Source
- List & Pettit (2006), LIT-705, read in full in NOTE-550, from the authors’ revised version of 11 January 2006. Page numbers are the manuscript’s. Read as the lawful substitute for Group Agency (LIT-391), which stays Deferred.
- List (2014), LIT-700, read in full in NOTE-554.
What was actually shown
The setting. A group’s constitution is a function from its members’ judgments to the group’s judgments on an agenda of propositions closed under negation. List and Pettit, LIT-705, take the court example: P, Q, R and “R iff (P and Q)”. The constitution has universal domain, accepting every profile of complete, consistent and deductively closed individual judgments. The group is robustly rational if its judgments are complete, consistent and deductively closed for every such profile (p. 9). Because the constitution is itself members’ dispositions, the group’s judgments supervene on what the members judge and do, and nothing non-individual is added (p. 6).
The impossibility (pp. 9–12).
- Proposition 1: robust group rationality is inconsistent with majoritarian supervenience, the group judging each proposition as the majority does. The paper sketches the proof with the discursive dilemma: three individuals, each consistent, whose majorities accept P, Q and the biconditional and reject R.
- Proposition 2: it is inconsistent with uniform proposition-wise supervenience, the same function, for every proposition, of the members’ judgments on that proposition. Proved by Pauly and van Hees (2003); the proof is omitted.
- Proposition 3: it is inconsistent with proposition-wise supervenience at all, even if the function differs by proposition, provided it depends on more than one individual and respects unanimity. Proved by Dietrich and List (2005); the proof is omitted. “The group’s judgment on a particular proposition cannot generally be a function of the group members’ individual judgments on that proposition” (p. 12).
The results are stated for the court agenda and said to hold for wider classes (n. 3): Proposition 1 for agendas with a minimal inconsistent subset of three or more propositions, 2 for “minimally connected” and 3 for “strongly connected” agendas.
The possibility (pp. 13–17). Under the premise-based procedure, majority votes on the premises and the conclusion derived by the logical constraint, the group is robustly rational and its whole set of judgments is a function of the members’ whole sets: set-wise supervenience (Proposition 4, whose proof is given). Members’ judgments on the conclusion are then insufficient to fix the group’s (“weak discontinuity”) and not even needed (“strong discontinuity”) (p. 14). Under the distributed premise-based procedure each premise is settled by its own subgroup, so each member need contribute “a single judgment on a single proposition” and no member’s judgment on the conclusion enters at all (Proposition 5, pp. 16–17, with details in List 2005). The group’s judgment is still wholly fixed by its members’. So a group agent is “nothing mysterious” and yet “may hold judgments that are not directly continuous with its members’ corresponding individual judgments” (abstract).
The extension to kinds of attitude. List, LIT-700, divides collective attitudes by their supervenience base. Aggregate attitudes, summaries such as a poll, supervene proposition-wise and are rational only contingently. Common attitudes, held by all with common awareness, supervene on an enriched proposition-wise base and are consistent if the members are. Corporate attitudes, a group agent’s, satisfy holistic supervenience: “the collective (here corporate) attitude on any proposition p supervenes on the individual attitudes across a web of interconnected propositions and/or other non-attitudinal contributions by the individuals” (§4.4). He cites the 2006 paper and the book for why: “Except in special cases, no aggregation rule can robustly secure consistent and complete corporate attitudes on a set of interconnected propositions and also make the collective attitude on every proposition a function of the individual attitudes on the same proposition”. The kinds are almost logically independent: a corporate attitude need be neither any member’s nor commonly known (§5).
What this does not say
- It does not hold without its conditions. Proposition-wise rules survive if the domain is restricted, completeness or closure relaxed, a dictator or a constant rule used, or the agenda shrunk (n. 9). Real groups with shared priors may restrict their own domain. The account is a claim about robustness across every rational profile, not about what a given group does on a given day.
- It does not say a group’s judgments are anything over and above its members’. Supervenience is kept throughout. What fails is proposition-by-proposition dependence, not dependence. List calls the result “supervenience without reducibility” by analogy with non-reductive physicalism; what more than holistic supervenience “not reducible” amounts to, the paper does not say (NOTE-554, Limitations).
- It does not reach attitudes other than categorical judgments by proof. The theorems concern binary judgments; List and Pettit say little generality is lost for plans (pp. 4–5), and List names “aggregate fears and hopes, common fears and hopes, and corporate fears and hopes” (§1), but neither paper proves the impossibility for degrees of belief, desires or emotions. That is why the claim is stated for judgments.
- It does not say group agency can arise without members who sustain it. List and Pettit doubt it can: their §6 argues that no constitution could guarantee robust rationality whatever members’ reasons, without central direction and deference, and concludes that “at this margin of excitement, alas, the intelligibility runs out” (p. 21). That is an argument from the absence of a theory, offered as such, and it is not part of this account.
- It does not say the non-aggregativity is emergence. One reading the record has offered is that the failure of proposition-wise supervenience is non-aggregativity in Wimsatt’s sense, and so emergence compatible with reduction: List and Pettit’s homogeneity condition, invariance of the group’s judgments under permuting the members (p. 15), is an invariance-under-rearrangement condition, and proposition-wise supervenience is close to invariance under decomposing the agenda and reaggregating. That reading is the filing agent’s conjecture, not either author’s. Wimsatt’s paper (LIT-707) is unread, his conditions are known here only from his later summary, and whether they apply to a mapping from judgment profiles to judgments at all is unchecked. It is kept out of the claim.
Connections
- The authors’ replies (LIT-730, NOTE-582; THEORY-536). List and Pettit say that weakening universal domain or completeness complicates the formal arguments, but that rules meeting proposition-wise supervenience stay “extremely limited” and exclude “some of the most natural methods of aggregation” (p. 4). That supports this account beyond the strict conditions its title states, on the authors’ word and not by proof, so this account does not lean on it. The replies also confirm that the premise-based procedure is one illustration, not a recommendation. THEORY-536, their account of group attitudes as real and non-redundant, is compatible with this supervenience.
- Genuinely collective emotion. Huebner’s test for a group’s own emotion, a state “that could not be exhaustively characterized in terms of intentional states of individuals in aggregation”, is the informal counterpart of the failure of proposition-wise supervenience, and his navigation crew, routines each sensitive to one constraint and integrated on a chart, has the shape of the distributed premise-based procedure. THEORY-125 states his account. No relation is declared: this is a theorem about rational judgments, and it is not shown to carry over to emotions, whose coherence conditions are not deductive closure.
- THEORY-043 (minds of minded agents) finds the readings agree that
agency survives composition in a thin, functionalist sense, with List
(LIT-401) granting group agency. This account supplies, for judgment, a
principled structural condition on that agency: a collective of minded
members holds robustly rational judgments only if its organisation routes
them holistically. It says nothing about experience, which is where that
account’s dispute lies, so it does not meet its
promote_when. - Korsgaard (THEORY-065) grants collective agents agency in the same sense as individuals, because a group can achieve the unity of agency by conforming to its constitutive norms. This account says what such conformity costs in judgment: the group must sometimes judge against the majority of its members on a proposition to stay consistent across its agenda. The two are compatible.
- Social choice. Sen’s impossibility, THEORY-027, is the nearest result in the record: a different aggregation problem, over preferences and rights rather than logically connected judgments, with the same form of conclusion that conditions each reasonable alone cannot all hold.
Note of 2026-10-19: the court-agenda results filed apart
On the court agenda, majority voting fails robust rationality (Proposition 1), the premise-based procedure secures it with set-wise supervenience (Proposition 4), and the members’ judgments on the conclusion neither fix the group’s nor need to enter (weak discontinuity, Proposition 5). These are now THEORY-725, Active, with the proofs written out. What keeps this THEORY Proposed is the general claim its title states: that no proposition-wise rule, uniform or not, can do what the premise-based procedure does (Propositions 2–3). The source cites that claim without proving it.
Note of 2026-10-19: Proposition 3 re-derived, and the title narrowed
The record now proves Proposition 3, which LIT-705 cites from Dietrich and List (2005) without proof. The full steps are in record/promotion-audit/decisions/THEORY-127.yaml (batch 023, ADR-047 v2). In outline, with independence, unanimity, universal domain and robust rationality:
- For each literal a, let W_a be the family of coalitions C such that the group accepts a whenever exactly C does. By independence this is well defined. C ∈ W_a exactly when N∖C ∉ W_¬a.
- If a minimally inconsistent set contains a and ¬b, then W_a ⊆ W_b. Let C hold a together with the rest of the set, let everyone else hold ¬b together with the rest, and apply unanimity and consistency.
- The court agenda {P, Q, R, R↔(P∧Q)} has six minimally inconsistent sets: {¬P, R, B}, {¬Q, R, B}, {¬P, ¬R, ¬B}, {¬Q, ¬R, ¬B}, {P, Q, ¬R, B} and {P, Q, R, ¬B}. Through them every literal reaches R and R reaches every literal, so all eight W_a are one family W.
- The triple {¬P, R, B} stays consistent with two members negated, {P, ¬R, B}. A profile built on it shows that W is closed under intersection.
- W is then an ultrafilter on a finite set, so it is generated by one member, who dictates.
The uniform case, Proposition 2, follows. A uniform rule must be self-dual. It must also respect unanimity, because the unanimous profile {P, Q, R, B} would otherwise yield the inconsistent {¬P, ¬Q, ¬R, ¬B}.
The title used to say “interconnected propositions” without conditions, and that is false. On {p, q, p↔q} the judgment sets form an affine subspace over GF(2). There the issue-wise parity rule x₁ ⊕ x₂ ⊕ x₃ is independent, unanimous and robustly rational, and it depends on three members. So the title now names the court agenda and the two conditions the proof uses:
- every literal reaches every other by conditional entailment;
- some minimally inconsistent triple stays consistent with two of its members negated.
Whether these coincide with LIT-705 n. 3’s “strongly connected” agendas is not claimed. The extension to corporate attitudes beyond judgment, which LIT-700 proposes, is QUESTION-152.