Contribution
The paper moves the question of adaptive or oppressed preferences in consent from the content of a preference, and from its bare causal history under injustice, to the mechanism by which it was formed. It names seven mechanisms (Sanctions, Beliefs, Association, Choice, Imitation, Amor Fati, Attention) and proposes one criterion to sort them: how much room, and how many good resources, the mechanism leaves for the person’s own free agency. The main new result is a verdict: preferences formed by Attention (extended contact with the object) can ground fully effective consent even when an unjust agent forced the contact. This correctly handles cases that Enoch’s (2020) historical account gets wrong.
Key insight
An unjust force that pushes you toward a preference leaves your consent effective when it only shapes what you come into contact with and leaves you free to value it. It diminishes your consent when it takes away the room to decide, as with sanctions, or the materials to decide well, as with a manipulated body of evidence. What matters is the agency exercised in forming the preference, not whether injustice was on the causal path. Because that mechanism is rarely visible, even to the person consenting, much ordinary reliance on consent is made under moral uncertainty.
Assumptions
Premises and framing (an argument by cases, not a formal model):
- Consent’s standard effect. In a textbook case, consent to X (i) makes X permissible for the recipient and (ii) makes a third party’s intervention to prevent X impermissible (p. 1). The standard defeaters are coercion, too few options, lack of information, impaired capacity and unwaivable duties. There can also be defeaters of those defeaters (p. 1, fn. 1).
- Scope. Only one defeater is studied: the provenance of the preferences on which the consent is based. The paper offers no theory of how this combines with the other factors (p. 2). Its claims are meant to cover promising, requesting, offering, gifting and joint projects as well as explicit consent (p. 2).
- Effectiveness is graded. Permissible acts differ in how good they are, and impermissible acts in how bad. “Diminished” effectiveness means changing the moral facts less, or needing more “scaffolding” (Kukla 2021) to make X permissible (pp. 8–9).
- The relevant injustice is external. It is “societal, governmental, or interpersonal” (p. 8).
- Judgments about the cases are data. The author says up front that he does not accept every permissibility judgment in the literature’s examples (p. 2).
- Empirical premises about preference change are cited, not tested. They include value change after moving country (Bardi et al. 2014), choice-induced preference change (Brehm 1956, with the Chen & Risen 2010 / Alós-Ferrer & Shi 2015 dispute noted in fn. 5), and the “first is best” nudge (Carney & Banaji 2012).
Key results
What it argues:
- §2.1 The perfectionist content-based account fails in both directions. (a) Consent can be ineffective although the preference’s content is admirable. The Angel in the House’s self-sacrifice would be praised in a privileged person. The variant that locates the defect in the unequal distribution of goods produced by everyone’s preferences fails too, because the self-sacrificing preferences are satisfied by that distribution (p. 3). (b) Consent can be effective although its object harms the consenter, within the limits of unwaivable duties (Hills 2003).
- §2.2 Enoch’s (2020) historical account is rejected. On that account a preference is nonautonomous if it was shaped by injustice toward the agent in a way “sensitive to the unjust circumstances being, well, unjust”. The counterexamples are Lone Activist (René comes to prefer solo campaigning under a regime designed to produce exactly that preference) and Pressured Parent (Holt, pressured into parenthood, comes to value it). Both satisfy Enoch’s condition, yet consent grounded in their preferences seems effective (p. 4).
- §3 Seven mechanisms (pp. 4–8). Attention works in three ways: (1) epistemic, where contact reveals features one already valued; (2) creative, where one brings the object under a normative concept, a valued non-normative concept, or a newly coined concept, and so partly fixes that concept’s extension by a free act of judgment (Murdoch 1970, Kant’s “free use of the imagination”); and (3) reweighting, where contact shifts the weights on dimensions of value. Murdoch’s M. and D. illustrates all three (p. 8).
- §4 First contention (p. 8). Consent is diminished for Sanctions imposed by unjust external forces and for Beliefs formed in an unjustly manipulated evidential environment. It is not diminished for Attention even when the contact was unjustly imposed. This also explains Enoch’s kidnapped pianist without appeal to “sensitivity to injustice”.
- §4 Criterion (p. 9). The less space and the fewer or poorer resources for the free exercise of agency in forming the preference, the more the effectiveness of consent is diminished. The degree of pressure also matters: mild scolding differs from ostracism. In Frankfurt-style cases, what counts is how the preference was actually formed, not how it would otherwise have been formed.
- Other mechanisms (pp. 9–10). Association can ground effective consent; its failures are really restricted-option failures (Gerver 2026). For Choice, the verdict follows one’s view of nudges: if nudges reduce free agency, Choice-formed preferences cannot ground effective consent, and if they do not (Thaler & Sunstein 2003), they can. Imitation can ground consent, except under forced cognitive overload combined with controlled exemplars. Amor Fati is not adjudicated.
- §5 Conclusion (p. 10). The moral facts are sensitive to features of preference formation that the person seeking consent often cannot see. So uncertainty about whether consent is effective is “ubiquitous and quotidian, and, as a result, very urgent”. This is posed as a puzzle and not resolved.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Consent’s effectiveness can be defeated by the provenance of the preference it rests on, even when the preference is real and reflectively endorsed | weak | informal argument from cases (§1: Food Inequality, Class Inequality, Angel in the House, Medical Treatment); the author himself flags that he does not endorse all the literature’s judgments |
| C2 | The perfectionist content-based account is wrong in both directions | moderate | counterexample (Angel in the House) plus the self-satisfaction reply to the distributive variant, §2.1 |
| C3 | Enoch’s (2020) historical, injustice-sensitivity account misclassifies cases | moderate | two counterexamples (Lone Activist, Pressured Parent), §2.2; the second is conceded to be “trickier” |
| C4 | Preferences formed by Attention ground effective consent even when the contact was unjustly imposed | moderate | intuitions about René, Holt and Enoch’s kidnapped pianist, plus the three-route analysis of Attention, §§3–4 |
| C5 | Preferences formed by unjust Sanctions, or by Beliefs from an unjustly manipulated evidential environment, diminish consent’s effectiveness | weak | asserted as the “first contention” (p. 8), supported by the §1 examples; no separate argument |
| C6 | The sorting criterion is the space and resources left for free agency in forming the preference | weak | offered as a “hypothesis” fitted to C4–C5 (p. 9); “free agency” is not analysed |
| C7 | Every mechanism named falls on one side of the division, and the paper explains why (abstract) | weak (overstated) | Association, Choice and Imitation are treated, the last two conditionally; Amor Fati, the mechanism the adaptive-preference literature is about, is never sorted |
| C8 | Moral uncertainty about the effectiveness of received consent is pervasive and urgent | weak | inference from C4–C6 plus the opacity of mechanisms, §5; no argument about how to act under it |
Method
Conceptual analysis by cases, in the style of analytic moral philosophy. The paper sets up examples, rejects two candidate accounts by counterexample, builds a taxonomy of mechanisms using cited empirical psychology and literary and philosophical illustrations (Woolf, Murdoch, Eliot), and fits a criterion to the verdicts. There is no formal apparatus.
Concepts
- Effective consent: consent that switches the moral facts as in the textbook case, (i) permission for the recipient and (ii) impermissibility of third-party intervention. It comes in degrees (pp. 1, 8–9).
- Defeater / defeater of defeaters: a factor that stops consent from having its standard effect, and a factor that restores the effect despite the defeater (p. 1).
- Mechanism of preference formation: one of Sanctions, Beliefs, Association, Choice, Imitation, Amor Fati or Attention (pp. 5–6). The paper says the list is not exhaustive.
- Attention: forming or changing a preference through extended contact with its object, by epistemic, creative or reweighting routes (pp. 6–8).
- Free exercise of agency (space and resources): the sorting criterion. “Space” is room to decide oneself whether to form the preference. “Resources” are, for example, access to alternative ways of thinking (p. 9). It is left unanalysed.
- Quandaries / hard choices: options that are incomparable or too costly to compare (Chang; Walden 2024), and so open to shaping through Choice (p. 6).
Connections
It builds on Enoch’s 2020 “False Consciousness for Liberals, Part I” (Philosophical Review 129(2)), which it corrects, and on Walker 1995. That Enoch paper is not held in this record; LIT-117 is Enoch’s earlier, unrelated “Agency, Shmagency”. The paper draws on the adaptive-preferences literature (Khader 2011, Elster 1983, Superson 2005), on nonideal consent theory (Kukla 2021, Tadros 2021, Gerver 2021/2026, Gardner 2017), and on Murdoch’s “The Idea of Perfection” for Attention. None of these is in the record.
Account of agency held. Agency here is a person’s capacity to freely determine their own preferences, and it can be exercised creatively: concepts are partly applied by fiat, and weights are reset. What limits it is the space others leave and the resources available. The account is gradualist (more or less space and resources) and actualist (Frankfurt-style counterfactual sanctions do not matter). Agency is not analysed, only used as the criterion that decides the cases. The agency tag is justified. The central thesis is a claim about when a preference is the product of the agent’s own agency, which is squarely “autonomy, control” in the tag’s blurb. The ethics and society-and-governance tags are justified too.
Bearing on the record
It supports no THEORY document in this record and contradicts none. It is the natural home for a future theory note on the autonomy of shaped preferences.
For ML practice it carries no instruction. It has conceptual relevance for anyone reasoning about whether preferences induced by recommender systems or persuasive AI can ground meaningful consent. On its criterion, systems that curate exposure (Attention) would be less damaging to consent than systems that manipulate the evidence a user sees (Beliefs) or exploit default ordering (Choice, via “first is best”). But the paper says nothing about machines, and that application would be an extrapolation. Nothing belongs in the Anthology.
Note of 2026-10-16: THEORY-040 now cites LIT-202 as a source.
Limitations
- The criterion is explicitly “rough” (p. 5) and a “hypothesis” (p. 9). “Free agency”, “space” and “resources” are not defined, so different readers can apply it to the same case and disagree.
- The abstract overstates what is delivered. Amor Fati, the paradigm mechanism of the adaptive-preference literature, is never classified, and Choice gets only a conditional verdict.
- Every verdict rests on intuitions about constructed cases. The author concedes that Pressured Parent is “trickier” and that he does not share all the literature’s judgments.
- There is no account of how provenance interacts with the other defeaters, which is deliberately out of scope (p. 2).
- The closing puzzle of moral uncertainty is raised and left unanswered: “Let me end in a rather unsatisfying way” (p. 10).
- The empirical claims about preference change rest on cited social-psychology findings, and the paper notes that one paradigm is disputed (fn. 5).
Open questions
- Where does Amor Fati (adaptive preference proper) fall under the criterion? One test: do preferences formed to avoid inevitable dissatisfaction leave “space” for agency?
- Can “space and resources for free agency” be made precise enough to sort nudges (Choice) without taking a side in the nudge debate?
- How should one act on consent under the pervasive uncertainty the paper identifies? Is there a moral-uncertainty decision rule, perhaps linked to Kukla-style scaffolding?
- Does an Attention-formed preference remain effective when the attended object is itself curated by the unjust force? This case, where Attention and Beliefs blur, is not considered.
Corrections to the seeded skim
- The dossier and the LIT summary omit the paper’s unifying criterion, which is its main positive thesis: what sorts mechanisms is “the space and resources those mechanisms leave for the free exercise of agency by the person forming the preferences” (p. 5; stated as a “hypothesis” on p. 9). The summary gives only the verdicts on Sanctions, Beliefs and Attention.
- The dossier says “Two Attempts” examines the historical account “next”, as though it were endorsed or left open. The paper rejects both: the perfectionist content-based account fails in both directions (§2.1), and Enoch’s (2020) historical account is refuted by two counterexamples, Lone Activist (René) and Pressured Parent (Holt), whose preferences are formed under injustice aimed at them and “perfectly sensitive to the injustice”, yet ground effective consent (§2.2, p. 4).
- The skim did not reach the other four mechanisms. The full text treats them as follows (§4, pp. 9–10). Association: the preference can underpin effective consent; where it is exploited (a toxic long relationship), the consent fails because the options are an unacceptably restricted package (citing Gerver 2026), not because of how the preference was formed. Choice: the verdict is conditional, following whatever one thinks about nudges and agency. Imitation: it can underpin consent, since offloading to others is a free choice, unless unjust cognitive overload forces the shortcut while the imitable preferences are also controlled. Amor Fati is listed (“what of Association, Choice, Imitation, and Amor Fati?”, p. 9) but never adjudicated, although the abstract promises to explain “why each mechanism falls on its side of the division”.
- The venue should read “Philosophy and Phenomenological Research 113(2): 251–261 (Crossref); the PDF read carries Early View pagination 0:1–11”. The PDF also shows a revision date (Revised: 13 May 2026) that the dossier omits.
- The dossier says “The Effectiveness of Consent” covers pp. 8–9; it runs pp. 8–10. It adds a Frankfurt-style case (p. 9: actual formation matters, not counterfactual formation) and Kukla’s “scaffolding” point (effectiveness is graded, and diminished consent may be made effective with scaffolding), neither of which the dossier records.
- The dossier’s observation that the printed keywords (“Bayesian | chance | logic | philosophy of mathematics | probability”, p. 1) are a publisher error is confirmed.
- The agency tag is justified, and more strongly than the dossier knew: the criterion that decides every case is a claim about agency.