Contribution
It put a measure of self–other overlap into the empathy–altruism paradigm and showed that, in its own data, the measure took over the predictive work that empathic concern had been doing. Before it, the egoistic alternatives to Batson’s hypothesis (aversive arousal, social disapproval, guilt, sadness, empathic joy) had been ruled out one at a time. This paper named a nonaltruistic alternative that Batson’s own model granted would defeat it, if it held: “without a distinct self and other … it is not possible to detach altruism from egoism” (p. 482). It tested that alternative together with personal distress and sadness in one model.
Key insight
When we help those close to us, what predicts how much we help is how far we see ourselves in them, not how much compassion we feel. Compassion rises with closeness only because closeness also raises the sense of oneness, and compassion is the felt signal of that oneness. On this view helping a close other is neither selfish nor selfless. Once self and other are merged, the authors argue, that dichotomy “loses its meaning” (p. 491).
Assumptions
- The self is malleable and can extend into others. Self-conceptions “can be made to shift temporarily, flowing across established boundaries” (p. 482). The merging is “conceptual, not physical”: “it is conceptual identities that are merged” (p. 482).
- Two routes to oneness. Perspective taking, which comes with attachment, “facilitates the symbolic merging or expansion of the self into the other”. Or cues of genetic commonality (kinship, friendship, familiarity, similarity) produce felt oneness, as inclusive fitness would predict (p. 483). The paper does not test which route operates.
- Imagined need and imagined help stand in for real need and real help. Participants rated “what they were currently feeling” after focusing on the scenario, and chose what they “would be willing” to do (pp. 483–484). The authors defend this by noting that their pattern reproduces the one from active-helping paradigms (p. 491).
- Regression control identifies the mediator. The claim that oneness and not empathic concern “mediates” helping rests on hierarchical regression and, in Study 3, a path model with covarying, not causal, links among the psychological variables (fn. 3).
- Oneness is defined as dyadic. Footnote 1 reserves “oneness” for “that which occurs in a dyadic relationship with a specified other” and keeps it apart from “we-ness” as group membership or social identity (p. 483).
Key results
- Design (pp. 484–487). Participants described a same-sex near stranger, acquaintance, good friend or closest family member, then imagined a need and chose one of seven costed helping options. Costs were weighted from pilot ratings. Study 1 used eviction (n = 90; weights 0.0 to 8.3). Study 2 used the person’s death leaving two children without a home (n = 74; up to “raise them as you would your own”, 9.2). Study 3 crossed closeness with three levels of need: a phone call, eviction and orphaned children (n = 263; 4 × 3). Empathic concern was Batson’s four adjectives (sympathetic, compassionate, softhearted, tender). Personal distress was alarmed, worried and uneasy; sadness was sad, low-spirited and heavy-hearted. All were rated after the helping choice. Oneness was the IOS scale plus a 7-point item on “the extent to which they would use the term we” (p. 484), placed before or after the other measures (order had no effect).
- Closeness raised helping, empathic concern and oneness (Table 1). Study 1 helping went from 1.20 for a near stranger to 6.89 for family, and oneness from 1.52 to 4.57. Study 1 F values: helping F(3, 82) = 33.28; empathic concern F(3, 82) = 5.37; oneness F(3, 86) = 30.58. Empathic concern rose less steeply than oneness. In Study 1 it was 3.04 for a near stranger and 4.36 for an acquaintance, and did not differ between acquaintance, friend and family.
- Both predicted helping (pp. 485–486). Study 1: r = .45 for empathic concern and r = .76 for oneness. Study 2: r = .33 and r = .53.
- Oneness removed empathic concern’s effect (pp. 485–488). In
hierarchical regression (gender, then empathic concern, then distress
and sadness, then oneness), empathic concern predicted helping at step 2
and still at step 3 in every test. At step 4 it became non-significant,
while oneness was significant:
- Study 1: empathic concern b = .11, F < 1; oneness b = 1.10, F(1, 74) = 48.27.
- Study 2: empathic concern b = .37, F(1, 57) = 1.37, ns; oneness b = .57, F(1, 57) = 10.39.
- Study 3, phone call: oneness F(1, 65) = 13.39, unique variance 7.7% against 0.8%.
- Study 3, eviction: oneness F(1, 50) = 27.15, 14% against 0.7%.
- Study 3, orphaned children: oneness F(1, 50) = 16.13, 13.5% against 0.9%.
- The pattern holds for near strangers (p. 490). Analysing only near strangers in Study 3, empathic concern predicted helping at step 2 (F(1, 72) = 14.57) and lost significance once oneness was entered (b = .08, F < 1; oneness F(1, 54) = 10.26). “In no instance within any of the 12 relationship conditions across our three studies was empathic concern still predictive of helping after the influence of oneness was extracted” (fn. 5).
- Path model, Study 3 (pp. 488–489). Oneness was “the only explored predictor to mediate the effects of the manipulated variables on helping”: χ²(12, N = 236) = 16.70, p > .15, CFI = .995. A model treating the mediators as consequences of helping fit worse. Closeness, need and their interaction also affected helping directly, so “other psychological constructs” are at work (p. 490).
- Oneness and empathic concern have different profiles (p. 491). Severity of need raised empathic concern but not oneness, and empathic concern was tied to distress and sadness while oneness was not. The empathic concern index was reliable (α = .89; oneness .83). This is the authors’ answer to the objection that oneness is just a better measure of empathy.
- Need moderated closeness (p. 487, Figure 1). The effect of closeness grew with severity of need, interaction F(6, 243) = 10.33. Family never significantly exceeded good friends, which the authors note is “not completely consistent with a strict evolutionary model” (p. 490).
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Relationship closeness raises empathic concern, perceived oneness and willingness to help | strong within the paradigm | Table 1; ANOVAs in all three studies |
| C2 | With oneness controlled, empathic concern no longer predicts willingness to help, while oneness does with empathic concern controlled | moderate | five hierarchical regressions and the near-stranger subanalysis; but cross-sectional ratings, with empathic concern rated after the helping choice |
| C3 | Oneness, not empathic concern, mediates the effect of closeness and need on helping | weak to moderate | one path model with covarying mediators (fn. 3); “mediation” is statistical, on hypothetical helping |
| C4 | Oneness and empathic concern are distinct constructs, not one construct measured twice | moderate | different relations to need severity and to distress and sadness (p. 491) |
| C5 | Empathic concern is an emotional signal of oneness | weak; interpretation | inferred from C2–C4; “bidirectional causal influences” suggested and not tested (fn. 3) |
| C6 | Helping under oneness is nonaltruistic: not selfless, and not to be called egoistic either | conceptual | argued from Batson’s own concession that altruism needs distinct self and other (pp. 482, 490–491) |
| C7 | In-group favouritism and recategorization effects share oneness as a common cause with dyadic helping | speculation | cited studies only (p. 492) |
Concepts
- Oneness — “a sense of shared, merged, or interconnected personal identities” (p. 483), with a specified other. Measured as the IOS scale (seven pairs of increasingly overlapping circles) combined with how far one “would use the term we” to describe the relationship (p. 484). Distinguished from “we-ness” as social identity (fn. 1).
- Empathic concern — Batson’s other-oriented emotion of compassion, tenderness, softheartedness and sympathy (p. 481).
- Nonaltruistic — the authors’ word for helping when self and other are merged. It is not selfless, and they decline to call it egoistic, because “the traditional dichotomy between selfishness and selflessness loses its meaning” (pp. 490–491).
- One-egoistic-factor-at-a-time approach — the authors’ criticism of Batson’s programme. Ruling out one egoistic motive in one situation “does not rule out egoism due to other factors there” (p. 484), so several nonaltruistic motivators should be tested at once.
Connections
The paper answers Batson’s empathy–altruism hypothesis (Batson 1991; Batson & Shaw 1991) with Aron and Aron’s self-expansion and IOS work and Davis et al.’s finding that perspective taking merges self and other. It draws on Hamilton’s inclusive fitness for the claim that “the self in self-interest can lie outside of one’s body” (p. 482). It quotes Batson and Shaw’s doubt that merging occurs “except perhaps in some mystical states” (p. 482). It places its result beside Batson et al. (1995), which also treats empathic concern as a signal, of “valuation of another’s welfare”. The difference is that here the valuation follows from the self being “implicated in the other” (p. 491). Batson et al.’s reply in the same issue (LIT-1700) is unread. The SEP entry (NOTE-1274) and Miyazono and Inarimori (NOTE-1277) reported this paper before it was read; both are borne out on what they report. Miyazono and Inarimori’s “seven-point scale” of oneness is the paper’s “we” item.
Bearing on the record
- CLAIM-491: partly. The claim’s note of 2026-10-30 calls this paper its best evidence, “read only at second hand”, and it is now read. It supports the claim’s empirical form. In every test, helping tracked oneness and not empathic concern, and the authors’ “nonaltruistic” is the claim’s middle position, not selfishness. It falls short in three ways. Oneness is measured as dyadic and kept apart from group identity (fn. 1), so the paper bears on the claim’s “others” and only by speculation (p. 492) on its “group”. The helping is hypothetical. And on CLAIM-492’s demand, the paper separates oneness from empathic concern (C4) but not from closeness or caring in general: closeness drives oneness hardest, and nothing shows oneness present where caring is absent or the reverse.
- CLAIM-494: against it in part, not decisive. In this paradigm empathic concern did not predict helping once oneness was controlled, even for near strangers, the targets Batson’s studies use (p. 490, fn. 5). That is the opposite of what the objection asserts. But the objection’s evidence is Batson et al.’s reply (LIT-1700), which answers this paper, and that reply is unread. What this reading adds is the paradigm the reply criticised. The helping is imagined. Empathic concern was rated after the helping choice (fn. 6). And the case for mediation is regression on correlated ratings, with a path model whose links among mediators are covariances (fn. 3). Whether merging “was controlled” in the sense CLAIM-491 means turns on the IOS and “we” items, which the SEP argues test overlap, not identity (NOTE-1274).
- QUESTION-300. The paper’s answer to which measure captures the self’s extent is IOS plus “we”. Its defence that this is not empathy under another name is the different-profiles argument (C4).
- THEORY-1114. The paper’s footnote 1 keeps oneness apart from social identity, the construct self-categorization theory studies. It cites Brewer and Gardner (1996) and Turner (1987) for in-group favouritism through self–group merger only as a parallel (p. 492).
- No instruction for machine-learning practice; nothing for the anthology.
Limitations
- Hypothetical helping. Every outcome is a choice about an imagined person in an imagined plight, and no help was given. The authors argue the pattern matches active-helping paradigms (p. 491). That is an argument from similarity of results, not a test.
- Order and timing. Empathic concern, distress and sadness were rated after the helping choice, the reverse of Batson’s usual order. The authors flag this and ask for more data (fn. 6). Oneness was counterbalanced.
- Statistical, not experimental, mediation. Nothing manipulated oneness apart from closeness. The path model treats the mediators as covarying and the authors note that “bidirectional causal influences” were likely but could not be tested (fn. 3).
- Shared variance. Oneness and helping were both strongly driven by the closeness manipulation (oneness F ≈ 30, helping F ≈ 33 in Study 1), while empathic concern varied less across closeness. The authors’ reply (fn. 5) is that the pattern held within each of the 12 closeness conditions, which addresses the between-condition confound but not shared method variance within them.
- Sample. Introductory psychology students at one university, rated anonymously.
- Scope of the conclusion. The authors deny that oneness is “the only influence to be seriously weighed” (p. 490), and their in-group and cross-cultural extensions rest on cited work only.
Open questions
- Does oneness predict helping when it is manipulated directly, holding closeness fixed? And does it predict real help, given with a cost?
- Do the IOS and “we” items measure a self extended into the other, or closeness and caring described in spatial and pronoun terms? A measure that comes apart from caring, as CLAIM-492 asks, would settle which.
- Does Batson et al.’s reply (LIT-1700) find empathy’s effect with the same oneness index, or with different merging measures? The SEP’s account suggests different ones.