The Mistake Account of Manipulation

Robert Noggle, Central Michigan University (Philosophy, Anthropology & Religion)

ORCID 0000-0002-4682-3825

Assigned DOI: doi.org/10.3998/jpe.9951

Abstract

This paper discusses the four main elements of a theory of manipulation which I call the Mistake Account, and which I defend in my 2025 book on manipulation. This theory claims that manipulation is influence that operates by inducing a mistake into the target’s psychological states or processes. It is often effective due to imperfections in human reasoning, including the shortcuts we use to make good decisions rapidly most of the time. It is often bad because it harms or undermines the autonomy of the target, but it is always bad because it induces mistakes, and mistakes are bad.

---

At one time or another, we’ve all been manipulated. Manipulation is often easy to recognize—though all too often we only recognize it after we’ve fallen for it. But what exactly is manipulation? Why do we fall for it? Why do we feel mistreated when we do? I try to answer these questions in Manipulation: Its Nature, Mechanisms, and Moral Status (Noggle 2025), and will try to do so more briefly here.

To think about manipulation, it helps to start with examples:

1. In Shakespeare’s play Othello, Iago plays on Othello’s emotions to incite anger and jealousy towards his new bride Desdemona.
2. A pickpocket jostles his victim, misdirecting his attention away from the fact that his wallet is being extracted.
3. In the film that gives this tactic its name, Gregory gaslights Paula into distrusting her own sound judgment and listening to him instead.
4. Adolescents use peer pressure to get one of their fellows to shoplift.
5. A teen nags his parents to get them to upgrade his perfectly adequate cell phone.
6. A scoundrel charms his victim into craving the scoundrel’s approval, which he makes conditional on complying with his demands.

These seem like clear cases of manipulation. But what do they have in common that makes them all examples of the same phenomenon?

Well, none of these examples of manipulation are forms of rational persuasion, good-faith bargaining, or coercion. Thus, it might seem tempting to define manipulation as influence that does not fall into one of those other categories. But this simple answer faces problems. Consider these examples:

• Leading a large tour group by showing the people at the front where to go and counting on the rest to use the human tendency to “follow the crowd” to get there too.
• Cheering up a sad friend by taking her to a comedy film.
• Getting someone to feel guilty about having harmed someone.

None of these influences fit the standard descriptions of rational persuasion, bargaining, or coercion. Yet they do not seem like obvious cases of manipulation; someone who described them as manipulation might reasonably be accused of hyperbole, or using the term ‘manipulation’ in a way different from how we use that term in ordinary conversation to, among other things, express moral disapproval of (or at least discomfort with) a form of influence. Clearly, we need a more sophisticated account of manipulation.

Several accounts of manipulation have been proposed. However, they all face serious problems. For example, some people define manipulation as hidden influence. However, some forms of manipulation—like nagging and peer pressure—operate in plain sight. Some people define manipulation as influence that bypasses reason. However, philosophers don’t agree on what reason is, and how things like appeals to emotions relate to reason. This makes the “bypassing reason” view unclear and incomplete, and likely to remain so for the foreseeable future. Some people define manipulation as something like deception. However, some forms of manipulation—like nagging and peer pressure—don’t seem to have much in common with deception.

This last idea—that manipulation is related to deception—plays an important role in the theory of manipulation that I defend. So, let’s turn now to that theory, which I call the Mistake Account. It comprises four main elements: building on the trickery account; an account of manipulative pressure; drawing a parallel between deception and manipulation; and an account of the moral status of manipulation.

Building on the Trickery Account

Many years ago (Noggle 1996), I argued that manipulation involves getting someone to adopt a mental state that falls short of a normative ideal for that kind of mental state. For example, Iago’s behavior is manipulative because it gets Othello to feel misplaced or inappropriate anger and jealousy. On this view, manipulation is a broader category of which deception is a special case: Whereas deception tricks someone into adopting a faulty belief, manipulation tricks someone into adopting a faulty mental state of any kind—belief, desire, emotion, pattern of attention, etc. I call this the Trickery Account.

The Trickery Account properly identifies many forms of manipulation. Playing on the emotions involves getting someone to adopt an inappropriate emotion. Iago is a prime example, of course. Another is a con artist getting someone to feel inappropriate empathy for a non-existent Nigerian prince. Misdirection involves inducing a faulty pattern of attention, as when the pickpocket gets his victim to pay more attention to being jostled than to his wallet being removed. Gaslighting involves getting someone to have inappropriately low confidence in her own sound judgement.

The Trickery Account also accommodates something I call the Dual Use Phenomenon. Many examples of clearly manipulative influence have non-manipulative counterparts. Inducing empathy for a non-existent Nigerian prince is manipulation because empathy in such a case is misplaced. But it is not manipulation to try to induce someone to feel appropriate empathy for real people suffering undeserved misfortune. The pickpocket’s jostling is manipulation because it directs the victim’s attention away from the important fact of his wallet being removed. But it is not manipulation to direct your attention toward something important, like a piano that’s about to fall on you. Gaslighting is manipulation because it gets someone to adopt an inappropriate distrust of her own sound judgement. But it is not manipulation to get an intoxicated friend to distrust his impaired judgement that he can drive home safely.

So, the Trickery Account gets a lot right. The Mistake Account retains what the Trickery Account gets right. What makes it different from (and superior to) the Trickery Account is its ability to account for manipulative pressure.

Manipulative Pressure

In an outstanding paper, Marcia Baron (2003, 40) observes that an important form of manipulation “has at its core pressure to comply,” which “may involve a threat that does not rise to the level of being coercive”. She makes a compelling case that manipulation sometimes involves pressure that falls short of what we find in coercion.

Nagging is perhaps the simplest example of manipulation involving pressure. Here the pressure is exerted by the annoyance forced upon the target for refusing to do what the manipulator wants. Manipulators can also harness conflict aversion, that is, the discomfort people often feel with social conflict, to exert pressure. For example, a manipulator might make it difficult to refuse a request by pre-emptively thanking you for agreeing, thus making it more awkward to refuse. Or a manipulator might harness reciprocity by doing a trivial or unwanted favor for someone before asking for a much bigger favor. Here the pressure takes the form of the discomfort people typically feel when they refuse a request from someone who just did something for them.

The forgoing examples employ what we might call negative manipulative pressure, since they involve disincentives for refusing to do what the manipulator wants. But manipulative pressure can also be positive, involving incentives for agreeing to do what the manipulator wants. In a charm offensive, the manipulator gets the target to crave the reward of the manipulator’s approval and then makes that approval conditional on doing what the manipulator wants. A tactic I call “conditional flattery” offers a reward in the form of a compliment that only applies if you do what the manipulator wants. For example, “someone with your organizational skills would be an outstanding chairperson for this (burdensome) committee.”

The Trickery Account lacks a natural way to handle manipulative pressure, since pressure seems different from trickery. For example, nagging is such an unsophisticated tactic that it would seem silly to call it trickery. Nor does it seem to require getting the target to adopt a faulty mental state. It does get the target to feel annoyed, but since nagging is genuinely annoying, it’s difficult to see such annoyance as a faulty mental state. In short, pressure does not seem to either resemble or require anything aptly described as trickery or as inducing a faulty mental state.^[1]

The second element of the Mistake Account is meant to solve this problem. To see how it works, consider an example. A teen wants his parent to replace an adequate cell phone with a more expensive one. The parent declines, on the grounds that it is not a good use of the family’s financial resources right now. Not satisfied, the teen nags. The resultant irritation eventually leads the parent to give in and purchase the new phone, despite continuing to regard this as a poor use of family resources.

Nagging often works. But how? If the nagging were not just annoying but debilitating, then it might change the parent’s cost-benefit ratio enough to make buying the phone optimal. But this “hyper-nagging” would be more like coercion than manipulation. That’s because coercion involves disincentives large enough to make complying with the coercer’s demand the target’s best, or least bad, option. “Your money or your life” is coercion. “Your money or I will call you a banana-head” is not.

We commonly distinguish manipulation from coercion, and manipulative pressure is commonly said to fall short of being coercive.^[2] If this is correct—if manipulative pressure does fall short of being coercive—then it must involve disincentives that are too small to make compliance the target’s best, or least bad, option. Thus, if the parent is manipulated rather than coerced, we must imagine him giving in not because doing so is optimal, but despite it being sub-optimal.

But now we face a puzzle: If manipulative pressure does not make it optimal for the target to do what the manipulator wants, then how does it work? Why would I comply with the demands of a manipulator who threatens me with something that’s not bad enough to make refusing worse than complying? The answer is that manipulative pressure works when and because the target makes what philosophers call an akratic choice, that is, one made against the chooser’s better judgement. If the parent judges it better to suffer the nagging than to buy the phone, but buys it anyway, then he chooses akratically. If I agree to an onerous task to avoid the fleeting social awkwardness created by a manipulative exploitation of reciprocity or conflict aversion, then I choose akratically.

Similarly, we should say that positive manipulative pressure involves incentives that are too small to make compliance optimal for the target, and which induce compliance only when and because the target chooses akratically. A manipulative charm offensive works when the target akratically chooses the manipulator’s approval despite it not being worth what must be done to get it. Conditional flattery works when the target akratically agrees to some task in order to claim a compliment that, while nice, does not compensate for the burden of the task. Thus, it appears that manipulative pressure—whether positive or negative—works when and because the target makes an akratic choice to comply in response to an offer or a threat that is too small to make complying optimal.

Whatever else is true of akratic choices, making one is clearly a mistake.^[3] Seeing this reveals a deep commonality between manipulative pressure and manipulative trickery: Both work by getting the target to make a mistake. It is clearly a mistake to choose akratically. And it is clearly a mistake to adopt a faulty mental state. Thus, we can accommodate manipulative pressure by replacing the Trickery Account’s idea that manipulation induces faulty mental states with the broader idea that manipulation induces mistakes. Thus, we can retain what the Trickery Account gets right, while recognizing how pressure can also be manipulative. This is the second element of the Mistake Account, and the one which gives it its name.

The book contains an entire chapter detailing various mistakes involved in manipulation. Here’s the short version: Beliefs are mistaken when they are false, but this criterion also applies to other belief-like states, such as impressions, which might not fit certain definitions of “belief.” Emotions are mistaken either by failing to fit the facts (as when Othello is jealous of Desdemona despite having no reason for jealousy) or by being too strong or too weak for the circumstances (as when someone is excessively fearful of a small, well-behaved dog). Mistakes of attention and weighting occur when the amount of attention paid to some fact, or the amount of weight placed on it in decision-making, is disproportionate to its actual importance. The various forms of manipulative pressure all involve the mistake of akratic choice.

There’s also a chapter on psychological vulnerabilities to these mistakes. Here, again, is the short version. Humans face a huge problem: We must often make decisions quickly. However, decision-making is an open-ended process without well-defined endpoints, and our cognitive capacities are limited. Our solution is a division of labor. It is commonly accepted among psychologists that two different kinds of processes contribute to human decision-making.^[4] Unconscious Type 1 processes rapidly identify what is relevant in the situation and often recommend a response. Slower, conscious Type 2 processes might ratify this recommendation, or they might initiate deliberation based, at least initially, on what the Type 1 processes identify as relevant. This division only works because Type 1 processes are fast. They are fast because they take shortcuts. These shortcuts create vulnerabilities to errors, many of which can be exploited by manipulators. These include the well-known cognitive and decision-making biases, such as the availability heuristic; framing effects; the status quo bias; social proof heuristics, and the immediacy bias that often underlies akratic choices. But we should not assume that manipulation works only through Type 1 processes. Human capacities for conscious reasoning are far from perfect, and their imperfections can also be exploited by manipulators.

Despite its shortcomings, our imperfect rationality is an amazing achievement. It allows creatures with mushy brains, developed by the blind watchmaker of evolution, to make largely good decisions in time to meet the challenges of a hostile environment. When we do get manipulated—and we all do—we should remember that much of our vulnerability to manipulation is the price we pay for our ability to get things right promptly most of the time. This speed-accuracy trade-off increases our vulnerability to manipulation, but it’s also what allowed our ancestors to overcome the challenges that threatened their survival.

The Parallel with Deception

A third element of the Mistake Account is the parallel it draws between manipulation and deception; in particular, it claims that deception is a subtype of manipulation. I do not claim that we have strong and uniform intuitions that deception is a subtype of manipulation. Instead, I regard it as more like a discovery. It emerges when we think about how to describe what happens when someone is manipulated. Consider this example:

Snow Flurries: Maddie does not want Tim to drive to a party tonight. She sees that the weather forecast predicts light snow flurries. She tells Tim that snow is expected, without saying “flurries” or mentioning that no accumulation is expected. She then ominously recounts graphic tales of accidents on slippery roads. This makes Tim too afraid to drive to the party.

This seems like a clear case of manipulation. But what, exactly, is the correct description of what happened to Tim? Did Maddie get Tim to have a mistaken fear of driving tonight? Or did Maddie get Tim to have a mistaken belief about the likelihood of hazardous road conditions tonight, and this belief made him afraid to drive? Note that on this second description, Maddie deceives Tim, since she gets him to have a false belief.

Now, if deception is different from manipulation, then whether Maddie manipulated Tim depends on whether she got Tim to have a false belief. If so, she deceived Tim; if not, she manipulated him. But it seems implausible for the question of whether Maddie manipulated Tim to depend only on whether Maddie caused Tim to have a mistaken fear, or whether she caused him to have a mistaken belief that caused that fear. Depending on how we specify the details of the case (and how we settle certain difficult philosophical questions about the nature of mental states), these descriptions may be little more than two different ways of describing the same set of psychological facts. And even if there is a real difference between the two, it seems too subtle to be the deciding factor about whether manipulation has occurred.

Similar remarks apply to other forms of manipulation. When someone gives in to manipulative pressure, does she choose against her better judgement? Or does she irrationally and temporarily change her belief about what it is best to do.^[5] Does the victim of gaslighting adopt an inappropriately low confidence in her own judgement, or the belief that her judgment is untrustworthy? If deception and manipulation are different things, then determining whether manipulation as opposed to deception took place in such cases requires determining which description is correct. Often, these will be little more than two ways to describe the same set of psychological facts. And even when there are psychological facts that make one description more accurate than the other, it is difficult to believe that they are significant enough to determine whether a given case is one of deception or manipulation, if we insist on seeing them as different phenomena.

However, if we treat manipulation as a broader category which includes deception, then classifying an influence as manipulation will never depend on whether to attribute a false belief to someone who acquires excessive fear, or doubts her own judgement, or chooses the lesser good. Thus, I think we should accept that deception is a subtype of manipulation. Once we do this, we can extract some very useful work from the parallel between these two concepts. This is because our intuitions about deception are often more clear and more uniform than our intuitions about manipulation. Thus, when we lack clear and uniform intuitions about some aspect of manipulation, we can use the concept of deception as a template to help us decide, in a principled way, what to say about manipulation.

For example, we can use this parallel to answer the following question: Should we define manipulation in terms of what the manipulator regards as a mistake, or in terms of what really is a mistake? Is Iago’s behavior manipulative because Iago believes that it’s a mistake for Othello to be jealous, or because it really is a mistake for Othello to be jealous? Consider these statements:

(1) Danielle acted deceptively toward Tammy.

(2) Marvin acted manipulatively toward Tony.

Whether (1) is true depends on Danielle’s beliefs. It is true only if Danielle tried to get Tammy to believe something that Danielle believed was false. If Danielle believes that there is candy in the cupboard, she acts deceptively if she tries to get Tammy to believe that there is no candy in the cupboard. This is so even if Danielle herself is mistaken about the cupboard’s contents.

I recommend treating (2) similarly to (1), so that its truth value depends on what Marvin believes is a mistake. Someone acting either manipulatively or deceptively is trying to induce a mistake, and whether a person is trying to do that depends not on what actually is a mistake, but on what the influencer believes is a mistake. These considerations suggest the following definition:

M acts manipulatively toward T if and only if M tries to influence T’s behavior or mental state by causing T to have what is, from M’s perspective, a mistake in T’s psychological states or processes.^[6]

Next, consider these statements:

(3) Danielle deceived Tammy.

(4) Marvin manipulated Tony.

Statement (3) means that Danielle succeeded in her attempt to get Tammy to believe something false. This is only true if Danielle got Tammy to believe something which Danielle believes is false and which really is false. So (3) would be true if Danielle got Tammy to believe that there is no candy in the cupboard when both (a) Danielle believes there is candy in the cupboard, and (b) there really is candy in the cupboard. If Danielle is mistaken and there really is no candy in the cupboard, so that she inadvertently got Tammy to believe something true, then she still acted deceptively. But she did not deceive Tammy, because Tammy did not acquire a false belief.

I recommend treating (4) similarly to (3), so that (4) is only true if Tony made what Marvin believes is a mistake, and if that thing really is a mistake. Thus, to act deceptively or manipulatively is to attempt to induce a mistake. To manipulate or to deceive someone is to succeed in that attempt. Thus:

M manipulates T if and only if:

(1) M acts manipulatively toward T, and

(2) Each of these success conditions is met:

a. What, from M’s perspective, is a mistake really is a mistake;

b. T makes (approximately) this mistake; and

c. T’s making this mistake has the (approximate) effect on T’s behavior or mental state that M intended.^[7]

So far, I’ve been assuming that manipulation requires intent. That is usually what we say about deception. But not always. We sometimes call inanimate objects deceptive, even though they lack intentions. Sometimes, we do this because the object was designed to induce false beliefs. Thus, we call a secret door deceptive because it was designed to induce the false belief that it’s just an ordinary bookcase. This usage is derivative—the deceptiveness of the object derives from the deceptive intentions of its designer. But we also sometimes call something deceptive even if it was not designed to induce false beliefs. We call conditions in the eye of a hurricane deceptive because it seems like the storm has passed, but the other half is on its way. “Looks can be deceiving,” we may say, even when no one made the thing look that way. This looser, metaphorical usage indicates a tendency to induce false beliefs—even if no one intended for that to happen.

We can identify parallel derivative and metaphorical uses of the term ‘manipulative.’ We commonly describe tactics, forms of influence, or devices like dark patterns as manipulative in the derivative sense when they are deliberately designed to induce mistakes. But it also seems reasonable to allow the term ‘manipulative’ to have a looser, metaphorical sense, indicating a tendency to induce mistakes, even if no one intended for that to happen. This is important because some technologies, like artificial intelligence, might develop tendencies to induce mistakes without being designed to do so, and without having their own intentions. We might decide that such systems are manipulative in the same metaphorical way that we call the eye of a hurricane deceptive.

The Moral Status of Manipulation

Another puzzle that becomes less puzzling when we embrace the parallel between manipulation and deception concerns the moral analysis of manipulation, which is the fourth element of the Mistake Account.

Clearly, the term ‘manipulation’ often has a negative moral connotation. This connotation suggests that manipulation is pro tanto immoral, that is, that there is always some moral reason to refrain from manipulation, even though that reason can sometimes be outweighed by opposing moral reasons. Some people define ‘manipulation’ more broadly, to include influences that do not seem pro tanto immoral. If you define ‘manipulation’ this way, then please treat my use of the term as applying only to those cases of manipulation that do seem pro tanto immoral. What accounts for the pro tanto immorality of (these forms of) manipulation?

Well, manipulation often harms the target; when it does, this fact may be the most morally significant one. But manipulation remains pro tanto immoral even when it does no harm. Using tactics like those Iago uses on Othello to get a friend to be angry enough at an abusive partner to leave a dangerous relationship might do no harm on balance. It might even be morally justified. But it’s still pro tanto immoral, because it would have been better, morally speaking, to protect the friend from abuse without resorting to manipulation.

Manipulation often undermines the target’s autonomy. But suppose that Margaret helps her friend Tyler kick his smoking addiction by getting Tyler to be disgusted that tobacco leaves sometimes get bird droppings on them. Since the tobacco is washed during processing, this disgust is misplaced, and inducing misplaced disgust is manipulation. But this manipulation enhances Tyler’s autonomy (at least as autonomy is commonly understood), since it helps free Tyler of the addiction so he can act on his authentic desires. It might even be morally justified. But it’s still pro tanto immoral, since it would have been better, morally speaking, to help Tyler without resorting to manipulation.

So, if its effects on well-being or autonomy do not fully explain the pro tanto immorality of manipulation, what does? I contend that the normative status of manipulation rests on the normative status of mistakes. Manipulation is bad because mistakes are bad.

This is easiest to see with beliefs. Beliefs are often said to “aim” at truth; their function is to reflect or represent accurate information (Searle 1983; Humberstone 1992; Smith 1987). Consequently, beliefs are governed by a norm of accuracy, so that believing falsely is a kind of failure. Such a failure is a misfortune (Fried 1978, 62–64; Williams 2004).

This is not to say that a person is always harmed by a false belief. Indeed, some false beliefs protect the believer from harm. But the fact that believing something false can be less bad overall than knowing the truth does not mean that it’s not a misfortune. It just means that sometimes it’s the lesser evil.

The Mistake Account extends this analysis to other mistakes. It is a mistake to ignore something important, to feel an unfitting emotion or one that is too strong or weak, to react too strongly to an incentive or disincentive, and so on. These various mistakes are also misfortunes, since they are ways in which something has gone wrong. Something has gone wrong when an emotion is unfitting, excessive, or deficient for the circumstances. Something has gone wrong when someone fails to pay attention to something important. Something has gone wrong when someone acts against her judgement about what is best to do. Having something go wrong is a misfortune, even though there can be worse misfortunes. And it is pro tanto immoral to inflict a misfortune on someone, even if doing so is sometimes justified as the lesser evil.

But why are mistakes misfortunes? The answer lies in our status as rational agents. To be a rational agent—even an imperfect one—is to care about getting things right: Believing what is true, having emotions that are fitting and non-excessive, paying attention to what’s actually important, doing what we judge best, and so on. Of course, having good reason to care about getting things right does not preclude us from caring about other things as well, or from sometimes ranking other cares above getting things right. But to place no value at all on getting things right is to reject one’s status as a rational agent. So, our status as rational agents (even imperfect ones) gives us a reason to regard mistakes—and thus manipulation—as bad.

But how bad? Here again, the parallel between manipulation and deception is helpful. It suggests that the moral status of any given instance of non-deceptive manipulation is similar to that of a relevantly similar instance of deception. The argument for this claim resembles the argument for the parallel between manipulation and deception. When we can describe a case of influence equally well as inducing or not inducing a false belief, it seems implausible that its moral status would depend on which description we choose. To describe an (intentional) influence as inducing a false belief is to describe it as a case of deception. So, if a case of manipulation can be redescribed as a case of deception, and if it's implausible to think that this redescription changes its moral status, then we can use our moral intuitions about deception to guide our assessment of the original case of manipulation.

As with deception, some instances of manipulation are justified by their consequences. Presumably, authorities are morally justified in manipulating a terrorist into revealing the location of a hidden explosive. Less dramatically, getting drivers to slow down before a dangerous curve by placing lines across a roadway that get closer together before the curve is manipulation if it works by creating a mistaken impression of increasing speed. But it seems morally justified because of its potential life-saving benefits. Similarly, as with deception, some instances of manipulation are morally trivial, such as creating a false impression of enthusiasm when teaching dull material, or using gentle peer pressure to encourage a shy friend with a great singing voice to join in at karaoke. Although these instances of manipulation are less than morally ideal, their immorality is sufficiently minor to be easily outweighed.

Conclusion

Here then are my short answers to the three questions with which we began. Manipulation is influence that introduces mistakes into the target’s psychological states or processes. It works because of the imperfections in human reasoning, including the cognitive and decision-making shortcuts that allow us to make prompt decisions that are pretty good most of the time. It is often bad because it causes harm or undermines autonomy. But it is always bad because it induces mistakes, and doing that is not an ideal way to treat another person, even if, in the real world, it is sometimes the least bad option.

References

Baron, Marcia. 2003. “Manipulativeness.” Proceedings and Addresses of the American Philosophical Association 77 (2): 37–54. https://doi.org/10.2307/3219740.

Fried, Charles. 1978. Right and Wrong. Harvard University/Belknap Press.

Humberstone, I. L. 1992. “Direction of Fit.” Mind 101 (401): 59–83.

Kahneman, Daniel. 2013. Thinking, Fast and Slow. 1st edition. Farrar, Straus and Giroux.

Noggle, Robert. 1996. “Manipulative Actions: A Conceptual and Moral Analysis.” American Philosophical Quarterly 33 (1): 43–55.

Noggle, Robert. 2025. Manipulation: Its Nature, Mechanisms, and Moral Status. New Topics in Applied Philosophy. Oxford University Press.

Searle, John R. 1983. Intentionality: An Essay in the Philosophy of Mind. Cambridge University Press. https://doi.org/10.1017/CBO9781139173452.

Smith, Michael. 1987. “The Humean Theory of Motivation.” Mind (Oxford) XCVI (381): 36–61. https://doi.org/10.1093/mind/XCVI.381.36.

Williams, Bernard. 2004. Truth and Truthfulness: An Essay in Genealogy. 1.3.2004 edition. Princeton University Press.

Tags  

Manipulation gaslighting deception autonomy