1

Manipulation is, and has always been, a pervasive feature of human life. And yet, the attempt to understand what manipulation precisely is and what role it may legitimately play in our lives has proven very difficult. Perhaps this should not be surprising: that which surrounds us on all sides is often most difficult to discern and evaluate properly. To make things yet more difficult, manipulation is also a tremendously diverse phenomenon – and this too adds to the challenge of apprehending it. But, is it not the case that we all know manipulation rather intimately? Indeed, we do; but this too is no recipe for easy understanding: the things most difficult to comprehend are those farthest from our minds, but no less those closest.

The succession of daily manipulations begins the second you wake up. Let us conjure up an ordinary morning. You have set the time on the alarm clock to be 10 minutes fast, to create a false sense of urgency that motivates you to rush out of bed. You get up and brush your teeth. The toothpaste contains an inert whitening material (titanium dioxide), inserted with the sole purpose of creating a connotation of cleanliness, so that you form a subconscious belief that a white toothpaste would clean your teeth well – indeed, make them white. You turn on the morning news. The choice, order, and framing of items are determined by various agendas – agendas that are mostly never announced, yet shape your views in conformity with them. On the news you see the country’s leader speaking. The precise wording of his address, his intonation and emphases, the background behind his podium, the precise timing of his address, and so on, were all meticulously doctored by professionals to have maximal psychological impact on you. The news is interrupted by advertisements that subconsciously affect your interest in random products, by repeating their brand names in combination with stimuli that research indicated you should find pleasant. You get dressed, choosing the clothes that project the image of yourself that you want to impress on potential collaborators you will meet this morning at work. Opening the morning mail, you find in it letters from various organizations soliciting your support, by using pictures and stories that – while taking care not to be too obvious – intend to stimulate pity, alarm, or other feelings, that invoke your generosity. Your little daughter refuses to change from her pajamas; to solicit her cooperation, you tell her that if she can dress faster than you, it will show that she is now finally “truly a grown up.” Your daughter’s unexpected refusal to wear anything but pajamas takes up too much of your time, so you text-message your colleague, informing her you’ll be late to your meeting. However, knowing her sensitivities, and concerned that she might feel disrespected, your text-message provides a somewhat more dramatic description of the cause of your delay. You are now finally ready to leave the house, when you remember to apply deodorant that, while concealing your true smell, was successfully marketed with the promise to express “the true you.” In the stairs, you pass by your always-grouchy landlord. Cognizant of the importance of keeping on his right side, you manage a big smile and a cheerful “Good morning!” It is but an ordinary morning, and you are an ordinary person. It has not been an hour since you woke up, yet you have been immersed in one manipulation after another. Manipulations engulf us.

As this depiction of waking into a daily stream of manipulations intimates, manipulations are a pervasive feature of almost every aspect of social life. Parents manipulate children to defuse irrational defiance, and children manipulate parents to get their way. Romantic relationships produce creative manipulations to woo and impress one another. Doctors manipulate patients to comply with important treatment, and patients manipulate doctors to get the prescriptions they seek. Lawyers manipulate witnesses, jurors, the court, and each other. (They could hardly do their job well otherwise.) Advertisers’ main task is to excel at manipulating their target audience into preferring the products they promote. Shops routinely use manipulative marketing tactics to boost sales; and much website design is an attempt to manipulate users to remain on a webpage for as long as possible. Indeed, to the extent that consumerism is unavoidably manipulative, our culture is manipulative in its core. Employers manipulate workers toward greater efficiency, and workers manipulate employers to increase benefits. Coaches manipulate athletes to extract maximum performance, and athletes manipulate coaches to go easy on them. News media manipulate their readers, viewers, and listeners by presenting as objective news that is biased by agenda, by responsiveness to what their subscribers are willing to ingest, by tit-for-tat relations with their sources and interviewees, by financial pressures, and so on. Press secretaries, charged with the vital role of increasing government transparency, manipulate as a matter of principle, so as to always present policies as just, important, and timely. Politicians routinely manipulate the public, the civil service, journalists, and each other. And we all manipulate ourselves to stick with a diet, overcome fears, maintain a good view of ourselves, or what have you. This is obviously no more than a sample.

Manipulations are not only prevalent in human interactions, they also have significant, and often decisive effects on what we all end up believing, choosing, doing, and feeling. We should therefore want to understand them properly. This book will offer a theory of manipulation. The philosophical analysis of manipulation involves two main questions: (1) What is manipulation, or: what does “manipulation” mean? (2) What is the moral status of manipulation, and by virtue of what does manipulation have the moral status it has?

The second part of the book is dedicated to the latter question. This basic moral question about manipulation calls for special concern given the intriguing observation that there is perhaps no other behavior that is both highly morally suspect and yet practiced by everyone, mundanely, and in infinite ways. It is therefore incumbent upon us to evaluate sensitively the role that manipulation plays in our lives. Would our lives be better or worse if manipulation were virtually eradicated (except as last resort)? How realistic is it for society to proceed efficiently without manipulation? If it is not, is this a human tragedy? If it is indeed a tragedy, do we nonetheless have a moral duty to endure it? These are rather mighty questions, and we will set them aside until the second part of the book. In the first part, starting now, we focus on the first question: what exactly is manipulation?

2

Manipulations, or manipulative actions, are a subset of all actions aimed at influencing people. They aim at influencing our beliefs, preferences, intentions, choices, desires, attitudes, emotions, or moods in the direction desired by the manipulator. This much is more or less uncontroversial. The tough question is how precisely to delineate the subset of influences that comprises all and only manipulations. (Delineating the precise subset is certainly not made easier by the fact that the initial set – influence – is such a fundamental and extensive category of human interaction that it can be interpreted as covering all of human communication and much causal interaction besides.)

The term “manipulation” is normally used in two principal senses. The first sense – a direct reflection of the etymology of the word – is that of “handling” (from the Latin manus = “hand”). In this sense, we say, for instance, that the orthopedist manipulated the fracture or that the worker manipulates heavy machinery. The second sense refers to influencing the psychology of others. While the first sense can apply to objects, the second sense is exclusively interpersonal or social. It is this second sense that is the focus of this work. The first sense will interest us only to the extent that it can help shed light on the second.

In a seminal philosophical paper focusing on manipulation, Joel Rudinow wrote: “I have encountered no one who has not immediately known what sort of thing I have in mind when I talk about manipulation between persons and who has not also had examples of it ready at hand.”Footnote ¹ This certainly rings true, but what we are seeking now is to understand the concept that underlies all such examples and our intuitive grasp of them.Footnote ² One central source of difficulty in pinning the concept down is the formidable diversity of forms of manipulation. “Manipulation” can refer to such actions as: conditioning, incentivizing, changing availability of options, providing irrelevant inputs or crowding out relevant inputs, framing information, keeping someone in the dark, insinuating, using leading questions, using “code words,” using ambiguous language, invoking imagery, exaggerating, invoking anxieties, impressing, charm offensives, wearing the other down, stroking the other’s ego, ignoring the other, invoking temptations, making someone uncomfortable to refuse, inciting, precipitating guilt trips, saying what the other wants to hear, advancing (plausible) arguments that the manipulator herself does not adopt, devising elaborate plots in which the target has a determined role, reinforcing false beliefs, and the list continues. Now it is natural to wonder what one concept could unify such an extensive and heterogeneous array of interpersonal phenomena.

Given manipulation’s prominence in human life and its tenuous moral status, it merits systematic reflection – in particular, philosophical scrutiny. Yet, considering the magnitude of the phenomenon, philosophical literature on manipulation has been relatively scant. One can of course find historical discussions of the phenomenon of manipulation – Aristotle’s Rhetoric or Machiavelli’s The Prince easily come to mind – but this is different from the attempt to explain the precise meaning of the concept “manipulation” as well as manipulation’s moral status. In a sense, we could not expect much historical analysis of “manipulation” before the term even existed, and the first recording of it in English (in the relevant sense of psychologically controlling or maneuvering other people) is as recent as from 1828.Footnote ³ First philosophical papers on the meaning of manipulation started appearing only in the 1970s,Footnote ⁴ and they remained a trickle until the 2000s. In the first decade of the new millennium, Richard Thaler and Cass Sunstein developed their influential work on “nudging” in public policy,Footnote ⁵ and this arguably gave a push to research on manipulation in philosophy as well. And yet, the philosophical field is still in its infancy (there is to date only one philosophical monograph in English that attempts to provide a general account of manipulation, albeit very partial in scope,Footnote ⁶ and but one anthology of papers).Footnote ⁷ This book’s task is to provide a comprehensive philosophical theory of manipulation.

The first step in constructing a theory on the meaning of manipulation is to review the state of the art of thinking about the concept. This should help us get some sense both of what advancements in understanding have been made and of what remains to be achieved. How should we go about doing this? Reviewing critically every attempt at definition found in the literature would be needlessly technical, tedious in the extreme, and might actually obfuscate the larger picture. Instead, what seems more reasonable is to review the main perspectives on the meaning of manipulation (then, from the angle of each perspective, we can discuss select definitions). This will give us important orientation as to the state of current understanding and, no less importantly, present an overview of the enduring puzzles. My analysis will highlight the web of difficulties and lacunae in current thinking about manipulation, and will offer concrete insights into the ways in which the phenomenon keeps slipping away from the best attempts at defining it. This in turn will motivate the alternative conceptualization that I will present in the next chapters.

A helpful place to start is with two general characterizations of manipulation – one negative, and one positive – that capture common insights about the concept. The first (negative) characterization is found already in Rudinow’s seminal paper: “Most people,” Rudinow writes, “would distinguish manipulation from persuasion, on one hand, and from coercion, on the other.”Footnote ⁸ Manipulation, the slogan goes, is “neither coercion nor rational persuasion.”Footnote ⁹ (We can understand why, given the formidable diversity of kinds of manipulation described above, one would be tempted to characterize manipulation merely negatively, i.e., by what it is not.) The second (positive) characterization is that the nature of manipulation is either that of “trickery” or of “pressure.”Footnote ¹⁰ Now the trickery account models manipulation on deception, while the pressure account models manipulation on coercion. Hence, the combination of these two schemes presents us with three poles against which to assess manipulation: rational persuasion, coercion, and deception. (Coercion appears in both schemes; we therefore discern three poles instead of four.) These three concepts form the three cornerstones in manipulation’s conceptual space, as it emerged in the literature. Sapir Handelman, for example, opens his book on manipulation by stating the basic background fact that manipulation is “not exactly coercion, not precisely persuasion, and not entirely similar to deception.”Footnote ¹¹ Our attempt to understand manipulation should start by elucidating the affinities and contrasts between manipulation and each of these three concepts. In this chapter, we will examine how far we can advance by using this comparative route. The main bulk of this chapter will be an examination of the literature on the contrast between manipulation and rational persuasion. Subsequently, I will analyze the nature of manipulation in comparison to deception and then to coercion. Finally, I will examine how the concepts combine.

3

The most central effort by far in attempting to understand the nature of manipulation has concentrated on contrasting it with influence by means of rational persuasion. The intuition behind this distinction is ancient – traced back at least to the Platonic distinction between dialectic and rhetoric – and it undoubtedly captures a prominent feature of the phenomenon of manipulation. We would like to understand how precisely this generic contrast illuminates the nature of manipulation. I will now embark on a review of the central ways in which this question has been answered, by examining the most influential and promising definitions advanced in the literature. I will point to their merits, but also emphasize their significant drawbacks. My intention in this critique is less to refute any definition but rather to expose the anatomy of the many difficulties with which the search for a definition of manipulation is riddled. (Refutation may well be an unreasonably high bar in this domain: Since we expect some vagueness in accounting for social phenomena, showing that definitions fall short of providing necessary and sufficient conditions does not amount to refutation.) Understanding the difficulties should be the first step in attempting to overcome them.

“Manipulativeness,” Marcia Baron writes, is the kind of influence that “reflects either a failure to view others as rational beings, or an impatience over the nuisance of having to treat them as rational.”Footnote ¹² What does this kind of treatment amount to more specifically? Lawrence Stern claimed that manipulation is “subverting or by-passing another person’s rational or moral capacities for the sake of some result.”Footnote ¹³ Since Stern, the idea that manipulation is an influence that either “bypasses” or “subverts” rational decision-making has been widely adopted by thinkers on the subject, and seems in fact to be the most popular account of the distinguishing feature of manipulation. Manipulation, in other words, is the kind of influence that either fails to engage its target’s rational capacities, or if it does engage them, does it in a way that undermines their proper functioning. In yet other words, there are two ways in which manipulative influence deviates from rational persuasion: either it is nonrational (it circumvents its target’s rational faculties altogether), or alternatively it is irrational (it engages those faculties but in a way that subverts standards of rationality).

At times more emphasis seems to fall on the “bypassing” aspect. “Manipulation occurs when one influences another by bypassing their capacity for reason, either by exploiting nonrational elements of psychological makeup or by influencing choices in a way that it is not obvious to the subject.”Footnote ¹⁴ Similarly, “steering policies are manipulative […] because they deliberately circumvent people’s rational reasoning and deliberating faculties, and instead seek to influence their choices through knowledge of the biases to which they are susceptible.”Footnote ¹⁵ At other times, authors stress the “subverting” aspect: “manipulation is intentionally and successfully influencing someone using methods that pervert … [their] decision-making process.”Footnote ¹⁶ Similarly: a manipulative nudge “perverts people’s rationality and thus makes them less rational.”Footnote ¹⁷ Usually, however, as in Stern’s formula, bypassing and subverting are equally the culprits: manipulation “influences people’s choices in ways that circumvent or subvert their rational decision-making processes.”Footnote ¹⁸

The “bypass or subvert” definition is rather crude, but despite its crudeness, or perhaps precisely because of it, it manages to capture adequately a surprising range of cases. To illustrate, a manipulator can influence her target via means such as the following: conjoin an unexciting object she wants to sell with an exciting stimulus, so as to make it seem more attractive; act repetitively so as to lower the target’s level of attention to her actions before abruptly changing course in the direction of the desired influence; flatter the target, to soften his objection to her request; ignore the target, as a means to make him seek her closeness; assume and project some symbols of authority that would invoke the target’s disposition to comply with her request; make a small helpful gesture toward the target, only to elicit his disposition to reciprocate at the moment when she needs much greater help from him; whisper messages in her target’s ear while he is falling asleep so that he subconsciously internalize her messages; set a high (though not outrageous) asking price that acts as a cognitive anchor and helps raise the final offer the target makes; present item x, which she wants the target to choose, alongside random item y that is of lesser quality, so that the joint evaluation of x-alongside-y will tilt more favorably toward x; present an excessive load of (relevant) information that will surpass the target’s ability to assess the fine details; say something true but in such a way that the target can be expected to interpret it falsely; present evidence for correlation but treat it as evidence for causation, and so on. These examples are very different, yet in all of them the manipulator influences her target by means that either bypass rationality or subvert it. These manipulations make use of: conditioning, appeals to unfitting emotions, elicitation of reflex psychological reactions, subconscious influence, exploitation of cognitive biases, overwhelming the target’s information processing capacity, false implicatures, and inducing invalid forms of inference. This diverse sample gives some idea of the broad scope of manipulations covered by the “bypass or subvert” view.Footnote ¹⁹

The “bypass or subvert” view (at least as typically presented) cannot be the right theory of the meaning of manipulation, however. Here I will treat “bypass” and “subvert” separately. The central problem with the bypass view of manipulation is that it is hugely overinclusive: acts of influence bypass reason without invoking anyone’s idea of manipulation. Examples are endless and easy to come by. You smile to the clerk when you arrive at his desk to get service; or, you dress up smartly on the morning of your job interview. Such behaviors are routine, normative, and unremarkable, yet they are intended to influence others (to give you better service, to make them see you in a more favorable light) while bypassing their reason. Admittedly, under certain descriptions such behaviors may be manipulative, but the relevant fact for us is that often or normally they are not. To take the example of dressing up for a job interview: clothes do in fact predictably help to impart desired impressions. Since creating the right impression is the trivially reasonable and responsible thing to do on a job interview, and since behaving otherwise (wearing just whatever one “authentically” feels like that morning) would be silly, maladaptive, and could even be disrespectful, it would not normally cross our minds to employ the term “manipulative” to describe such sensible and normative behavior – on the contrary: we would rather call it “dressing appropriately.” Other examples of influencing people not through rational persuasion, such as making our suffering or vulnerability manifest so as to invoke a measure of empathy from others, may be crucial conditions for social coexistence. Again, under some descriptions we may deem them manipulative, but the point is that often we do not see them in that light. Hence, influence that bypasses rational persuasion cannot offer a sufficient condition for manipulation (i.e., there is intentional influence through nonrational-persuasion that is non-manipulative).

The insufficiency of “influence by nonrational-persuasion” as a definition of manipulation has deeper roots in the very nature of language. The content, or literal meaning, of a proposition (John Austin’s “locutionary meaning”) underdetermines its communicative significance (“illocutionary force”).Footnote ²⁰ If for instance I say “Go on, do as you said!” this may be meant and understood as either a command, or a plea, or a cynical statement to the effect that I know that you cannot in fact do as you said, and so on. Which of those is the case is determined by various performances that accompany the literal content and are not as such rationally assessable. If such performances are included in the view of manipulation as bypassing reason, then we reach a veritable reductio ad absurdum: all of language is manipulation. It is conceivable that we could salvage the “bypass” view if we supplemented it with an appropriate philosophy of language (including a successful logic of illocutionary inference) but this would not salvage the sufficiency of the “bypass” view, as we know it – indeed, precisely the opposite. (Performances, we should add, are communicative even unconnected to any proposition, and this too is often not manipulative. You laugh when I tell a joke to signal to me that you are not angry anymore. This non-propositional communication may well bypass rational deliberation, and yet it is a conventional form of human communication.)Footnote ²¹

It seems that the “subvert” criterion of manipulation is less vulnerable to criticism, for to the extent that we deliberately subvert others’ rational decision-making, don’t we thereby necessarily manipulate them? Postponing the answer to that for a while, the “subvert” view is at any rate problematic in that subverting reasoning is not a necessary condition for manipulation. Consider the following illustration:

As Moti Gorin rightly claims, this is a case where “the rational capacities of the manipulee are in no way inhibited from fulfilling their function.”Footnote ²³ Claudia Mills claimed that “a manipulator is interested in reasons not as logical justifiers but as causal levers. For the manipulator, reasons are tools, and bad reasons can work as well as, or better than, a good one.”Footnote ²⁴ Gorin sees the truth in Mills’ first sentence, but shows that her second sentence is not the necessary conclusion: Good reasons too can function as manipulative levers, when they proceed from ulterior motives. Once we understand that the determining factors here are the influencer’s ulterior motives, we can easily construct endless analogous examples of manipulation that do not subvert their target’s rational capacities.

The “bypass or subvert” view of manipulation seems inadequate, as it is both overinclusive and underinclusive (it fails to provide both sufficient and necessary conditions). At best – to the extent that we believe it can be salvaged – it is highly underdeveloped.

4

Gorin himself attempts to provide a more accurate theory. In the heart of his theory is the idea that “manipulation is a form of interpersonal influence that deliberately fails to track reasons.”Footnote ²⁵ Gorin provides a definition that expresses this idea, and that attempts to capture the meaning of manipulation precisely:

Gorin’s definition is a thoughtful and more accurate elaboration of the “bypass or subvert” view. Let us start by understanding better what this definition says. First, we can appreciate its sophisticated structure: it has a necessary core which is then supplemented by four mutually exclusive criteria, each of which when added to the core renders it sufficient. Second, we can see that the four criteria can be divided into two groups: the fourth, and then the first three. Criterion (4) is a version of the “bypass” intuition. It refers to “means of influence that by their natures do not reliably track reasons.” They are not “considerations – good or bad – in the light of which one acts.” Examples of such means that Gorin provides include framing effects, setting powerful defaults, or playing background music that is known to affect the target in a given way.Footnote ²⁷ In contradistinction to (4), criteria (1) to (3) involve providing reasons to the manipulator’s target. Criterion (1) is Gorin’s “subvert” criterion: influencing the target to respond (by acting, believing, etc.) in ways that are not well supported by reasons. Criteria (2) and (3) express Gorin’s intuition about the pertinence of ulterior motives to manipulation. Criterion (2) holds that the form of “bullshitting” that consists of providing reasons without caring whether they are good reasons is manipulative, even if the reasons provided turn out fortuitously to be true.Footnote ²⁸ Criterion (3) deserves a little more elucidation. We may ask: What is meant by saying that A is motivated by B’s sufficient reason to x, but that A nonetheless leads B deliberately to x in light of some other reason? If A leads B to x in light of some other reason, then in what sense is A nonetheless “motivated by B’s sufficient reason to x”? The idea seems to be that A views the fact that B has sufficient reason to x as a constraint on A’s influencing B to x, such that A would refrain from influencing B to x had A assessed that B lacks sufficient reason to x. This distinguishes (3) from (2), since in the latter, “the manipulator just happens to believe that her manipulee has sufficient reason to do what the manipulator intends she do.”Footnote ²⁹

Gorin’s definition is an important step forward in the attempt to understand manipulation. It is therefore instructive to examine how difficulties crop up with virtually every element of it. I will divide my critical discussion into two in the following way. First I will explain how Gorin’s definition is extensionally incorrect (i.e., it fails to cover the right scope), then I will argue that even if it were extensionally correct, it would still be undermined by significant indeterminacy.

The extensional inadequacy should be clearest with respect to criterion (4), which more or less repeats the overinclusiveness of the “bypass” view. Gorin’s explanation of why “means of influence that do not reliably track reasons” are as such manipulative is that it is not enough that our actions are in fact in accordance with what we have most reason to do; it is also imperative to do the right thing because it is the right thing to do, that is, out of the right reasons. Otherwise a person is in the unfortunate state of being “detached” from an important aspect of reality, namely, the reasons that should govern her action. But this suggests an implausibly rigorous idea of self-governance. Giving a friendly pat on the shoulder to your colleague to encourage her in a difficult moment does not by any stretch make you a manipulator, although this kind of influence does not reliably track reasons. We encountered similar examples when we discussed the bypass view, and anyone can come up with myriad others. Against Gorin’s hyper-rationalistic view, we could even say that only in cases of some communicative disorders do people need reasons to be explicitly “tracked” in every instance of communication. Criterion (4) is accordingly overinclusive. Much normal influence is not supposed to track reasons.

Regarding criteria (1) to (3), we should first point out that they can be graded according to their level of adherence to rational standards: While in criterion (1) the manipulator provides bad reasons, in criterion (2) the manipulator provides good reasons, but merely incidentally, and in criterion (3) he is already committed to providing good reasons (though has concomitant ulterior motives). Since criteria (1) to (3) form a scale, from harsher to milder manipulation, we can assess extensional adequacy by checking whether we can find false positives regarding criterion (1) or false negatives in connection to criterion (3); in other words: we ask whether there are acts of influence that fit criterion (1) yet are not manipulations, and whether there are acts of influence that are still milder than criterion (3), and so are not deemed manipulations according to Gorin’s definition, but nonetheless are. In fact, we can find such examples on both sides of the (1)–(3) scale.

Actions that fit criterion (1) yet are not manipulations come to mind easily when we think, for instance, of situations where rational persuasion and rational deliberation are not the primary goals of the interlocutors.

Despite complying with Gorin’s first criterion, no one here is manipulated: they are rather engaged in a cooperative project, in which being influenced by some bad reasons is implicitly acceptable as a side effect. What I have here described as a one-off situation can also characterize various sorts of consensual long-term relationships (where both sides prioritize maintaining the convenience of the status quo between them over the accuracy of the reasons they get about the nature of the situation). As always, beyond a certain line, such a relationship would be described as indeed mutually deceptive and manipulative, but the point is that in its milder forms, it is the stuff of normal healthy relationship (the opposite of which would involve compulsive drilling into the other’s precise intentions and reasons underlying almost every mode of presentation of every issue).Footnote ³⁰

Moreover, we sometimes persuade others through bad reasons not only negligently (as with the couple described above) or subconsciously, but intentionally and explicitly, and that too may not result in manipulation, as long as the bad reasons function as elements of a good argument. Nice illustrations of the necessity of providing false explanations as part of an effort to achieve good understanding are provided by Maimonides:

When we observe the process of learning as one integral whole, then to the extent that the use of deliberately inaccurate or even contradictory reasoning is inevitable for the effort to secure optimal comprehension, we would tend not to view the offering of bad reasons as manipulative.

In criterion (3) the manipulator is committed to providing good reasons, yet does this out of ulterior motives. Presenting this as the mildest form of manipulation through exchanging reasons creates some false negative judgments (i.e., falsely judging cases of influence as non-manipulative). Consider the following case that fails to exhibit even the mildest criterion (3):

Michael has no ulterior motive in providing a good advice (reason) to Rachel. In this case criterion (3) does not hold. Moreover, not only is Michael providing a good reason to Rachel, in a straightforward manner, and while sharing her aim, but he is also attempting to prevent her from irrationally rejecting it. Nonetheless, Michael is acting manipulatively while giving Rachel his advice.

Having discussed the extensional inadequacy of Gorin’s definition, I will now argue that even if it were extensionally correct, it would still be plagued by unacceptable indeterminacy. Gorin’s definition presents four possible ways to “fail to track reasons.” This raises the question of which is to be employed in any given instance of influence. Since the definition cannot offer good guidance in this respect, it leaves much indeterminacy in attempting to decide whether a given act of influence is a case of manipulation or not. For the present discussion, let us grant that objective criteria determine clearly whether we should use criterion (4) or one of the rest (though in Chapter 2 we will see that this too is problematic). I will now concentrate on the question of when we should employ criteria (1)–(3). To wit, if we employ the more stringent standard of communication of criterion (3) when we in fact should employ the more lenient (1), then we will diagnose manipulation where there is none, and vice versa. Let’s attempt to understand this using the example of two versions of the same scenario of influence.

The relevant criterion for Party1 is arguably criterion (1), and according to it there is no manipulation, while the relevant criterion for Party2 is criterion (3), and it does detect manipulation. The important point to notice is that if we employed criterion (3) in Party1, then we would diagnose Party1 as a case of manipulation, which would be a wrong diagnosis. Now how did I determine that in Party1, criterion (1) is relevant, whereas in Party2, criterion (3) is relevant? Party1 and Party2 differ with respect to the significance of ulterior motives in influencing others. In Party1 it is Dina who approaches David with a question, and David responds with a good advice. In such a scenario, the fact that David has ulterior motives is of no relevance; that David answers Dina’s question by providing a good reason is sufficient to rule out manipulation. Party2 is different in two ways. In Party2 it is David who approaches Dina; this makes his ulterior motive much more pertinent and casts it in a completely different light. In addition, David knows that Dina is interested in him and so that his telling her that she should come to the party to which he is going is likely to have significant overtones for her beyond the good advice about the value of networking at the party; this too makes his ulterior motive (interest in Dina’s friend) loom large. These considerations require adopting the more stringent criterion (3). Crucially, the fact that we should adopt criterion (3) in one scenario but not the other is not a function of any formal properties of the definition and scenarios; it is a function of our sensitivity to what aspects of interpersonal communication are significant for the assessment of manipulation. Nothing in Gorin’s definition of manipulation can hint that the question of who initiates the communication might be relevant for the appraisal of manipulation, nor can the definition alert us to whether this or that psychological dynamic between the communicators adds “significant overtones” to the act of influence or not. Rather, these insights, and many others like them, belong with a psychological aptitude that we must possess if we are to use the definition effectively. This means that (in some such cases) it is not the definition that determines our identification of manipulation, but rather our prior intuitive insight into the nature of “manipulation” that determines our ability to use the definition accurately. Gorin’s definition cannot therefore be telling the full story about manipulation.

(Moreover, if all people are strict rational egoists – as many thinkers have believed the case to be – then we have an ulterior, egoistic motive operating almost whenever we offer good advice – reason – to another. Clearly, it would be unhelpful to claim then that all rational advice is manipulative. We must rather have insight into which ulterior motives are relevant for manipulation and which are not. Gorin’s definition is, unavoidably, silent about this.)

Lastly, another feature of Gorin’s definition is also susceptible to the same kind of criticism that I have just expounded. Gorin’s definition repeats the claim that manipulative influence is “non-coercive.” How instructive is it to simply glue this extra condition to one’s discussion of “failures to track reasons”? The answer, here too, is that to the extent that the we can distinguish manipulation from coercion objectively, independently from our understanding of manipulation, then this element of the definition would be benign. Such, however, is not the case, as I will explain when I discuss coercion below.

My purpose in analyzing Gorin’s definition as well as others in this chapter is, again, not to refute but to provide insight into the anatomy of the difficulties involved in arguably the best attempts to define manipulation. I further want to show that these difficulties are not restricted to far-fetched thought experiments, but arise in realistic scenarios, and that they pervade almost every aspect of the definitions. This suggests that a serious rethinking of our understanding of the concept of manipulation is in order.

5

Perhaps the most compelling view of manipulation in the literature was developed by Robert Noggle. According to Noggle, manipulations are actions that deliberately attempt “to lead their victim astray.” He writes: “there are certain norms or ideals that govern beliefs, desires, and emotions. […] manipulative action is the attempt to get someone’s beliefs, desires, or emotions to violate these norms, to fall short of these ideals.”Footnote ³² More specifically, manipulative actions lead astray when they cause their target to have beliefs that are false or irrelevant, when they condition the target to acquire desires not supported by his beliefs or precipitate akrasia in him, when they induce in the target inappropriate emotions, that is, emotions that are based on false beliefs or that are unfitting to the situation or misrepresent its salient features, when they misdirect their target to pay more attention to that which is less relevant for the choice at hand, and so on. While deception leads astray by causing faulty beliefs (specifically), manipulation is the more general category, leading astray over the entire spectrum of mental states. Manipulation subverts deliberation – it is “the deliberate attempt to induce – trick – a person into relying on a faulty mental state in her deliberation.”Footnote ³³ In a way, Noggle ends up distilling this diagnosis to one word: manipulation causes its target to react in a “mistaken” manner.Footnote ³⁴

Noggle’s view presents a clear and significant advantage over the views we have considered. We saw that the “bypass” view was implausibly over-inclusive, and that that problem was not solved by Gorin’s definition. Noggle’s account, on the other hand, has the resources to cope with that difficulty, given that it expands its scope from influencing beliefs to influencing emotions, desires, or any other mental state: When influence not through rational persuasion (i.e. which bypasses reason) comes in contexts where influence is not supposed to be through reasoning, then such influence indeed would not constitute “falling short of the ideals” that are relevant for mental states such as emotions or desires, and therefore would not constitute manipulation. Noggle can thus speak helpfully of nonmanipulative nonrational persuasion:

Noggle’s formulation is uniquely useful in judging when influences that are not argumentative (and there are plenty of those) are manipulative. For instance: if, by heaping one melodramatic tragedy upon another, a kitschy movie manages to extort our tears out, then it may well be manipulative, in that it makes us have faulty emotions.Footnote ³⁶ Accordingly, Noggle’s view does not fall in the trap (of the crude “subvert” view) of diagnosing manipulation only when bad reasons are provided (to this extent it shares the strength of Gorin’s view).

Despite the significant merit of Noggle’s view, my analysis below will attempt to show that it too suffers from major difficulties. I will first focus on ways in which Noggle’s criterion for manipulation is not necessary and then on ways in which it is not sufficient.

Consider the following case, exhibiting the “salami technique.”

Daphne planned and executed a grand plan in which Gabby, in the role of her pawn, was led to choose as Daphne wanted her to choose. Daphne manipulated Gabby, yet, crucially, this did not involve Gabby compromising her rational decision-making, either locally (i.e. in any specific step) or globally (as Gabby was okay with D in the first place). In other words, while Daphne manipulated Gabby, Gabby was not led astray. Having one’s mental states fall short of their ideal functioning is not a necessary condition for manipulation.

Let us get a better sense of this important point by reviewing another example:

The manipulative beggar understood the mind of his target and maneuvered him shrewdly, so that the father became a compliant pawn in the beggar’s scheme. Yet, importantly, the father acted in accordance with his values and did not fall short of ideals for beliefs, desires, emotions, or any of that. Manipulation does not necessitate “being led astray” and “falling short of ideals.” Manipulation, it is important to state, is not necessarily a zero-sum game, where the successful manipulative act must ipso facto undermine the manipulee’s mental states.

Perhaps the most important category of cases where manipulation does not necessarily entail compromising mental states is when manipulation operates primarily on the target’s environment rather on the target herself. I will refer to the former as “indirect” manipulation (the latter being “direct” manipulation).Footnote ³⁷ Noggle’s view was developed specifically for direct manipulations (in his words: to exclude cases where “the victim’s circumstances, rather than the victim herself, are altered”).Footnote ³⁸ Such a restriction is, however, at odds with common usage, difficult to maintain (“direct” and “indirect” blend into each other), and at any event, became unacceptable in an age where most discussions of manipulation involve nudging via “choice architecture” (i.e. designing the choice environment so that options would be presented in ways that would have predictable impact on decision-makers’ choices).Footnote ³⁹ And indeed, in more recent writing Noggle includes indirect influence under “manipulation,” arguing that “influencing choice by increasing the salience of a fact not relevant to the choice at hand is a form of trickery that renders the nudge manipulative.”Footnote ⁴⁰ Analyzing Thaler and Sunstein’s famous Cafeteria case (where customers are influenced to choose the healthier food, when it is placed first on the shelf and at eye level), Noggle claims that the targets of influence fall short of ideals of rationality, given that they react to features of the choice situation that are substantively unimportant. I believe that Noggle’s thesis is not convincing. The cafeteria customer who chooses the salad directly in front of his eyes is not behaving in a substandard manner. The fact that people focus more attention on that which is more salient in their perceptual field is an indispensable default strategy for any creature with limited perceptual and cognitive resources. In addition, satisficing can be rational, especially when what’s at stake is spending the precious minutes of one’s short lunch break on finding some food that will be slightly yummier than what is in front of one’s eyes. Here, again, we need not posit a necessary zero-sum game between the manipulative influence and the standards exhibited by its target. (To the extent that Noggle insists on expanding his thesis of falling short of ideals to such a weak sense of it, the new problem that will ipso facto arise is that it will become correspondingly more difficult for his definition to offer a sufficient condition for manipulation, as I discuss below.)

Above I argued that it is not true that manipulation necessitates falling short of ideals for mental states. It will be instructive to see how this works also in the reverse direction, namely, that it is not true that if ideals for mental states are upheld, then manipulation is ruled out. This latter claim is instructive because it highlights a central aspect of Noggle’s position, as compared to most expressions of the “bypass or subvert” view. We can formulate the point through the distinction between process and outcome. Most expressions of the “bypass or subvert” view see the process of influence as the crucial factor, such that if, for example, you influence through the mode of presenting information (instead of through the information itself), then you manipulate. Noggle’s view, in contrast, focuses on the outcome aspect, so that even if you influence through the mode of presentation, yet this serves to consolidate ideal mental states (rather than make them faulty), then you do not manipulate. This is an important insight and is often true. To take a trivial example: If I highlight parts of the text to make it easier for you to concentrate on the main ideas, then I didn’t manipulate you.Footnote ⁴¹ The point that is important for us to emphasize here is that although Noggle’s claim is often true, it is sometimes not true: We can influence a person to act in conformity with ideals for believing, desiring, and emoting, yet achieve this manipulatively. Let us consider two examples that Noggle mentions.

Noggle argues that “if a psychologist uses conditioning to instill desires that conform to the patient’s beliefs about what there is reason to do, then she is engaged in therapy rather than manipulation.”Footnote ⁴² I believe Noggle is mistaken in dichotomizing: therapy and manipulation are sometimes two sides of one dynamic. Consider for instance Freud’s view on transference-love, a process in which the therapist implicitly invokes transference of childhood sexual emotions (originally directed at the parent) to himself, as an indispensable element in psychoanalysis.Footnote ⁴³ It seems clearly mistaken to deny that a process in which one person (the therapist) uses covert seduction, by cunningly taking on the role of another person (the patient’s parent), so as to transfer feelings of love toward him, which would assist in unearthing secrets that the patient painstakingly tries to hide (the “resistance” phenomenon) is manipulative. We may well wonder: if this is not manipulation, what is? The fact (let us assume it is) that this process helps the patient behave in ways that conform to what (she thinks) there is reason to do cannot change this diagnosis. The point here is that influence that realizes ideal outcomes can nonetheless be manipulative by virtue of its process. Consider a second example. Noggle writes: “[I]t does not seem manipulative to try to make someone feel guilty for doing something that is clearly immoral.”Footnote ⁴⁴ This is indeed usually true, but, importantly for us, sometimes it is not. Making someone feel guilty can be done manipulatively, even if it is indeed rational for that person to feel guilty. Societies have norms for how you “rebuke thy neighbour” (Leviticus 19, 17): by whom and toward whom it is admissible, under which circumstances, over what transgressions, using what language, how far it is one’s business to toil so the other internalizes the criticism, and so on. It is indeed possible to rebuke one’s neighbor for a veritable transgression, yet to “rub it in” manipulatively. More generally: “Leading astray” by making the other “fall short of ideals” (outcome) may not be the case, yet the process may nonetheless render influence manipulative.

What is most important for us to learn in all this goes beyond a criticism of Noggle’s position. We have just seen that, pace Noggle, an assessment of the outcome of influence is insufficient for determining manipulation. Earlier we saw that an assessment of process is similarly insufficient to determine manipulation. (Recall, e.g., Maimonides’ example. Moreover, not only is it the case that a misleading process may not generate manipulation; it is also the case that an immaculate process may generate manipulation; for instance, if one presents a valid formal proof of his argument, anticipating that his interlocutor will fail to follow yet will shy away from admitting this – to others as well as to himself – and so will end up confused.) Hence, neither process by itself nor outcome by itself can determine manipulation. Might it be the case that the conjunction of both conditions is required for manipulation? This would be an overdemanding position, which would restrict the scope of “manipulation” unreasonably. We have already seen examples where each is not necessary, so it would seem wrongheaded to insist that both are necessary. The lesson is that the formal conditions (offered by suggested definitions) cannot define manipulation. We need some prior intuition about what kind of interaction is manipulative, in order to alternate successfully between the criteria of process and of outcome so as to capture the phenomenon we in fact intend to capture. It is revealing and important to note that this conclusion – the necessity of some kind of practical understanding that is prior to the defining criteria – is the same conclusion we reached when confronted with deep indeterminacies regarding the application of Gorin’s definition. A practical understanding of manipulation appears to be a precondition for making more accurate sense of it through our definitions.Footnote ⁴⁵

I have argued that Noggle’s definition fails to offer a necessary condition for manipulation; I will now argue that it also fails to offer a sufficient condition. This means that influence which intentionally causes its target to have “faulty mental states” may nonetheless not be manipulative. We have in fact already encountered an example for this in Bad Date, where the unlucky couple operated under an implicit mutual agreement of sorts that lowered the bar of the normativity of causing faulty mental states in each other. More important is the fact that examples to that effect are everywhere. In fact, Noggle himself says so: “Common forms of ‘impression management’ – dressing well for a job interview, staging a house for sale, etc. – employ non-rational mechanisms to create positive impressions, yet it is not clear that they merit the label ‘manipulation’.”Footnote ⁴⁶ And this is so despite the fact that the resulting positive impressions are not rationally warranted. (We may notice that the level of “falling short of ideals” is here no lower than that in Thaler and Sunstein’s Cafeteria case, which Noggle does want to consider manipulative!) The only sensible way to approach this difficulty in the definition of manipulation is if we give up speaking of “ideal” mental states, and instead make do with acceptable norms for them. This comes out clearly in Cass Sunstein’s working definition of manipulation: “I suggest that an effort to influence people’s choices counts as manipulative to the extent that it does not sufficiently engage or appeal to their capacity for reflection and deliberation.”Footnote ⁴⁷ To not be manipulative, influence does not need to evoke ideal responses, “sufficient” ones are enough. Importantly, whether influence appeals “sufficiently” to people’s reflective and deliberative capacities does not reflect some a priori standards but rather prevailing norms.Footnote ⁴⁸ The point I wish to stress here is that far from solving the problem under consideration (that Noggle’s definition does not offer a sufficient condition for manipulation), it recreates it in a new sense: What determines manipulation now becomes a vast set of conventional practices regarding influence, a “form of life” that cannot be captured by a definition.Footnote ⁴⁹ When influencing, we bypass or subvert the other’s ideal reflection or deliberation much of the time; whether this exceeds conventional norms and amounts to manipulation or not is determined by endless contextual factors. This is all the more true with respect to emotions and desires, where norms are yet murkier. Inducing akrasia or phobia (examples mentioned by Noggle) indeed involves falling short of relatively clear standards of rationality, but this is rather the exception. Think instead of such examples as influencing one’s partner to give another chance to the relationship or influencing a friend to leave what they are doing and join you on some adventure. Standards that determine the point at which such forms of influence cross an elusive threshold of “leading astray” and become manipulative are highly contextual – they emanate from and express the natures of particular relationships, of scenario types, and of the idiosyncratic dynamics between the personalities involved. As the great Michelangelo put it aptly: “What makes another wise, leads me astray.”Footnote ⁵⁰ Our examination of whether Noggle’s definition can provide a sufficient condition for manipulation ends thus with a yet deeper suspicion that definitions as such are perhaps too coarse to allow useful application.

This concludes my review of perhaps the most promising definition of manipulation, offered by Noggle, and more generally, of the lessons we can learn from examining the concept of manipulation in light of its purported contrast with standards of rationality for persuasion, deliberation, or for any mental state. My examination was by no means comprehensive; it rather attempted to shed light on the difficulties that current definitions face, to offer a tangible sense of the width and depth of these difficulties, and simultaneously to hint at possible alternative directions (which we will consider in the next chapters).

6

I now turn, much more briefly, to the other two main concepts that, according to the existing literature, form manipulation’s conceptual space: deception, and coercion. I start with deception.

That there are affinities and similarities between manipulation and deception is beyond doubt. What I want to assess here, however, is whether there is some relatively simple and sweeping way in which the concept of deception can illuminate that of manipulation, and that can therefore be incorporated into a definition of manipulation. (Compare to the simple and sweeping claim that manipulation “bypasses or subverts rational deliberation.”) There are two principal ways in which that could be the case: if all manipulations are deceptions or if all deceptions are manipulations. I will explain and reject both in turn.

If all manipulations are deceptions, then the nature of manipulation is deceptive, and we then expect to learn something significant about manipulation by examining deception.Footnote ⁵¹ This view is false, however – manipulations can be nondeceptive in various ways. Manipulations are not deceptive if they communicate the truth, the whole truth, and nothing but the truth, or simply if they do not communicate any falsehood.Footnote ⁵² Here are a couple of illustrations.

1. (1) It is an empirical fact that when asking someone to do us a favor, we have better chances to succeed if we provide a reason.Footnote ⁵³ Hence, if you ask the person in front of you in line: “Excuse me, I have five pages. May I use the Xerox machine?” you’d be much less likely to succeed than if you say: “Excuse me, I have five pages. May I use the Xerox machine, because I’m in a rush?” But what if you are not in a rush, you are not prepared to lie, and yet you nevertheless want to pass the person in front of you? In that case, you can ask: “Excuse me, I have five pages. May I use the Xerox machine, because I have to make some copies?” Research has shown that this works just as well! Influencing people to do you a favor, by using the ruse of providing the semblance of a real reason is manipulative. Yet it is no deception – because fully truthful – to say that you want to pass them because you have to make some copies.

2. (2) Reporting things exactly as they are, but releasing the details in a strategic order is normally no deception; yet it can be enough to have the effect the manipulator wants to achieve. For instance: It is in your interest that X be furious at Y. You know that X will immediately suspect Y for the blunder and be furious at her, so only a few sentences along your report you say that Z is the responsible agent. This is enough to achieve your goal: as you correctly predict, X gets incensed at Y instantaneously, and his anger, once kindled, cannot be fully quelled by learning that Z in fact did it. The manipulation here is based not on untruth but rather on understanding the occasional futility of the truth. Jonathan Swift described precisely that futility of a delayed truth, in those words: “like a man who hath thought of a good repartee when the discourse is changed, or the company parted; or like a physician, who hath found out an infallible medicine, after the patient is dead.”Footnote ⁵⁴ The futility of the truth can be very handy for a manipulator, without needing to resort to any deception.

3. (3) You know that X is very sensitive to flattery; it makes her bend over backwards to please the flatterer. Now that you need her help, you are careful to flatter her on any occasion you can, even for little things, so as to extract the maximum out of her. To the extent that your flattery, though copious, is always over things you truly find positive, then your manipulation is not deceptive: you communicated no falsehood.

It has been argued that perhaps manipulation is always deceptive because it conceals its intentions.Footnote ⁵⁵ This, however is not true, both because the intentions of the manipulator can be transparent to the target and still work (think for instance of manipulative seduction), and because concealment of one’s intentions is quite often not deceptive: there is no universal norm of full disclosure of one’s intentions to others, nor anything remotely close to this.Footnote ⁵⁶

We saw that not all manipulations are deceptions; perhaps, however, all deceptions are manipulations?Footnote ⁵⁷ If so, then although manipulations as such cannot be explained as deceptions (since some of them aren’t), we may nevertheless suspect that deception has an important explanatory role, in that it is the paradigm manipulation (hence, as would be expected, every instance of it is manipulative). This is indeed the position taken by (what Noggle calls) the “trickery” views of manipulation. Noggle, accordingly, is explicit that a constraint on our construction of the concept of manipulation is that it follows structural elements of the concept of lying.Footnote ⁵⁸

I believe that not all deceptions are manipulations, and accordingly that deceptions as such are not the paradigm of manipulation. Consider the following example.Footnote ⁵⁹

I do not think that Smith manipulated the passerby. We here do not start from presupposing that all deception is manipulation; rather, we ask: Is there anything distinctly manipulative about Smith’s treatment of the passerby? Sure, Smith caused the passerby to have a “faulty mental state,” but we have already seen that this is not a sufficient condition for manipulation, and virtually every other feature of manipulation is missing. At this stage of our investigation, we lack an adequate terminology to account for what is missing, but I will give some intuitive ideas.

A hallmark of manipulativeness is that the manipulator gains control of her victim. She “penetrates into his mental machinery” and is thus able to “operate him from within.” In stark contrast, the pure and simple lie of answering “no” instead of “yes” does its deceptive job straightforwardly, without any need to “penetrate into the mental machinery” of its victim. Manipulation requires at least the minimal interpersonal sophistication needed to understand how to harness the other’s psyche to perform as the manipulator wishes. In contrast, when a liar (e.g., grumpy Smith) says “no,” although the true answer is “yes,” the misleading impact is an automatic function of language; it requires devoting zero attention to the liar’s victim or to how circumstances affect the victim’s information processing. It requires zero understanding of the other’s psyche, and therefore also zero planning of how to maneuver the other. As Sapir Handelman noticed perceptively: “The liar simply misleads the target. These cases miss the unique sophistication of manipulation.”Footnote ⁶⁰ In effect, even the “faulty mental state” induced in the passerby is such merely in a rudimentary sense: although the passerby now (assuming he believed Smith) possesses a false belief, the lie did not cause any “subversion” of his deliberative capacities (which functioned adequately in response to false information). Given that Smith did not “gain control of the passerby’s inner psychological mechanisms” (e.g., judgment), this does not resemble a typical case where a manipulator “pulls his victim’s strings.” Furthermore, a typical feature of manipulation involves the manipulator harnessing his victim to become a pawn in his scheme; but Smith is not interested in the passerby playing any role in any scheme of his. Smith just cannot be bothered with giving an iota of consideration to the passerby – and indeed he does not. Lying happens to be the most economical way for Smith to ignore and brush off the passerby. The extreme lack of attention to the other exhibited by Smith is the very opposite of the mindset characteristic of manipulating another person. The manipulator engages his target in his scheme; Smith’s lie does precisely the opposite: it holds off the other, it is a form of disengaging. It is instructive to emphasize that to the extent that we relied only on test cases like Grumpy, the very idea of manipulation would arguably not even come into being, as it would serve no useful distinctive classificatory purpose; that is to say, beyond the existing designation of such cases as falling under the category of “lying.” Lying is not a paradigm of manipulation.

Our brief discussion of deception strongly suggests that there is no simple way in which the concept of deception can be used to unlock the meaning of “manipulation.”

7

Many have remarked that manipulating, in one way or another, is a form of power. This leads naturally to the attempt to understand manipulation in reference to coercion. Indeed, when we speak of “A authoring decisions that absent A’s intervention would be authored by B,”Footnote ⁶¹ then this formula seems to apply equally to A as coercer or as manipulator – both infringing on the autonomy of their targets. Our question, therefore is this: can the comparison (affinity and contrast) to coercion offer a key to a straightforward definition of the concept of manipulation?

The one significant theory here is that manipulation and coercion occupy different segments on a continuum of pressure. Hence Marcia Baron describes manipulation as involving “pressure to acquiesce” to the manipulator’s demands, noting that this pressure is exerted by “a threat that does not rise to the level of being coercive but differs mainly just in the degree of resistibility.”Footnote ⁶² Cass Sunstein similarly asserts that “manipulation can deprive people of agency, resting on a continuum for which coercion is the endpoint.”Footnote ⁶³ Coercion offers a benchmark of forceful influence against which manipulation can be assessed. Placing these observations in the larger context of manipulation as “neither coercion, nor rational persuasion,” we get the neat idea that: “If rational persuasion exerts no pressure at all and works only by providing reasons, and if coercion exerts a high level of pressure, then it makes sense to think that manipulation exerts a moderate level of pressure – more than rational persuasion, but less than coercion.”Footnote ⁶⁴

Although this view encapsulates a grain of truth, I will argue that it offers no more than little help in understanding manipulation. The two problems with the view that coercion and manipulation (and rational persuasion) reflect gradations on a scale of diminishing pressure are: (1) that it is often irrelevant to understanding manipulation, and (2) that even when it is relevant, it is highly inaccurate.

First, regarding irrelevance: many manipulations seem to have nothing to do with pressure. If I want you to choose X, and so when I present you with your options, I mention X first (or last), because I know that the way options are ordered influences the chances of them being chosen, then I manipulate you using my knowledge of a simple cognitive bias; I do not influence by pressuring you. If, when I inform you of the facts, I frame them in a farfetched narrative that makes my misconduct seem more tolerable, then I manipulate your understanding in a way that is not illuminated by a comparison to coercion. If, when you ask me whether Nicky is at home, I answer that she is not home on Wednesdays (which is true, only that it is not Wednesday today but I happen to know that you think it is), then I deceive you with some type of double bluff, but again, this manipulation is not well explained by invoking the ideas of pressure or coercion. Many, arguably most, manipulations call for a different kind of account. Hence, the scope within which “pressure” and “coercion” have explanatory force is limited to a fraction of the manipulation phenomenon.

The next point is that even when manipulation does involve pressure, the view that coercion, manipulation, and rational persuasion reflect gradations on a scale from maximal to minimal pressure is inaccurate. This is so both because manipulation can exert more pressure than coercion, and because rational persuasion can exert greater pressure than manipulation. The first of these is our main interest here. Let us consider examples.

Imagine two scenarios. In the first, a gang member stops you on an alley and informs you that he likes your shoes and that if you do not give them to him there and then, he will beat you up. In the second scenario, a frivolous rich man who likes to toy with human beings walks up to you on the same alley, informs you that he loves your shoes, and teases you with a real offer of one million dollars if you hand them over right there and then. Now it is entirely conceivable that you find the offer of a million dollars much more irresistible than the threat of being punched. And yet, the threat of physical harm is normally taken to be coercive, while the teasing offer is viewed as manipulative. (In accordance, your consent to give your shoes to the gang member would standardly not be seen as valid, while your consent to give your shoes to the rich man would be considered valid.) Our understanding of coercion and manipulation does not covary neatly with the actual level of pressure. Similarly, “If you do X, you risk being arrested,” is taken to illustrate the coercive force of the state. In comparison, “If you do X, I’ll think you’re really cool!” coming from the girl you have a crush on, may well be a manipulative trick on her part to get you to do X. Nonetheless, the second of these may exert much stronger pressure on the infatuated teenager’s choice of whether to do X. Again, in contrast to the theory we are considering, our understanding of coercion and manipulation does not covary neatly with the level of pressure.

For the sake of the completeness of the picture, I should mention that the claim that rational persuasion constitutes a bottom limit for manipulation with respect to level of pressure is also untrue – and perhaps even more deeply so. “Persuasion is never controlling”Footnote ⁶⁵ is a false idea if it is meant to describe human communication, both from the side of the persuader and from that of the one persuaded. I will return to discuss this in Chapter 2. If true, then the upshot for our present discussion is that the second leg of the theory of a gradation of pressure from coercion to rational persuasion is also inaccurate. This reinforces the conclusion that the theory is a shaky basis for an understanding of manipulation.

Finally, even if coercion, manipulation, and rational persuasion did arrange themselves in an orderly linear continuum of pressure, we would still face a conceptual problem deciding where manipulation ends and coercion begins.Footnote ⁶⁶ To get a taste of this, consider two different criteria for drawing the limit. According to one criterion, manipulation ends and coercion begins where the pressure crosses a threshold of “irresistibility.”Footnote ⁶⁷ (I am setting aside now the problem of the essential inscrutability of that phenomenon.) According to a different criterion, the transition happens at the point where we judge that the balance of responsibility (or agency) shifts to the influencer.Footnote ⁶⁸ The problem (as a moment’s reflection shows) is that there is no reason to think that these parameters draw the distinguishing line in the same place.

If, as we see, manipulation cannot be illuminated well-enough – let alone defined – by sheer quantity of pressure, how do we draw the appropriate distinctions on this spectrum? I would like to make two pertinent observations, and draw a (tentative) general lesson from each. The first observation is that there are many relevant parameters that participate in distinguishing manipulation from coercion. We have just encountered two, and I will mention others. Consider for instance: While coercion is a means of overcoming resistance, manipulation is a means of undermining resistance.Footnote ⁶⁹ Coercion is control that depresses the target’s agency, while manipulation’s control stimulates the target’s mechanisms of agency to act in accordance with the manipulator’s plan (e.g., the good salesperson convinces the buyer that she in fact really wants the product). Coercing can typically be exercised almost mindlessly and mechanically (brute force or crude threat), while “the manipulator must know something about the motivations of the one being manipulated in order to change their inclinations to his advantage.”Footnote ⁷⁰ “Manipulation, unlike coercion, does not interfere with a person’s options. Instead it perverts the way that person reaches decisions, forms preferences or adopts goals.”Footnote ⁷¹ Coercion limits negative freedom, while manipulation interferes with positive freedom, or autonomy. Now each of these observations is rather vague or even inaccurate as such, but they are all relevant for constructing a comprehensive view of manipulation, in light of its contrast with coercion. If true, this observation suggests the conclusion that manipulation is much harder to define than usually assumed, and suspicion creeps that perhaps it is not definable at all.

The second observation starts from the insight that the same form of influence can, with small modifications, be either manipulation (“merely pressuring”), coercion (“overpowering”), or rational persuasion (rational bargaining). Consider the message “If you φ, I will ψ!”; for instance: “If you don’t accept that job offer, I will leave you!” Depending on small nuances of context, this statement can be interpreted as either an attempt at rational persuasion (reminding the other of the expected costs of his choices), or manipulation (exerting pressure), or coercion (giving a sense of outright threat or intimidation). Both the content and the form of the message are identical in all three alternatives. What would distinguish manipulation from the other forms of influence are therefore only elements of the context, and more specifically – parameters of the intersubjective dynamic between the influencer and the one influenced (how they assess each other’s intentions, emotions, seriousness, vulnerability, style, commitment, etc.). (A similar insight about interpersonal dynamics was intimated by my discussion of Grumpy.) Generalizing, we may hypothesize that a successful theory of manipulation will place greater emphasis on intersubjective dynamics than it has been accorded to date.

I conclude that the concept of coercion and the idea of a scale of pressure fail to offer a firm basis for defining the concept of manipulation. Although they introduce some helpful insights, they are far from offering either necessary or sufficient conditions for the concept we are seeking.

With this I conclude my discussion of the contributions of each of the three central concepts that have featured in explanations of manipulation – rational persuasion, deception, and coercion (or pressure). We saw that none of them offers a straightforward way to conceptualize manipulation. Much to the contrary, in fact: they all point to the complexity, contextuality, and the missing intersubjective dimensions involved in “manipulation.” So much so, that our analysis raises doubts about the very plausibility of searching for a definition of manipulation. In a related vein, our analysis led to the thought that we somehow must already have a certain understanding of manipulation, if we are to succeed in elucidating it.

8

We have encountered a wide range of perplexities in attempting to understand the nature of the relation between manipulation and each of the three main concepts forming its semantic field. Even if we solved all those perplexities, however, we would still face a second-tier challenge when examining the relations among the three elements and the question of how they combine to constitute manipulation. Without such understanding, our view of manipulation would be at best fragmentary.

Sapir Handelman writes: “manipulation dissected looks like a weird mixture of persuasion, coercion, and deception.”Footnote ⁷² Beyond identifying the ingredients that constitute manipulation, “weird mixture” is not really informative.Footnote ⁷³ It leaves open all questions regarding the combinations that do versus those that do not produce manipulation. Do all factors participate in all instances of manipulation or are there manipulations in which the contribution of one or more factor is zero? Are there combinations that preclude the possibility of manipulation? What are they, and what grounds this preclusion? Since there are very different possible combinations, what makes all those that do generate manipulation equally instances of manipulation? Is there a simple zero-sum game between the ingredients or can we increase the level of one without it coming at the expense of others? More disturbingly: it is not clear what is meant in claiming that the three ingredients “mix” with one another. Persuasion, deception, and coercion are very different phenomena, and so we need a serious explanation of how at all they can combine (for instance: persuasion provides information while deception conceals information) – and even more precisely and demandingly: combine so as to produce manipulation. Finally, we should also ask whether there aren’t other factors that participate in the constitution of manipulative actions.

The only serious attempt to provide a unifying theory has been Robert Noggle’s, so we should inspect his proposal.Footnote ⁷⁴ Noggle’s theory is premised on the observation that manipulations can be divided into two groups: the “trickery” group, which is fashioned after deception, and the “pressure” group, which is fashioned after coercion. Since each group explains well some cases but not those of the other group, Noggle entertains the possibility that “there are two irreducibly different forms of manipulation.”Footnote ⁷⁵ (The definition of “manipulation” would then be disjunctive.) This, however, would be problematic inasmuch as the underlying unity of the phenomenon that we purport to describe would remain obscure. Not surprisingly, therefore, Noggle writes: “The fact that we use the term ‘manipulation’ to refer both to trickery tactics and pressure tactics cries out for explanation.”Footnote ⁷⁶

Noggle’s attempted solution involves subsuming pressure-based manipulations under the trickery view, by claiming that they too induce their target to adopt a mistaken mental state. His account of manipulative pressure reads: “A manipulates B into doing X if A creates either non-optimalizing incentives for doing X, or non-coercive disincentives for failing to do X.”Footnote ⁷⁷ Let us first explain this clearly. A wants B to do X, which B does not want to do. To secure B’s cooperation, A can either make doing X very profitable for B or make the cost of not doing X very high for B. In both of these cases, A creates conditions under which it becomes rational for B to do X: in the former, X becomes B’s best option, in the latter, X becomes B’s least bad option. Manipulative pressure, Noggle argues, fulfills neither of these two options. Rather, “manipulative pressure can only succeed when the target mistakenly chooses the lesser good.” What is the nature of this mistake? Noggle answers: “successful manipulative pressure relies on the target’s propensity to act akratically”;Footnote ⁷⁸ in akrasia, we choose what is convenient at the moment rather than what is best supported by reasons. The manipulator harnesses the target’s tendency to irrationally overreact to an incentive or disincentive under pressure, to get the target to choose the action that the manipulator wants the target to choose. Conclusion: “manipulative pressure and trickery-based manipulation are […] attempts to influence the target’s behavior by getting her to make a mistake. Trickery-based manipulation involves attempting to get the target to adopt a mistaken belief, desire, emotion, or other mental state. Manipulative pressure attempts to get the target to respond akratically to an incentive or disincentive in a way that is disproportionate to its actual magnitude.”Footnote ⁷⁹

In reflecting on Noggle’s suggestion, consider first the following example: “Your partner wants to go on a family camping trip but you don’t. While you’re discussing it, your partner calls out to your children ‘Hey kids! Who wants to go on a camping trip?’ The children cheer. You correctly judge that it’s better to go on the camping trip (despite the drawbacks) than to disappoint your children.”Footnote ⁸⁰ The method of manipulation in this example is to corner you into a socially awkward situation. You are pressured, yet not tricked, as you understand precisely what is happening. Now this is obviously not a case of coercion, but it seems equally wrong to argue that you made a mistake by capitulating: The decision to not upset the kids does not seem akratic, but rather rational, even wise. The point is that it would be wrong for Noggle to argue that given that the response is rational, then there was no manipulation. To the contrary, this does look like a clear case of pressure manipulation, yet one that does not involve making a mistake. More generally, the manipulator here changed the circumstances in which the partner is called to operate – we can treat this as a case of “indirect” manipulation – and manipulations of this kind can often be pulled through without subverting rationality, as we have already seen (recall for instance Salami and Shrewd Beggar). This therefore provides a template for constructing many parallel counterexamples to Noggle’s suggestion.

Noggle’s proposition is problematic also in another sense. His analysis relies on disproportionality (between the intensity of incentive and response), yet this factor by itself does not overlap well with the existence or inexistence of manipulation. Two incentives can be equally strong (both objectively and subjectively), yet linking the one with a specific option will have a manipulative character while linking the other would not. For instance, in certain scenarios, offering an incentive of a financial type would not seem manipulative but offering an incentive of a sexual type would seem manipulative (there are imaginable scenarios between sexual partners where the opposite would be true), despite the incentives being of similar magnitudes. The same kind of consideration also holds regarding disincentives. Whether pressure to choose a certain alternative is manipulative or coercive does not depend solely on the magnitude of the disincentive, but also on its character. For instance, under some circumstances the retraction of encouragement can be on a par with (i.e., its psychological effect of similar magnitude as) the retraction of welfare support, yet the former is much more likely to seem manipulative, and the latter coercive. Moreover, the former may exert greater psychological pressure (being accordingly less resistible) than the latter, and yet the diagnosis of manipulation versus coercion would not change. (Notice that – not surprisingly – this criticism echoes our criticism above of the view that manipulation can be explained strictly by reference to the intensity of pressure.)

Consider next the following case. You are wavering between options x and y. I, who know you well and heard you explaining your present “hesitations,” understand immediately that you are wavering for no other reason than that you have a hard time making final decisions, and that you’ll end up choosing x (both by extrapolating from many previous decisions of yours, and because everyone can see clearly that x is the best choice for you by your own lights). Since I care about you, I want you to choose already, before the choice might no longer be available. I therefore rush you and prod you in ways that I know will make you anxious and irritated, until, sweating under the pressure, you shout “Okay, okay, okay, I’ll choose x!” and indeed you do. Influence here bypassed reason – it was a pressure manipulation. Yet, pace Noggle, it did not involve your making a mistake, nor did it involve akrasia (or even “reverse akrasia”).

In the above example, the pressure manipulation did not lead its target to accept a less-than-optimal option. It is possible, moreover, that a target of manipulative pressure will indeed be led into accepting a less-than-optimal option or a less-than-coercive proposal, and yet that that will not amount to making a “mistake.” I can here merely hint at this deeper challenge to Noggle’s unifying theory of “manipulation.” Social life is a grid of pressures within which we all normally and routinely live. Healthy adaptability requires a measure of readiness to succumb to social pressure even when it seems not optimally rational. There is virtue in having the appropriate disposition to be swayable in the appropriate moments. Importantly (and relevantly for the present point), this swayability is not the conclusion of a second-level cost-benefit consideration (e.g., how much energy it is rational to expend on resisting pressure), but rather a well-cultivated adaptation to the social grid, which no doubt includes an appropriate measure of self-doubt regarding one’s own judgments, a general understanding that others see things that you don’t (which they cannot explain to you at the moment), and a healthy intuition about when to place trust in others. If this is true, then succumbing to some pressure manipulations, despite the “disproportion” involved, may not always count as a mistake. (I return to discuss this point in Chapter 5.)

As we have seen, the attempt to understand manipulation encounters major obstacles when assessing it vis-à-vis each of the main concepts that form its semantic field and when attempting to understand the conditions under which these elements interact to give rise to manipulation. (In Chapters 2 and 3, I analyze “manipulation” as a cluster concept that is identified through perception; this will offer a solution to the problem of providing a formula for the interaction of the different components of manipulation.)

9

One last point regarding the general orientation of this investigation is that we will be seeking a purely descriptive (non-moralized) concept of manipulation. Allen Wood explains:

We do not reject the possibility of a moralized concept of manipulation. Rather, it is natural, and arguably methodologically advisable, to first seek a morally neutral account of a concept, and only if this seems unfeasible, to then inspect the possibility of a moralized (thick) concept. Hence, we here explore the possibility of a non-moralized (purely descriptive) concept of manipulation, while remaining open to the possibility of revising this orientation if the results of our exploration suggest such a need. As we will see, however, our analysis will not suggest the necessity to revise the assumption of a non-moralized concept. Quite the opposite: Chapters 4 and 5 will attempt to show that manipulation is not inherently morally bad – and this, if true, will rule out the possibility of a moralized concept.

And now, after reviewing the basic terrain, let us finally move on to chart an alternative topography.