Contribution
It separates three things that informal talk and the social sciences both call “what the group thinks”: a summary of members’ attitudes, an attitude all members share knowingly, and the attitude of a group agent. It gives each a definition, a supervenience base, a social role and a rationality profile, and shows that they come apart in all but one direction. By the author’s description it is a review, not new results; its contribution is the disambiguation.
Key insight
How “reducible” a collective attitude is, is fixed by how wide its supervenience base is. A poll’s verdict on p depends only on what people think about p. A common belief that p depends also on what people think others think about p. A group agent’s view on p can depend on what members think about everything else, and on what they do, because the group must stay consistent across its whole agenda. Only the third needs a group agent, and only the third can come apart from what any member believes.
Assumptions
- Functionalism about attitudes (§4.2). Beliefs and desires are states defined by their role in an agent; so a group agent’s attitudes are “whichever states of the organized collective play the relevant functional roles”. This answers the charge of circularity in defining corporate attitudes by group agents and group agents by their attitudes.
- A thin definition of agency (§4.2). Beliefs, desires or goals, and a capacity to act on them, with no condition of consciousness and no normative content (List & Pettit 2011, ch. 1).
- Common awareness as an infinite hierarchy (§3.1), with the caveat that a dispositional or primitive definition would do.
- Rationality licenses agency ascription (§4.5). For groups, unlike humans, the burden of proof is on whoever ascribes agency; only local irrationality is compatible with it.
Key results
- Aggregate attitudes (§2). Defined after Quinton (1975): to ascribe a mental predicate to a group “is always an indirect way of ascribing such predicates to its members”. Formed by direct aggregation (votes, polls) or behavioural aggregation (prediction markets: the price of a $1 bet on p reflects the market’s credence in p, §2.2). They depend on the rule (Riker: no “will of the people” independent of procedure). They satisfy propositionwise supervenience (§2.5). Majority belief can be inconsistent (Table 1). Market beliefs tend to coherence only because incoherence is an arbitrage opportunity (§2.6).
- Premise-based rules produce potential corporate attitudes (§2.5). Rules that derive conclusions from voted premises violate propositionwise independence; their outputs are better seen as coherent corrections, or candidates for corporate attitudes, than as aggregates.
- Common attitudes (§3). Every member holds it, believes every other member holds it, and so on. Coordination needs them (the coordinated- attack problem cannot be solved by finitely many messages, §3.2). Offices, currencies and rituals depend on them. They satisfy enriched propositionwise supervenience (§3.3) and inherit consistency from unanimity (§3.5).
- Non-reductionistic rivals (§3.4). Searle’s we-intentions (a special mode) and Gilbert’s plural subjects (a new subject) play the role of common attitudes “in very different ways”. List leaves we-attitudes out until shown necessary, judges plural-subject attitudes “more akin” to corporate ones, and says invoking a plural subject for every joint action “seems to give rise to an unnecessarily rich ontology of subjects”.
- Falsely attributed common attitudes (§3.6). Clauses (2), (3), … can hold without clause (1). This is pluralistic ignorance (Prentice & Miller 1993; O’Gorman 1975 on white Americans’ estimates of support for segregation). It can sustain bad equilibria, and public pledges can break them (Mackie 1996 on footbinding).
- Corporate attitudes (§4). A group agent’s, guiding its action. Not reducible: a group’s attitude on p cannot generally be a function of members’ attitudes on p, by the List–Pettit impossibility result (§4.4). They satisfy holistic supervenience, on “individual attitudes across a web of interconnected propositions and/or other non-attitudinal contributions”. The relation is “supervenience without reducibility”, as in non-reductive physicalism.
- Three lessons (§5). (1) An aggregate attitude need involve neither a common nor a corporate one. (2) A common attitude need not involve a corporate one, and a falsely attributed common attitude need not involve an aggregate one. (3) A corporate attitude need involve neither.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Collective-attitude ascriptions are ambiguous among aggregate, common and corporate attitudes, and the ambiguity causes confusion | moderate | examples throughout; no case of confusion is analysed in detail |
| C2 | Aggregate attitudes supervene proposition-wise and are rational only contingently | strong | by definition; Table 1; arbitrage argument for markets |
| C3 | Common attitudes supervene on an enriched proposition-wise base and are consistent if members are | strong | by definition, §§3.3, 3.5 |
| C4 | Corporate attitudes supervene holistically and are not straightforwardly reducible | strong as to the supervenience; moderate as to “not reducible” | the List–Pettit theorem, cited (§4.4); the reducibility claim is by analogy to non-reductive physicalism |
| C5 | Plural subjects and we-modes are not needed to explain coordination | weak | §3.4; asserted with reference to Chwe and Perea, not argued |
| C6 | The three kinds are almost logically independent | strong | §5, by construction and examples |
Concepts
- Aggregate attitude — a summary of members’ attitudes produced by an aggregation rule or statistical criterion (§2.1); Gilbert’s “summative”.
- Direct / behavioural aggregation — attitudes elicited and summarised, or read off behaviour such as prices (§2.2).
- Common attitude — held by all members as a matter of common awareness (§3.1).
- Falsely attributed common attitude — higher-order clauses hold, the first-order clause does not (§3.6).
- Corporate attitude — an attitude held by the collective as an intentional agent (§4.1).
- Propositionwise / enriched propositionwise / holistic supervenience — the three supervenience bases of the three kinds (§§2.5, 3.3, 4.4).
Connections
- List & Pettit 2006 (LIT-705). Extended; declared on the LIT. The 2006 set-wise supervenience becomes “holistic supervenience” here, now admitting non-attitudinal contributions.
- The book (LIT-391), unread. Cited for the definition of agency (ch. 1), for supervenience (ch. 3) and for strategy-proofness (ch. 5).
- Gilbert (LIT-492, LIT-701) and Searle (LIT-490; LIT-482 read). §3.4 is the paper’s one engagement with them. It answers neither in detail. On Gilbert’s collective guilt feelings the taxonomy would ask whether the plural subject has the rational structure a corporate attitude needs. List does not raise it.
- Huebner (LIT-702). His holism/collectivism distinction maps onto aggregate-or-common against corporate. His McCain–Palin campaign is a candidate corporate fear that is explicitly not an aggregate of members’ varied emotions, which is List’s lesson 3 applied to emotion.
- Schmid (LIT-697), unread. His title’s “corporate emotions” are, by his abstract, a we-mode phenomenon, the kind List sets aside. Whether Schmid’s corporate emotions are List’s corporate attitudes is a question for a reading of Schmid.
- Von Scheve & Ismer (LIT-696). Their I-mode and We-mode collective emotions are, in List’s terms, aggregate or common attitudes, and their We-mode borrows the we-mode List declines to need.
Bearing on the record
- Candidate THEORY (proposed, not filed): “collective emotion” names three different things, aggregate, common and corporate emotions, and the disagreements among Gilbert, Huebner, Schmid and the social psychology of convergence are partly about different kinds. Only corporate emotions are candidates for being a group’s own state. Sources: this paper, LIT-702, LIT-696. Promote when Gilbert and Schmid are read and their accounts placed with their own words.
- No instruction for machine-learning practice.
Limitations
- By its own statement a review. The theorem it relies on is cited, not proved.
- Emotions are mentioned once, in a list (§1). Whether affective states, which are not obviously propositional or rationally constrained, divide the same way is not examined.
- “Not straightforwardly reducible” is supported only by analogy to non-reductive physicalism; what more than holistic supervenience it amounts to is not said.
Open questions
- Do corporate emotions exist, and would they need the rationality profile that corporate beliefs need? An emotion’s coherence conditions are not deductive closure.
- Is pluralistic ignorance about emotion (each member thinks others are angrier) a falsely attributed common emotion, with the same equilibrium effects?