Manipulation and the Demands of Agency

I. Introduction

Many philosophers think manipulation is wrong, when it is, because it induces a distinctive response in its target’s practical reasoning, describable in non-moral terms: for example, it circumvents, non-rationally influences, or subverts it. This is the Reasoning View. I previously argued that the Reasoning View is both over- and under-inclusive—in the sense that it over- and under-generates verdicts of wrongfulness—and that it misunderstands the sense in which wrongful manipulation threatens autonomy.¹ My Reductive View holds instead that manipulation is wrong, when it is, because it influences its target’s reasoning (though not by providing additional reasons) in a way that infringes one or more of their other rights. Wrongful manipulation threatens our autonomy in the sense that it invades the boundaries of our sovereign domain. Thus, the wrong of wrongful manipulation is non-basic: it is best explained in moral terms.

Massimo Renzo has recently challenged the Reductive View and defended a novel version of the Reasoning View that aims to avoid my objections.² For Renzo, wrongful manipulation is wrong because it undermines someone’s efforts to exercise their practical agency well; and exercising practical agency well requires not only doing what one has reason to do, but doing so for the right reasons. Here I argue that the Reductive View survives Renzo’s criticisms. I then engage Renzo’s positive picture. First, I argue that his reformulated Reasoning View solves the over-inclusiveness problem only at the cost of introducing a deeper, explanatory problem. I then argue that his view rests on a far more demanding picture of practical agency than he recognizes—one on which the causal history of our reasoning determines its adequacy—and, as a result, does not solve the problem of under-inclusiveness.

II. Responding to Renzo’s Objections

Renzo’s first charge against the Reductive View is that it counts too much conduct as wrongful manipulation. He offers the following case:

Gossip You know who my biological father is. I do not. Unwilling to let a good bit of gossip go to waste, you tell me, even though it is generally agreed in our society that this information should not be shared unless someone explicitly requests it. As a result, I am so distressed that I decide against better judgment to contact my biological father.³

According to Renzo, the Reductive View classifies your conduct in Gossip as wrongful manipulation, since you alter my reasoning by means that infringe my rights. Although Renzo agrees that you wrong me, he thinks your behavior is not properly characterized as wrongful manipulation, since it is not designed to induce a response in me.⁴

But the Reductive View is not primarily in the business of classifying conduct as manipulation, wrongful or otherwise. In fact, the view is compatible with virtually any account of what manipulation is. Suppose Renzo is right that manipulation requires a certain aim. Then defenders of the Reductive View should hold that wrongful manipulation is influence that aims at inducing a response in someone’s reasoning in ways that infringe their other rights. Questions would then arise about the aim’s content: must I aim at infringing your rights, or is it enough to aim at inducing a response, where my inducing that response infringes your rights? While interesting, these questions are outside the Reductive View’s scope.

Put another way, the Reductive View cannot be said to count too much (or too little) conduct as manipulation, for it does not count anything as manipulation (or not manipulation) in the first place. The Reductive theorist and her opponent share an interest in explaining the phenomenon of wrongful manipulation, but the Reductive theorist is content to let the opponent delimit the scope of manipulation however they like. Once some salient phenomenon is on the table, the only question the Reductive View answers is what makes it wrong, when it is. My earlier statement of the Reductive View may have obscured this point, for that formulation had descriptive content: it implied that manipulation alters someone’s reasoning, not by providing additional reasons for them to act. However, as I noted there, the definitional component of the view is dispensable.⁵ The view’s core contention is that wrongful manipulation is wrong because it infringes the target’s other rights. If your conduct in Gossip is not wrongfully manipulative because it is not manipulative, then the view renders no verdict about it, other than that it shares a wrong-making feature with wrongful manipulation—the same feature it would share if it improved my reasoning.⁶

Renzo’s second objection to the Reductive View amounts to the same thing. He offers the following case:

Deceased Mother During a business negotiation, your counterpart arranges for someone who bears a striking resemblance to your deceased mother to be present. This makes you so distraught that you cannot effectively exercise your practical reasoning.⁷

At first, Renzo objects that, although your counterpart’s act is wrongfully manipulative, there is no independent right it infringes. He quickly anticipates my reply: the relevant right is that against being gratuitously upset. We should understand this not as a right against the state of affairs in which you are upset, but as a right against your counterpart’s gratuitously upsetting you—the correlate of his directed duty not to so upset you.⁸ On the Reductive View, the wrongness of your counterpart’s act is explained by his infringement of that duty.

Anticipating this reply, Renzo shifts his objection. He reiterates the worry that, unless we assume that your counterpart’s act is designed to induce a specific response, it is not manipulative. He says that the Reductive View’s “conditions for wrongful manipulation are met,” yet there is no manipulation.⁹ But again, the Reductive View provides conditions for manipulation’s wrongness, not its classification. If your act is not wrongfully manipulative because it is not manipulative, the view renders no verdict about it.

As the discussion of Gossip and Deceased Mother clarifies, the Reductive View reduces the wrong of wrongful manipulation to the wrong of infringing other rights. In each case of wrongful manipulation, the act’s wrongness consists in some other wrong. Like other reductive projects, the Reductive View of manipulation invites the question of whether it explains away its target phenomenon or instead vindicates it by showing that it rests on firmer ground. I take it to do the latter. One may of course be skeptical of reduction in general, in which case one will suspect that the view has eliminated the wrong altogether; but that is a challenge to all reductions, not the Reductive View in particular.¹⁰

A true counterexample to the Reductive View would be a case in which someone’s conduct is wrong, and is manipulation, but does not infringe the target’s other rights. Renzo’s third objection aims to supply such a case:

Jealous Guy Yoko and Guy are exclusively dating. Yoko consents to Guy’s making her jealous, since she once did the same to him. Guy flirts with someone at a party, exclusively to make Yoko jealous. Flustered by his conduct, Yoko embarrasses herself at the party, making a scene by getting angry at the host for no reason.¹¹

In Renzo’s view, Guy’s flirting is wrongful—and wrongfully manipulative—insofar as making Yoko jealous serves no good purpose; however, it involves no independent wrong against Yoko. I disagree: Guy plainly infringes a further duty he owes to Yoko. Against this idea, Renzo claims that Yoko is not wronged by Guy’s flirting because she “has consented to this treatment” and thus “has waived whatever rights normally ground her claim against [it].”¹² But consenting to someone’s Φ-ing does not mean consenting to anything they do by means of Φ-ing. Yoko consents to Guy’s making her jealous, but not to his gratuitously bringing about what is bad for her. As her partner, Guy owes her concern regarding how she fares. This duty is outside the scope of Yoko’s consent. And by foreseeably causing her to embarrass herself for no good reason, Guy infringes it, even if he does this by doing something she has consented to.

One might object that, while Guy has a duty not to embarrass Yoko, he has no duty not to foreseeably cause her to embarrass herself.¹³ I do not see why. The concern Guy owes Yoko is concern for how she fares, including through her own actions. As her partner, he must consider how his conduct affects her, even when the causal path runs through her agency. Compare: I should not air a grievance I am entitled to air right before my partner’s important work dinner if I know it will rattle them into floundering in front of their colleagues. How they react is up to them; but if I can easily wait for a better moment, I should. Likewise for Guy: if he can foresee that making Yoko jealous right now will fluster her into embarrassing herself, and he could easily refrain from doing so, then consciously bringing about what is bad for her wrongs her.

Still, one might be skeptical that Yoko has any right against Guy’s foreseeably causing her to embarrass herself. Perhaps she waived the right, giving broad-scope consent to Guy’s making her jealous, whatever the foreseeable consequences. If so, then the Reductive View cannot say that Guy wrongfully manipulates her; but then, it is also less plausible that he does.

Or perhaps Yoko never had the right to begin with. (Perhaps the foregoing has been unconvincing.) In that case, the Reductive View in its original form cannot say that Guy wrongfully manipulates Yoko. However, there are various fallback positions that maintain the spirit of the Reductive View while vindicating the idea that Guy wrongfully manipulates Yoko. One might hold, for instance, that manipulation is wrong, when it is, because it alters someone’s reasoning in a way that independently wrongs them, regardless of whether it infringes a right. (This would require rejecting correlativity.) Or one might hold that manipulation is wrong, when it is, because it alters someone’s reasoning in a way that is morally wrong without wronging anyone, or is relationally defective without being wrong—as callousness is relationally defective. Surely, even if Yoko has no right against Guy’s gratuitously bringing about what is bad for her, there is something morally unsavory about his doing so—whether it wrongs her without infringing a right, is morally wrong, or is relationally defective. At the very least, he acts badly as a partner. The availability of these positions shows that the case puts no pressure on us to locate the wrong of Guy’s act in the quality of Yoko’s reasoning.¹⁴

III. Evaluating Renzo’s Positive Picture

Renzo’s proposal is a version of what I call the Quality of Reasoning View. In Renzo’s words, manipulation is wrong, when it is, because it “undermines (or perhaps negatively interferes with) its target’s efforts to exercise their practical agency well so as to adequately respond to their reasons for action.”¹⁵ His view has two distinctive features, designed to avoid my main objections to the Quality of Reasoning View.

The first is an appeal to convention. I argue that the Quality of Reasoning View over-generates verdicts of wrongfulness because it is often not wrong—not even pro tanto wrong—to undermine someone’s practical agency.¹⁶ This is most evident in adversarial and competitive contexts like sports and games, negotiation, litigation, marketing, and sales. Without wronging me, you can flatter me in a business negotiation or trash-talk me in a basketball game to subvert my practical reasoning. Attorneys can paint their witnesses in a favorable light to make jurors trust them more than is warranted; street vendors can use attractive salespeople to bring in customers; supermarket owners can strategically place products to boost their own profits.

From these cases, I conclude that we have no general obligation against subverting others’ reasoning, contra the Quality of Reasoning View. But that is not all. Although reasoning-subverting influence is often not wrongfully manipulative in these contexts, it can be. While it is fine to use flattery to induce practical mistakes during a negotiation, it is wrongfully manipulative to induce those mistakes by revealing someone’s private information to embarrass them or, as above, by gratuitously upsetting them with a deceased parent’s look-alike. Moreover, while an adversary may subvert reasoning through flattery, a loving partner or friend arguably cannot. From sets of examples like these, I conclude that the shape of the wrong of wrongful manipulation tracks an underlying landscape of rights. We have no general right against reasoning-subversion, but rather a right against others subverting our reasoning (or improving it) by means that encroach on our independently specified sovereign domain—a domain whose contours depend on relationships, among other things. That is what the right against manipulation amounts to, on the Reductive View.

Renzo draws a different conclusion: that although we have a general right against reasoning-subversion, it is sometimes pared back. Where we lack the right, we are not wronged by the conduct.¹⁷ A question arises: why do we lack the right when we do? Renzo joins me in rejecting explanations based on consent and forfeiture. As I argue, these tools are insufficiently fine-grained to explain the highly specific ways in which our rights are pared back in adversarial and competitive contexts. (Recall that we do not lack the rights wholesale, but with respect to certain tactics and against certain people.) Instead, Renzo appeals to social convention. As he puts it, “certain practices and conventions carve out exceptions in this general right and its correlative duty.”¹⁸ I lack a right against your subverting my reasoning with trash-talk in a basketball game because such behavior is conventionally tolerated in basketball.

Renzo’s solution is attractive. Among other things, it explains why the shape of our right against manipulation is largely insensitive to what particular individuals want or take it to be. Nevertheless, it faces a problem. If the Reductive View is right about which cases constitute wrongful manipulation,¹⁹ then we observe a remarkable pattern: across a diverse range of contexts, our social conventions systematically carve out exceptions to the general right against reasoning-subversion in precisely those cases where reasoning-subversion would not infringe other rights. This striking coincidence demands explanation. Why do our conventions perfectly track an independent order of rights?

The problem is not that our conventions are binding only if we can independently explain their particular shape. Renzo’s view may well solve the over-inclusiveness problem, even without such an explanation. The problem is that it does so at the cost of introducing an explanatory bug. It posits a general right whose contours are determined by social convention, but whose contours just happen to track an independent order of rights. On the Reductive View, it is no mystery that subverting reasoning is wrongfully manipulative exactly where it infringes other rights, for the wrong of wrongful manipulation is nothing over and above its infringing those other rights.

One might reply that our conventions are frequently shaped by morality. Facts about what counts as a rights-infringement may shape our conventions, which in turn shape our right against reasoning-subversion. This response avoids the coincidence problem but creates a new one: if underlying facts about rights do the real explanatory work, then why posit a general right against reasoning-subversion at all? The moral facts alone can explain why reasoning-subversion is wrongfully manipulative when it is—as on the Reductive View—without requiring a detour through a general right whose contours are modified by convention. Absent independent grounds for believing in such a right,²⁰ we should prefer the more parsimonious Reductive View.

Renzo then addresses the under-inclusiveness problem. Contra the Quality of Reasoning View, I argue that one can wrongfully manipulate someone by improving their reasoning. The Reductive View provides a formula for constructing examples: imagine that someone’s reasoning is bolstered in a way that independently wrongs them. My central case is this:

Phone Call India and Sebastian are discussing how to share childcare duties. India, who tends to undervalue her needs, is likely to agree to an unequal distribution in Sebastian’s favor. Knowing that India tends to value her needs more appropriately when she is angry, Sebastian intentionally angers her by ensuring that she overhears a phone call in which he divulges to his mother some of the marital issues they have agreed to keep private. This angers India, and she appropriately values her needs in their conversation, advocating an equal distribution of responsibilities.²¹

I claim that Sebastian improves India’s reasoning. He makes her appropriately weigh her needs and advocate an equal labor distribution on grounds that support doing so. Renzo disagrees. He thinks Sebastian gives India good reason to be angry; however, since being angry about someone’s indiscretion does not rationally support valuing one’s needs more, he finds that India responds inadequately to her reasons.²²

Renzo treats this case as symptomatic of a deeper disagreement between us. Following Raz, he distinguishes between “conforming” with reasons—performing the action one’s reasons call for—and “complying” with reasons—performing that action for the reasons that make it called-for. Renzo thinks that India conforms but does not comply with her reasons, so Sebastian does not improve her reasoning. Quality of Reasoning theorists can therefore overcome the under-inclusiveness problem by recognizing that both conformity and compliance are necessary for adequately responding to reasons.

Renzo’s diagnosis is puzzling. Let us grant that good reasoning requires compliance, not mere conformity. Phone Call was meant to be a case in which India not only conforms but complies with her reasons, advocating a fair distribution on grounds that it is fair. The point was that our moods and emotions affect which reasons we are moved by. (This phenomenon is perfectly general: I may be more readily moved by mathematical proofs when I am happy because this makes them more pleasant to follow.) If Sebastian knows that India is more readily moved by considerations of fairness when she is angry, then he can change her decision by making her angry. This does not mean that she advocates an equal distribution of labor on grounds that she is angry.²³

The real source of disagreement between Renzo and me is deeper. We should distinguish between the reasons for which someone acts and the reasons why she acts for those reasons. India’s reason for advocating an equal labor distribution is that it is fair; the reason why she is so moved is that she is angry. Renzo overlooks this distinction. Consider his discussion of the warning labels on cigarette packages. He claims that his view classifies some labels as wrongfully manipulative and others not, depending on their design. Brightly colored labels are wrongfully manipulative, he says, since “the mere fact that the label is, say, bright red, rather than grey or brown, does not provide one with a reason to quit smoking,” whereas labels with gruesome pictures of lung disease are not wrongfully manipulative, since “the revulsion smokers might feel at the sight of those pictures is a good reason to quit smoking.”²⁴ In his view, the key difference is that smokers who see red labels do what they have reason to do—quit smoking—but for reasons that do not support quitting.

But a smoker who quits because they see a red warning label does not typically quit on grounds that the label is red. Rather, their reason for quitting is the content of the label itself—that smoking causes lung disease. Perhaps they would never have noticed the label if it had been grey or brown. If so, then the reason why they quit on those grounds is that the label is red. Likewise, the smoker who quits because of a gruesome label quits on grounds that smoking causes lung disease, because the label gruesomely depicts it.

With this distinction in view, we can reformulate Renzo’s proposal. He seems to think that, when someone (Φ-s on the ground that P) because Q, whether they reason well depends partly on the content of ‘Q.’ Whether a smoker reasons well depends on why they are moved to quit by the prospect of lung disease; whether India reasons well depends on why she is moved by reasons of fairness. As a first pass, we might say that good reasoning requires Q to be a reason to Φ. But as a general requirement, this condition is clearly implausible. Suppose I believe it will rain on grounds that the weatherman said so, and I trust him because of his focused demeanor. His focused demeanor is not a reason to believe it will rain; nevertheless, I reason perfectly well.

Instead, we should interpret Renzo as saying that Q must rationally support (Φ-ing on the ground that P). Perhaps the fact that depictions of lung disease are gruesome rationally supports being moved by the prospect of lung disease, while the fact that those prospects appear on a red label does not. Returning to the original case, Sebastian subverts India’s reasoning because he leads her to be moved by reasons of fairness only because she is angry, where her anger does not rationally support being moved by those reasons. Renzo acknowledges that his account of manipulation comes at a cost: that it “rests on a fairly demanding picture of the kind of practical reasoning required to exercise our practical agency well.”²⁵ But the picture is far more demanding than he recognizes. It requires embracing not only that good reasoning entails compliance with reasons, but that good reasoning is a matter of the reasons why we comply with our reasons.

Should we accept this demanding requirement? On the one hand, its general structure is common in moral psychology. Consider action from virtue: the courageous person defends his city not because such acts are courageous, but because his city needs him—yet it is crucial that he is moved by this consideration because he is courageous. We may be less sure what to say if he lacks a stable disposition toward courage and instead simply happens to be moved on this occasion by the needs of the city. On the other hand, despite its familiar structure, Renzo’s requirement differs crucially. On his view, the content of ‘Q’ concerns external facts about the world, not the principles to which an agent is committed or her stable character dispositions. Thus, despite the structural parallel, the analogy remains unclear.

Moreover, Renzo’s view makes good reasoning exceedingly demanding. Our reasoning is pervasively shaped by factors that do not rationally support it: museum visits make me more receptive to aesthetic arguments; good moods make me more receptive to practical advice. His requirement therefore threatens to make good reasoning vanishingly rare. Worse, the demand iterates: if the reasons why we act for good reasons must themselves be grounded appropriately, regress threatens, and virtually all psychological interventions undermine reasoning quality.

Renzo attempts to address this problem in responding to a variant of Phone Call:

Happy Place India tends to stress out about conversations regarding childcare duties and therefore to make mistakes in prudential reasoning. To combat this issue, Sebastian ensures India is calm before their discussion. He puts on relaxing music, avoids distressing topics, and recalls happy memories. As a result, India relaxes and gives proper weight to her interests, advocating an equal labor distribution on grounds that it is fair.²⁶

Sebastian does not wrongfully manipulate India in Happy Place. But India bases her proposal on reasons of fairness because she is relaxed; and being relaxed does not rationally support being moved by reasons of fairness. Does Sebastian therefore impair India’s reasoning? Renzo claims not: he says that Sebastian merely “aims at removing psychological impairments that get in the way of India’s efforts,” rather than “using India’s anger as a motivational force.”²⁷ That is an acceptable answer, as far as it goes. But it raises the question of why India’s baseline psychological state—her tendency toward anxiety—should count as an impairment, while her anger-induced clarity should count as a faulty response. Both states influence which considerations move her, and both are shaped by factors (her personality, Sebastian’s interventions) that do not rationally support her motivation. Renzo provides no principled way to draw this line.

Perhaps Renzo can avoid the problem at hand by focusing on better and worse reasoning, rather than good or adequate reasoning.²⁸ I have argued that Renzo is committed to thinking good or adequate reasoning requires not only conforming to and complying with one’s reasons, but doing so because of some factor that itself provides rational support for being so moved. This makes good reasoning vanishingly rare. What if Renzo instead made this a requirement of perfect or ideal reasoning? It is no embarrassment to say that perfect reasoning is vanishingly rare. He could then hold that manipulation is wrong, when it is, because it makes its target’s reasoning worse, or brings it further from perfection.

The trouble with this proposal is that reasoning which fails Renzo’s condition need not fall short of perfection at all. Consider Happy Place: being relaxed does not rationally support being moved by reasons of fairness, yet India’s reasoning is not merely adequate—it is ideal. If one insists that Happy Place involves imperfect reasoning, then a second problem arises. What makes it imperfect is that India’s motivation is causally shaped by a factor—her emotion—that does not rationally support her being so moved. But we can never avoid being shaped by such factors. Thus, the standard for perfection does not just make perfect reasoning vanishingly rare—it makes it practically impossible. A standard for perfect reasoning that no actual reasoning could satisfy is not a very attractive ideal.²⁹

This debate mirrors epistemological debates about the significance of the causal history of our beliefs. Our moral, religious, and political convictions would differ dramatically with different upbringings or evolutionary pasts. It is natural to think that, if our belief-forming processes—like cultural conditioning and natural selection—are insensitive to truth, we should lose confidence in our beliefs.³⁰ However, specifying which causal information raises these worries is notoriously difficult. Critics argue that any attempt to do so imposes impossibly demanding standards, requiring us to step outside our own cognitive perspective to validate our fundamental reasoning methods—a requirement that, if consistently applied, would undermine virtually all our beliefs.³¹ Likewise, Renzo’s causal-history requirement imposes an impossibly high practical standard, threatening to undermine virtually all our practical reasoning.

Escaping the under-inclusiveness problem therefore requires settling foundational questions that extend far beyond the manipulation literature and have proven notoriously difficult. Why does India reason inadequately in Phone Call but adequately in Happy Place? Without a principled way to draw this line, Renzo’s view is far too demanding. This creates tension within his broader project: if good reasoning is as demanding as his view suggests, it is hard to see how we could be obligated to protect others’ reasoning quality. We are not, for example, generally obligated to protect others’ virtue. Indeed, the more demanding the standard for adequate reasoning is, the less obvious it is that we have even a general duty not to act in ways that predictably undermine it.

Acknowledgments

Thanks to Massimo Renzo for helpful comments on the first half of this paper, much of which appeared as a contribution to the blog PEA Soup. Thanks also to Kieran Setiya, Bennett Eckert-Kuang, Julian Perilla, and Quinn White for useful conversation and feedback on earlier drafts, and to two anonymous reviewers for their insightful comments.

Competing Interests

The author has no competing interests to declare.