It would be difficult to overstate the importance of the concept of autonomy in the realm of bioethics. Consider, for example, the case of a person who refuses a potentially life-saving vaccine because they falsely believe that it is particularly dangerous; a person who is clinically depressed who wants to end their life; or a person who refuses life-saving medical treatment for religious reasons. In all of these cases, it seems that questions surrounding the autonomy of the individual and their preferences ought to be at the forefront of our deliberations. If the individual is autonomous, then their preferences count in a way that they otherwise will not. Autonomy, when it is present, is often thought to trump considerations having to do with well-being, but when it is absent, then considerations of well-being seem to be all that remain. Thus, the absence of autonomy seems to warrant paternalism to some degree, and its presence sometimes seems to require us to allow significant apparent harm to occur.
In Autonomy, Rationality, and Contemporary Bioethics, Jonathan Pugh provides us with a detailed account of autonomy that addresses an impressively wide range of questions and problems often faced by autonomy theorists. He focuses throughout most of the book on defining local, decisional autonomy – which concerns the autonomy of our individual choices and the mental states that caused them – and his account is one according to which both theoretical and practical rationality are required for autonomy. After summarizing what I take to be five of the main requirements of this account, I will then discuss how Pugh's theory addresses two big problems that theories of autonomy often face.
The first requirement that must be met for our choices to be autonomous is that they must be motivated by beliefs that were formed in theoretically rational ways. Theoretical rationality involves norms like “being responsive to evidence in sustaining one's beliefs, drawing logical implications from matters of fact and probability, and holding broadly consistent and coherent sets of beliefs” (p. 22). The reason why theoretical rationality is required for autonomy on Pugh's account is that autonomy is concerned with individuals more or less successfully pursuing their genuine ends; having true beliefs tends to make us more successful in this pursuit; and theoretical rationality tends to lead to true beliefs. This condition establishes that a person will lack autonomy if, for example, they reason that a particular procedure will probably result in their death because (1) it involves the administering of anesthesia and (2) anesthesia sometimes results in death. Since this inference is faulty, then their refusal to undergo the procedure on the basis of the resulting belief will lack autonomy.
One might worry that such a requirement is overdemanding; after all, given the prevalence of cognitive biases, it seems likely that many of the various beliefs that motivate our day-to-day choices were formed in theoretically irrational ways. Pugh has two responses to this kind of concern. First, he argues that we cannot override a patient's choice in all cases in which they seem to lack autonomy – for a variety of reasons – so a rationalist account such as his will not lead to overruling patients’ choices too much. Second, he argues that global autonomy does not require that most of one's individual choices be locally autonomous. Still, if such a requirement conflicts with our intuitions about cases by deeming choices non-autonomous even when they seem autonomous, the concern about overdemandingness might remain.
The second requirement of Pugh's theory of autonomy is that we must hold what he calls ‘decisionally necessary beliefs’. These are the beliefs an agent must hold ‘if her action is to be appropriately connected to the pursuit of her intended end’ (p. 132). While this condition is not, strictly speaking, a matter of theoretical rationality, it is motivated by the same considerations that motivated Pugh's requirement for theoretical rationality – namely, both theoretical rationality and holding decisionally necessary beliefs are important for ensuring that our values and motivations are reflected in our choices. This requirement establishes that a person lacks autonomy if they consent to a procedure without understanding that it comes with some risk, or if they aren't aware of certain significant effects of the procedure (e.g., they don't realize that a vasectomy results in infertility). It also helps to motivate Pugh's explanation of what is required for informed consent.
This discussion of decisionally necessary beliefs is one I would have liked Pugh to spend more time on. Intuitively, it seems to me, holding a particular belief can only plausibly be necessary for autonomy if the relevant information is, in some sense, available to us (e.g., perhaps our physician has the relevant information). To see this, consider the case of someone who has a poorly understood illness and whose physician suggests one invasive, expensive, painful, or otherwise inconvenient treatment option after the next. None of these are effective, and none of them had any chance of being effective given the actual cause of the illness. But while these options are unsuccessful in all nearby possible worlds (to use one of Pugh's tests for decisionally necessary beliefs), we have no way of knowing this since the actual cause of the illness is unknown to us.
It seems that, on Pugh's view of decisionally necessary beliefs, the patient's choice to undergo these treatment options would not be autonomous since they lack the knowledge they would need in order for their choices to connect up with their motivations: undergoing a burdensome procedure without knowing that it is entirely unrelated to the illness you aim to treat, after all, seems relevantly similar to getting a vasectomy without realizing that it will lead to infertility. But it seems implausible, at least to me, that their choice to undergo treatment counts as non-autonomous. This is especially the case given how many unknowns no doubt exist in medicine today and how many experimental treatments people seem to autonomously choose. One potential way around this problem could be to maintain that autonomy is only undermined by lacking decisionally relevant information when someone else intentionally withholds that information, or when one ought, in some sense, to have access to it, but Pugh explicitly rejects this approach. So I would be interested to hear more about what Pugh thinks about cases like these and, if he does agree that the choice to undergo treatment is autonomous, to hear more about how his account can explain that.
The third requirement for autonomy on Pugh's account brings practical rationality into the discussion. He claims that an individual's preferences must be held on the basis of beliefs which, ‘if true, would give the agent reason to pursue the object of the desire’ (p. 142). So while Pugh maintains that autonomy is consistent with being motivated by some false beliefs (so long as they are theoretically rational), this third requirement establishes that autonomy is inconsistent with being motivated by desires that are not reasonable to hold on the basis of one's beliefs. To use one of his examples, if a person were to rationally conclude that being thin is more valuable than being healthy and thereby prioritized thinness over their health, this would be an autonomous choice according Pugh (though he suggests that it is perhaps unlikely that such a belief could be held in a theoretically rational way). However, if a person were to believe that smoking is bad for them and on this basis come to desire to smoke, then their choice to smoke would be lacking in autonomy since it would not be practically rational for them to have this preference on the basis of their beliefs.
The fourth requirement for autonomy – that the individual's motivating preferences cohere with what Pugh calls their ‘character system’ (p. 142) – is also related to practical rationality. Pugh maintains that, in order for a preference or the choices it motivates to be autonomous, there must be ‘a sufficient reason for the agent to maintain that preference in the light of other competing preferences and theoretically rational acceptances’ (pp. 57–58). Importantly, decisional autonomy only requires that an agent have sufficient reason to maintain her preference; it does not require that an agent have what would be the absolute most reasonable preferences for her to possess given her character structure.
This allows for individuals to make autonomous choices even when they are ambivalent in certain ways. Take, for example, someone who is conflicted about what choices to make with respect to end-of-life care. Suppose that they have been committed to pursuing less invasive treatment options, up until the point at which invasive treatments have become necessary. If the agent changes their mind and chooses a more invasive option, even to their own surprise, this will be an autonomous choice on Pugh's view insofar as it is consistent with central elements of who they are. This clarification helps to prevent Pugh's account from being too demanding, since we often make apparently autonomous choices toward which we feel ambivalence.
The fifth requirement for local, decisional autonomy that I will mention is that our choice must not be the result of coercion. I found Pugh's discussion of coercion particularly enlightening, as it is extraordinarily difficult to determine when and why a limited range of options should undermine autonomy, whether the cause of our limited range of options should matter for autonomy, and whether certain offers should count as coercive. Having a limited range of options is certainly not always autonomy-undermining. This is crucial for us to accept in the context of bioethics, since patients will often have a very limited range of treatment options but still seem to be capable of autonomously choosing between them. However, a limited range of options can undermine our decisional autonomy, according to Pugh, if our options are limited because we are being intentionally coerced by another agent. The intentions of the other agent matter here, because in intentional coercion, all of the decision-maker's relevant options are shaped by the coercer in order for them to use the decision-maker's decisional autonomy for their own ends.
The coercer, then, is distinct from someone making a mere offer, as making an offer to someone does not involve punishing them if they refuse to take it. The person making an offer is providing an additional option, but is not altering any of the decision-maker's current options. In contrast, the coercer undermines the autonomy of the decision-maker by reshaping their option set entirely. Thus, while coercion does not undermine theoretical or practical rationality, Pugh maintains that it does undermine decisional autonomy by “robbing the recipient of the freedom to maintain their status quo situation” and thereby involves the “domination of one party's choice domain in the interests of another” (p. 110). This means that, for example, offering someone money for remaining in a clinical trial does not undermine their autonomy (so long as the researchers are not responsible for the person's need for money in the first place).
Pugh's rationalist account of autonomy provides thoughtful and compelling solutions to at least two big problems that any theory of autonomy must contend with. First, agents sometimes make choices that seem like they must be lacking in autonomy given how bizarre the agent's motivating preferences seem. Consider, for example, someone who chooses to refuse life-saving treatment because the treatment will temporarily be somewhat painful. One potential explanation for why the agent lacks autonomy is that the agent has the wrong values, and that being autonomous requires us to have certain values. But Pugh rejects this kind of substantive approach to autonomy because, on his view, paternalism may be warranted in cases in which autonomy is lacking, and he thinks that paternalism will be inappropriate in cases in which we simply hold different values from those which the agent holds. This rejection of substantive views, then, will only be compelling to those who share Pugh's stance on the relationship between paternalism and autonomy.
His alternative solution to the problem of apparently bizarre preferences is that the person's choice will lack autonomy if it is motivated by beliefs that were formed in theoretically irrational ways (e.g., faulty inferences), or if their desires are practically irrational (i.e., not properly related to their reason-giving beliefs, or incoherent with their overall character system). This means that the same apparently bizarre preferences and choices will sometimes be autonomous and sometimes not, depending on the process by which the agent formed the preference and made the choice. This gives physicians reason to try to better understand their patients’ motivations for their choices and to encourage thoughtfulness and reflection in their patients, which is a nice feature of his view.
Pugh's view also addresses a second difficulty, which is that of providing responses to cases that are both principled and in line with our intuitions. Often, providing intuitive responses to cases requires theorists to make ad hoc lists of things that undermine autonomy without providing a deeper explanation of why these things undermine autonomy. Pugh's account provides such explanations. For example, things like deception and psychological manipulation will undermine autonomy on Pugh's view insofar as they lead us to hold theoretically irrational beliefs, to lack decisionally necessary beliefs, or to have practically irrational desires. This allows him to explain why not all external forces are autonomy-undermining (e.g., rational persuasion). He also maintains that impersonal forces, like mental illness, can undermine autonomy for the very same reasons, but that they need not; thus, he warns us against assuming that all patients with mental illness thereby lack autonomy. Interestingly, Pugh also seems to argue that adaptive preferences are autonomy-undermining because they undermine practical rationality. He claims that when we favor a particular option because it is the only option – as opposed to favoring it because of its good-making features – then we are forming preferences that are not backed by our reason-giving beliefs. In addition, adaptive preferences often involve self-deception, which may make it difficult for us to reflect on and maintain the consistency of our set of beliefs and preferences.
There is much to like about this book. It covers a wide range of topics – many of which I could not include here – like the relationship between local and global autonomy, how negative and positive liberty relate to autonomy, and why autonomy is valuable. Pugh also includes nuanced discussions of how to apply his theory to real-life cases, including discussions of the various epistemic barriers to assessing the autonomy of others. His careful discussions shed light on a number of pressing problems, and the theoretical tools he provides us with are valuable contributions to conversations around autonomy, both inside and outside of the field of bioethics.
