Contribution
The chapter separates two ideas of agency, natural and normatively constituted, and argues that the second captures two features of ordinary action-attribution that the first struggles with. It restates the argument of Self-Constitution in short form. It reads Bratman as able to take the normative view on board. It then argues that the laws of reason in general, theoretical as well as practical, are “the laws of mental activity”. That is what makes collective agency possible “in the most literal sense”.
Key insight
The unity that makes a movement someone’s action, and not something that happens through them, is not given by having one mind. Self-consciousness about one’s own attitudes undoes the unity an animal has, and leaves “a heap” of impulses and perceptions that must be put back together. Following the norms of reason is how that is done. So the unity of an agent is the same kind of thing whether the agent is one person or a group: something achieved by conforming to constitutive norms, not something found.
Assumptions
- Kant’s two imperatives are constitutive of willing. “When you will an end … you commit yourself to staying on the track of that end in the face of possible changes in your circumstances.” The categorical imperative is the law of autonomy, and the hypothetical imperative is the law of efficacy.
- Success in action is internal to agency. “An agent does not just effect something … an agent undertakes to effect something, and to undertake to effect something is to set yourself up for success or failure.” A deviant success, such as a ricocheting bullet, is a failure as agency.
- Animal perception has “teleological content”. Non-human animals perceive things as “to-be-eaten”, “to-be-fled”. This is offered as a belief, with a short evolutionary argument.
- A constructivist, not substantive-realist, view of reasons. It is assumed here, and argued elsewhere.
Key results
- §1.1–1.2. Agency, individual or collective, needs unity, because an action is “a movement, or the effecting of a change, that is backed by the agent as a whole”. Ordinary attribution carries the “activity implication” and the “identity implication”.
- §1.3. Self-constitution explains both, separately. Activity comes from the form of the principles (self-determined efficacy). Identity comes from their content (practical identities: friend, teacher, parent, citizen). The account is indifferent to causal history: “It is the content of the principles we follow, not the metaphysical story about how we came to follow them, that explains the activity implication.”
- §1.4. A naturalist can explain the identity implication through a Lockean reading of intentions, which the chapter attributes to Bratman. Doing so means “adopting the self-constitutionalist view that our identities are constituted by the choices that we make when those choices are governed by the laws of reason”. The Lockean view needs only that the agent conform to her principles, not that she choose them. So the activity implication still needs self-determination.
- §2.1–2.2. Self-consciousness of one’s own attitudes, unlike mere awareness of one’s experiences (which mirror-test animals may have), sets the task of unification. Theoretical reason unifies a conception of the world, which also unifies the mind as its knower. “To be active, and to be under the norms of reason, are one and the same thing.”
- §2.3. Being one natural mind “sets you the task” of normative unification, but it is not the only thing that does. Talking and acting together set the same task, so “nothing … prevents people from acting together in the most literal sense of the word”.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Ordinary attribution of action implies activity and identity, which a causal-pathway account does not explain on its own | argued, by examples and by the free-will debate | §1.2 |
| C2 | Conforming to the categorical and hypothetical imperatives gives action the form of self-determined efficacy, and that is what activity is | argued; depends on Kant’s claim that the categorical imperative has determinate content, flagged as “controversial” | §1.3 |
| C3 | Whether an agent is active depends on the content of her principles, not on how she came to hold them; responsibility is “not that deep” | asserted, with a footnote analogy (a republic’s sovereignty does not depend on how it was founded) | §1.3, notes 13 and 15 |
| C4 | A naturalist about agency who explains the identity implication through Lockean intentions has taken on self-constitution | argued, as a reading of Bratman | §1.4 |
| C5 | Reflective self-consciousness of one’s attitudes breaks given unity, and the laws of reason are the principles of re-unification | argued, with the evolutionary speculation marked as such | §2.2 |
| C6 | Unity of agency is an achievement, so collective agents are possible in the same sense as individual ones | argued from C5; the mechanisms of collective unification are not given | §2.3 |
Concepts
- Natural vs normatively constituted agency: agency identified by a causal pathway, against agency whose capacity “consists in or depends on the existence of certain normative relations” and whose success “depends on conformity to the norms in question”.
- Activity implication: in acting, the agent’s own efficacy, “spontaneous in the Kantian sense”, is at work.
- Identity implication: an action evinces the agent’s “essential self”, which is why love, hate, gratitude and resentment are appropriate responses to it.
- Self-constitution: making oneself into an agent, and into the particular agent one is, by choosing under the formal norms of agency.
Connections
The chapter’s main interlocutor is Bratman. It cites Intention, Plans, and Practical Reason (LIT-593), Faces of Intention and Structures of Agency (LIT-590), and reports a seminar exchange of spring 2009 on post-hypnotic intentions. It sets Gilbert’s normative account of shared intention (her “What is it for Us to Intend?” and “Shared Intention and Personal Intention”) beside the political model of Rousseau and Plato. The record holds Gilbert’s earlier “Walking Together” (LIT-477), Deferred and unread. It draws on Davidson and Velleman for the natural conception (LIT-295 is Velleman on action), and on Kant and Plato for the normative one. Its claim that self-consciousness of attitudes sets the normative problem is the same claim the Tanner Lectures (LIT-564) make under the name of “the reflective structure” of the mind.
Bearing on the record
- THEORY-040 (manipulation). C3 is a clear instance of a structural account choosing the first option for activity. A manipulated agent who follows the right principles is active, whatever their origin. For deep responsibility, though, Korsgaard concedes that history matters (note 15). The reported Bratman exchange is second-hand evidence that Bratman also bites the bullet. Korsgaard’s own reply is that our objection to the hypnotic case “is not after all wholly based on its origin” but on its expected ephemerality. That suggests a non-historical condition, durability or integration over time, that THEORY-040 does not consider. Worth a line in its connections. Whether it escapes manipulation (a manipulator can install durable states, as the Demetrius case shows) is exactly what Korsgaard leaves “puzzling”.
- THEORY-029 (the regress). Korsgaard’s stopping point is not an attitude but conformity to formal norms that constitute agency. That is a candidate for THEORY-029’s refutation condition, “a stopping point that is specified independently of agency”. It is specified by the norms of agency itself, so it may count as the dilemma’s second horn. The record should weigh it when THEORY-029 is next revised, together with the Tanner Lectures’ wanton argument (NOTE-465).
- A THEORY, filed as THEORY-065 (Proposed). The unity that makes a movement an agent’s action is achieved by conformity to constitutive norms, not given by the causal unity of a mind, and the same holds for collective agents. Sources: this chapter and LIT-564. The rival in the record is the natural conception, as Bratman holds it (LIT-567). It is Proposed until Self-Constitution (LIT-597) or the reply to the Bratman volume is read.
It carries no instruction for machine-learning practice.
Limitations
- The manuscript was read, not the published chapter.
- Its central Kantian premise, that the categorical imperative has determinate content, is assumed (“suppose … as Kant supposes we can”).
- The account of reason as unification is said to be “lifted” from another paper and is given here in compressed form.
- The collective-agency conclusion is stated in a paragraph. How a group observes “the rules of normative unification together” is not worked out.
- It does not say what unifies a natural mind (“I am certainly not going to advance a view”).
Open questions
- Does C3’s indifference to history survive the Demetrius case, a durable installed identity, or must the account add a condition on how the content of one’s principles was acquired?
- How does a group constitute itself as a single agent by following norms, and do Gilbert’s joint commitments (LIT-477) supply the mechanism?