Personal financial incentives in health promotion: where do they fit in an ethic of autonomy?

Abstract

Aim  This paper reviews the ethical controversy concerning the use of monetary incentives in health promotion, focussing specifically on the arguments relating to the impact on personal autonomy of such incentives.

Background  Offering people small amounts of money in the context of health promotion and medical care has been attempted in a number of settings in recent years. This use of personal financial incentives has attracted a degree of ethical controversy. One form of criticism is that such schemes interfere with the autonomy of the patient or citizen in an illegitimate way.

Methods  This paper presents a thematic analysis of the main arguments concerning personal autonomy and the use of monetary incentives in behaviour change.

Results  The main moral objections to the uses of incentives are that they may be in general or in specific instances paternalistic, coercive, involve bribery, or undermine the agency of the person.

Conclusion  While incentive schemes may engage these problems on occasion, there is no good reason to think that they do so inherently and of necessity. We need better behavioural science evidence to understand how incentives work, in order to evaluate their moral effects in practice.

Introduction

Recently, much policy attention has been focussed on the use of personal financial incentives to effect changes in health related behaviour. Examples include schemes aimed at helping smokers to quit, children to eat healthily, and schizophrenic patients to adhere to their medication. These schemes attract considerable public discussion, usually involving a moral argument against their use. Incentivizing people this way may involve – critics allege – rewarding bad behaviour, reducing intrinsic motivation, treating ‘good’ citizens unfairly, endangering the professional‐patient relationship, coercing the vulnerable and so on. On the other hand, incentive schemes all place individual choice and responsibility for personal behaviour at the centre of policy, and it is arguable that they are more respectful of autonomy than alternative, more frankly coercive or structure‐ rather than agency‐focussed approaches to population health. I do not here discuss the justice‐related concerns about selective targeting of the poor or equity impacts, but concentrate on personal autonomous decision making.

In this paper, I will sketch a framework for appraising the moral status of incentive schemes. I do not intend that this will be applicable directly to particular schemes; rather it will help us understand the moral debates about incentive schemes and their contours somewhat better.

Consider the smoker who wishes to quit, but either cannot get started, or having made attempts to quit, keeps falling back into her old habit. Or the sedentary university professor who knows he should take regular physical exercise, but the call of the email is too strong. Here, we have two examples of individuals who have quite definite behavioural changes they want to make, and health goals they wish to realize, but who find it hard to achieve these changes in a robust way. Consider further that the smoker finds herself pregnant, and the professor is recovering from a heart bypass operation. Not only do the people in question have an interest in their behaviours and health statuses, it is arguable that other people do, and moreover that there is a social or State interest here too. This additional interest may derive from protection of innocent others from harm, or from justice‐based considerations about health inequalities or the fair distribution of scarce resources, or from other reasons. In some situations, the motivation to change behaviour may originate not in the person’s own occurrent preferences, but in externally framed interests. I may either not have preferences in respect of some behaviour or be ignorant about the consequences of acting on my occurrent preferences. For instance, I may be indifferent between sugar and artificial sweetener in my drink, if they taste the same; and I may be ignorant about their relative health benefits. The State might take a view that artificial sweetener is safer for one’s health in general, and ‘nudge’ me towards artificial sweetener, while leaving sugar an open option. For instance, it might subsidize artificial sweetener or tax sugar and allow the price mechanism to influence my choice.

Recently, many researchers and policy makers in health care, public health and health promotion in both developed and developing economies have taken an interest in ways of positively shaping the environmental influences on choice. ¹ , ² , ³ , ⁴ Traditional approaches to behaviour change include information provision (which is of limited effectiveness) and prohibition or regulation of ‘bad’ behaviours. ⁵ Prohibition and regulation can be effective, but have both pragmatic and moral difficulties associated with them: the costs of ensuring compliance can be high; their burdens may lie unequally and unfairly on different sections of society; they can be relatively insensitive to individual differences; and their restrictions on individual liberty can be difficult to justify. ⁶

The idea behind actively shaping the environmental influences on choice is that while the individual patient or citizen retains ultimate decision‐making and deliberative authority over their own conduct, contextual and situational factors can be modified so as to facilitate making choices which are coherent and acting consistently with those choices. The moral concept of the authority of the patient in making decisions in line with his preferences and values, without undue influence, domination or coercion, is usually termed ‘autonomy’. Personal autonomy is currently the central moral value in contemporary debates about health promotion. This is for various reasons: the influence of the law relating to consent; the influence of political liberalism; and the origins of health promotion ethics in medical ethics. Thus, debate in this area tends to focus on how to promote health in ways consistent with personal moral autonomy and political liberty. ⁵ Modification of the context of choice could involve a micro‐level strategy tailored to individuals’ own preferences and identified decision‐making biases as, for instance, is used in some web‐based schemes involving self‐incentivization. Or it might involve a meso‐level strategy, where individuals in small‐ or medium‐sized groups come together to regulate each others’ behaviour (as in some school or employment‐based incentive schemes). Or, most commonly perhaps, it might involve a macro‐level strategy. In a macro‐level strategy, a coordinating body (such as a government, local council, charity or profession) intervenes to change the environment for relatively large numbers of people, in ways which the majority would endorse.

An example of a macro‐strategy might be the prohibition of smoking in public places. This is a policy which commands widespread support, even among smokers. It does not involve banning smoking as such. But it does involve reducing the harm to third parties by moving smoking outside; and this is the major rationale for the policy. ⁷ However, as a side‐effect, it changes the default option for a smoker from smoking at will, perhaps only semi‐consciously, to needing to make an active decision to go and smoke. A consequence of this may be that the smoker cuts down or quits smoking. And this may well be consistent with what the smoker actively and explicitly wishes to do, but finds it hard to do where temptation is within easy and constant reach. ⁸

As it happens, while these effects on smoking behaviour are welcomed by most public health practitioners and policy makers, the ban on smoking in public places is normally justified by appeal to the need to prevent harm to unconsenting third parties, rather than by appeal to the health benefits to smokers. This is because of a worry that appeal to the benefits to smokers may involve either frank paternalism, or the so‐called ‘tyranny of the majority’. ⁹ We can split the concern here into two: is the choice of end paternalistic (i.e., the valuation of something as a harm and the desire to reduce that harm, where that harm might be considered as self‐inflicted)? And is the selection of means to reduce that harm paternalistic? Paternalism is normally considered a bad thing, in that it involves overriding personal autonomy in the best interests of the person, rather than letting her decide for herself. This is bad, under normal circumstances, because it involves overriding the person's considered values, substituting someone else's values illegitimately, and perhaps treating the person as a mere means to someone else's ends. The debate about incentives and other means inspired by the behavioural sciences should really focus on the latter, rather than the former. The former problem is generic to any public health activity, and it is inherent in any policy‐making involving social coordination. What we could call the new behavioural public health is no different. But what is different about it is its approach to understanding and intervening in the ways people make decisions and behave.

Take another smoking example. Until recently, if I wished to give up smoking, my resolve was continually challenged by exposure to highly visible and attractive tobacco product advertising, reminding me of the pleasures and glamour of smoking. Now that this advertising has been almost completely eliminated, I no longer have these cues all around me, and I am that much less frequently reminded of my dormant but not extinct desire to smoke. I am still at liberty to smoke, if I wish to, and I am well aware of the brands and forms of tobacco product available to me, because this information is easily obtained at point of sale or by word of mouth. A smoke‐free life has become that much easier, without anything but a trivial impact on my effective liberty. There may be arguments about the freedom of speech of the tobacco manufacturers and vendors, and it might be argued that the quality of my decision‐making vis‐à‐vis choice of brand or form of tobacco product has been reduced. But the former is not relevant to individual autonomy on the part of the smoker, and the latter is almost certainly irrelevant where there are other kinds of information available and where there is in fact very little to choose between products. ⁷ , ⁸

This example shows how a change to the environmental cues to behaviour may have a beneficial impact on people’s ability to stick to a behaviour change they will and endorse, without significant impact on their liberty. Contrary to the intuitive idea that changing environments to change behaviour must be inherently underhand and autonomy‐undermining, we have here an example which shows how such environmental changes may be autonomy‐enhancing, by allowing the agent to stick more closely to his or her settled choices and decisions. On the other hand, the example of tobacco advertising also reminds us that the subtlety of the ways environments for choice may bias those choices can involve shaping behaviours in ways which are not salient to people’s conscious decision making and which might indeed subvert their most deeply held or deliberatively endorsed values and preferences. ¹⁰ This represents an important challenge to individual autonomy. And, in at least some cases, where agents reject the policy’s goals, but the policy makes it harder for them to stick to their preferences, this may additionally involve unfairness. ¹¹

One type of intervention to support personal behaviour change which has received considerable attention is the use of personal financial incentives. Money incentives – as we shall see – are intended to reinforce the desirability or feasibility of behaviour changes which a patient may wish to perform but for one reason or another fails to do in the absence of an incentive. They bring long‐term goals into the short term decision horizon, for instance. ¹² In this paper, I will give a preliminary account of the moral issues which arise in the use of such incentives in health promotion, paying particular attention to considerations of personal decision‐making autonomy. It is not my intention here to give conclusive arguments about the merits and disadvantages of this type of intervention, or specific instances of it, but rather to give an overview of the types of arguments which arise.

A case study

Consider a 16‐year‐old young woman, Holly, who has been offered Chlamydia screening, and has either declined it or failed to attend her appointment. The local health‐care provider (in England, the Primary Care Trust), noting low rates of uptake of the programme in this agegroup, has established a scheme whereby if young adults between the ages of 12 and 18 come for screening, they will be given a £10 mobile telephone credit. The scheme is a general scheme for the agegroup: the health‐care provider decides that selective entry criteria which exclude the sexually inactive are inefficient and put participants off. Learning of this scheme, Holly presents herself for screening. There are three broad concerns here: coercion, bribery and undermining her autonomy.

Coercion

The ethical arguments here are diverse. A first argument, often mentioned in media debates about this type of incentive scheme, is that it involves coercion. The reasoning is that Holly has changed her behaviour in response to the offer. But for the offer, she would – it is assumed – not have come for screening. It is further assumed that her non‐attendance reflects a considered choice on Holly’s part, and that therefore she has good reasons not to attend. The offer of a financial incentive has overborne her considered choice, essentially making her do something that all other things considered she does not wish to do.

One difficulty here is that there is no standard account of coercion which is a commonly agreed upon account of what it involves. ¹³ While there are a range of philosophical accounts of coercion, these do not map tidily onto either the ordinary language usages of the term or even onto the legal usage. Another problem is that the argument may well prove too much: a frequently used counter‐argument here is that on this account, any paid employment involves coercion. Moreover, it rests on the questionable assumption that Holly’s non‐attendance did reflect a considered choice, as opposed to inertia, forgetfulness, or having other things to do which were more pressing or enjoyable. Finally, it is possible that Holly may not object to screening, so much as not see its relevance; and further, that as she sees it as irrelevant to her, she therefore thinks it is being done for others’ benefit alone (including the benefit of the professionals). And hence, it might be that she expects to be paid to do something which benefits others but not her. On this account, she might think it more coercive to persuade, cajole or harass her into attending for screening, than to be paid to do so.

As we can see, the coercion theory is very difficult to make out, purely on the basis of making sense of an individual’s response to an incentive. It may be that some version of the coercion theory can be sustained, under specific conditions. One way it may work is that if the subject is offered a significant amount of money, conditional on completing a particular series of actions, then what is most salient to them is not the value of the actions in themselves, but the fear of loss of the reward. Behavioural economists point to the well‐documented phenomena of fear of anticipated regret. ¹⁴ Coercion typically works by forcing an agent to do something through fear of the consequences of not doing it. To force you to do something, I must arrange that the costs to you of not doing so are both large and frightening to you. And I must arrange that your welfare is dependent on my will, so that you have an interest in keeping me happy, on pain of your feared loss. So, in the incentive situation, if the structure of my offer of money is such that you come to consider the money as ‘already’ yours, then my proposal not to pay you unless you do as I ask is framed as a loss. And if the loss is big enough, you may feel coerced to comply. The scale of the loss here is to some extent subjective: one element of some acts of coercion is that they work by taking advantage of pre‐existing needs, so that offers of even small amounts of money might coerce somebody in poverty. This would be explicable in two, possibly interacting, ways: first, the degree of desperation induced by severe poverty might make even a small incentive much‐needed; and second, even if the recipient is not desperately poor, the utility of a small amount of additional money can still be predicted to be much greater than the utility of the same amount of money to someone much better off, because of the well‐attested phenomenon of the diminishing marginal utility of money. ¹⁵ , ¹⁶ These questions are ripe for empirical investigation as much of our ethical argument here depends on the specific features of the psychological mechanisms in play, and as much of our policy choice will turn on how and when these mechanisms work as well as when interventions based on them are morally justified.

Bribery

Suppose Holly does not feel coerced, and no impartial observer would consider her choice to be coerced. Another moral difficulty with her decision to take part in screening may then arise. This is the claim that she has been bribed to take part. There are two different moral concerns here. One is that she has been paid to do something which she should have been doing anyway. This is the type of bribery that concerns us in connection with paying bribes to public officials merely to do their jobs in a timely and professional way. The other is that she has been paid to do something which she should not have been doing anyway. This is the type of bribery that concerns us in connection with paying bribes to public officials to secure (unfairly) a benefit to the payer of the bribe.

Some critics of inducement to Chlamydia screening may disapprove of the screening programme as such. Here, we have an unusual case, since there is presumably nothing wrong in itself with diagnosing and treating a curable infectious disease (assuming, for the sake of argument, that Chlamydia screening is safe, has low false positive and negative rates, and that the treatment is safe and effective). ¹⁷ But the critics may consider that the screening programme has deleterious effects on the morals of young people, perhaps by contributing to a culture of ‘safe sex’ in teenagers and young adults rather than ‘no sex’ (abstinence before marriage). This is a standard type of objection to ‘harm reduction’ programmes. ¹⁸ This is a debate all on its own. Here, let us suppose that there is a moral objection to screening, held strongly by some. Those people will see an inducement to participate as corrupting Holly’s morals, by inducing her to participate in an immoral programme, and by getting her to change her mind, or at least her behaviour, for money.

Individuals who have no such moral objection to Chlamydia screening might object to the inducement scheme for precisely opposite reasons. They might see it as bribery, in that Holly should participate for non‐monetary reasons: because it is in her interest to do so, and because it promotes public health by allowing adequate infection surveillance and control. Many critics of incentive schemes dislike this apparent feature they share, of paying people to do what they should be doing anyway. ¹⁹

This concern can also be analysed further. First, there is the concern that payment may weaken internal motivation, and second, that it may undermine prosocial behaviours. ²⁰ The weakening of internal motivation concern is that we may shift people from doing things because they are the right thing to do, or because it is in their own best interests, to doing things to get external rewards. If I do something for the money, the thought goes, I am orientating my practical reason towards the reward, and overlooking the reasons which properly ought to motivate me (my health or the public good, for instance). ²¹ I may even come to expect reward, so that I am less likely to act on intrinsic motivations than I was before. Indeed, this attitude might come to infect my decision making outside the context of this particular incentive scheme and become a more general feature of my expectations and motivations. This takes us to the second concern that I may be more likely to act in ways which benefit others only where there is some tangible benefit in there for me. This has sometimes been argued in the context of payments to research subjects in social survey research; if no payment is offered at all, then a certain response rate might be typical across society at large, but if some payment is offered, not only may response rates in surveys which do pay fall (if people think the payment is insufficient), they may fall further in surveys which do not pay, as there is an expectation that researchers should pay and that payment signals that the research is ‘really’ just in the interest of the researcher and not in the general public interest. This concern is sometimes studied by economists under the label ‘motivation crowding out’. ¹⁹

How are we to evaluate these arguments? The first observation, again, is that largely they depend on testable empirical hypotheses, and there is an extensive literature in psychology and behavioural economics on just these hypotheses in other contexts. ²² Secondly, as with the coercion objection, the bribery objections – they are really as I have shown here a family of objections rather than just one – may prove too much. There are many activities where payment is necessary and expected, and where the remunerated behaviour is in the interest of the agent or the public, but we do not label the payment bribery and evaluate its moral status as such. Philosophy lecturers typically expect to be paid for their work, for instance, even if they would philosophize without payment and even if doing philosophy is in the public interest. To make sense of the bribery claim we need a much finer grained and contextually sensitive account of the moral wrong the bribery claims are trying to identify. Partly this depends on the texture and context of the transaction: many parents would think it wrong to pay their child to do their homework, and wrong to pay them to clear Granny’s garden of fallen leaves, but acceptable for them to be paid by a healthy adult neighbour. On the other hand, Granny might well give the child a token reward for being a good boy, having tidied her garden, being careful to make it clear that it was a gift, not a payment for work. And grasping the differences between reward and payment, and contract and gift is no trivial matter, especially for a child. This is a very subtle sociological problem, which has had relatively little recent work performed to investigate it empirically. ²³

Undermining autonomy

The purported moral wrongs in coercion and in bribery are in a sense external threats to autonomy. They rest on the assumption that the agent is autonomous, and address her as such. Holly, it is assumed, knows that she should go to be screened, but cannot be bothered, and has to be bribed to do what she should. Or, she is a fine upstanding woman, who rejects the offer because she has firm moral objections to it. But the offer of money corrupts her resolve, either through need of the inducement, or fear of the loss, or some other reason to do with being distracted by a powerful extrinsic motivation.

Some criticisms of inducements see them as undermining autonomy itself. On the above account, Holly’s autonomy is not undermined, so much as suborned. But if she were to feel that actually she was not the author of her choice to be screened in any meaningful way, or that her status as a person was in doubt or under threat, we might want to say that her autonomy as such was undermined. Noting that in Holly’s case, given the nature of the intervention and the types of incentive used, this is unlikely, let us turn to cases where autonomy is already fragile: mental illness and drug dependency.

Notice first that there are forms of coercion which are so extreme as to involve ‘breaking the will’ of the agent. Torture is a typical example. ²⁴ , ²⁵ So far as I am aware no critic of incentives in health promotion alleges that incentives break the will of agents, and arguably even very large sums of money would not do so in the way that physical pain or intense and acute psychic distress do. What very large incentives might do is induce someone to do something which radically disrupts the narrative unity of their personhood. The standard example here is the ‘indecent proposal’, where someone may be induced to betray some important personal or normative commitment under the lure of large financial gain. ²⁶ There are two elements to the concern here: one is with personal integrity itself, and the other is with the behavioural change induced. In the health context, it is hard to imagine a personal behavioural change which would involve a radically disruptive change in someone’s self‐image, in ways which neither the person evaluating the decision prospectively nor the person looking back on the decision would endorse as being ‘their own decision’, but one example may be a schemes which offer sterilization to people with drug problems or large families (one such scheme being the widely publicised ‘Project Prevention’). ²⁷ Decisions to break an addiction may well involve a breach in the narrative unity of the self – I might well see ‘me as an addict’ and ‘me clean’ as truly different people. But this could be something I profoundly wish after the change, and profoundly endorse beforehand. ²⁸ , ²⁹ The most likely case involves inducing someone to take long‐term antipsychotic medication in a context where all things considered they dislike the side‐effects of the medication and see their unmedicated self as their true, authentic self. ³⁰ This is indeed a troubling problem in the ethics of psychiatry. But it should be stressed that the moral status of inducement here does not depend on this dilemma. The dilemma exists whether or not the inducement is offered, and insofar as we dislike an inducement in this context, it is arguably because it is an inducement to do something we dislike, or judge to be wrong, rather than because of anything inherently wrong in the inducement itself. ³¹

Perhaps this is too quick. It may be that treating someone with antipsychotic medication is only acceptable with their consent (for instance, the Mental Health Acts in the UK require the capacitous consent of even formally detained patients to electroconvulsive therapy, for instance). It may be that this is because the medication changes the personality in important respects. So if we are concerned that the inducement undermines the quality of the decision by the patient, then the moral justification for the treatment itself may be in question.

How may inducements undermine the quality of the decision? Recall the concern about extrinsic motivation. That was introduced as a concern about undermining people’s ability to do the right thing for the right moral reason. However, it may be that it applies here too, in a different way. If the moral justification of the treatment rests on the deliberative endorsement of the treatment (assume here that we are discussing long‐term treatment of the mentally ill in the community, rather than crisis treatment of someone acutely mentally ill), then what may be from a legal point of view a valid, capacitous consent may fail to be a morally valid consent. If it matters that the patient is doing it in full knowledge of the risks and implications of accepting treatment (and of the alternatives and their consequences), then a decision which is actually made with one eye on the money may not meet this test. In some cases this is clear: a patient who really does not want the treatment, but desperately needs the money, and consents specifically to get the money is perhaps better described as being coerced, as above. But consider the more subtle case of the patient who, perhaps distracted by the money, fails to deliberate carefully enough and consents in a spirit of ‘oh, all right then’. Has he been nudged in an insidious way? Has the focus on the mechanisms of his personality come to treat his person as merely phenomenal, in such a way that his moral autonomy has been undermined? ²¹

Much of the discussion of incentives in the policy literature builds on the work popularized in Sunstein and Thaler’s ‘Nudge’. In my opening section I described an approach to making smoking cessation easier by banning tobacco advertising. The idea was not to ban smoking as such, but simply to regulate the ways in which I can obtain information about it, so that I can obtain it easily should I wish but not have it forced on me should I wish to avoid it. This is a very crude approach to adjusting my ‘choice architecture’ as Sunstein and Thaler put it. More subtle ways of doing so involve building on evidence from psychology and behavioural economics about the ways in which my decision making may be erratic, irrational or self‐subverting. These experimentally inspired (and sometimes experimentally evaluated) interventions aim both to understand how we manage to make choices which are incoherent with our stated preferences or the values of our best selves, and to restructure the conditions under which we make choices so that they are not subverted in the same way. However, it is arguable that the strategy is troubling. Typically, these interventions do not work by unbiassing our decisions and choice frameworks. Instead, they work by using the same biases to produce choices which fit with our stated preferences or values of our best selves. So the idea is not actually to improve the quality of our decision making, but to trick ourselves.

This may not matter. Although ‘Ulysses contracts’, for instance, are controversial, this is not because of how they work in the present, by making us pre‐commit to a certain course of action and making us stick to it. The idea of a Ulysses contract draws on the classical myth of Odysseus/Ulysses, who wished to hear the song of the Sirens without risking being lured to his death. He instructed his sailors to tie him to the mast, and not to untie him until they had passed the Sirens, whatever he might say thereafter, and no matter how forcefully he insisted. ³² , ³³ In psychiatry and dementia care, a Ulysses contract is a form of advance decision whereby the patient, while competent, binds himself to a certain treatment plan even if, later, he says (competently or otherwise) that he wishes to be released from it. It is because they bind future selves to the wishes of present selves in ways which the future selves may not truly endorse. A person can rationally sign a Ulysses contract, endorsing both the course of action committed to, and the mechanism of enforcement. They are vulnerable to the criticisms that people may genuinely change their minds and that the contracts are insensitive to changes in external circumstance. Now, consider the incentive scheme which seeks to nudge us into giving up smoking by offering us small short term rewards, and thus overcoming the weak influence long run future health states have over the present desire to smoke. If I consent to the incentive scheme’s structure, and to the plan to give up smoking, and to the psychological mechanism, then I have consented in full to intervention. But it is likely that absent a good explanation, I do not grasp the way in which the scheme works. And my consent may indeed by framed by factors other than deliberative endorsement of the smoking cessation intervention. It may be that my apparently autonomous decision making is being tweaked by this psychological sleight of hand.

In a health‐care ethics arena in which we place enormous moral and epistemological emphasis on autonomous choice, the idea that we ‘cheat’ autonomy by framing the context of choice in ways which the agent may not notice or take account of is controversial. There are a number of possible responses here. First, it is clear that autonomous choice, given the pervasiveness of cognitive biases, is empirically, if not normatively, much more complex and perhaps compromised than we ordinarily allow. Second, it is also true that many actors in the health field are making diverse uses of these cognitive biases in ways we often overlook or are unaware of – notably the behaviour of the food, tobacco and other industries and companies. In the light of the pervasiveness of cognitive biases, we should of course highlight the way that corporate actors intervene in ways which trade on these biases. But this will not eradicate those biases. Indeed, most likely nothing can. Given that, making good and morally careful use of them is sound policy. Moreover, if it can be performed in a way which commands reflective deliberative endorsement by citizens and patients, then arguably it is morally justified, provided it is fair, effective and efficient.

Conclusion

In this short paper, I have not been able to give a comprehensive moral evaluation of the use of incentives in health promotion and health care. In particular, I have not discussed the equity and social justice elements of the debate, and have concentrated almost exclusively on the issues relating to personal moral autonomy. Nor have I been able to examine in any depth what we can call the relational or constitutive elements of autonomy, particularly in the context of the impact on the doctor–patient relationship and the role that incentives might play in altering the terms of that relationship. ²⁹ , ³⁴ I have concentrated, in perhaps a rather traditional way, on the ethics of influencing what one person decides without deeper consideration of their social and political context. This is simply a first iteration, intended to open, rather than close, the question of when and how incentives may be ethically acceptable in health care. The somewhat trite argument that we allow people to be paid to work, so paying them to do other things too introduces no new moral problems is clearly false, because it neglects the social meanings of different spheres of life, and different ways of choosing. ³⁵ Nonetheless, too hasty a dismissal of incentives overlooks their potential benefits, and perhaps relies on a naive conception of the person in its own turn. What I have attempted to do, in the light of some of the behavioural science evidence, is show that the initial moral reactions widely reported in the literature and the mass media turn out on closer inspection to be neither so compelling nor so straightforward as they at first appear. Further analysis and empirical evidence are certainly necessary.

Funding

The research for this paper was part of the work of the Centre for the Study of Incentives in Health, supported by the Wellcome Trust, grant reference WT086031MF (Principal Investigators TM Marteau, RE Ashcroft, PH Dolan).