Respect for Humanity

Abstract

Philosophers have defined autonomy in a variety of ways. In this chapter, we present a Kantian account of personal autonomy as the capacity to set and pursue your own ends (which Kant refers to as “humanity”). We explain how this involves two distinct components: capacity and authenticity. To have autonomous capacities, you must possess baseline abilities (to form intentions, evaluate commitments, etc.); be free of external constraints; be free of certain cognitive inhibitions, and you must have a sufficiently wide range of options. In order for your desires to count as authentic, they must be free of manipulation and coercion, consistent with other motivational states, free from alienation, and they must not be adaptive preferences. After developing our conception of autonomy, we explain and defend Kant’s view that we are morally obligated to respect humanity in ourselves and in others. We conclude by briefly explaining why other moral theories have similar commitments to the moral significance of autonomy.

Every human being has a legitimate claim to respect from his fellow human beings and is in turn bound to respect every other. Humanity itself is a dignity; for a human being cannot be used merely as a means by any human being (either by others or even by himself) but must always be used at the same time as an end.

—Kant, The Metaphysics of Morals, 6:462

Let’s admit it, we in the consumer web industry are in the manipulation business. We build products meant to persuade people to do what we want them to do. We call these people “users” and even if we don’t say it aloud, we secretly wish every one of them would become fiendishly addicted.

—Nir Eyal (See Eyal (2012). This quote comes from an article Eyal wrote for Forbes. He uses nearly identical language in chapter 6 of Hooked. Cf. Eyal (2014, 164–65).)

2.1 Introduction

Many of us share the intuition that autonomy matters. You ought to be the author of your own life story. And if you were to discover that your actions and desires are being manipulated by someone else, you would probably feel uncomfortable about that fact. In what follows, we will argue that this is precisely what is happening with our mobile devices.

But before moving on to our moral arguments about technology and digital minimalism, we must begin by establishing the ethical framework that we will use throughout the book. We do not want to assume that everyone shares our Kantian commitments. Instead, we would like to defend them as independently plausible moral views. We begin with autonomy. Regardless of the moral theory that you happen to hold, you probably believe that autonomy is morally significant in some way. But how should we define autonomy? And why exactly does it matter morally? Our aim of this chapter is to answer those questions.

We understand autonomy as self-government: it is a rational agent’s capacity to set and pursue her own ends. This may sound uninformative at first, but it will be fleshed out considerably in the next few sections. Once we have established a definition, we argue for a moral obligation to respect autonomy. This conclusion should start off with some intuitive appeal. It is fairly commonplace to acknowledge violations of autonomy as moral wrongs. That’s why people often think that there is something morally objectionable about manipulation and coercion. Such acts undercut an agent’s ability to set and pursue her own ends. Finally, we will show why certain technologies, like smartphones and social media, threaten to undermine our autonomy in Chap. 3.

2.2 Humanity: Autonomy as Rational Agency

When Kant talks about “autonomy,” he is referring to moral autonomy, which he understands as an agent’s capacity to be governed solely by a moral law that proceeds from her own rational will. He defines (moral) autonomy as “the property of the will by which it is a law to itself” (G 4:440). According to Kant, the moral law springs from one’s own reason, so when we obey the moral law, we are submitting to a principle that we have given ourselves. For Kant, morality and autonomy are two sides of the same coin: morality consists in self-legislation through reason.

This idea of moral self-legislation harmonizes nicely with the Greek origin of the word: “autos” meaning “self” and “nomos” meaning “law.” But Kant’s lofty ideal sets the bar too high to match the usage in contemporary moral debates. Nowadays, when a moral philosopher refers to “autonomy,” it typically means something quite different and much less demanding. More often than not, ethicists are talking about personal autonomy rather than moral autonomy. Personal autonomy is the ability to be in charge of our own lives, to act and reflect on the basis of our own beliefs and desires, and to pursue things in accordance with our own conception of what is good for us. This is radically different from Kant’s concept of autonomy according to which acting autonomously is synonymous with acting morally.¹ Throughout the book, whenever we talk about “autonomy,” we are almost always talking about personal autonomy. We will say very little about Kant’s idea of moral autonomy.

But it is not as if Kant makes no mention of personal autonomy.² On the contrary, he talks a great deal about the importance of rational agency. But given his special emphasis on moral autonomy, Kant uses a different word to refer to this capacity. The word he uses is “humanity.”³ Kant refers to “humanity” throughout his moral philosophy, and he tends to gloss it as “the capacity to set oneself an end” (MS 6:392).⁴ Kant thinks that this capacity is what sets rational beings like us apart from the rest of the animal kingdom. Our inclinations do not necessitate our actions. Whatever desire we may have, we are capable of rejecting it for the sake of another end. What is more, we have the ability to reflect on the value of our chosen ends and revise those choices; we are free to pursue “any end whatsoever” (MS 6:392). You might set out to become an investment banker, because you wanted to make a lot of money. But then you get to college and decide that you do not, upon reflection, care about getting rich, so you decide to pursue a career in academic philosophy. As a rational being, you get to set your own ends, whatever those might be.

That is the idea of autonomy that we will use throughout the book. Simply put, autonomy is the ability to set and pursue your own ends. Of course, we will need to say more about what it means for ends to be considered your “own,” but this is a good starting place for the definition. Philosophers have employed a wide variety of metaphors to express the core idea of autonomy. To be autonomous is to be the author of your own life story (Raz 1986, 369). Autonomy requires sovereignty, the capacity to govern yourself. Autonomy means putting yourself in the driver’s seat rather than being a mere passenger.

But metaphors can take us only so far. One way to progress beyond the metaphors is to examine the contrast class. We can develop a better sense of autonomy by considering examples of actions that lack autonomy. Sometimes we act for the sake of desires that are not truly our own. We may be coerced, manipulated, or in the throes of a powerful addiction. In such cases, we demonstrate an autonomy deficit by acting on “alien” desires. We act for reasons that are not our own, desires that were implanted in us (to employ yet another metaphor). In the next section, we will develop our account of autonomy further by considering different explanations of this phenomenon. What, then, are alien desires, and how are they inconsistent with autonomy?

2.2.1 Alien Desires

If someone pointed a gun at you and demanded that you turn over your wallet, it is fairly obvious that your decision to hand over the money was not autonomous. You were being coerced. Coercion can be understood quite simply as being forced to make a particular decision on the grounds that you have no acceptable alternative.⁵ Other kinds of autonomy deficits are not so simple to grasp. It is trickier to explain what is going on when an unwilling drug addict acts against her better judgment or when a brainwashed cultist sells her belongings and moves to the compound.

What exactly is going on when someone acts on an alien desire? Many of the contemporary theories of personal autonomy were developed with an eye toward offering explanations of what is going on in those scenarios. According to one popular family of theories, the above actions should be understood as involving a lack of coherence between the agent’s motivational states. For the unwilling addict, she may feel a strong desire for the next hit, but there is an important sense in which she does not want to do what she ends up doing. Harry Frankfurt (1971) offers a model that explains this behavior in terms of first-order and second-order desires. First-order desires refer to things that we want in the ordinary sense: you want to go for a walk, eat a piece of chocolate cake, or read the next chapter of this book. These are things that you want in the first-order sense. But at the second-order level, you could talk about whether you want to have those desires or not. You could also talk about whether or not you want to act on your first-order desire. Perhaps you are trying to avoid sugar, so you want to stop having a first-order desire for the cake. You do not want to act on your first-order desire to eat the cake. Those are second-order desires and volitions. For Frankfurt, the unwilling addict lacks freedom (or autonomy)⁶ because there is a tension between her first-order and second-order desires. She wants to use heroin, but, at the second-order level, she wants to stop having that first-order desire.

There are several theories of autonomy that require something similar to the kind of coherence that Frankfurt put forward in his influential paper. The unifying feature of “coherentist” theories of autonomy is this. An agent’s action is considered autonomous “if and only if she is motivated to act as she does because this motivation coheres with (is in harmony with) some mental state that represents her point of view on the action” (Buss and Westlund 2018). For Frankfurt, the coherence in question concerns the agent’s first-order and second-order desires. Gerald Dworkin offers a similar model according to which a “person is autonomous if he identifies with his desires” when engaging in higher-order reflection (Dworkin 1981, 212). Dworkin says that autonomy consists in “a second-order capacity of persons to reflect critically upon their first-order preferences, desires, wishes, and so forth and the capacity to accept or attempt to change these in light of higher-order preferences and values” (1988, 20).

Gary Watson suggests a slightly different approach. He argues that autonomy should be understood in terms of consistency between the agent’s desire and her evaluative judgments about what is most worthwhile. Watson is skeptical about characterizing autonomy in terms of desires. He says that agents “do not (or need not usually) ask themselves which of their desires they want to be effective in action; they ask themselves which course of action is most worth pursuing” (1975, 219). When someone acts compulsively (and thus lacks autonomy), Watson would explain this by saying that “the desires and emotions in question are more or less radically independent of the evaluational systems of these agents” (1975, 220). Considering the case of a kleptomaniac who steals compulsively, Watson concludes that “it is because his desires express themselves independently of his evaluational judgments that we tend to think of his actions as unfree” (Ibid.).

One important thing to note about these accounts is that the harmony in question often appears to be synchronic. It could be evaluated at a particular moment in time, looking only at a single timeslice of an individual.⁷ In the moment when the addict uses, we could ask, following Frankfurt, whether or not her first-order desire is consistent with her second-order volition. Or, following Watson, we could ask if her action coheres with her evaluative judgments. In contrast to this approach, Michael Bratman (2003)⁸ argues for an account that involves the agent’s long-term plans about her life. Agents form a variety of plans throughout their lives, and one of the central tasks of practical reason is to fit different plans together in a consistent and stable way. To act autonomously, we must act in a way that is consistent with the long-term plans that were the result of a deliberative process.

In addition to emphasizing the role of long-term plans, Bratman also mentions the importance of the historical processes that gave rise to the agent’s plans. He suggests that a “full story about human autonomy will also need to appeal to some sort of historical condition that blocks certain extreme cases of manipulation, brainwashing, and the like” (2003, 175–76). To see why this is an issue of vital concern, let us return to the example of the brainwashed cultist. Imagine that she forms her desire to move to the compound only after undergoing some powerful forms of manipulation. Perhaps she was subjected to an extreme form of operant conditioning. If the manipulation was successful, it is perfectly conceivable that her first-order and second-order desires might be in harmony. Not only does she have first-order desires to sell her belongings and move to the compound, she wants to have those desires and she wants them to determine her will. It seems clear, however, that we should not regard her decision as autonomous. These desires were the result of manipulation, so it would be a mistake to consider them autonomous.⁹

It is for precisely this reason that Dworkin (1976, 1988) stresses the importance of procedural independence when the agent identifies with her desires. It is not enough to merely identify with her desires; this identification must not be the result of manipulation. To address concerns like these, we might want a view of autonomy that requires us to evaluate the process through which the desire was developed. John Christman defends just such a view. In order to regard a desire as autonomous, Christman suggests that we must ask whether or not the agent resisted (or would have resisted) the development of the desire if she were to attend the process of its development (Christman 1991, 11). Furthermore, he adds that the lack of resistance must not take place “under the influence of factors that inhibit self-reflection” and that the self-reflection must be (minimally) rational and involve no self-deception (Ibid.). We can safely assume that the cultist would repudiate the development of her desires if she were to understand the manipulative processes that shaped them.

But there are those who believe that Christman’s historical approach does not go far enough when it comes to rooting out influences that undermine autonomy. One thing that all of the above approaches have in common is that they are “content neutral” and “value neutral.” They are not committed to any particular theory of the good, and they make no judgments about the content of the decision. Each of those views would allow us to judge that a person is autonomous with respect to a desire even if we think that the desire in question is bad for the agent. They include no substantive requirements about attitudes or capacities (like self-trust or self-respect) that must be present when the agent identifies with her desires. They would also make it possible to view a desire as autonomous even though it was shaped by conditions of profound injustice. For instance, consider the situation of a woman in a very conservative society in the Global South, where oppressive gender norms have made it common for women to deprive themselves of food (even to the point of starvation) in order to make sure that their husbands and sons can eat far more than they need.¹⁰ Similarly, we are likely to have reservations about the abused wife’s preference to stay with her violent husband.¹¹

Cases of this kind might pose a problem for content-neutral, or “procedural,” models of autonomy. The woman who chooses to starve herself in order to overfeed her husband and son might get a pass on all of the procedural views discussed above. Using the Frankfurt/Dworkin model, she might approve of her first-order desire when she engages in higher-order deliberation. With Watson, she might argue that her action is perfectly consistent with her evaluative judgments. She might genuinely believe that it is more important to nourish her husband and son. On Bratman’s view, we might find that her action is consistent with her long-term plans. Even Christman, whose view is explicitly concerned with the process of desire formation, might lack the resources to explain why her decision is not fully autonomous. When the woman looks back on the process that developed her desire, she might find moments in her upbringing when her mother taught her these values or when societal pressures reinforced them, but, absent any outright manipulation this might not be enough for her to repudiate it as an alien desire.¹²

But it stretches credulity to suggest that her desire to starve herself is autonomous. We should not think of her as the author of this desire. For precisely this reason, a number of philosophers—particularly feminists—have been critical of content-neutral, procedural views of autonomy. According to the procedural view, a desire is autonomous as long as the procedure that one uses to endorse the desire (e.g., by higher-order reflection) is independent of things like manipulation and coercion. Feminist views of autonomy go further, as they frequently stress the claim that the autonomous person must have “substantive” independence as well (see Stoljar 2000; Benson 1987; Oshana 2006; Mackenzie and Stoljar 2000). Substantive independence may take many forms. For instance, autonomous choices should not take place in a social context in which your set of options is highly constrained by things like oppressive gender norms. You must have a sufficiently wide range of options (Brison 2000). What is more, we cannot simply look at the procedure the agent uses to endorse her desire. We might ask questions about the content of the desire itself. Or substantive theories may require that the agent’s conclusion not demonstrate a lack of self-respect or self-worth (Benson 1994). It has also been argued that full autonomy requires that a preference not be causally influenced by injustice (Enoch 2020).

Feminist critics have also argued that we inherited a flawed view of autonomy from the Enlightenment (especially from thinkers like Kant). They criticize the “atomistic” view on the grounds that it asks us to conceptualize the autonomous person as a fully self-sufficient individual who is bereft of any dependence on social relations. Natalie Stoljar suggests that this view “is abstracted from the social relations in which actual agents are embedded. Such a conception of the self is associated with the claim that autonomous agents are, and ought to be, self-sufficient, which in turn is associated with the character ideal of the ‘self-made man’” (Stoljar 2018). And since gender norms tend to involve expectations that women ought to value certain social relations (such as the family), this view “denies women, in particular, the social and political advantages associated with the label ‘autonomous’” (Ibid.).

As a result, some feminist philosophers have defended “relational” views of autonomy. According to these views, a person’s autonomy is not compromised by the fact that she is not fully self-sufficient. No one is or ought to be an island; we can be autonomous even if we recognize the moral significance of interdependence and social relationships. There is no denying that we are, to some extent, products of socialization. And given our dependence on one another, we ought to promote social arrangements that facilitate the development of autonomy. We should avoid restricting agents’ choices in such a way that their seemingly voluntary choices demonstrate autonomy deficits, like the woman who “voluntarily” chooses to starve herself. Thus, in the political context, feminists who champion autonomy see it “as a valuable kind of individual freedom that political arrangements ought to promote” (Stoljar and Voigt 2022).

But this is precisely what leads some political and moral philosophers to hesitate from endorsing a substantive view of autonomy. According to certain views of liberalism, public reason requires us to refrain from offering justifications that are wedded to substantive value commitments that may not be shared by other members of our liberal democracy (Rawls 2005). The motivation for this reluctance is that we typically want to avoid forcing someone to live by another person’s conception of the good. We ought to offer justifications that any “reasonable”¹³ person could accept, regardless of the comprehensive doctrine to which she is committed.

On the flip side, there are those who are skeptical about this conception of public reason. There are some issues with the idealization that is involved in asking whether or not a “reasonable” person would consent to a particular claim. The basic idea behind public reason is that we want to offer justifications that citizens will consent to because they view them as reasonable. The authoritative force of policies under liberalism is grounded in the consent of the citizens; this is the source of their legitimacy. But, at times, this tends to abstract away from actual, non-ideal humans, because they may be unreasonable in a variety of ways (racist, sexist, etc.). So it requires us to ask questions about the consent of idealized, “reasonable” persons. But then it becomes unclear whether or not public reason can do the very thing it set out to do. It appears to obviate the role of consent. Joseph Raz (1990) puts this in the form of a dilemma. Either political consent comes from actual human beings, in which case it might be so weak that it cannot yield the very basic legal principles that we want from it (such as those barring racial discrimination). Or else, we must abstract from real citizens and derive consent from idealized reasoners, but then it becomes unclear what role consent is really playing. After all, the motivation of liberalism’s commitment to public reason is to ground the government’s legitimacy in the consent of the governed.¹⁴

For now, there is no need to settle these difficult questions. But it is important to highlight the tradeoff between procedural and substantive views of autonomy. This may not matter much in moral contexts where political issues are not at stake. Two people could agree that autonomy matters morally even if one of them holds a procedural view and the other prefers a substantive view. So this issue is of less concern for the present chapter. But the value or content neutrality of autonomy becomes much more important when considering the coercive power of the state, an issue that we will address in Chap. 6.

At this point, it would be helpful to take stock of these views in a way that will allow us to draw on them later. First, we can separate procedural, content-neutral views from substantive ones that involve commitments to particular values or that condemn certain social arrangements. Within the family of procedural views, there are several that involve questions about the coherence of the agent’s motivational states. Such views might require us to ask about the agent’s desires at a particular time or they may require us to take a deeper look at the person over time.

These are the procedural views that we have discussed so far:

• Frankfurt/Dworkin: An agent is autonomous with respect to a desire just in case she approves of her desire when engaging in higher-order reflection. What is more, the higher-order deliberation must be free of external manipulation and coercion.

• Watson: An action is autonomous if and only if it is consistent with the agent’s evaluative judgments. It accords with her deeply held commitments about what is most worthwhile.

• Bratman: An action is autonomous just in case it harmonizes with her long-term plans.

• Christman: Autonomous desires are ones that the agent would not reject if she were to rationally reflect on the development of the desire.

Finally, it is important to note a subtle distinction that is present in these summaries. We can and should distinguish between different domains of autonomy. At times, it may be important to ask whether or not a particular desire or action is autonomous, and in other cases we may inquire about autonomy on a larger scale and ask what it means for a person to be autonomous. Procedural views could be fruitfully applied in each domain.¹⁵

It would be difficult to provide a similar overview of substantive theories of autonomy because there are too many of them, and the details vary widely. At their core, substantive theories reject the content neutrality of procedural approaches by proposing certain constraints. There is a distinction between theories that constrain the “content of autonomous preferences, values, or commitments (strong substantivism)” and those that merely put constraints on “the self-affective attitudes required for self-governing agency (weak substantivism)” (Mackenzie 2022, 36). A strong substantive theory might rule out certain options entirely (for instance, the choice to be a subservient housewife), whereas weak substantive theories may restrict themselves to asking whether or not the agent’s decision was fully autonomous because her deliberation was undermined by a lack of self-trust or self-respect (see, e.g., Govier 2003; McLeod 2002).

Our aim in this section is not to defend or attack any of these views of autonomy. On the contrary, we are open to all of these accounts, and our definition of “humanity”—the ability to set and pursue one’s own ends—is meant to be broad enough that it is consistent with a variety of approaches. So although our view of autonomy is not wedded to any particular account, it will be useful to draw on various features of these accounts in order to explain why a particular action or desire is at odds with the agent’s ability to set and pursue her own ends. In some instances, it may be useful to invoke Frankfurt’s language of second-order desires. In other places, we may deploy Christman’s historical model, or we might use Watson’s approach and ask questions about the agent’s evaluative judgments. We did not provide explanations of these views in order to settle the question of which one does the best job of defining autonomy. Instead, these accounts will provide a conceptual toolkit that we will draw on throughout the book in order to better understand how smartphones and the attention economy undermine our autonomy. In the next section, we revisit our definition of “humanity” in order to provide a more concrete understanding by means of the views we have explored in this section.

2.2.2 Humanity Revisited: The Ability to Set and Pursue One’s Own Ends

Recall once more, that our definition of autonomy throughout the book will correspond to what Kant has in mind when he uses the word “humanity.” This is the rational agent’s capacity to set and pursue her own ends. We may begin by noticing that this definition comprises two importantly distinct elements: capacity and authenticity. The capacity condition refers to the agent’s ability to set and pursue ends, and the authenticity condition requires that the ends be, in some sense, her own. It is common for accounts of autonomy to note this distinction and to regard these two conditions as separately necessary and jointly sufficient for autonomy.¹⁶ If an agent lacks either one of them, then she is not fully autonomous. If she has both, then she is.

2.2.2.1 Agential Capacity

We will begin with capacity. For starters, to qualify as an autonomous agent, someone must fulfill the baseline criteria for rational self-government. The sense of rationality at issue here is fairly minimal. It does not require a person to be perfectly logical and consistent, but it does require basic rational competence: the capacity to make fitting inferences from available information, to evaluate and reconsider one’s commitments, to be self-reflective, and to respond appropriately to inconsistencies.¹⁷ In addition to the baseline criteria for setting ends, it is equally important that the agent satisfy the criteria for pursuing ends. She must be capable of forming intentions, adhering to the plans she sets (which requires the ability to merge plans), adapting to changes in her environment, making use of available resources, etc.¹⁸

It is in virtue of these baseline capacities that we do not think of infants or young children as fully autonomous. They may have the potential to become autonomous as they grow older, but, at early stages of development, children lack the capacities necessary to be autonomous. It is precisely because of this lack of capacity that paternalistic constraints are appropriate when applied to children but often inappropriate when applied to adults. A small child might want to play with sharp knives or eat nothing but ice cream, but we are under no obligation to respect such choices.

But this is only the starting point. Once the baseline criteria have been satisfied, there are additional questions about the capacities an agent must have in order to successfully exercise her rational agency. Paternalistic interventions, such as the parent’s restriction of the child’s dinner options, are only one of many ways that someone might undermine another person’s sovereignty. We can easily imagine adults, who possess all the rational capacities that infants lack, but who are unable to successfully exercise their capacities because of external constraints.¹⁹ For instance, someone may be held hostage, coerced by threats of violence, or misled through deception. There are many ways to prevent people from successfully exercising their capacities, and some of these are fairly obvious.²⁰ Other constraints may be more subtle, however.

Once again, feminist perspectives on this issue are rich with examples of autonomy deficits that are difficult to diagnose. Trudy Govier (2003) and Carolyn McLeod (2002) defend the claim that “self-trust” is necessary for autonomy.²¹ Someone might satisfy the baseline criteria for rational agency but nevertheless be unable to successfully exercise those capacities because those around her (or various traumatic experiences) have made it difficult for her to trust her own judgment. So although she has rational capacity in principle, she routinely defers to the judgment of others in practice because she lacks self-trust. Benson (1994), drawing on the film Gaslight, provides an example of a woman whose husband—a physician whom she trusts—convinces her that she is “unstable” and “hysterical.” As a result, she begins to feel helpless and disoriented, further eroding her self-confidence. Benson concludes that the woman’s ability to make her own decisions has been severely compromised, even though her rational faculties are fully intact.

This gives us yet another condition to consider when evaluating an agent’s rational capacities. So far, we have established that rational agency requires certain baseline capacities (reasoning with evidence, forming coherent plans, etc.), freedom from external constraints (being held hostage, coerced by threats of violence, etc.), and freedom from influences that would undermine the agent’s ability to successfully exercise her rational capacities (gaslighting, deceit, etc.). But this still might not exhaust the set of conditions required for autonomy.

Consider the example of a boy in West Virginia in the late 1800s. He is bright, talented, and successful in school. But almost everyone he knows is a coal miner, and he grows up believing that coal mining is the only way for him to make a living. When he comes of age, he is free to choose between one of three different coal mines. He lacks the financial means to move from the area (and he has no clue what he would do for a living if he left). So he weighs his options, evaluating the different pay and hours, and he chooses to work at one of the mines.²²

Someone could say that he was free, in some abstract sense, to set and pursue any ends that he wanted. There were no immediate or apparent constraints on his decision. No one held him hostage or put a gun to his head. But if his freedom amounts to nothing more than choosing between one of three coal mines, we are likely to think that this falls short of full autonomy. We may agree with Susan Brison (2000) in thinking that in order to make autonomous choices, “we need to have a range of significant options to choose from, and so we need to live in a society that makes these possible” (2000, 283).²³ She points out that a wide variety of things might promote or inhibit our access to a range of options. This would include “personal, familial, social, political, and economic relations with others” (Ibid.). Furthermore, she rightly recognizes that we must be able “to perceive the availability of these options and to recognize their achievability by us” and that this requires us “to live in a culture in which the norms and expectations do not preclude such recognition” (Ibid.). The boy in West Virginia may have felt his options were restricted for any number of reasons. It may have been the result of pressure from his family and friends. Maybe he recognized that becoming a physician is an option for some, but he felt that such a thing was not a possibility for him.

This gives us a fourth condition of rational capacity. It is not enough to have a choice; one must choose between a sufficiently wide range of options.²⁴ For this reason, Rubel et al. (2021) distinguish between local and global autonomy. To have local autonomy, there must be a domain in which the agent is free to “make decisions about actions with immediate effect and may be able to ensure that those decisions comport with their values” (2021, 23). Rubel et al. consider the employee Ali, whose employer routinely assigns her to work on projects that she does not enjoy, and she has very little say over what she does at work. But, in general, she is a very talented person who could easily find another job and who is perfectly capable of leaving. With respect to her job, Ali lacks autonomy, but she is autonomous in the more general sense. She has global autonomy but lacks local autonomy (Ibid.). They contrast Ali with Bari whose global options are severely restricted. Bari is all but forced to care for her siblings and older relatives. She cooks and cleans for them. But she has incredibly wide latitude when it comes to fulfilling those obligations. So Bari has local autonomy but lacks global autonomy (Ibid.).

The fourth condition (having a sufficiently wide range of options) underscores an important sense in which autonomy must be seen as relational. No one’s capacities are developed in a vacuum. Culture, upbringing, and education are inextricably linked to the development of our capacities. They shape our sense of what is possible and what we can achieve. Indeed, even Kant, who is often maligned for defending an excessively atomistic view of autonomy, was sensitive to this point. As we will explain in Chap. 5, Kant argues that the cultivation of “humanity” through education is a moral imperative. Although it is true that Kant thinks each individual is morally obligated to cultivate her own capacities, he recognizes that this is something that we must undertake collectively, a point he repeats throughout his writings on religion, history, and anthropology.²⁵ What is more, he sees the collective perfection of the entire human species as the final end of history.²⁶ In his lectures on pedagogy, he says, “The human being can only become human through education. He is nothing except what education makes out of him. It must be noted that the human being is educated only by human beings, human beings who likewise have been educated” (UP 9:443).²⁷ This is one of the many ways that our autonomy is promoted (rather than inhibited) by social interdependence.

With those four conditions in place, we should have a much better understanding of the capacities required for autonomy. It requires

• (1_a) baseline competence (the ability to form intentions, make coherent judgments, etc.),

• (2_a) a lack of external constraints,

• (3_a) the absence of cognitive inhibitions (such as self-doubt), and

• (4_a) a sufficiently wide range of options (with the ability to pursue them effectively).

Of course, all of these things come in degrees. External constraints can be more or less severe. Questions about self-doubt, self-trust, and baseline competence will rarely be answered by a simple yes or no. You may have days when you are racked with self-doubt and imposter syndrome, but you may be teeming with confidence at other times. Or you may experience both of those feelings at the same time but with respect to different domains. Some students are totally confident in classes that require them to write essays, but their “math phobia” causes them to experience self-doubt in others. So the presence or absence of these conditions of autonomy is a matter of degree.

2.2.2.2 Authenticity

For now, this should be enough to understand what kinds of capacities matter when it comes to autonomy. Authenticity is a separate issue, but, as we will see, they are closely related. For instance, interfering with someone’s capacities (e.g., by gaslighting them) could lead them to adopt inauthentic or alien desires. To understand authenticity, it would be helpful to recall the lessons from the previous section. Your authentic desires are ones that did not result from manipulation or coercion. They are generally consistent with your long-term plans (Bratman), your higher-order reflection (Dworkin), and your evaluative judgments about what is most worthwhile (Watson). Finally, you should not feel alienated from your desire when you reflect on the process through which the desire was developed (Christman).

Once again, these things come in degrees. All of our desires are subject to some amount of external influence. We should not make the mistake of thinking that desires are one’s own only if they were developed in the complete absence of outside influence. That would set the bar too high. So where should the line be drawn? Perhaps we should begin with the clearest cases and work our way toward those that are more subtle and difficult. Arguably the clearest and most compelling case would be coercion. As noted earlier, when you decide to hand your money to the armed robber, it is very clear that this desire is not authentically yours. This desire is being forced upon you because the alternative is unacceptable (Wood 2014).

So when it comes to the first condition of authenticity (being free from manipulation and coercion), it is not hard to see why desires that result from coercion are inauthentic. There is a very clear sense in which you are being compelled to adopt someone else’s desire. Manipulation is more difficult to grasp. For starters, there are fundamental debates about what manipulation even is. Two people might agree that manipulation is at odds with autonomy even though they have radically different ideas about how to define the concept. Yet again, it would be helpful to begin with some clear cases of manipulation.

Consider Iago’s manipulation of Othello. Iago feels that Othello unfairly passed him over for promotion, so Iago formulates a plan to manipulate Othello into demoting his lieutenant, Cassio. Iago accomplishes his aim by deceiving Othello in a variety of ways. First, Iago acquires the handkerchief that Othello gave to his wife, Desdemona. Iago lies to Othello, telling him that he saw Cassio with the handkerchief (implying that Cassio is having an affair with Desdemona). This sparks jealous rage in Othello, and Iago convinces him to eavesdrop on Cassio’s conversation. Othello overhears Cassio talking about his affair with his mistress Bianca. Iago deceives Othello once more by leading him to believe that Cassio was talking about Desdemona. Othello becomes so angry that he orders Iago to kill Cassio.

It seems safe to say that Iago manipulated Othello. We can also see how Iago accomplished his aim: via deception. He lied to Othello about the handkerchief and about the meaning of Cassio’s conversation. So the first lesson that we can draw from this case is that you can manipulate someone by deceiving them. And although Iago deceived Othello by telling many outright lies, this is certainly not the only form of deception. Allen Wood (2014) claims that telling flat-out lies is manipulative because it “feeds the person false information, on the basis of which he makes choices the person presumably might not have made if he had known the truth” (35). In addition to outright lies, Wood goes on to note that “Other, subtler forms of deception—misleading, encouraging false assumptions, fostering self-deception—do this in more devious ways” (Ibid.).

In addition to deception, someone might manipulate you by pressuring you in various ways: guilt, shame, browbeating, social pressure, etc. Now, if the pressure is so extreme that they leave you with no acceptable alternative, then this would rise to the level of coercion. But manipulation often falls short of coercion. Consider the following case. Sofia is a college student who wants to major in philosophy. After thorough deliberation, she decides that her only criterion for choosing a major is studying what she loves, and she is certain that her passion lies in philosophy. When she breaks the news to her parents, they try to talk her out of it. After rational persuasion fails, they turn to emotional appeals, telling her, with tears in their eyes, that they would be heartbroken if she does not follow in her mother’s footsteps to practice medicine. They even threaten to stop paying her tuition. Assuming that Sofia has an acceptable alternative (such as going to a cheaper college or taking out student loans) this would not amount to coercion. But it certainly seems manipulative.

Wood encourages us to notice what is distinctive about manipulative “inducements.” He says that they “offer the person ‘the wrong sort of reason’—a sort of reason that the person would not endorse on reflection, if behaving rationally and operating with normal, healthy motivations intact” (Ibid.). The young woman had already decided that the only thing that mattered to her was following her passion. Pleasing her parents and avoiding student loans may be reasons to study medicine, but, for Sofia, they are not the right sort of reasons.²⁸ Seeing the response from her parents provokes so much emotional distress that Sofia’s judgment is clouded and she ends up studying medicine. She later regrets the decision, and she feels that she acted on the basis of reasons that she could not rationally endorse.

These two examples are meant to show two possible means of manipulation. Iago manipulates Othello by deception, and the parents manipulate their daughter by using emotional appeals to get her to act on the basis of reasons that she does not reflectively endorse. It would be helpful to consider one last class of examples before offering a general account of manipulation. Critics of persuasive advertising frequently argue that marketing is morally problematic insofar as it undermines our autonomy by manipulating our desires.²⁹ It is outside the scope of our project here to weigh in on that debate, but it would be helpful to consider one of the most popular examples. Commentators frequently cite the (apocryphal)³⁰ story of a New Jersey movie theater that drastically increased popcorn and soda sales by subjecting moviegoers to subliminal advertisements. The moviegoers did not know that they saw an advertisement at all, but it caused them to get up during intermission to purchase refreshments. Roger Crisp argues that the moviegoers are not acting autonomously when they do this; he quips that they are acting “automatonously” (1987, 413; emphasis added). That is, their behavior resembles that of an automaton rather than that of an autonomous agent. This is because the subliminal advertisement induced their actions without engaging their rational capacities in any way. They were not able to reflect on their reasons because the influence was affecting them at a subconscious level.

The debate concerning persuasive advertising is ongoing, and there is no consensus in the literature about whether or not advertising is manipulative. The mere fact that advertising has an influence on our desires is insufficient for establishing the claim that it manipulates us. Something further must be established. Anne Cunningham (2003) argues that some critics of advertising have created an unreasonably demanding standard for autonomy. She offers an alternative approach for addressing concerns about autonomy and advertising. She suggests that we should consider a desire autonomous (or alien) by evaluating it according to the model defended by Noggle (1995). In Noggle’s view, alien desires are the product of “discordant quasi-beliefs” (1995, 65). Quasi-beliefs are representational states that can function like beliefs insofar as they promote desires or actions, but they lack several important properties belonging to “straightforward beliefs.” Straightforward beliefs are integrated into our broader web of beliefs; “are formed by cognitive epistemic means and are subject to cognitive epistemic control” (60); are available to us upon introspection; and are ones that we are, other things being equal, willing to assert to ourselves (Noggle 1995, 59–60). Quasi-beliefs lack all of these features, and a quasi-belief is discordant just in case it is inconsistent with some straightforward belief.

To see how this would work, imagine a case where advertising is capable of achieving what the critics suggest that it is guilty of doing. For example, let’s say Rosa’s desire to drink Coca-Cola was brought about because the ads full of young, beautiful, happy people successfully led her to associate Coca-Cola with her unconscious desire for youthful exuberance and joie de vivre. If the desire driving her purchase is truly unconscious, then she would not be able to rationally deliberate over her choice or endorse it through higher-order reflection. The desire for Coca-Cola was brought about by manipulating her unconscious desires. Using Noggle’s model, we could say that her action was prompted by a discordant quasi-belief. She surely does not believe that drinking Coca-Cola will bring back her youth or make her life as fun as the advertisement, but that is precisely the content of her quasi-belief. Of course, those who defend advertising, like Cunningham (2003) and Arrington (1982), argue that advertising rarely does what the critics claim. But if it did accomplish what critics like Crisp (1987) and Wood (2014) suggest, then we can see why it would be inconsistent with autonomy.

Examples of manipulation extend well beyond deception, emotional appeals, and subliminal/subconscious advertising. In some cases, all that is required is the presence of an influence that leads to a decision that falls short of the agent’s ideal form of rational deliberation. For instance, consider the realtor who bakes cookies before the open house.³¹ Of course, prospective buyers are likely to feel welcomed by such a pleasant smell, but if Hanlon were to make an offer on the house on the grounds that the smell brought back fond childhood memories of feeling cozy in his abuela’s house, then it is easy to see how he is falling short of his ideal for deliberation. The fact that the house smelled like cookies was not a good reason for buying it.³²

In each case described above, manipulation involves influencing people’s choices in ways that interfere with their capacity for rational decision making. Manipulation makes it more difficult for the agent to make choices that are consistent with her evaluative judgments. Deception does this by tricking people into acting in ways that they would not want to act if they had accurate information. Emotional manipulation works by giving people inducements they would not endorse on reflection.³³ If advertising is indeed manipulative in some cases, it accomplishes this by changing desires through means not open to introspection or rational consideration. We could conclude, as Allen Wood does, that manipulation circumvents or undermines rational agency in some way (Wood 2014).

It is important to avoid overgeneralizing about manipulation. We would not want to suggest that all forms of non-rational influence are manipulative. That would set the bar much too high for autonomy. In many cases, we may appeal to someone’s emotions without manipulating them. You might want to influence the behavior of a close friend, so you decide to explain how their actions have hurt you in the past. Assuming that your friend cares about you, their emotional response to your suffering should count as an appropriate reason to change their behavior. You have influenced their behavior by means of an emotional appeal, but you have not manipulated them. It should also be pointed out that manipulation can be covert or overt. Othello covertly manipulates Iago, but Sofia’s parents make an overt effort to manipulate her. Covertness is not a necessary condition of manipulation.³⁴

With these views in hand, it is not difficult to see how manipulation and coercion are at odds with autonomy. Given that autonomy requires authenticity, our chosen ends must be, in some sense, our own. When we are coerced or manipulated, we act for reasons that are not our own. As we mentioned above, acts of coercion and manipulation are often seen as morally wrong, but it is worth noting that they are not wrong by definition.³⁵ As Roger Crisp points out, a skillful actor deftly manipulates our emotions, but this is not morally objectionable (1987, 414). Allen Wood says the same is true of coercion. A police officer might point a gun at the bank robber, coercing him to stop what he is doing, but coercion of this kind does not seem to be unjust or immoral (2014, 27).

In some cases, we can employ the local/global distinction to understand why a manipulative or coercive action is not immoral. For instance, parents may rightly manipulate their child into eating a healthy, balanced diet. They manipulate her into trying a wide variety of foods. A kindergarten teacher might manipulate (or even coerce) a child into sitting quietly and listening. In both cases, the parents and teacher are doing something that is at odds with the child’s local autonomy.³⁶ With respect to that particular choice, the child’s autonomy is constrained. But both actions can be seen as promoting the child’s global autonomy. By learning to enjoy a variety of foods, the child will have more dietary options in the future. Her choices will be less constrained than if she were to become a picky eater. Similarly, by making the student sit down and listen, the teacher is helping the student develop her capacities and thus promoting her autonomy.

This sums up the first condition of authenticity. Authentic preferences are ones that are free of manipulation and coercion. Just as we argued above, when discussing the conditions of agential capacities, most of these conditions should be understood as existing on a spectrum rather than as binaries. The more that a preference is the result of manipulation, the less autonomous it is. If questions of this kind are seen as a binary, the conclusions run the risk of being implausibly strong. Once more, we must avoid overgeneralizing about manipulation and autonomy. The account would be implausible if it condemned every attempt to exert non-rational influence on someone.

The second condition of authenticity is one that was discussed extensively above in the context of “alien” desires. Your authentic desires and preferences are ones that are consistent with your deep commitments. They demonstrate a certain kind of internal coherence that shows that you are the author of your own life story. We may side with Frankfurt and Dworkin by understanding this coherence in terms of first-order and second-order desires. With Watson, we may use the model of evaluative judgments, or we could follow Bratman by understanding this in terms of the agent’s long-term plans. Desires and preferences are authentic to the extent that you identify with them.

In addition to identifying with our desires, authenticity also seems to require the absence of alienation. At first glance, it may seem that non-alienation is already a feature of internal coherence accounts. But critics like Christman insist that they are distinct. One important feature of Christman’s account is that he regards mere “identification with” a desire as too low of a bar for personal autonomy.³⁷ This is why he argues that the agent must evaluate the historical process through which the desire was developed. In some cases, an agent might identify with a desire (e.g., she regards it as her own when engaging in higher-order reflection), but she would feel alienated from it if she were to reflect on the historical development of that desire.³⁸ Christman writes:

Consequently, I suggest that the proper test for the acceptability of the characteristic in question is one where the person does not feel deeply alienated from it upon critical reflection. Alienation is not simply lack of identification, in that I can fail to identify with a trait but not be alienated from it. I can simply be indifferent to it or undecided about it. Alienation is a stranger reaction; it involves feeling constrained by the trait and wanting decidedly to repudiate it. (2011, 143)

He goes on to note how alienation goes beyond identification models in that it involves both “judgment and affective reaction” (144). He contrasts this model with identification accounts that focus too narrowly on cognitive judgments. This is the third condition of authenticity. Your desire is authentic only if you are not alienated from it. Furthermore, you would not feel alienated from the desire if you were to reflect on the process that gave rise to it. The brainwashed cultist is a helpful example. She might identify with her desire to move to the compound, but she would likely repudiate the desire if she were to reflect on the manipulative processes that shaped her preferences.

Finally, as we explained above, there are those who criticize Christman for not going far enough when it comes to issues involving oppression and adaptive preferences.³⁹ To see why, we can return to the case of the woman who chooses to starve herself in order to overfeed her husband and son. This woman might identify with her desire, and she might not repudiate the desire when reflecting on the processes that developed it. In the autonomy literature, this is often referred to as an “adaptive preference.” Her preference is “adaptive” in the sense that it was shaped by restrictive conditions that limited her set of options.

Indeed, the broadest accounts of adaptive preferences describe them in just such a way. According to those views, adaptive preferences are ones that were shaped, at least in part, by constraints placed on someone in virtue of their circumstances. But when using such a broad definition, it is difficult to see why anyone would regard them as problematic. As Rosa Terlazzo points out, her preference for a wheat-based diet is likely the result of growing up in the US, and she might very well prefer a rice-based diet if she had grown up in China (2016, 206). But that preference does not present any moral concerns. Many of our preferences are adaptive in that sense. Terlazzo notes that we might feel quite differently about a different set of cases:

[T]he case of the woman who thinks that her husband is justified in beating her, the poor person who has no desire for an education, the woman who wants to undergo breast augmentation to make herself more attractive to men, or the man who refuses treatment for depression because he takes stoicism to be central to masculinity (Ibid.).

Now let us ask: what is it, exactly, that distinguishes the problematic cases of adaptive preferences from ones that are not so troubling?

According to one popular view, adaptive preferences are worrisome only in cases where the preference is bad for the agent in some way.⁴⁰ It is bad for the abused wife to believe that she deserves to be hit; it is bad for the woman to starve herself; and so on. When Aesop’s fox tells himself that the grapes are sour simply because they are out of reach, his preference does not seem to be adaptive in the problematic sense.⁴¹ After all, he might very well be better off believing that the inaccessible grapes would taste bad. Otherwise he may agonize over what he is missing. Similarly, Nussbaum provides an example of a child who comes to prefer a career in philosophy after learning that she was not cut out for being an opera singer (2001a). Again, this does not seem problematic. So what exactly is distinctive about the adaptive preferences that are morally worrisome? One way of explaining why some adaptive preferences are bad for agents would be to suggest that adaptive preferences are instances of autonomy deficits. But this account runs the risk of vicious circularity. Which adaptive preferences are the bad ones? The ones that undermine autonomy. Which ones undermine autonomy? The bad ones.

In order to escape this circle, we must provide a procedural account of adaptive preferences that would explain why some adaptive preferences undermine autonomy while others do not. Terlazzo (2016) defends a view of autonomy that is mostly procedural but which includes a couple of substantive constraints. Recall that substantive theories of autonomy are ones that do not remain content neutral when making judgments about autonomy. For instance, if a feminist view rules out the possibility of autonomously choosing to be a subservient housewife, then it would not be content neutral; it would be substantive.⁴² Interestingly, Terlazzo’s view of autonomy does not rule out any preferences with respect to the value of their content, but it does require the agent to choose from a set of valuable, live options. In addition to requiring basic capacities and continence, she argues that a preference is autonomous if and only if:

• (1_b) the agent reflectively endorsed the preference “at some point in its formation,”

• (2_b) her reflection took place “in the presence of recognized alternatives,”

• (3_b) “at least some of these alternatives were valuable ones,” and

• (4_b) “some of these valuable alternative options were live ones (that is, they were ones that X could reasonably see herself exercising, given her current values and ambitions)” (Terlazzo 2016, 215).

She uses this view of autonomy in order to argue that adaptive preferences should be understood as core preferences that lack autonomy precisely in virtue of failing to meet one of these conditions.⁴³

According to this view of adaptive preferences, we do not need to worry ourselves with questions about whether or not the preference is bad for the agent. We can also avoid making substantive claims that abandon content neutrality. Terlazzo does not require autonomous preferences to be ones that promote value for the agent; she only requires that the agent make her choice among a set that includes a few valuable, live options. So when a kidnapping victim develops Stockholm syndrome, we can safely say that her preference to stay with the kidnapper is adaptive in the problematic sense. She made this choice only after judging that she had no hope of escape. According to the well-being view of adaptive preferences, it might be hard to make such a judgment. It is difficult to judge whether or not the preference is bad for the victim. Perhaps Stockholm syndrome was a coping mechanism that allowed her to survive the traumatic situation. In that sense, the preference might have been good for her (at least in a local sense).⁴⁴

This view allows us to render an intuitively plausible verdict about the examples we mentioned. On the one hand, we can see why the preferences of the abused wife or the kidnapping victim are not autonomous. On the other hand, we can explain why there is nothing wrong with the fox’s preference to pursue food other than the grapes. The fox chooses to abandon his preference for the grapes only after realizing that they are unattainable. Not only is this perfectly rational, it is consistent with autonomy.⁴⁵ As long as the fox had other live options (e.g., berries that are closer to the ground), his choice was not constrained in a problematic way. Similarly, we do not regard it as an autonomy deficit when an American consumer chooses between many varieties of apples but cannot easily acquire durian or rambutans. Adaptive preferences are worrisome only in cases where we judge that the agent came to have a preference precisely because she did not have any other valuable, live options.

In sum, this gives us four conditions for authentic ends. Authentic preferences are ones that are:

• (1_c) Free of manipulation and coercion,

• (2_c) consistent with the agent’s motivational states,

• (3_c) free from alienation, and

• (4_c) not adaptive.

We are not suggesting that these categories are perfectly discrete. There is considerable overlap and interplay between them. For instance, a preference that results from manipulation or coercion (condition 1_c) is likely to be inconsistent with some of the agent’s motivational states (condition 2_c) and it is also likely to spark feelings of alienation (condition 3_c). What is more, there are several connections between the conditions of capacity and the conditions of authenticity. One of the requirements of autonomous capacity is that the agent’s reasoning must be free of influences that undermine her ability to deliberate (3_a). It would be helpful to recall Paul Benson’s example of the husband who is gaslighting his wife, convincing her that her faculties are unreliable so that she defers to his judgment. In the previous section, this was characterized as a threat to her agential capacities. In the absence of self-trust, the wife is unable to successfully deliberate and choose for herself. But the wife’s decision to defer to her husband also lacks authenticity. It was the product of manipulation. The husband’s action has a twofold effect on her autonomy; it undermines both her capacity and her authenticity.

So we do not mean to suggest that there is no overlap between these eight conditions (four for capacity, four for authenticity). What is more, we do not want to claim that each of these picks out a necessary condition. This may be true of some of the conditions. When it comes to global autonomy, it is fairly clear that the baseline capacities outlined in the previous section are necessary conditions. That’s why infants and small children lack autonomy. This means that some of these conditions are necessary, but this may not hold for all of them. It would be fair, however, to regard them as jointly sufficient conditions of autonomy.

If someone satisfies all of the conditions of authenticity and capacity, it is safe to say that she is autonomous. But what should we conclude when an agent or an agent’s preference fails to satisfy one or more of these conditions? Perhaps the safest claim here would be that every violation of a condition counts as prima facie evidence that the agent’s autonomy has been undermined. This means that the failure to satisfy one of these conditions is evidence of an autonomy violation but it is not necessarily constitutive of a violation.⁴⁶

To sum up, throughout the book we will understand autonomy along the lines of what Kant meant by humanity. It is a rational agent’s capacity to set and pursue her own ends. It involves a variety of conditions for both capacity and authenticity. It requires that the agent have the ability to set and pursue ends, and it requires that the ends be, in some sense, her own.

2.3 Why Humanity Matters Morally

Kant is probably the most well-known and ardent proponent of the idea that we are morally obligated to respect autonomy. Indeed there are few thinkers in the history of philosophy who put such a high premium on rational agency. Kant famously characterizes the moral law as an obligation to respect this capacity in others and in ourselves. This has come to be known as the categorical imperative’s “formula of humanity.” In the Groundwork he describes the moral imperative in this way:

• Formula of Humanity: So act that you use humanity, whether in your own person or in the person of any other, always at the same time as an end, never merely as a means (G 4:429).

Kant argues that there is something distinctive about the value of rational agency that sets it apart from everything else in the world. Because human beings are rational agents, we have the capacity to act morally, and it is in virtue of this capacity that we have what Kant calls “dignity”—a value that is elevated above any price.

Given that this book is a work of Kantian ethics, we could simply assume that Kant is right about the value of humanity and move on to the moral duties that follow from this principle. But that would be too quick. We can (and should) ask whether or not Kant is right. Why should we think that rational agency (humanity) is so important? What is distinctive about humanity that merits this elevation above everything else?

Our aim in this section is to present a compelling case for the value of humanity and to explain how Kant derives moral duties from this idea. We begin in the next section by stepping away from Kant to explore the common intuition that autonomy is morally important and that we have a duty to respect it. We then return to Kant and explain how he arrives at the formula of humanity. We conclude by showing how other moral theories (such as consequentialism and virtue ethics) should arrive at similar conclusions about the moral significance of autonomy. Although Kant believes that autonomy has intrinsic value, other theories might see it as instrumentally valuable. For classical utilitarians, such as Bentham, pleasure is the only thing that is intrinsically good, and autonomy is instrumentally good only insofar as it is a means to pleasure. We will revisit this idea in later chapters when we defend the existence of moral obligations to be autonomous with respect to technology and the attention economy. We believe that Kantian ethics does the best job of capturing what is morally worrisome about the attention economy: if autonomy is non-instrumentally morally valuable and the attention economy undermines autonomy, then the moral concern is quite clear. But, as we will argue in later chapters, other moral theories can reach the same endpoint even if they take a different route.

2.3.1 The Commonsense Intuition

Most readers are likely to share the intuition that there is something valuable about autonomy. Other things being equal, an autonomous life is better than one that lacks autonomy. This probably rings true in your own experience. You want to be the author of your own life story. When it comes to major life decisions, you would rather be in the driver’s seat than ride along as a mere passenger. You would feel wronged if someone made those decisions without your consent or if they undermined your capacity to make those choices for yourself. To see why this is compelling, we can contrast autonomous lives with similar ones that lack this element of self-direction.⁴⁷ It would be better for Sofia to choose her own career rather than to have her parents decide for her. Similarly, people should be free to choose their own life partners rather than be forced into arranged marriages.⁴⁸ This is also true in mundane situations. Imagine relinquishing all control over your diet and letting random strangers decide what you will eat for every meal. The value of such freedoms might be so obvious that they hardly need any defense.

Indeed, there are many situations in which the moral significance of autonomy is already taken as a given. There might be no domain in which the value of autonomy is more obvious than the field of medical ethics. Although “respect for autonomy” is only one of the four widely accepted principles of bioethics (the other three being beneficence, nonmaleficence, and justice), autonomy is often given pride of place.⁴⁹ Even those who regard these principles as “equal” sometimes say that autonomy should be seen as “first among equals” (Gillon 2003).⁵⁰ It would be hard to overstate the importance of respect for autonomy in medical contexts. When a patient (who has decision-making capacity, i.e., the ability to understand relevant information, appreciate one’s situation, reason about options, and communicate choices)⁵¹ refuses lifesaving treatment, physicians are required to respect the decision. This is what leads bioethicists to conclude that “autonomy is assumed to be more valuable than human life” (Brudney and Lantos 2011).

The importance of autonomy is also recognized in a variety of political contexts. For instance, within the tradition of liberalism, there is widespread opposition to paternalism. In spite of their many differences on other issues, Kant and Mill converge on this point. In On Liberty, Mill articulates the famous harm principle, as he argues that the state should interfere with an individual’s liberty only in order to prevent harm to others (1988, 9). But he thinks that we should not use the state’s coercive power to prevent individuals from harming themselves. Similarly, Kant thinks we cannot rightly force other people to live according to our vision of what would make them happy. Interestingly, Mill’s opposition to paternalism bears a striking resemblance to Kant’s. As we will see in Chap. 6, Kant sees individual freedom as the only suitable justification of the state. The government exists to promote and safeguard the freedom of individuals.

In his “Theory and Practice” essay, Kant goes so far as to say that paternalistic government would amount to “the greatest despotism thinkable.” He writes:

No one can coerce me to be happy in his way (as he thinks of the welfare of other human beings); instead, each may seek his happiness in the way that seems good to him, provided he does not infringe upon that freedom of others to strive for a like end which can coexist with the freedom of everyone in accordance with a possible universal law (i.e., does not infringe upon this right of another). A government established on the principle of benevolence toward the people like that of a father toward his children—that is, a paternalistic government (imperium paternale), in which the subjects, like minor children who cannot distinguish between what is truly useful or harmful to them, are constrained to behave only passively, so as to wait only upon the judgment of the head of state as to how they should be happy and, as for his also willing their happiness, only upon his kindness—is the greatest despotism thinkable (a constitution that abrogates all the freedom of the subjects, who in that case have no rights at all). (TP 8:290–91)

In general, we tend to think that paternalistic interventions are justified only when someone lacks decision-making capacity. That is precisely why we do not allow small children to refuse medical treatment even though we must respect the decisions of capacitated adults who decline treatment. Similarly, we allow students to drop out of college, but education is compulsory for young children. Autonomy is the decisive factor in each of these situations.

And even when we put aside the state’s coercive power, we tend to find something objectionable about the idea of being subject to the will of another person. This concept lies at the heart of contemporary republican views of freedom. Proponents of this view, such as Philip Pettit, have argued that freedom consists in the absence of domination. To be free is to act in such a way that your decisions are not subject to the arbitrary will of another person (2014, 30).⁵² Drawing on an intellectual tradition that goes all the way back to the Roman Republic, this view of freedom paints a stark contrast with the kind of domination that would have characterized the lives of slaves. Once again, Kant made valuable contributions to this way of thinking. In the margins of his copy of Observations on the Feeling of the Beautiful and Sublime, Kant says that there is “no misfortune more terrible to him who would be accustomed to freedom than to see himself delivered to a creature of his own kind, who could force him to do what he wants” (2005, 11).⁵³ When faced with a choice between slavery and risk of death, Kant thinks there would be no reservation about “preferring the latter” (Ibid.).

All of this goes to show that there is fairly widespread support for the idea that autonomy is morally significant. People generally share the intuition that we should respect autonomy (at least in the vast majority of situations). This is true in medicine, interpersonal contexts, and in the realm of politics. But this falls short of a philosophical justification. Even if this idea enjoys a great deal of intuitive support, we ought to ask what justifies it. What grounds the moral significance of autonomy? In the next section, we give a brief overview of Kant’s reasons for thinking that autonomy matters morally. We explain why he picks humanity as the one and only “end in itself.”

2.3.2 Kant on Humanity as an End in Itself

In order to understand why Kant identifies humanity as an end in itself, we must begin by explaining the philosophy of action that underpins Kant’s moral philosophy. In the Groundwork, Kant puts forward the view that every intentional action should be analyzed in terms of a maxim, which he defines as “the subjective principle of volition” (G 4:401). Whenever you perform an intentional action, you are acting on the basis of a maxim. Whether you explicitly articulate it or not, your maxim can be characterized in terms of an act that you perform in some set of circumstances in order to achieve some end. For example, after seeing rain in the forecast, you pack your umbrella in order to avoid getting wet. Of course, you do some things without maxims (e.g., sneezing), but those are not intentional actions.

With this view of maxims in place, Kant gives his first formulation of the moral law, as he tells us that we should

• Formula of Universal Law: act only in accordance with that maxim through which you can at the same time will that it become a universal law (G 4:421).

He then shows how to derive duties from this formula by explaining why we cannot will certain maxims to be universal laws (e.g., making a false promise to secure a loan, and refusing to ever help those in need). Immediately after this discussion of the formula of universal law, Kant starts to ask what this would look like for human beings, who always act for the sake of ends. Since he sees the moral law as valid for all rational beings, he acknowledges that it would also necessitate the actions of being with a “holy will” or “divine will” (G 4:414). But when it comes to human beings, the moral law takes the form of an “ought” precisely because it does not necessitate our actions. Our compliance with the moral law is not guaranteed.

He then distinguishes between two different forms of “oughts.” First, there are hypothetical imperatives, which tell us that we ought to do something only because we desire something else. You ought to study for the LSAT if you want to go to law school. You ought to crack the eggs if you want to make an omelet. Kant wants us to notice the crucial limitation of hypothetical imperatives. They command us to do something only insofar as we desire the end in question. If you do not want to go to law school, you do not have to study for the LSAT. If you do not want an omelet, you do not have to crack the eggs. This is why he refers to the moral law as the categorical imperative. Unlike hypothetical imperatives, which depend on you willing some end, the moral law commands obedience from all rational agents. It does not matter what you want or who you are: the demands of the moral law are universal.

This is why Kant thinks that the moral law must be grounded in reason rather than anything empirical (e.g., human nature, or desire). Kant thinks we would never discover universal moral laws if we tried to ground them in something like human nature. If we based morality on what people happen to want, we would end up with contingent laws that hold only insofar as someone has that particular desire. But it’s important to recall that human beings always act for the sake of ends. This is why Kant follows his discussion of the universal law with questions about the possibility of an objectively valuable end. He wants to know if there is any end that has absolute worth for all rational beings. If there is, then it would be a suitable foundation for a universal law.

Kant answers his question in the affirmative by suggesting that humanity (rational nature) is the only thing that could be seen as an end in itself. He writes, “Now I say that the human being and in general every rational being exists as an end in itself, not merely as a means to be used by this or that will at its discretion; instead he must in all his actions, whether directed to himself or also to other rational beings, always be regarded at the same time as an end” (G 4:428). In order to understand why Kant thinks this, we must ask what is distinctive about rational agency. How does the value of humanity differ from the value of other things and why does Kant think this is the case?

The value of humanity is distinctive in three ways.⁵⁴ It is objective, unconditional, and non-fungible. An objective value is one that holds for all rational agents.⁵⁵ This is contrasted with relative value, which requires the presence of a contingent inclination. For instance, the value of a chocolate cake is merely relative, because it has value only insofar as agents desire it. Not everyone has this inclination (even if most of us do), so it cannot make a claim on the wills of all rational agents. This is what Kant means when he says that humanity is an “end in itself.” It is something whose value does not depend on anything else (such as a desire). The value of humanity is also unconditional; it is not something that you could forfeit under any conditions. No matter what you do, you deserve respect as a human being. Finally, it is non-fungible, which is to say that human beings have the kind of value that does not admit of exchanges for things of equal value. Kant says in the Groundwork that “everything has either a price or a dignity. What has a price can be replaced by something else as its equivalent; what on the other hand is raised above all price and therefore admits of no equivalent has a dignity” (G 4:434–35).

It is also important to distinguish between ends that we are trying to bring about from ends that already exist. The omelet is an end that you are trying to bring about. It does not yet exist, but you would like to make one in order to satisfy your desire. It is an end that must be effected. The same goes for the law degree. Humanity, by contrast, is an “existent end” (G 4:437). The fact that humanity is valuable does not mean we are obligated to produce more of it. Thus, the value of humanity for Kant is not analogous to the value of pleasure for Mill. Kant is not arguing for something like Parfit’s repugnant conclusion.⁵⁶ Instead, Kant is saying that we must recognize the value of human beings who already exist. As Allen Wood explains, “When I seek to relieve someone’s suffering, my end to be effected is a certain state of affairs (the person’s comfort); but I also act for the sake of a human being, whom I value in some way and is thus an existent end of my action” (1999, 116). Wood also shows how people sometimes express respect for an object, such as a flag, by taking off their hats. Those who wish to recognize the value of the flag do so by expressing respect for it, not by making more flags.

Those are the distinctive features of humanity’s value. Rational agency is distinct from everything else because its value is objective, unconditional, and non-fungible. It is also to be seen as an existent end, rather than an end to be effected. This answers the first of the two questions. We now have a better understanding of what Kant thinks is special about humanity, but we have not answered the justificatory question. Why does Kant think humanity merits this elevation above everything else?

We can approach this conclusion, as Kant does, by process of elimination. He rules out other candidates for objective value in a widely discussed section of the Groundwork (G 4:428).

Kant begins by considering things that we value—the “objects of inclination.” Some of us value chocolate cake, omelets, law degrees, or books about Kantian ethics. But we can immediately see why these are not suitable as grounds for categorical value. First, not everyone values them. Second, these ends are valuable only insofar as someone desires them. Their value is instrumental rather than intrinsic. They are valued for the sake of something else (namely, the satisfaction of a desire).

This leads us to the next candidate. Since the objects of inclination do not have intrinsic value, perhaps it is the inclinations themselves. It is not the objects of desire that are good, it is desire-satisfaction that is good. As we saw earlier, however, we often repudiate our desires for one reason or another. The unwilling addict does not regard her desire for the addictive drug as a good thing at all.⁵⁷ On the contrary, she would rather be free of that inclination entirely. When your doctor tells you to avoid sugar in order to keep your insulin down, you no longer regard your desire for cake as something that is good.

For Kant, it is only by means of your rational agency that your inclinations become valuable. Because you are someone with the capacity to set and pursue your own ends, you evaluate your inclinations, and you decide which ones you should act on. It is for precisely this reason that you must regard your rational agency as the source of value. Reason confers value on your inclinations by making judgments about which ends matter to you. Korsgaard sees Kant’s reasoning in this passage as a “regress upon the conditions” (1996, 120). We start by asking why the objects of inclination are valuable. They have value because of inclinations. Why do those inclinations have value? The answer is that rational agency has endowed them with value. When we ask why rational agency has value, we must realize that we have reached the end of the story. This is the “unconditioned condition.” It is the bedrock on which all other value rests.⁵⁸

Kant argues that we cannot possibly fail to recognize the value of our own rational agency. If we did, we would collapse into incoherence. Whenever we perform any action whatsoever, we demonstrate a commitment to the value of our rational agency. We cannot escape the fact that we act for the sake of reasons, and this very fact about our rational agency demonstrates the value of humanity as an end in itself. In this way, Kant’s argument foreshadows Mill’s point about how “questions of ultimate value do not admit of proof” and this is something that is “common to all first principles” (1957, 44).⁵⁹

Indeed, in Kant’s argument about the value of rational agency one can also hear echoes of Aristotle’s defense of the principle of noncontradiction. It would be impossible to give a proof of the principle on noncontradiction, since any proof would presume the principle and thus beg the question. So Aristotle challenges his opponent to say anything at all. If the utterance picks out any object whatsoever, then it demonstrates a commitment to the principle. But what if our interlocutor refuses to speak? If so, then Aristotle thinks that his actions would betray a commitment to the principle. Even if he does not say anything, he must do something. And when he does, we could ask why he walks to the city of Megara instead of walking “into a well or over a precipice” (2001, Metaphysics 1008b16). By doing anything whatsoever, the skeptic shows that he does not really reject the principle.

The same can be said for Kant’s defense of humanity as an end in itself. Rational agency is the only tool we have for deciding which things have value and which do not. What is more, we cannot possibly opt out of rational agency. We inevitably act for the sake of reasons. When we do so, we demonstrate our commitment to the value of humanity as an end in itself. As we explained above, this means that its value is objective and unconditional. Everyone is compelled to recognize the value of rational agency. Unlike omelets and law degrees, we all acknowledge the value of rational agency. We do so—whether explicitly or implicitly—every time that we act.

But another skeptical challenge remains. What could be said to the egoist who argues that each individual recognizes herself as a source of value but fails to recognize the value of other rational beings? The Kantian response to the egoist is to point out the incoherence of the claim. You cannot possibly regard yourself as having a claim to objective value without recognizing the value of other beings whose rational agency is identical to your own. This is why the Golden Rule is effective even when it is used on small children. The egoistic child takes a toy from her friend who is smaller and weaker. So you ask her, “How would you like it if a bigger child took your toys?” Of course, the child should recognize that she would not like that at all. But she offers an egoistic rejoinder: “But that’s not me. They should not take my toys, but it’s fine for me to take their toys.” The inconsistency is obvious. How can the child think that her desire or consent places limitations on others without recognizing that others can make equally legitimate claims on her? If Kant is right to suggest that the value of humanity is objective, then we must see that it holds for all agents, both as makers of claims and as recipients. We are both rulers and subjects in the kingdom of ends.

To summarize, Kant’s argument goes like this. We begin by asking how anything could have value. Rational agency is the only answer. We acknowledge this fact (even if implicitly) every time that we act. That is, we regard ourselves as having objective value insofar as we are rational agents. On pain of inconsistency, we must recognize this in others as well: “But every other rational being also represents his existence in this way consequent on just the same rational ground that also holds for me” (G 4:429). This leads us to the conclusion that the value of humanity is objective and unconditional. It is the only end that could possibly make a valid claim for all rational wills, and this means that it can serve as the ground of the categorical imperative.

Once Kant reaches this conclusion, he immediately presents the formula of humanity: “So act that you use humanity, whether in your own person or in the person of any other, always at the same time as an end, never merely as a means” (G 4:430). But even if we grant Kant’s point about the value of humanity, we might wonder how that is supposed to lead to the formula of humanity.

Perhaps the best way to understand the formula of humanity is to think about how it differs from consequentialist or “teleological” moral theories. Kant’s ethics is often characterized as “duty-based” or “deontological,” and this is typically done to illustrate the contrast between Kantian ethics and consequentialist ethics. This contrast is commonly used to depict Kantian ethics in an uncharitable light. For example, consequentialists often argue that deontological ethics requires us to adopt a “fetishistic attitude toward rules against lying” while utterly ignoring the moral relevance of consequences, even when those consequences are disastrous (Wood 2007, 260). But this is an unfair caricature of Kant’s view.⁶⁰ It would be a mistake to think that Kant dismisses the moral relevance of consequences, and it is important to underscore the importance of teleological thinking in Kant’s ethics.⁶¹ Indeed, this discussion of humanity should have made it clear that Kant is deeply sensitive to the idea that there is something of fundamental value (humanity, defined as the rational capacity to set ends), and that this existence of this value entails moral obligations for us.

So although Kant does not ignore the moral relevance of consequences and ends, his ethical system is importantly distinct from those that ground moral obligation in fundamentally teleological principles. For Kant, ethics cannot ultimately reduce to doing whatever brings about the best result. It was noted above that humanity’s value must be recognized as an “existent end.” This means that our moral duties do not involve bringing about such an end; they involve demonstrating the appropriate kind of respect for the humanity that already exists. As Korsgaard puts it, the escape from consequentialism ultimately involves abandoning the idea that “the business of morality is to bring something about … The primal scene of morality, I will argue, is not one in which I do something to you or you do something to me, but one in which we do something together. The subject matter of morality is not what we should bring about, but how we should relate to one another” (1996, 275).

When it comes to thinking about how we should relate to one another, Kant insists that we must treat each other with respect. If we recognize the value of humanity as something that is fundamentally greater than anything else, then we must express the appropriate attitude toward it when we act. It is the only thing whose value is objective, unconditional, and non-fungible; so, our actions should demonstrate respect for humanity.⁶² Imagine that someone gives you their mother’s ashes in a fragile vase. They tell you that this object has great importance to them, and they need you to hold onto it while they travel. If they were to return home and find you tossing it back and forth with a friend, they could rightly suggest that you failed to treat the object with the kind of respect it deserves. If you were to reply by pointing out that the vase did not break (so there were no harmful consequences), this would not be enough to show that your action was morally permissible.

Of course, in a case of that kind, we might think that the vase is not the thing that fundamentally matters. What matters in such a case is your friend and the fact that you betrayed their trust. Showing respect for the vase is the right thing to do only because you should respect your friend. To respect the humanity of another person is to recognize that their capacity to set and pursue ends is every bit as important as your own. You should not interfere with their rational capacities and you must refrain from doing things that would undermine their ability to set and pursue their own ends. This is why Kant thinks it is wrong to make a false promise to the banker. When you make a false promise to the banker about repaying a loan, your action expresses disrespect for the banker. If you recognize that rational agency is a faculty that deserves respect, then you cannot, on pain of inconsistency, disrespect the ends of other agents as if they did not matter.

The false promise to the banker involves treating someone as a mere means. This is a powerful way of explaining what is compelling about the formula of humanity. To treat someone as a mere means is to ignore or override the moral weight of their consent.⁶³ And such actions are almost always morally wrong; it is objectionable to use someone in such a way. Of course, this does not rule out every action that treats people as a means. We do that all the time in morally permissible ways. You treat your Uber driver as a means of transportation, the person behind the counter is a means of procuring certain goods, and so on. But in those cases, you do not treat people as mere means because you also respect their ends. The Uber driver and the store employee are being paid for their labor, so we cannot say that you have done something that is inconsistent with their consent.

But treating people as mere means is not the only way to fail to show respect for them. Kant thinks that you fail to respect someone as an end in themselves when you express contempt for them (MS 6:463). He thinks that such an action would be wrong even if it does not use the other person in any way. This means that there are multiple ways of failing to respect humanity. In some cases, the failure to respect humanity involves treating someone as a mere means; in other cases, it simply involves failing to show adequate respect for humanity as an end in itself. If we recognize the objective and unconditional value of humanity, then we are morally required to demonstrate respect for humanity when we act. If this is the ground of our moral obligations, as Kant thinks it is, then we can see how his ethical system differs from consequentialism. Our moral duties do not stem from requirements to bring about the best outcomes; they come from an obligation to express the appropriate kind of respect with our actions.

To bring this back to the issue at hand, we are now in a much better position to understand what the formula of humanity demands of us when it comes to respecting the autonomy of rational agents (including ourselves). If Kant is right about humanity, then there is something very special about our capacity to set and pursue ends. This capacity is something that we should cultivate and foster in ourselves. We ought to safeguard it against things that threaten to weaken our capacity to be in charge of our own lives. We must also respect humanity in others and refrain from doing things that would undermine their ability to set and pursue their own ends. To do so is to fail to show the kind of respect that all rational beings deserve in virtue of their rational agency. In the next chapter, we will argue that our relationship with technology threatens to undermine our autonomy. But before moving on, we would like to explore a handful of other moral theories in order to show why they are also likely to commit to the moral significance of autonomy.

2.3.3 The Centrality of Autonomy

Of course, Kant’s ethical theory isn’t the only one that values autonomy. Moral theories of all stripes have reasons to care about autonomy. But their reasons will differ. For starters, why should utilitarians value autonomy? Classical utilitarians are committed to hedonism—the claim that pleasure is the only thing that is intrinsically good, and pain is the only thing that is intrinsically bad. So it is not obvious why a utilitarian should care about autonomy. Nevertheless, classical utilitarians like Mill and Bentham were indeed committed to the moral importance of autonomy. For Bentham, the value of autonomy is merely instrumental. When Bentham talks about liberty (or political sanctions) in his writings, he makes it very clear that pain and pleasure are the only real values at stake (1907, ch. 3). He thinks that it is often the case that someone feels pain of one kind or another when they suffer a loss of liberty.

Given what Mill says in Utilitarianism, it would be reasonable to expect him to concur with Bentham on this point. He makes it clear that happiness should be understood as nothing more than “pleasure and the absence of pain” (1957, 10). He then goes on to explain why things like fame, money, and power are good only insofar as they promote happiness. Given his commitment to hedonism in Utilitarianism, it is somewhat surprising to see what he says about autonomy in On Liberty. First, he clarifies what he has in mind by autonomy (though he does not use that word):

A person whose desires and impulses are his own—are the expression of his own nature, as it has been developed and modified by his own culture—is said to have a character. One whose desires and impulses are not his own has no character, no more than a steam engine has a character. (1988, 57)

When Mill reflects on why the capacity to make our own choices is morally significant, we might expect him to cash it out in terms of pleasure or happiness. But he seems to suggest otherwise when he writes, “If a person possesses any tolerable amount of common sense and experience, his own mode of laying out his existence is the best, not because it is the best in itself, but because it is his own” (1988, 64, emphasis added). Here, Mill sounds like he is claiming that autonomy has intrinsic value. Being in charge of your own life matters not merely because you will make the best choices (the ones that maximize happiness) but because they are your own.⁶⁴

This means that consequentialists have two ways of committing to the moral importance of autonomy. It could be that autonomy matters instrumentally insofar as it promotes pleasure, or autonomy could have final value. Of course, consequentialists who commit to the final value of autonomy are not required to hold the Kantian view that autonomy is the only thing that has this kind of value. Some consequentialists, such as G.E. Moore, have pluralistic theories of the good that include things like beauty, knowledge, and friendship. So consequentialists could simply include autonomy as one of the many goods that we are morally obligated to bring about through our actions.

As for eudaimonist virtue ethics, perfectionist theories, and ethical egoism, the accounts are going to be fairly similar. Like utilitarianism, these theories are, at bottom, teleological. They begin with an account of the good, and ethical prescriptions are then made on the basis of promoting that good. Aristotelian virtue ethics, for instance, starts off with an account of human flourishing (eudaimonia). For Aristotle, happiness does not merely consist in pleasure and the absence of pain; instead, he believes that happiness requires the possession of certain objective goods. For most contemporary Aristotelians, autonomy is likely to be on their list of objective goods.⁶⁵ The same is true for Aristotle himself who regarded the exercise of one’s own practical reason as partially constitutive of happiness. Perfectionist theories are quite similar to virtue ethics. Perfectionists ground moral and political prescriptions in an account of what is required for the proper development of human nature. Once again, both historical and contemporary perfectionists tend to value autonomy on the grounds that the exercise of autonomy is a fundamental component of human nature and happiness.⁶⁶ The same goes for ethical egoists, who typically count autonomy among the goods that rationally self-interested individuals ought to secure for themselves.

Finally, Scanlon’s contractualism also places a great deal of weight on rational agency. Indeed, the core principle of contractualism is grounded in the equal moral status of persons in virtue of their capacity for rational agency. Scanlon writes,

[R]especting the value of human (rational) life requires us to treat rational creatures only in ways that would be allowed by principles that they could not reasonably reject insofar as they, too, were seeking principles of mutual governance which other rational creatures could not reasonably reject. This responds to the problem of selecting among reasons in a way that recognizes our distinctive capacities as reason-assessing, self-governing creatures. (1998, 106)

Respect for rational agency is a core commitment for contractualists. It is in virtue of this respect that contractualism obligates us to act according to principles that no one could reasonably reject. As Ashford and Mulgan explain, “According to contractualism, morality consists in what would result if we were to make binding agreements from a point of view that respects our equal moral importance as rational autonomous agents” (Ashford and Mulgan 2018).

Of course, this list of moral theories is not exhaustive, but it should be sufficient to demonstrate that there are many paths to the same conclusion. When it comes to recognizing the moral importance of autonomy, we might be, as Parfit suggests, “climbing the same mountain on different sides” (2011, 385). A wide variety of ethical theories share this commitment to the value of autonomy. The capacity to be in charge of one’s own life is something that matters morally. There are very good reasons for believing, just as Kant did, that we ought to respect the capacity of rational agents to set and pursue their own ends.

2.4 Conclusion

Our aim in this chapter was to answer two questions. What is autonomy and why does it matter morally? Throughout the book, we take autonomy to be synonymous with what Kant referred to as “humanity”—the capacity to set and pursue one’s own ends. As we explained in Sect. 2.2, humanity comprises both a set of capacities and a set of conditions for authenticity. Exercising your humanity requires you to have certain capacities to set and pursue ends, and it requires that the ends be, in some important sense, your own. In each case, we identified four conditions. In order to have the capacities required for autonomy, you must have the following:

• (1_a) baseline competence (the ability to form intentions, make coherent judgments, etc.),

• (2_a) a lack of external constraints,

• (3_a) the absence of cognitive inhibitions (such as self-doubt), and

• (4_a) a sufficiently wide range of options (with the ability to pursue them effectively).

And, in order for your ends to be considered your own, they must be:

• (1_c) free of manipulation and coercion,

• (2_c) consistent with your motivational states,

• (3_c) free from alienation, and

• (4_c) not adaptive.

When it comes to understanding why autonomy matters morally, we offered a wide range of reasons derived from a variety of moral theories. Of course, our leanings are Kantian; we believe that autonomy has intrinsic or final value. Kant argues that we are morally required to respect this capacity both in ourselves and in others. This is expressed very clearly in the humanity formula of the categorical imperative. We then showed why other moral theories, such as consequentialism, virtue ethics, and contractualism are also committed to the claim that autonomy is morally significant. So it is with this conclusion in mind that we turn to the topic of the next chapter. If autonomy matters morally, as we believe it does, then we ought to be concerned about our relationship with technology, especially with devices like smartphones.