ill's rejection of paternalism forms part of his liberty principle, but it is such an important part that he explicitly spells out his anti-paternalism. immediately after stating that the prevention of harm to others is the only legitimate basis of interference with the freedom of individuals.
His own good, either physical or moral, is not a sufficient way-rant. 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. 
But Mill's anti-paternalism does not extend to “children and persons under age” (On Liberty, 137). He also allows us forcibly to prevent a person from crossing an unsafe bridge (152). But here he assumes that the person is unaware of the dangerous condition of the bridge, and that he does not wish to fall into the river. It is his ignorance which justifies a temporary restriction of his freedom. Interference is justified only if there is not time to warn the person about the danger. Mill generalizes by claiming that interference is unjustified unless the person is “a child, or delirious, or in some state of excitement or absorption incompatible with the full use of his reflecting faculty” (152).
In trying to understand the nature and scope of Mill's anti-paternalism, it is useful to bear in mind two different aspects of a person's conduct: the decision-aspect and the consequence-aspect. The former refers to the different ways m which an agent's decision to act in a particular manner is vitiated or impaired, or his consent to certain acts is not “full and free”. Thus a person may commit a dangerous act without realizing, or fully appreciating, its danger. This is the case with the person crossing the unsafe bridge. On the other hand, the consequence-aspect refers to the undesirable consequences of a person's act, as for example the fact that the act will harm him or produce other undesired effects. [109/110]
The cases which Mill excludes from the scope of his anti-paternalism are cases in which both the decision -- and the consequence-aspects are present. Thus when Mill refers to children, it is evident that he believes that they lack certain capacities that normal adults have. Children, especially very young children, have not developed sufficiently their emotional and intellectual capacities, and they also lack the degree of knowledge and experience needed for meaningful choices.
An analysis of Mill's arguments and examples will show that he is opposed to strong paternalism but that he favours a degree of weak paternalism [cf. Feinberg, Paternalism]. Strong paternalism maintains that we are justified in interfering to prevent a person from harming himself even when his decision is fully voluntary or totally unimpaired. Weak paternalism is the doctrine that we are justified in interfering to prevent a person from harming himself only when there is a defect in his decision to engage in the self-harming activity. Weak paternalism therefore justifies intervention when both the decision- and consequence aspects are present, whereas strong paternalism justifies intervention by reference to the consequence-aspect alone.
An agent's decision to perform a particular act may be affected by a variety of factors which reduces the significance to be attached to the decision. A brief discussion of some of these factors will help in the understanding of weak paternalism.
First, a person may lack some relevant knowledge. This can take several forms. An agent may do something without being aware of the harmful consequences of his acts, and there may be reason to believe that had he known these consequences, he would not have acted in the way he did. An obvious case of this is the taking of a medicine with harmful side-effects unknown to the patient. But a person may also be lacking in knowledge not with respect to the consequences of his act, but to the nature of the act. An example of this would be a patient whose modesty is outraged by a doctor who pretends that he is giving her some special medical treatment. Lack of knowledge is often, as in this case, the result of fraudulent or misleading representation by others.
Secondly, there is lack of control. Sometimes a person may be aware of the consequences of his act, but because of [110/111] temporary emotional unbalance, may be unable to appreciate the full significance of these consequences, and to exercise rational judgement with respect to them. Emotional unbalance may be caused by grief, distress, or severe strains. There is also another type of factor. Some drugs are severely addictive, and though the initial decision to take them is freely made, the subsequent “choices” to continue taking them are impaired by the agent's addiction.
Thirdly, there is undue influence. This can be exerted in different ways, in some of which there will be disagreement about whether or not undue influence was really present. Coercion is an example of undue influence, but there are also the pressures of economic inducements, and sometimes of customs and traditions. A person who would not otherwise perform an act may do so because of economic inducements. This is common enough, and does not normally raise questions about undue influence. There was, however, an interesting case in which an air-hostess, guilty of a breach of regulations, consented to be caned by the manager of her company as an alternative to dismissal or being grounded, which would have meant the loss of her “flying pay”. The manager administered six cuts on her buttocks with a light cane. He was convicted of assault in spite of receiving her written consent (McCoy, cited by Hughes, 683-84). The occasions on which economic inducements may be regarded as impairing a person's decision depend on a number of factors like the degree of reluctance of the consenting victim, the relative bargaining positions of the parties, the relationship between the parties, and the nature of the available alternatives. No sharp line can be drawn between the situations in which undue influence is present and the situations in which it is not. The pressures exerted by customs and traditions are also sometimes of such great strength as to be treated by some as a kind of “undue influence”. Thus Glanville Williams observes that the consent to duelling was often given with great unwillingness and solely out of fear of being called a coward. When duelling was legally prohibited, unwilling duellists had an honourable excuse for not fighting, and were thereby freed from “the tyranny of custom” (78).
Persons whose decisions are substantially impaired in one of the ways specified above are likely later on to regret [111/112] their decisions to engage in harmful conduct. Interference will, in retrospect, be accepted and appreciated. But the subsequent consent of the subject is not a necessary condition for the justification of weak paternalism. Suppose, for example, that the person crossing the unsafe bridge condemns our interference. This does not make the interference wrong. What it does is to make any further or subsequent interference wrong. But the original interference is still justified so long as it was not known at the time that the subject would not consent [for discussions of the connection between paternalism and the consent of the subject, see: Carter, Hodson]. Weak paternalism insists on always maintaining some contact with a person's own preferences and values. It promotes each person's own good as defined by that person himself. It is not a cloak for enforcing the values and preferences of the person interfering or society at large.
However, in some cases, because of the nature of the impairment, it is not possible to discover what a person's real preferences and values are. This is true of those special categories of persons children, the mentally subnormal, and the mentally ill who lack the capacities of normal adults. We may not know how particular persons would choose if they ceased to be children, or to be mentally ill, or if they were not mentally subnormal. There is here the danger that interference may be designed to shape and develop their values and preferences in line with those of the interfering party. The subject who is interfered with in this way will come to approve of the intervention; this problem is discussed by: Rawls, 249-50; Murphy, 482-83; Carter, 136-39. In order to reduce the dangers of such manufactured consent, we need something like the Rawlsian device of choosing in a hypothetical state called the original position (Rawls, ch. 3). In Rawl's theory of justice, rational and mutually disinterested persons choose from behind “a veil of ignorance” the principles which are to govern the basic institutions of their society. The veil of ignorance deprives them of knowledge of their particular talents and abilities, their class interests or social positions, their conceptions of the good or their particular moral and religious views, their psychological make-up, and the stage of development of their society. It does not, however, rob them of information about general psychological laws, and general truths about the world, or the nature of human societies, or human nature. Persons in the original position will allow some [112/113] interference with their conduct if they turn out to be children, or adults who lack certain normal capacities.
In adopting the device of the original position for the limited purpose of helping to determine the scope of weak paternalistic intervention in the conduct of special categories of persons, we are not committed to Rawls's ambitious policy of deriving principles of justice from the original position. Rawls applies his theory to the problem of paternalism, see 248-50; see also Richards, 192-5, and Murphy.
There is a much greater likelihood of agreement about interference on weak paternalistic grounds than about the fundamental principles of justice. For example, there will be agreement about the need to interfere to prevent harm, and to protect and develop the capacities for choice between alternative patterns of conduct. But there will also be disagreement, and in particular about how children are to be treated as they grow older.
It is possible that the accumulation of evidence will show that children, except when they are very young, do not lack many of the relevant capacities of normal adults (on this point see Houlgate). If this is correct, then paternalistic intervention in children's conduct will in many cases be unjustified. But even so, there seems to be one important qualification. If we think that relatively young children are capable of making informed choices about sexual matters, we should then be prepared to tolerate sexual relations between children of about the same age. But it does not follow that we should also tolerate the same type of sexual relations between a child and an adult. If we left children completely free to enter into any relationship with anybody they wish, they are likely to be exploited by some adults with much greater experience.
This brief discussion of paternalism is sufficient to show that Mill is quite consistent in accepting weak paternalism while at the same time rejecting the right of society to impose its value on individuals. There are at least three differences between the two doctrines.
I. Weak paternalism involves the protection of individuals from harming themselves in situations where their decisions impaired. On the other hand, in order to enforce society's Zaitpvalues the law may interfere even when a person's decision to engage in an activity is fully voluntary and clearly Informed. Thus to prevent freely consenting adult homosexuals [113/114] from indulging in homosexual activities in private simply because these activities violate the shared morality of society, is to enforce this morality, but the prohibition cannot be justified on weak paternalistic grounds.
2. Weak paternalism is not concerned to punish moral wickedness. Indeed paternalistic intervention only takes place when the agent's decision is impaired in some way, from the moral point of view, this is generally regarded as reducing his moral culpability. A person who acts as a fully free moral agent in violating the accepted morality of society is outside the pale of paternalistic intervention, but, on the other hand, he is precisely the sort of person against whom society may wish to enforce its shared values.
3. The basis of paternalistic intervention is confined to an appeal to the interests of particular persons who are to be prevented from harming themselves. This is as true of strong Paternalism as it is of weak paternalism. But the enforcement of society's shared values often relies on appeals to all sorts of more general considerations, including, as we have noticed, the protection of the distinctive institutions of society. Paternalism focuses attention constantly on the interests of the person with whom we are interfering. The enforcement of society's shared values is only too likely to sacrifice these mterests to the often unreasoning and unreasonable anger and hatred of the community.
Whereas Mill is prepared, to some extent, to accept weak paternalism, he seems to be absolutely opposed to strong paternalism. One argument Mill uses against strong paternalism is often cited:
But neither one person, nor any number of persons, is warranted in saying to another human creature of ripe years, that he shall not do with his life for his own benefit what he chooses to do with it. He is the person most interested in his well-being: ... with respect to his own feelings and circumstances, the most ordinary man or woman has means of knowledge immeasurably surpassing those that can be possessed by any one else. The interference of society to overrule his judgement and purposes in what only regards himself must be grounded on general presumptions; which may be altogether wrong, and even if right, are as likely as not to be misapplied to individual cases, by persons no better [114/115] acquainted with the circumstances of such cases than those are who look at them merely from without. 
Certainly Mill is here claiming that generally a person knows his own interests best. But it has often been argued that this is only a defeasible general presumption, and that elsewhere, in Principles of Political Economy, Mill himself recognized cases in which the government knows a person's interests better than the person himself. One such case is the provision of education. Mill argues: “The uncultivated cannot be competent judges of cultivation. Those who most need to be made wiser and better, usually desire it least, and if they desired it, would be incapable of finding the way to it by their own lights” (Collected Works. Vol. III, 947). Mill is discussing the limits of the doctrine of laissez-faire in social and economic life generally. Government must step in to provide services that individuals will not be able to provide for themselves as effectively. But Mill does not argue that the government is justified in coercing normal adults to do certain things for their own good. His position is that the government should provide services which people are free to use if they so choose.
So Mill does not jump from the acknowledgement that individuals are sometimes not the best judges of their own interests to the advocacy of strong paternalism. He is convinced that coercive interference tends to stunt the person's capacities, and is at least to some extent degrading.
To be prevented from doing what one is inclined to, or from acting according to one's own judgment of what is desirable, is not only always irksome, but always tends, pro tanto, to starve the development of some portion of the bodily or mental faculties, either sensitive or active; and unless the conscience of the individual goes freely with the legal restraint, it partakes, either in a great or in a small degree, of the degradation of slavery. [Collected Works. Vol. III, 938]
But when a government merely provides individuals with various facilities, and leaves them free to avail themselves of these facilities, it does not restrict their freedom. Mill therefore justifies non-coercive government activity.
It is of course part of the price of allowing individuals freedom to choose for themselves and to conduct experiments in living that they sometimes make mistakes and harm themselves. But it is not conducive to the development of [115/116] individuals that they should be fenced in from all sorts of harm. However, it is plausible to argue that some self-harming actions should be treated differently. In order to develop oneself in any direction one must preserve a minimum of intellectual and physical powers and capacities. If, through his risky actions, an individual kills himself, or destroys these powers and capacities, then he removes the very prerequisites of his self-development. To prevent him from harming himself in these ways is to preserve opportunities for his future development, and to leave him with the ability to choose between different patterns of life. So, it may be argued, some cases of strong paternalism help rather than hinder the process of self-development.
To this Mill's reply is that a person may place such a high value on what he does that the risk of death or grave harm is worth taking: “no one but the person himself can judge of the sufficiency of the motive which may prompt him to incur the risk” (On Liberty, 152). He should be warned of the danger and of the risk he is taking, but once he knows this but still persists in going ahead, he should be allowed to do so.
But it is sometimes said that, for example, compelling people to wear seat belts is only “a trivial piece of compulsion” and that the risks of not wearing seat belts are disproportionate to the benefits. On the other hand, it is conceded that the risks involved in mountain climbing may be proportionate to the benefits it gives; these two examples are used to illustrate the same point by three different persons: Dworkin, Paternalism, 125; Regan, 200; Glover, 179-80. Mill's point is directed precisely against this type of argument. For many of those who drive without seat belts will disagree that their case is very different from that of the mountain climber. They will claim, with some justice, that they value not wearing seat belts much more than is made out. By rejecting their own assessment of their values we run the risk of evaluating the benefits of an activity by our own standards rather than by their standards. If people who are aware of the risks involved in not wearing seat belts still persist in not wearing them, it may be that they find wearing them a bigger nuisance and a greater hindrance to their enjoyment than we do, or because they attach less importance to the risk of harm relative to the satisfaction of their present wants. We admire those who climb dangerous mountains, or those who sail alone round the world, and [116/117] because of this we think that the benefits they obtain are worth the risks they run. Similarly, when people defy their doctor's “orders” and continue to work hard we understand and perhaps even approve of this if we appreciate the value of the work. But where we disapprove of an activity, or cannot appreciate it, we tend to think that the agent himself derives little benefit from it. In these ways the practice of strong paternalism easily becomes a cloak for the imposition of our values on those who are coerced. But if we do not allow paternalism to slide into the enforcement of our values on others, then we have no clear basis for distinguishing between activities we are prepared to permit and those we wish to suppress.
Moreover, the desire to seek danger is often the product of a certain type of personality and temperament that some people take pride in cultivating. To interfere with their risky conduct is to deny them the opportunity to pursue their own plans of life. Friends and relatives may sometimes interfere out of love and concern. But a person can break away from them, and even in the last analysis invoke the law to restrain them. But when the state itself interferes, then there is no escape.
The consistency of Mill's rejection of strong paternalism is called into question when he refuses to allow a person voluntarily to sell himself into slavery. Mill's argument is worth quoting:
By selling himself for a slave, he abdicates his liberty; he foregoes any future use of it beyond that single act. He therefore defeats, in his own case, the very purpose which is the justification of allowing him to dispose of himself.... The principle of freedom cannot require that he should be free not to be free. It is not freedom to be allowed to alienate his freedom. 
Mill's argument is unclear. He allows us to give up part of our freedom, as we constantly do when we enter into contracts. Nor is it the case that in any transaction that involves giving up some of our freedom, the freedom that we gain must be greater than the freedom that we lose. For freedom is not the only valuable end, and sometimes a net loss of freedom, [117/118] voluntarily contracted, is more than made up by a gain in, other benefits like income and material comforts. Mill has said nothing to indicate that he would prohibit such exchanges of freedom for other goods. Indeed in discussing the Mormon institution of polygamy, he explicitly argues against intervention even though the principle of liberty is violated: “far from being coutenanced by the principle of liberty, it is a direct, infraction of that principle, being a mere riveting of the chains, of one half of the community, and an emancipation of the other from reciprocity of obligation towards them.” (148) Intervention is not justified because the marriage is voluntary since many women “prefer being one of several wives to not being a wife at all”. So slavery is different, not because it involves a sacrifice of freedom, but because the freedom is given up completely and permanently.
We can imagine a situation in which a person has voluntarily sold himself into perpetual slavery. After a few years he greatly detests his condition and very much regrets having sold himself as a slave. His health suffers from the demanding work required of him, and he now refuses to obey the slaveholder any longer. The slave-holder invokes the law. Sanctions will be used to enforce the terms of the contract. The law thereby not merely permits, but also uses sanctions to support, the slave-holder's harming of the slave. Now this example seems to fall between cases of self-harming conduct and cases of conduct harmful to others. The example is not a clear case of the harm done by one person to another with his full consent. For in these cases when A harms B, B consents to the harm and does not withdraw his consent at the time the harm is inflicted. But perpetual slavery is also not a clear case of conduct which harms others in the sense which allow state to interfere. In the clear cases, one person harms another who has at no time consented to be harmed.
It is possible that Mill wishes to treat at least some cases in which one person harms another in conformity with earlier contractual agreement, but against the person's present wishes, as conduct causing harm to others against their will and therefore as conduct coming within the legitimate scope of legal intervention. If this is the case, then interfering in such contracts is not a paternalistic act but one designed [118/119] to prevent harm to others. Since the prevention of harm to others is not a sufficient condition for intervention one has to balance the advantages and disadvantages of interfering against those of not doing so. In the case of perpetual slavery, the harm done to a subsequently unwilling person is very grave, and there will be sufficient reason for not recognizing the slavery contract. But if there is a “slavery” contract, renewable at frequent intervals, and imposing limits to what may be required of the slave without his existing consent, this should be enforceable; for a discussion of the distinguishing features of slavery, see R. M. Hare, "Slavery". For although the individual would still be giving up his freedom, the contract will not be radically different from other freedom-limiting contracts, and at regular intervals he has the option of ceasing to be a slave. It is also important to note that the argument I have suggested on Mill's behalf does not prohibit a person from voluntarily becoming a lifelong slave of another person. But it supports the refusal to give legal recognition to contracts for perpetual slavery.
have suggested that selling oneself into slavery can be brought under Mill's principle which justifies the prevention of harm to others. There is another line of argument which stretches the category of conduct which harms others without their consent even further, and I wish now to explain and resist this argument.
D. H. Regan claims that some acts which are normally regarded as self-harming can be treated as acts which harm another person (Sect. iii). What we ordinarily regard as different time-segments of the same person can more appropriately be viewed as different persons. Thus the motor-cyclist at the time of his choosing to ride without a helmet, and the motor-cyclist who later meets with a serious accident, are two different persons. If there are two different persons, then the act of riding without a helmet is really an act which harms another person, the motor-cyclist's later self. So intervention in the act comes within the scope of Mill's liberty principle. This account draws on Derek Parfit's work on personal identity; cf. Parfit, "Identity," and Parfit, "Later selves;" see also Bernard William's comments, in Williams, "Persons." Parfit refers to the criterion of psychological connectedness for distinguishing between persons. This considers the degree of similarity in the psychological characteristic at different times. [119/120]
However, Regan seems to adopt a narrower criterion. He writes: “the motor-cyclist is a different person, in the relevant respect, if he is no longer the sort of person who would ignore his future wellbeing for the sake of small increments of present utility” (205). So if the motor-cyclist is likely, after being banned in an accident, to regret the past act of riding without a helmet and to change his attitude towards risk, then there is a different person who has been harmed. But since a person's attitude towards risk is only one of his psychological characteristics, it is likely that a change here is not accompanied by a change in other psychological characteristics. This leads to a situation in which for some purposes the same person is present at two points in time, but for other purposes there are two different persons. Regan accepts this. He gives the example of someone who, at about the same time that he embezzles, also commits an unrelated act of aggravated assault. Suppose that after ten years he has grown more honest and would no longer embezzle. However he has not changed in his attitude towards physical aggression. Regan thinks that it is inappropriate to punish him for embezzlement, but proper to punish him for assault. This situation can be handled easily enough by the resources we ordinarily use of saying that one and the same person is to be punished for assault but not for embezzlement. But if we adopt Regan's criterion of when we have the same person, we face the problem of determining the relation between the assailant and the person who is not punished for embezzlement. They are not the same person, but there is no way of punishing the assailant which does not also adversely affect the other person. If the assailant is imprisoned, the other person will also find himself in gaol. If the assailant is fined, then the resources of the other person are also reduced. All their desires, interests, and values are the same. Even Siamese twins are not so closely bound together.
Similarly, if we apply Regan's criterion of personal identity, we would have to say that the motor-cyclist, who rides without a helmet, is not the only person affected by punishment. An ordinary person has many projects, not all of which may fit into a single, over-all plan. For example, he has made promises which he intends to keep, and these promises are [120/121] unrelated to his desire to ride without a helmet. By punishing him for riding without a helmet, the state may deprive him of the opportunity to keep some of his promises. Punishing the motor-cyclist will therefore also interfere with the freedom of an innocent person, the promise-maker. This seems unfair. So even if we accept Regan's view that the motor-cyclist's act can harm another person, the same reasoning also provides us with a basis for not punishing him, and it is not clear how we are to balance the case for punishing the guilty motor-cyclist against the claims of the innocent promise-maker.
Punishing the motor-cyclist is of course a preventive measure taken to reduce the harm caused if there is an accident. So the real offence of the motor-cyclist is that of engaging in a dangerous activity which is likely to cause harm to another person, namely, his later self. It is of course perfectly proper to prevent someone from performing a dangerous act even though not all such acts will actually cause harm to others. For example, not every single act of driving at high speed through a busy street will cause harm. It is enough that every such act is likely to cause harm. However, in the ordinary case of dangerous driving, although actual harm to others may not be inflicted, there are actual persons who are likely to he harmed, and who therefore need protection. But in the case of the motor-cyclist, if there is no accident, then it is not just the case that he does not harm another person. There will also be no other person to be harmed. The accident both gives birth to and harms the later self. But still, there are other analogies to draw on. It is certainly not the case that the only type of harm to others is harm to actual persons who already exist at the time of the relevant act. Environmental pollution may not harm the existing population, but it will harm future generations. No doubt in this case the people who will be banned will come into existence independently of the harmful act, but it is unclear whether this is a significant difference.
But there is one important difference. Death is not a relevant harm in the case of the motor-cyclist. For if the accident causes death, then it does not produce a later self who is harmed. Death, at least instant death which leaves no time for regret, ensures that there is no birth of the later self, [121/122] and so death makes the accident self-harming but not harmful to another person. This point is recognized by Terry S. Kogan who adopts Regan's position. Kogan argues that the account of later selves cannot be used to justify state intervention in suicide.
In order for Mill's justification of state intervention to apply intertemporally, it is necessary that one make a reasonable prediction that future selves will disidentify with the present self. But where it is clear that there will be no future selves, such prediction becomes meaningless. Since there will be no future intertemporal “others”, suicide is a purely self-regarding action. [841-42]
But the same point applies to all risky actions which cause death, or severe brain damage in which the person, while still alive, is unconscious or is incapable of formulating new values and plans of life, and is therefore incapable of renouncing his old values and attitudes. From such a brain-damaged being no later self can emerge to be harmed. It is paradoxical that activities which are likely to cause these types of serious harm fall outside the scope of Regan's and Kogan's new defence of paternalism, whereas activities which are likely to cause lesser harm can be interfered with.
Kogan also claims that there is evidence to support the view that a person will reject his earlier goals as a result of the traumatic experience of a serious accident (837). If this is true, then it presumably applies even when a helmet or seat belt was worn. The accident may, for example, be caused by driving at high speed, or by the failure to pay sufficient attention, and the person now regrets and rejects his previous attitudes. If it is the trauma of the serious accident which causes the change in him, sufficient for a later self to emerge, then this change can come about even when he had taken the precaution of wearing a helmet or seat belt. Now those who support the compulsory wearing of helmets and seat belts do not of course claim that the rate of accidents thereby reduced. What is rightly claimed is that when there are accidents, the wearing of helmets and seat belts reduce the extent of harm caused. No doubt death and severe brain damage will be among the serious harm that will be reduce. But we recall that on Regan's and Kogan's account, if there is death or severe brain damage, then there is no harm to[122/123] others as there are no later selves. This being the case, the probability that the wearing of seat belts and helmets reduces the incidence of these types of harm constitute a reason against requiring them to be worn. For when the harm caused is death or severe brain damage, the victim is the present self and not a later self. Accidents which result in these types of harm do not therefore cause harm to others, but are purely self-harming. If the wearing of seat belts and helmets reduces the number of deaths and cases of severe brain damage, then it increases the harm caused to others by reducing the number of self-harming actions. This is a surprising but inescapable implication of an appeal to the idea of harm to later selves.
Last modified 1 May 2015