Contribution
A new formulation of team reasoning built for cooperation for mutual benefit, not for a team’s collective good. It argues that the economist’s standard device, keeping individual rationality and putting the pro-social motive into the preferences, cannot say what trust is. It argues that Bacharach’s team reasoning cannot either, and gives a schema that can.
Key insight
Intending that each of us benefits is not intending the benefit of us as one entity. “Intending to promote the common interests of team members is not the same thing as intending to promote the collective interests of the team” (p. 154). The first is cooperative. The second can ask a member to lose so that the team gains. A team reasoner of the first kind need not see her act as a sacrifice for a social end, nor as “adopting a collective goal that transcends her private interests” (pp. 163–164).
Assumptions
- Payoffs measure each player’s interests as she judges them, not all-things-considered utility (p. 145). A player who maximizes her payoff is “self-interested”; one who does not is “non-selfish”. This is what makes the paper’s distinctions statable.
- Maximin as the benchmark of benefit. Whatever a player can guarantee herself is not a product of cooperation (p. 157).
- Pure strategies; mixed strategies are set aside (notes 1 and 3).
Key results
-
The games (pp. 145–150, payoffs to A and B).
- Trust Game: hold gives (0, 0); send, then keep gives (−1, 5) and return gives (2, 2).
- Market Game: the same, but keep gives (−1, −1), so returning is in B’s interest too.
- Confidence Game: return benefits A at B’s expense.
- Dilemma Game, Hi–Lo Game and a three-player Snowdrift Game.
-
Social preferences fail on trust (§3).
- Inequality aversion or altruism can make B return. But it ignores that A’s send preceded it, and second movers are known to respond to the options the first mover removed (Falk et al. 2003).
- Guilt aversion (Battigalli and Dufwenberg) would have B confirm expectations in the Confidence Game too, where doing so is being conned.
- Under Rabin’s reciprocity, sending is neither kind nor unkind when return is expected, since (2, 2) beats (0, 0) for both. So B would keep: the practice of trust cannot be common knowledge in equilibrium.
-
Against Bacharach (§4). His team reasoning:
- maximizes a team utility, which needs interpersonal comparison and can trade members’ interests off;
- binds whether or not others identify, which leaves no room for reciprocity;
- treats individual rationality as the benchmark, a module that switches on only when individual choice fails;
- assumes teams can compute optimal profiles, where real cooperation follows existing and perhaps suboptimal conventions (rules of the road).
Bacharach makes group identification psychological and prior to rationality: “the question ‘Who am I?’ (or ‘Who are we?’) is logically prior to rational choice” (p. 153).
-
Mutually beneficial practice (§5).
- Condition 1: every player does better than her maximin payoff.
- Condition 2, for three or more players: every subgroup benefits from its complement’s participation. This excludes free riding: in the Snowdrift Game, (dig, dig, wait) fails it.
-
The Schema of Cooperative Team Reasoning (pp. 161–162).
- P1: there is common reason to believe s* is a mutually beneficial practice.
- P2: there is reciprocal reason to believe each player will choose her part of s*.
- P3: there is reciprocal reason to believe each player endorses and acts on the schema.
- So, C: I should choose my part of s*.
Endorsing the schema is the analogue of Bacharach’s group identification. P3 works like a contract that binds only when everyone has signed.
-
Conclusion (pp. 163–164). In the Trust Game the schema can lead the second mover against her interest, but she does not construe it as a sacrifice. In the Market Game it leads to what self-interest also recommends, but with the intention of mutual benefit. Contra Adam Smith’s butcher, market exchange need not express self-interest on either side.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Theories that keep individual rationality and make preferences social cannot represent the practice of trust as common knowledge | moderate: a proof for Rabin’s model and counterexamples for others; not every social-preference model is checked | §3 |
| C2 | Intending the common interests of members differs from intending the collective interest of the team, and only the first is cooperation for mutual benefit | moderate: conceptual, argued from cases | §4 |
| C3 | The Schema of Cooperative Team Reasoning represents trust and market exchange as joint action for mutual benefit | moderate: a schema shown to give the intended verdicts in five games | §5 |
| C4 | An action that is extensionally self-interested can be intentionally cooperative | moderate: follows from C3 in the Market Game | §6 |
Concepts
- self-interested / non-selfish: acting to maximize, or not, one’s payoff interpreted as one’s own interests as one judges them.
- social preferences: preferences that take direct account of others’ payoffs, beliefs or intentions.
- team reasoning: asking “What should we do?” rather than “What should I do, given what others will do?”. Not reducible to the second by redefining payoffs (p. 152).
- common versus collective interests: each member’s benefit, versus the benefit of the team considered as a single entity.
- mutually beneficial practice: a strategy profile meeting Conditions 1 and 2.
- reciprocal reason to believe: a chain of reasons to believe about the others, saying nothing about what one has reason to believe of oneself.
Connections
- Sugden 2003 (LIT-1782), unread. This paper revises Sugden’s own 1991 and 1993 papers and develops ideas from 2011, and the 2003 paper sits between them. It does not restate the 2003 schema, so it shows what Sugden kept and changed, not what the 2003 paper said.
- Salice and Miyazono (NOTE-1266) take team reasoning, via Bacharach, as one of the two parts of group identification.
- The group entity versus a category of members (TERM-096). Sugden’s distinction between collective and common interests is the same line drawn for interests. The collective interest belongs to the team as one entity; the common interests belong to the members severally.
Bearing on the record
- The owner’s thesis (CLAIM-491). The paper bears on it at three
points. These are inferences; Sugden does not discuss the extent of the
self.
- The thesis is closest to Bacharach. To identify with a group is to take it as one’s unit of agency, so that its good is the good one pursues. That makes “self-interest on a larger self” a fair gloss on Bacharach’s team reasoner.
- Sugden’s alternative is a third shape the thesis does not yet name. The cooperative team reasoner intends each member’s benefit, her own included, without a larger self whose interest could outweigh hers. It is not the thesis’s larger self, nor self-interest as usually read, nor sacrifice.
- Sugden rejects the device the thesis risks. Folding pro-social motives into an individual’s preferences, he says, “requires, as a matter of conceptual necessity, that an individuals’ pro-social motivations can always be represented as payoffs to her as an individual” (p. 145). Doing it by enlarging the individual is the same move.
- The triviality objection (CLAIM-492). Sugden makes the objection’s point about social preferences: “it is always possible to assume that the relevant player has a preference for behaving in exactly that way” (p. 148). He asks for non-trivial specifications, as CLAIM-492 does for the self.
- A dated note is added to the claim.
- No THEORY filed.
- No ML instruction.
Limitations
- Not the 2003 paper. What Sugden argued in “The Logic of Team Reasoning” is still unread.
- A sketch, by its author’s description (p. 156: “I will now outline such a theory”). No experiment tests the schema.
- Fairness is set aside (p. 160).
- Typos: “truest” for “trust” (p. 149); “Shared Iintention” in the bibliography.
Open questions
- Can an experiment tell a cooperative team reasoner from a Bacharach team reasoner? A one-shot Trust Game with and without assurance that the partner team-reasons would separate them by P3.