Contribution
A ten-page statement of Kane’s view for a symposium on Four Views on Free Will, with replies to Fischer, Pereboom, Vargas and McKenna. Its value here is that it states, in Kane’s own words, the two things the record had only at second hand: where chaos enters, and how he answers the objection that undetermined choices are luck.
Key insight
Free will is about the formation of the will, not only about free action: “It is about self-formation, about the formation of our ‘wills’ or how we got to be the kinds of persons we are” (p. 37). And the indeterminism in that formation is not a cause acting on its own but a hindrance inside a goal-directed effort, so success despite it is the agent’s (p. 42).
Assumptions
- Ultimate responsibility (UR). “To be ultimately responsible for an action, an agent must be responsible for anything that is a sufficient reason, cause or motive for the action’s occurring” (p. 36).
- A regress, stopped by indeterminism. UR would need an infinite regress of past responsible acts unless some acts had no sufficient causes or motives (p. 37).
- The neural hypothesis. Conflict is “reflected in appropriate regions of our brains by movement away from thermodynamic equilibrium—in short, a kind of ‘stirring up of chaos’ in the brain that makes it sensitive to micro-indeterminacies at the neuronal level” (p. 39). Two crossing recurrent networks carry the competing motives (p. 40). Stated, not argued.
- Parallel processing. The brain can pursue two cognitive tasks at once, so an agent can endeavour to make both choices (p. 41).
Key results
The paper argues; it reports no data.
- Three kinds of free act (p. 37): voluntary and uncoerced, compatible with determinism; done “of our own free will”, from a will the agent is ultimately responsible for forming; and self-forming or “will-setting” acts, which must be undetermined. The kinds are nested, and only the third must be undetermined.
- Frankfurt cases refute PAP but not the requirement of alternatives for SFAs, which need more than one robust option (pp. 38–39). Pereboom’s Tax Evasion and McKenna’s variants are, either way, not SFAs.
- The businesswoman (p. 39): torn between helping a victim and her meeting, she wills both, so whichever succeeds is her doing.
- Indeterminism as obstacle (pp. 39–42): the mathematician solving a problem despite neural noise, the assassin whose arm wavers, the husband breaking the table, and a Morse message getting through static. In an SFA the noise hindering each choice “is coming from her own will, since she also deeply desires to do the opposite” (p. 40).
- Macro-control (p. 43): we need not “micro-manage our individual neurons”; what responsibility needs is “macro-control of processes involving many neurons”.
- Diminished control, undiminished responsibility (pp. 43–44): “diminished control in such circumstances does not entail diminished responsibility when agents succeed in doing what they are trying to do”; three assassins at 50, 80 and 100% chance are equally guilty.
- “Self-formation is not a gift, but a struggle” (p. 44).
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Free will requires ultimate responsibility, which requires undetermined self-forming actions | moderate | regress argument (p. 37) |
| C2 | Frankfurt cases leave the alternative-possibilities condition on SFAs standing | moderate | case analysis of Pereboom’s and McKenna’s examples (pp. 38–39) |
| C3 | Conflict stirs up chaos that makes neural processes sensitive to micro-indeterminacy | weak | stated as a hypothesis, no evidence given (p. 39) |
| C4 | Indeterminism that hinders a goal-directed effort does not undermine responsibility for its success | moderate | the assassin, husband and Morse analogies (pp. 40–43) |
| C5 | Indeterminism lessens control but not responsibility | moderate | the three-assassins case (pp. 43–44) |
Concepts
- UR — as under Assumptions.
- Self-forming action (SFA), or will-setting act — an undetermined act by which an agent forms the will it later acts from (p. 37).
- Plural voluntary control — not named here; the doubling of effort, “to make both choices”, plays its role (p. 41).
- Macro-control — control of processes involving many neurons (p. 43).
Connections
- Fischer, Pereboom, Vargas and McKenna, the symposium’s other authors, answered at pp. 35–39. The record holds Pereboom’s four-case argument (THEORY-328) and McKenna and Coates on manipulation (THEORY-040).
- Austin 1961 and Anscombe 1971, the source of the obstacle examples (p. 40).
- The 1996 book (LIT-1985, Unavailable), where the argument is developed.
Bearing on the record
- CLAIM-533: the attribution confirmed, and the defeater located. The chaos is Kane’s (p. 39). His reply is the one the claim’s defeater names, the indeterminacy as “a hindrance or obstacle to the attainment of the goal” internal to the effort (p. 42). But the reply is about responsibility, not control: Kane concedes that indeterminism lessens control (p. 43). So it does not defeat the claim, which is about control; it moves the question to whether responsibility can do without the control the claim says sensitivity does not supply.
- CLAIM-542: noise relabelled as the agent’s own input. In an SFA, Kane says, the noise hindering each choice “is coming from her own will” (p. 40). In the claim’s terms that moves the noise channel into the agent’s own channel: two of the agent’s efforts, each the other’s disturbance. Then whichever wins is the agent’s, by construction, and the claim’s question of the agent’s share against noise has no outside noise to measure against. That is Kane’s escape, and it holds only if the indeterminacy is all endogenous to the conflict, which the paper assumes.
- TERM-100: UR is formation provenance. Kane’s central question is whether “the sources of our wills” can “be completely traced backwards to something over which we had no control, such as Fate … or social conditioning or hidden controllers” (p. 37). That is the formation sense, and type-2 freedom, acting “of our own free will” from a will one formed, is provenance-dependent in exactly that sense. CASE-191’s result, that formation provenance leaves no trace in a self-maintained capacity, is why UR has to be a historical condition.
- TERM-099: Kane’s self-formation is not the record’s. Kane’s SFAs are undetermined choices that set the will; the record’s self-formation is the production of a maintained capacity, with nothing said of determinism. The term already says so; the text confirms it.
- QUESTION-071 and THEORY-328. Kane’s UR is a historical answer to manipulation; Pereboom’s four-case argument is the reply that such a history need not separate a manipulated agent from an ordinarily determined one. The paper does not answer it.
- No instruction for machine-learning practice.
Limitations
- The neural picture is a hypothesis. No evidence is given for conflict stirring up chaos, for its sensitivity to micro-indeterminacy, or for the two crossing networks.
- The analogies carry the argument. The assassin and the husband have one aim and an external obstacle; the SFA has two aims, each the other’s obstacle, and the paper asserts rather than shows that the analogy carries.
- A summary. The full argument is in the 1996 book and Four Views, not read.
Open questions
- If the indeterminacy is partly exogenous, as neural noise generally is, how much of an SFA’s outcome is the agent’s? The paper’s construction assumes the question away.