NOTE-847: Harland — Beyond the Moral Influence Theory?

Contribution

A critique of Manuel Vargas’s Building Better Beings (2013). Vargas presents his agency cultivation model (ACM) as a rehabilitated moral influence theory (MIT) that escapes MIT’s objections by aiming praise and blame at a valuable form of agency: responsiveness to moral considerations. Harland argues two things. ACM is MIT under another name, because the agency it cultivates is just a disposition to behave morally. And Vargas’s counterfactual test of free will is epistemically redundant and inverts the relation between freedom and responsibility.

Key insight

If what makes an agent “free” is defined by what a well-functioning blaming practice needs, then freedom cannot be a reason to blame. To know whether S satisfies the test, you must already know whether blaming people like S is good policy. Once you know that, the test adds nothing. In Harland’s image, Vargas’s free will is “the whistle on an engine”: “it is actually our blameworthiness that makes us free”.

Assumptions

• Harland’s stipulations, not Vargas’s definitions. “External motivation is left undefined but we might stipulate the following”. He stipulates external motivation, intrinsic motivation and “Responsiveness to Moral Considerations: A person is responsive to some moral consideration M to the extent that he is intrinsically motivated to act in accordance with M” (pp. 407–408). The identification is supported by passages of Vargas (2013: 203, 208–209, 217), not quoted from him.
• Vargas’s own framing of the test. The similarity and proportion parameters are fixed by “the standards an ideal, fully informed, rational, observer in the actual world would select as at least co-optimal for the cultivation of moral considerations-responsive agency” (Vargas 2013: 214, quoted p. 418), and responsibility is “a function of what the justified norms of moral influence say” (Vargas 2013: 103, footnote 11).
• Classes of agents already sorted by their response to blame. The redundancy argument proceeds by classes: psychopaths, kleptomaniacs, boisterous boys, pupils with “rotten social backgrounds”, religious bigots, alcoholics. It assumes we can know how a class responds to blame before applying the test.
• Deontology as Nozick’s side-constraint view, and virtue as requiring “proper reasoning” (Timpe and Boyd), for the argument that non-consequentialists will not adopt ACM (§2.2).

Key results

• ACM’s replies to MIT’s objections are available to MIT (§1).
  • Deontological side constraints can be added to either.
  • The objections from conflating being and holding responsible, from the phenomenology of the reactive attitudes, and from backward-looking blame (Klaus Barbie) get replies that “can just as easily be made by supporters of MIT”.
  • Vargas’s claim that a future-oriented justification is “all the license we can hope for” (2013: 194) is at odds with his opening claim that praise and blame make sense “partly because of” their effects (2013: 3–4).
• The John and Fred cases (§2.1). John internalises, through fear of hellfire and ostracism, an aversion to homosexuality he cannot justify. Fred internalises, through punishments, an aversion to racism he cannot justify (“They just should!”). Both end intrinsically motivated. Vargas must deny John enhanced agency, since his norm “carries no genuine moral force”. But Fred came by his true norm the same way, so he “remains unworthy of the label moral agent”: “There is an inverse relationship between moral agency and the susceptibility of one’s moral convictions to external forces.” Fred is “a moral drifter”. Even so, “there are obviously good reasons for holding them accountable”; what fails is the claim that doing so “promotes their moral agency”.
• The dilemma for a richer notion (§2.1). A Dworkinian notion of moral agency (scrutinising one’s convictions against one’s others, the facts and logic) would distinguish ACM from MIT, but it “requires mental fortitude, intelligence, and creativity”, which praise and blame are not shown to enhance. “The richer the conception of agency employed, the less tenable ACM becomes.”
• ACM draws the line where MIT does (§2.2). Vargas’s example of moral responsiveness, perceiving “when a quiet friend is in need of consolation” (2013: 208), is shared by chimpanzees, bonobos, rats (as cited from Churchland) and infants of 18–30 months. So ACM, like MIT, counts some animals and toddlers as responsible agents and fit for “wholehearted resentment”.
• A defence of ACM (§2.2). Against Elzein, ACM does not treat anyone merely as a means, because it justifies a system whose benefits reach everyone it blames, following Kramer on deterrence.
• Why deontologists and virtue ethicists will not adopt ACM (§2.2). It grounds desert in consequences (against Pereboom’s basic desert and Moore’s retributivism); it invites Hegel’s “raising one’s stick at a dog”; it cannot separate blame from milder brainwashing or pills; and it treats compliance with duties as an end to maximise, a “utilitarian of rights” in Nozick’s phrase. “His account does not merely share MIT’s shortcomings; it just is MIT.”
• Vargas’s test (§3.1). S is a responsible agent for considerations M in circumstances C if S has the basic agential capacities and (A) the capacity to detect M (S actually detects it, or does so “in a suitable proportion” of worlds relevantly similar to C) and (B) the capacity for self-governance with respect to M (S is motivated accordingly, or is so in a suitable proportion of relevantly similar worlds, where S detects M, becomes motivated by it and acts accordingly) (Vargas 2013: 213–214). Free will is (A) and (B) jointly. Relevance and proportion are fixed by what is “co-optimal” for “fostering agency that recogni[s]es and suitably responds to moral considerations” (2013: 219–221): “a Goldilocks notion of capacity” (Vargas 2018: 120).
• Epistemic redundancy (§3.2). “We cannot determine how Jones would have behaved in the relevant circumstances until we have decided whether it is good policy to blame similar people for similar wrongdoings … once the matter is settled, we will be left with no reason to engage in any counterfactual reasoning.” “The work necessary to give the counterfactual test meaning makes that very test redundant.”
• Conceptual inversion (§§3.2–4). “A wrongdoer’s blameworthiness is logically prior to his free will.” “If freedom simply means belonging to a group of people whom it is good policy to blame, it is not really freedom in any recognisable sense of the term.” The scope is stated: “In short, if we embrace ACM, enquiries into an agent’s counterfactual freedom will become superfluous when considering the ethics of holding him responsible.”
• The diagnosis (§3.2). Vargas wants the test to do two jobs: pick out agents likely to be improved by blame, and preserve the folk intuition that responsibility requires the ability to do otherwise. His “philosophical conservation” principle is “entirely pragmatic”, based on the costs of revising beliefs. The result is “a much more profound shift in folk thinking than he recognises”.
• A note on psychopaths (footnote 9). On Vargas’s capacious notion, psychopaths who adjust conduct to prudential concerns, and might gain an affective drive against wrongdoing through blame, should count as moral agents, though Vargas exempts them.

Claims

id	claim	strength	support
C1	Vargas’s replies to the objections to MIT are equally available to MIT	moderate	§1, objection by objection
C2	Vargas’s responsiveness to moral considerations amounts to intrinsic motivation to act in accordance with them	moderate	a stipulation supported by Vargas’s capacious passages (2013: 203, 208–209, 217)
C3	Agency formed by internalising social pressure is not moral agency, even when the norms are true	weak	the John and Fred cases; intuition, set against a Dworkinian standard
C4	ACM and MIT draw the boundary of responsible agency in the same place	moderate	animal and infant examples, via Churchland and developmental studies
C5	Most deontologists and virtue ethicists cannot adopt ACM	moderate	Nozick’s side constraints; the virtue literature; the manipulation argument
C6	Given ACM, the counterfactual test is epistemically redundant: applying it presupposes the policy verdict it is meant to deliver	moderate	the parameter-fixing passages of Vargas 2013, 214–221, and worked class examples, §3.2
C7	Given ACM, ascriptions of free will follow blameworthiness rather than ground it	moderate	follows from C6 and Vargas 2013: 103
C8	Inquiry into freedom is irrelevant to the ethics of holding responsible	weak as stated in the abstract; moderate when restricted, as the body restricts it, to ACM and counterfactual freedom	§3.2

Concepts

• Moral influence theory (MIT). Responsibility practices are vindicated by their tendency to strengthen dispositions to behave in morally desirable ways (Nowell-Smith, Smart, Schlick, Dennett, Arneson, McGeer).
• Agency cultivation model (ACM). Vargas: moralised praise and blame “are justified by their effects, that is, how they develop and sustain a valuable form of agency”.
• External and intrinsic motivation. Motivation to comply with a norm that does, or does not, depend on expecting beneficial or adverse personal consequences. Harland’s stipulations, after Sripada and Stich.
• Moral consideration. Vargas: “a consideration with moral significance such that, were one to deliberate about what to do, it ought to play a role in those deliberations”; it may be “largely or perhaps exclusively affective”, and its detection “need not be conscious”.
• Moral drifter. An agent whose moral convictions are the product of “unthinking capitulation to social pressures”, whether or not they are true.
• Counterfactual freedom. Vargas’s free will: conditions (A) and (B) above, circumstantial (relative to M and C) and scalar (footnote 8).

Connections

The target is Vargas (2013) and his reply to critics (2015). The paper sides with Elzein (2013) against ACM on desert but against her mere-means objection, and engages McGeer (2015), Capes’s review (2016), Pereboom (2014) on basic desert, Fischer and Ravizza (1998), whose epistemic and control conditions Vargas’s two parts track, and Strawson’s “Freedom and Resentment”. None of these is held here except through other readings of the free-will line. Harland writes from the Kramer school (acknowledgements; Kramer on rights and on capital punishment).

Bearing on the record

• QUESTION-068 (Deferred; no answers: relation). It asks whether settling whose a will-organization is would settle accountability, or whether accountability needs conditions, such as normative competence and avoidability, that ownership does not supply. The paper bears on it in two ways, both indirectly.
  • Avoidability, on an influence account. Harland argues that once responsibility is justified by its effects, an avoidability condition calibrated to those effects is not an independent condition. It is fixed by the policy question. So on such accounts the question’s second option, a separate avoidability condition, collapses into the ethics of holding responsible. This is the account filed as THEORY-310. It does not touch accounts on which avoidability is a metaphysical condition, as in THEORY-118 and THEORY-119.
  • Ownership against moral agency. Fred’s norm is internalised and intrinsically motivating, yet Harland calls him “a mere receptacle of social mores” and still finds “good reasons for holding” him accountable. That is a candidate for the converse case “What would answer it” asks for: accountability where the will is arguably not the agent’s own. But the reasons given are forward-looking, the verdict is asserted rather than argued, and Harland does not discuss ownership. It does not survive the dispute between THEORY-118 and THEORY-119, because it uses neither.
• LIT-991’s filing note overstates in two places.
  • It says the paper criticises a responsiveness condition “stated apart from whose the will is”. The paper never raises ownership. Its complaint is that Vargas’s responsiveness is too capacious to distinguish moral agency from a disposition to behave morally.
  • It says the paper argues “that freedom does not bear on holding someone responsible, which cuts against the question’s ownership side”. The abstract says so unconditionally, but the body restricts it: “if we embrace ACM, enquiries into an agent’s counterfactual freedom will become superfluous”. It is a claim about Vargas’s freedom inside Vargas’s model, and it says nothing about ownership, so it does not cut against the ownership side.
• THEORY-122. Harland’s Dworkinian alternative, moral agency as reflective scrutiny of one’s convictions, is a normative-competence condition of the kind THEORY-122 holds that identification does not supply. The paper offers it as a dilemma for Vargas, not as its own view.
• THEORY-147. Harland separates holding someone accountable (there are “good reasons” to hold John and Fred accountable) from their being moral agents. That fits THEORY-147’s separation of accountability from other kinds of responsibility, but Harland gives no positive account.
• CLAIM-022. Unaffected. The paper says nothing about practical efficacy or ownership.
• No instruction for machine-learning practice; nothing for the anthology.

Limitations

• Second-hand Vargas. Every claim about ACM rests on Harland’s quotations and his stipulated reconstruction of responsiveness. Vargas 2013 and the 2015 reply are not held here. Footnote 7 records that McGeer read Vargas the other way (capacities “ontologically prior to, and conceptually independent of” their role in the system), and that Vargas rejected McGeer’s reading.
• The redundancy argument assumes the classes. It works class by class, with the class’s response to blame already known. On the reader’s view, not the paper’s: if the counterfactual test is what sorts individuals into classes, or if the parameters are calibrated once for the whole system and then applied to individuals by facts about them, the test keeps work to do. Harland’s teacher example calibrates per class (“boisterous boys”), which is the case most favourable to him.
• Thin empirical support. The animal and infant cases come from secondary sources (Churchland, Goodall, Bartal et al., Zahn-Waxler et al.). The reference for Bartal et al. gives Science volume 344; the article is in volume 334 (2011).
• The John and Fred cases rest on intuition about what moral agency requires, set against a Dworkinian standard that Harland does not defend.
• The abstract overstates the conclusion, as noted above.

Open questions

• Does Vargas’s 2015 reply, or later work, answer the redundancy argument, for instance by fixing the parameters once at the system level? Reading Vargas 2015 (Philosophical Studies 172:2659–2678) would settle what the test is meant to do.
• Is there a forward-looking account whose capacity condition is not calibrated to the system’s aims, and so escapes the inversion?