; this paper’s third-person result undercuts that assumption, as note 19 concedes. On the philosophical side it answers Rovane’s and Sider’s pessimism about the method of cases (§3, p. 297). None of these was read here except Williams.

Bearing on the record

• The rubber hand THEORYs: no bearing. THEORY-766 (the report difference), THEORY-769 (ownership as graded common-cause belief), THEORY-767 (interoceptive signals) and THEORY-768 (minimal selfhood as whole-body identification) are about the felt ownership and location of a body now. This study is about judgements of a person’s persistence over time, with no bodily stimulation and no ownership or location measure. It does not meet, partly meet or bear against any of their promote_when or refutation conditions. Nor does it touch THEORY-458 (ownership against agency). The only common ground is method: the authors’ “thought experimenter demand” is the vignette counterpart of the expectation problem THEORY-766 names, and here too it is hypothesised and not controlled.
• LIT-462, NOTE-387, THEORY-531 (Williams). THEORY-531’s claim is about what fear can rationally follow, and its promote_when says “Reports of what people would in fact fear cannot settle it”. This study collects reports of a different thing, whether one “will feel the pain”, and asks nothing about fear, so it neither meets nor refutes THEORY-531. Its vignette is also not Williams’s series: total elimination of distinctive psychology in one step, no torturer, no copied “memories”, no second person. What it does test are two empirical remarks in Williams (pp. 179–180). First, that the mentalistic verdict comes from a third-person telling and the bodily one from a first-person telling: these participants give the same answers in either person (C3). Second, that the swap verdict is produced by the experimenter’s choice of where to stop: memory still matters with no recipient (C4, weakly, since a full exchange was not tested). THEORY-531’s title says the two tellings “support opposite criteria”. As a fact about responses, the opposition is reproduced here, but it follows the case, not the grammatical person. A proposal only: THEORY-531’s “What this does not say” could add that the link between person of telling and verdict is Williams’s observation about his own two arguments, and that among these participants it does not hold.
• The record’s body/psychology split (THEORY-546, LIT-448, NOTE-374 §§3–7). The study sets a psychological criterion against a non-psychological one, and lumps bodily, organismic and soul views together on the latter side (pp. 296–297). Its two concrete cases do not separate brain from body from bare sentience. In the Blok case the brain survives in a new body and the original body is destroyed; erase the memories and participants say it is not Jim. In the Pain case the same brain (cured: the procedure “works exactly as expected”) and body survive with distinctive psychology gone; participants say “you” will feel the pain. So same brain, different verdicts. The difference between the cases is whether the original body survives and whether the question is identity (“still Jim”) or feeling (“will feel the pain”). The paper does not discuss this; the observation is mine. What it suggests is that the Pain item may be answered as a question about whether a subject of experience continues in that body, which a psychological theorist who counts bare sentience as psychology need not deny. One participant’s explanation says so outright (“not your sense of feeling”). The study therefore bears on the split as a matter of folk judgement, and favours psychological necessity in the abstract, but the concrete case that is supposed to oppose it does not cleanly test the bodily view.
• THEORY-417, NOTE-374 and TERM-045: which question. By Olson’s map the study asks the persistence question, but its abstract probe (“what makes a person the same person”) invites the characterization reading that THEORY-417 separates from it, and several free answers quoted (“that you have the same beliefs, values, and memories”) would fit either. The Pain item reads more naturally in TERM-045’s sense 1 (a subject that feels) than in sense 3 (persisting); the Blok item asks sense 3 directly. The paper’s “persistence of self” runs these together, which is the conflation ADR-024 exists to prevent; in this record’s terms the work is about personhood, with one item that may be about the self.
• Can folk judgement settle the metaphysical question? Not by itself, and the authors do not claim it does: their conclusion is conditional on intuitions informing the theory, and they say it is “hardly a decisive finale”. NOTE-374 records Olson’s view (§2) that the persistence question is not about the meaning of words and so not answerable a priori; and lists “thought experiments are evidence” among the debate’s assumptions. On that view, a survey establishes how a population applies a concept under various frames, and reaches the persistence condition only through a further premise that the concept is reliable about the fact. What it can do without that premise is narrower and still useful: show that an intuition philosophers cite is or is not widely shared (Williams’s Pain verdict is: about 70%), and test a proposed explanation of an intuition (Williams’s first/third-person explanation fails). That is the distinction QUESTION-048 keeps for reports of experience: a report establishes what is reported, and reaching what is reported about needs a further warrant. The parallel is mine; QUESTION-048 concerns experience, not identity.
• LIT-451 (Hume). Hume held that the “nice and subtile questions” of personal identity “can never possibly be decided” and are “grammatical rather than philosophical”. Frame-dependent verdicts with a 36% reflective minority are what that view predicts; the authors think aggregated, reflective judgement can still tip the balance. The pairing is mine.
• THEORY-447 and THEORY-476 (Parfit, Lewis) concern fission and what matters. The authors explicitly did not test fission or teletransport, since those concern sufficiency (n. 11), so nothing here bears on them.
• No instruction for machine-learning practice.

Limitations

• One population, small samples. Arizona introductory philosophy students; Ns from 28 to 54 per condition. The authors flag the homogeneity themselves (p. 307).
• The reflective result is marginal. p = 0.053, described as greater than chance, two-tailed by the authors’ own label. The forced-choice wording names “some of your memories” as the alternative and tells participants their answers are inconsistent, which is its own demand.
• The demand explanation is untested. No condition removes the “who else would feel it?” pressure while keeping the case concrete; the abstract question differs from the Pain case in many ways at once.
• Necessity only. Sufficiency of psychology, the half of Williams’s Lockean case that drives the body-swap verdict, is not tested.
• Construct of the Pain item. “Distinctive mental states” leaves sensation intact by wording, and the procedure “works exactly as expected”, so the brain is cured, not destroyed (item 2 asks whether “the infected brain will continue to function”). Agreement that “you will feel the pain” may express continuity of sentience, not rejection of psychological criteria. The adequate explanation quoted on p. 301 reads exactly that way.
• Within-subject Blok replication. Each participant saw both memory and no-memory versions, which invites contrast; the authors say a between-subject design “might well” differ (n. 16).
• Reporting. The cell counts for the N = 88 comparison are not given and do not follow from the reported samples; t(42) = 3.33 is given as p = 0.01, which is conservative. Figures were not viewed here.
• No measure of religious belief, though pilots drew religious answers and soul views are among the non-psychological options (n. 21).

Open questions

• Does the Pain verdict survive a vignette that eliminates all psychology, sensation included, or asks “will it be you” rather than “will you feel”? That would separate identity from bare sentience.
• Does Esch’s itch variant (n. 20) lower agreement? That would test vividness against demand.
• Is the 36% minority a stable individual difference (religiosity, personality), as note 23 asks?
• How do the results stand against later folk work: Berniūnas and Dranseika’s (2016) reply in the same journal, which argues that the folk do not take psychological continuity as necessary for personhood or identity (known here only from its PhilPapers record and a search snippet, not read), and Nichols’s later work with Strohminger on the moral self? Neither is held.