Informed consent and liberal individualism

In promoting informed consent to the central place it occupies in medical ethics, the discipline is in step with central currents in liberal political thought. Liberal thought is characterised, naturally enough, by its emphasis on the liberty of the individual. Normal adult human beings are conceived by liberals as having the right and the capacity to make choices that advance their own significant projects. The role of the state, with regard to this conception of the individual, is to promote and harmonise the choices of free individuals. The limits of each individual's liberty are defined by the rights of other individuals: each of us is free to choose how to live and how to act, subject to restrictions stemming from respect for other individual's identical rights.

For liberals, the primacy of individual liberty entails a respect for the private sphere. The role of the state is to allow each of us to pursue what Rawls2 calls our ‘conception of the good’; our notion of what kind of life is valuable (a religious life or a hedonistic life, a life of devotion to good works or to knowledge, a life centred around family and so on). Liberals believe that the state must be neutral with regard to conceptions of the good, neither favouring any nor restricting any (so long as they respect the rights of others to pursue their own conceptions of the good). Different liberals offer different justifications of this neutrality: perhaps we cannot confidently judge the worth of rival conceptions of the good, perhaps people have a moral right to be wrong, or perhaps a profusion of what Mill3 called ‘experiments in living’ is instrumentally valuable insofar as it allows for the assessment of different conceptions.

Regardless of the justification, liberals hold in common Mill's claim that ‘ over himself, over his own body and mind, the individual is sovereign’. 3 Mill argues that liberal thought entails state neutrality between conflicting conceptions of the good, and requires a renunciation of state paternalism. That is, the state may not interfere with individuals' actions, even to promote their own conception of the good. This claim is typically justified on epistemic grounds: individuals are best placed to judge for themselves how to pursue their conception of the good.

In rejecting paternalism, medical ethics extends the liberal conception of individual autonomy from the political sphere to the medical. Just as each of us has the right to pursue our own conception of the good without interference from the state, so, it is plausible to think, we have a right to pursue the good life as we see it without interference from medical professionals. Given the importance of health and life to almost all conceptions of the good (perhaps to all reasonable conceptions of the good), the protection of the medical sphere from unwarranted interference seems justified. Medical ethics might adopt Mill's words for its own:

[T]he only purpose for which power can be rightfully exercised over any member of a civilised community, against his will, is to prevent harm to others. His own good, either physical or moral, is not a sufficient warrant. He cannot rightfully be compelled to do or forbear because it will be better for him to do so, because it will make him happier, because, in the opinions of others, to do so would be wise, or even right.3

The idea that each of us has the right to pursue the good life without interference from others who regard our conception of the good as wrong or immoral is deeply attractive. Nothing in what I shall say here conflicts with it. However, there are grounds for thinking that some degree of paternalism might nevertheless be justified. Though it is unacceptable, on liberal grounds, to promote particular conceptions of the good or interfere with the pursuit of any reasonable conception, there may good grounds for some degree of paternalistic interference with individual choice when this interference can reasonably be expected to promote the pursuit of the good life by that very individual's own lights. We may be justified in interfering with choice when we do so to make people better able to pursue their own conception of the good.

John Rawls2 famously identified a class of goods he called primary goods. Primary goods are goods that every rational individual can be presumed to want. Possession of a sufficiency of primary goods is almost always useful and never burdensome for the pursuit of any reasonable conception of the good, so no matter what else we want, we should want a sufficiency of primary goods. Primary goods include basic rights and liberties, a sufficient income, freedom of movement and occupation, and so on. The state does not violate its neutrality in ensuring that all citizens have a sufficiency of these goods: on the contrary, it is obliged to provide them if it can, since in doing so it promotes individuals' ability to pursue their conception of the good. Extending Rawls's thought, I will argue that medical professionals are justified in a certain degree of (mild) paternalism insofar as that paternalism can reasonably be expected to promote the primary goods that all reasonable individuals want to possess, or to prevent individuals from taking steps that would interfere with realising their own conception of the good. In making these claims, of course, I take issue with Mill's view that we cannot interfere even for these reasons. Mill's view is plausible, I shall claim, only if human reasoning is well designed to allow us effectively to pursue our conceptions of the good unaided. I shall argue that this view of human reasoning is overly optimistic.

Limits on individual rationality

The Enlightenment, out of which liberal political thought grew, was impressed by the power of individual rationality. From Reformation theology, with its emphasis on the individual ability to establish communion with God without the mediation of priests, to the French and American revolutions, central currents of Enlightenment thought emphasised the power of individuals to make decisions for themselves. More than anything else, it was the rise of science that seemed to make this faith in reason plausible. Scientific thought advanced at an ever-quickening pace from the 17th century on, and scientific explanations of natural phenomena became increasingly powerful and encompassing. Enlightenment thinkers saw this success as the result of throwing off chains: chains of deference to Church and tradition, and the chains imposed by despotic authorities. Enlightenment, as Kant saw it, consists in ‘man's emergence from […] the inability to use one's own understanding without the guidance of another’.4

The success of science is indeed the most impressive epistemic achievement in human history. However, there is good reason to think that in identifying individual rationality unchained as the driving force of science, the Enlightenment overlooked an equally significant factor: the social organisation of science. Science is the massively successful epistemic enterprise it is, in important part, because it is a distributed enterprise.5 The distribution of cognitive labour occurs in two ways, one conflictual and the other cooperative. First, scientific claims are tested by researchers (typically groups of researchers: cooperative distribution of cognitive labour occurs within as well as between the units of knowledge production) working independently of one another, who have strong incentives to find fault with the work of rivals as well as to formulate and test hypotheses of their own. Second, researchers take on trust the results of this process, such that a claim that has been independently tested multiple times is very often simply accepted and incorporated into new work. An individual scientist is usually incapable of scientific research on her own: she needs access to the findings of others as well as to the specialised tools that others have built (physical tools like fMRI machines or computers, or mathematical tools like tests for significance). In most fields, she also needs to be embedded within a research group to actually do science.

Independent testing helps to ensure that scientific claims are neither fraudulent nor spurious: that data is not faked nor arises from irrelevant factors (say contamination of samples or order effects). Independent testing is also typically hypothesis driven: that is, it aims at attempting to replicate the experiment, and does so in the service of a particular interpretation of the data, thereby aiding in the refinement of the theories that explain the data. This hypothesis-driven methodology, when combined with relations of trust between scientific research groups, enables rapid scientific progress. Researchers do not need to build their theories from the ground up; rather, they aim to add incrementally to the work of their predecessors (science moves so fast that a scientist's predecessors, in this context, may be a group whose work was published last week).

Because science is deeply dependent on a distribution of cognitive labour, it is perilous to infer from the success of science as an epistemic enterprise to the reliability of individual human rationality. It is not because scientists are freed from all constraints that science works as well as it does: it is because scientists accept a range of constraints on what can be said and how it can be said, on what counts as evidence and what should be tested. If anything, the success of science might support the opposite conclusion to the one drawn by Enlightenment thinkers: we might take it to show the epistemic importance of tradition, where ‘tradition’ is understood as constraining not content but form.

On the one hand, then, we can explain the success of our best epistemic enterprises (in important part) by reference to their social organisation. On the other hand, as I shall now show, when we enquire into the powers of individual human reasoners the picture we are presented with is in many ways bleak. Human beings are, under a variety of conditions, systematically bad reasoners, and many of their reasoning faults can be expected to affect the kind of judgements that they make when they are called upon to give informed consent. In what follows, I will outline a small subset of the evidence about the limitations of individual human reasoning, with special emphasis on the kinds of limitations that might be expected to be relevant to situations in which informed consent is sought.

The fallibilities of human reason

Towards psychological realism in medical ethics

The motivations for making informed consent central to medical ethics were laudable, but as it currently conceived, it rests on implicit assumptions with regard to the capacities of normal human beings that may be unrealistic. At least insofar as the doctrine rested upon the supposition that normal human beings can be expected, unaided and in stressful and novel contexts, to make choices that contribute to the achievement of their own most cherished ends, it seems to be in trouble. Of course, individual autonomy is, very plausibly, important enough that we ought to promote it even if it predictably imposes costs on some individuals, but the evidence is accumulating that these costs are far higher than most people imagine. It may be possible to avoid much of this cost without unduly infringing on autonomy. Indeed, it may be possible to avoid these costs while actually increasing autonomy.

The simplest way to avoid these costs, it might be thought, is by ensuring that patients are given the chance to reassess their decisions. We may think that though patients may initially be overwhelmed by the stress of the decision facing them, if we allow them to reflect again, perhaps over the course of several days (when this is practicable) and change their minds, they will be able to avoid some of these pathologies. Indeed, this kind of strategy may be a part of the solution to the problems outlined above, but there is reason to think that it will be of limited usefulness by itself. There is a large literature on what happens to our judgements after we have made a decision. As a consequence of making the decision, our attitudes to the alternatives change: we come to think the option we have chosen is far superior to the options we have rejected. The relevant mechanism here seems to be cognitive dissonance: because we are aware of the attractive features of those options we have decided not to pursue, we experience dissonance, and dissonance can be resolved by changing our judgements. This phenomenon is known as the spreading of alternatives: alternatives that were initially thought of more or less equal (or at any rate, not very dissimilar) value come to be thought as very unequal after one is chosen. We tend to come to see the option we have chosen as very much better than those we have rejected.21 Lieberman et al22 showed that the spreading of alternatives does not require that subjects recall the option chosen. This evidence suggests that if patients do not change their minds about their initial judgements, this may not be because the initial judgement was the one that was really most in accord with their values.

Alternatively, we might look at ways in which we can prevent patients being subject to the biases and other pathologies in the initial context of choice by teaching them techniques to avoid it. This is known as debiasing in the psychological literature. There have been some successes with the application of debiasing. For instance, prompting subjects to conduct symmetrical memory searches seems to mitigate the effects of confirmation bias to some degree.23 In general, however, debiasing has not proved very effective, and is unlikely to be of very much use in the context of the informed consent procedure. For one thing, there are very many biases to correct for. For another, the context in which these decisions are made, with its inevitable stresses, is far from conducive to the application of these strategies, which are typically cognitively demanding. Though debiasing could be a part of the solution, it comes nowhere near to solving the problems outlined above on its own.

Debiasing is an attractive strategy because it avoids placing any pressure on patients. Given its limitations, however, we have good reason to look beyond debiasing, towards measures that are somewhat more coercive (inasmuch as they involve confronting and placing pressure on patients, while leaving the final decision in their hands). Many philosophers would balk at this suggestion, because they believe that putting pressure on patients infringes on their autonomy. This kind of worry has caused some thinkers to look for non-coercive alternatives. In a recent article constructing a parallel argument for what he calls ‘institutional prosthetics’ (the promotion of good social and individual choices by designing institutions so that the biases inherent in human psychology dovetail with the options that promote human goods), JD Trout24 argues that interventions that harness biases are minimally intrusive, and therefore are not paternalistic. He offers a test for whether an intervention is paternalistic: does it conflict with what would be chosen by fully informed and unbiased decision maker? While I endorse institutional prosthetics, and suggest that they may be helpful with regard to the kinds of political and social problems Trout has in mind and with regard to medical decision making (for instance, presenting statistics in a frequency format may assist patients in some circumstances), I do not think that they have a large role to play with regard to informed consent. The range of possible situations in which informed consent must be sought is too great for institutional nudges to help much. Moreover, I doubt that institutional prosthetics offer a superior solution to the more coercive measure I propose, even from the point of view of avoiding paternalism alone. It is not obvious, first, that institutional prosthetics pass Trout's test for absence of paternalism; furthermore, the test is not one that tracks a morally important property.

It is not obvious that institutional prosthetics pass Trout's test, because there may be reasonable disagreements about what it means to be ‘fully informed’. Is an agent fully informed if she is presented with all the relevant information in a format she is capable of grasping? Or does being fully informed require that she actually grasp the information? Or is the standard even more demanding: perhaps she is fully informed only if the information actually alters her beliefs and other propositional attitudes in the normatively correct way? To put the point in the context of informed consent, is an agent fully informed if she is (say) told about the phenomenon of hedonic adaptation, if she actually understands hedonic adaptation, or if she alters her attitudes towards her decision to take full account of hedonic adaptation? An institutional prosthetic, should one be available, that reliably leads agents to take full account of hedonic adaptation may reasonably be judged paternalistic, on the basis of Trout's test.

We can and should avoid the entire issue, by simply setting aside the question whether paternalism is involved in favour of focusing on the moral goods that antipaternalistic measures are supposed to protect. We should not fetishise antipaternalistic measures like the informed consent procedure. As Beauchamp25 reminds us, informed consent has an ethical rationale: it is designed to respect the autonomy of individuals. If we can redesign the informed consent procedure so that it is sensitive to the evidence regarding the fallibilities of human reasoning without compromising autonomy (perhaps even while increasing it), it would be unethical not to do so. Even if interventions fail Trout's antipaternalism test, they are ethically permissible if they do not violate the goods that antipaternalistic measures are designed to protect.

Since the role of informed consent is to protect and promote the autonomy of individuals, we can best approach the question of redesigning it by reference to the concept of autonomy. Unfortunately, there is no agreed-upon definition of autonomy in the philosophical literature; worse, there are ongoing controversies about central aspects of it. However, there is substantial agreement on core features of autonomy, by reference to which we can guide our reconstruction of informed consent.

The core of autonomy (as its etymology suggests) is self-rule: the autonomous individual is not ruled by another person, or in thrall to any institution or government. However, absence of rule by others is not sufficient for autonomy. The autonomous agent must actually rule, or be capable of ruling, herself. That is, she must be capable of shaping her life as she wants: in accordance with her values, her projects and her conception of the good. It is for this reason that, say, addiction impairs autonomy: the addict may not be ruled by another, but she has difficulty in shaping her life in accordance with her overarching values.26

Informed consent procedures are justified insofar as they protect autonomy: that is, insofar as they conduce to allowing agents to shape their lives in accordance with their own values. Once we recognise this, we also ought to recognise the moral urgency of reforming informed consent to take the fallibilities of human reasoning into account. Though some will see in the proposals I will advance a threat to autonomy, I will argue that just the opposite is true: the constraints I will suggest (or at least something along the lines to be proposed) can be expected to increase autonomy.

The many problems with human reasoning threaten our autonomy in two ways (some threats work one way, some another, and some might work in either or both). Either they cause cognitive illusions, causing us to misapply our values, or they cause our actions to be driven by attitudes that, while in some sense ours, should not be identified with our values.

Most of the evidence outlined above concerns cognitive illusions. When an agent misjudges which of two experiences is more pleasant (or less unpleasant), when it seems to her that an argument is fallacious only because (unbeknownst to her) she is motivated to reject it, when she responds in a certain way to a case due to strictly irrelevant aspects of the way it is framed and so on, she is subject to cognitive illusions. A cognitive illusion can be understood as analogous to a visual illusion: just as a visual illusion can cause us to make erroneous judgements by causing us to misperceive aspects of the situation in which we find ourselves, so a cognitive illusion can cause us to make erroneous judgements by causing us to misperceive features of our circumstances. When we are subject to cognitive illusions we may act in accordance with our own values and our own conception of the good, but we misapply them because the world is not as we take it to be. When an agent is subject to cognitive illusions, she is not ruled by another, but she fails nevertheless to rule herself: her actions cannot be expected to advance her goals in the ways she thinks they will.

The second way in which the fallibilities of human reason threaten autonomy is by bringing us to act in ways that do not reflect our values by causing attitudes of ours that we do not endorse to play a crucial role in our behaviour. Hyperbolic discounting might be understood along these lines. Hyperbolic discounters, recall, are subject to preference reversals: though she prefers good 1 to good 2 at almost all times, when the opportunity to consume good 2 is imminent her preferences reverse. When an agent's discount curves cross, there need not be any fact regarding which she is mistaken. She may recall very clearly that, and why, she usually prefers good 1 to good 2. She may even be well aware that she can expect to regret consuming good 2 (when good 2 competes with 1: eating fast food or smoking with health, for instance). She does not seem, therefore, to be subject to any illusion. However, her autonomy is compromised. Autonomy is a diachronic property of agents: an agent rules herself when she is able to exert her will across time.9 The agent subject to preference reversals is impaired in her autonomy because she cannot do this. Instead, she continually finds herself frustrating her own ends. She cannot effectively pursue health, say, because she regularly fails to go to the gym or refrain from smoking. She cannot even effectively pursue pleasure, because she regularly spends money on gym memberships that she could have spent on holidays and throws out the packet of cigarettes she bought only minutes before.

We also finding ourselves acting in ways that do not reflect our values when we are under cognitive load, tired or stressed. Under these conditions, our behaviour tends to be unduly influenced by environmental stimuli, especially temptations, and also by our implicit attitudes; attitudes that we have either as consequence of our enculturation or perhaps innately.19 These implicit attitudes are genuinely ours, in some sense, but they are not our values. When they have a content that diverges from the content of her conscious values, they are neither endorsed by her nor does she take herself to have reasons to act upon them (often she does not even take herself to have reason to act upon them at the precise moment they cause her behaviour: implicit attitudes typically cause by behaviour in ways that bypass our capacities for reflection upon what we are doing; sometimes, they cause us to misperceive the circumstances we confront and thereby cause cognitive illusions). They cause behaviour that conflicts with agents' plans and with their conception of the good and thus undermine their autonomy.

Since the fallibilities of human reasoning threaten to undermine autonomy, but the purpose of securing informed consent is to protect and promote autonomy, we have good reason to redesign the informed consent procedure in ways that help to avoid these fallibilities, even if the redesign reduces the scope for individual decision making in the procedure. We fetishise the procedure if we insist that the scope of decision making must be as broad as possible, even at the cost of a decrease in autonomy.

It would be a clear infringement of a competent patient's autonomy to have their decisions made by doctors, or anyone else to whom they have not granted this power, in accordance with values that are the doctors, or the hospitals, or what have you, and not the patient's. That would be rule by another: heteronomy. However it does not infringe the patient's autonomy if steps are taken to ensure that their decisions (a) reflect their own conception of the good and (b) promote the primary goods that all agents can be reasonably expected to want no matter what their conception of the good; this may remain true even if, left to his or her own devices, the patient would make different choices due to cognitive illusions or other influences.

What steps, concretely, ought we to take in reforming the informed consent procedure? Here I shall put forward some tentative suggestions: though I am confident that the evidence presented above demonstrates that we need to rethink informed consent, the question how this is to be done requires input from a variety of perspectives. Further, the suggestions I shall offer will be relatively abstract: they would require further work to bring them to the point of implementation (see box 1 for further details and suggestions).

Given the range of evidence that bears on how reasoning processes can be distorted, there is a case for introducing informed consent specialists. These specialists would receive special training in human reasoning and would be taught to be on the look out for the major pathologies so far identified. They might also be expected to communicate some of this information to patients. For instance, when they suspect that faulty affective forecasting might be distorting a patient's judgements (for example, in refusing an amputation), they might draw the phenomenon of hedonic adaptation to the patient's attention. There is some evidence that doing so tends to make people's affective predictions more realistic.27 Informed consent specialists might even express their disagreement with a patient's choice, saying that they will come to regret it. They might point out when the choice conflicts with primary goods that the patient can be expected to value, such as health, absence of suffering and length of life. Perhaps the patient should be asked, or required, to meet with people who have found themselves in circumstances like the one she is in; those who made the choice she has made and those who made a conflicting choice.

Of course it is possible that a particular patient has a highly idiosyncratic conception of the good: perhaps some kind of religious view to which they are deeply committed, which leads them to place little value on a good that most people view as primary. In cases like that, when they arise (and assuming the patient is competent), the patient's choice should be respected.

Harder cases arise when a patient is known, with reasonable certainty, to have a conception of the good with which their choice conflicts, directly or indirectly. Should we respect choices when we have strong grounds for believing that they are made as a result of a distortion of reasoning, but the patient remains obstinate in their choice despite directive counselling of the sort envisaged above? I think we should, though there are steps we can take, with regard to institutional design, that make it less likely that patients will persist with such choices. We might institute mandatory cooling off periods after informed consent, during which the patient is given the opportunity to change their mind. Procedures that are especially likely to give rise to later regrets might require longer waiting periods and more counselling.

It should be noted that there are no panaceas for the pathologies of human irrationality. Any strategy aimed at leading agents to make better decisions short of coercion will itself fall prey to the very problems that it tries to solve: agents will irrationally discount the advice of counsellors or the psychological evidence that is, adduced; they will take themselves to be exceptions to the claims made and so on. The strategies advocated here attempt to pull off a balancing act: respect patients' autonomy by leaving the final choice in their hands, since the conception of the good to be advanced is (almost always) theirs, and not ours, but at the same time raise the quality of their decisions by reducing the extent to which they are subject to cognitive illusions and to which they make choices that they can be expected to regret. No strategy that leaves the final choice in patients' hands entirely avoids the pathologies of human irrationality, but the kinds of strategies suggested can be expected to reduce their power.