Paternalism is defined in the following way, X behaves paternalistically towards Y: (1) only if X aims to close an option that would otherwise be open to Y, or X chooses for Y in the event that Y is unable to choose for himself: and (2) to the extent that X does so under to promote Y’s good. Paternalism covers both acts and omissions and may be successful or unsuccessful, and is distinguished from benevolence. This definition is contrasted with some alternatives that have been offered by others. On this account, paternalism need not be coercive, nor restrict liberty of action, though there is one sense in which it does restrict freedom. Paternalism need not be against the will, nor need it limit the autonomy of the paternalised, nor involve the violation of any moral rule. Finally, a paternalist may believe both that paternalistic intervention is for good of the target and that the target shares this view.
In this article, I compare and defend a particular definition against other definitions by testing general definitions against particular cases that I assume either clearly are or clearly are not cases of paternalism. Competing definitions are found to be either too wide or too narrow. Such a method is controversial. Some philosophers hold that political concepts are best understood only after normative argument rather than by linguistic or conceptual analysis. For example, some argue that disagreements over the meaning of freedom cannot be resolved ‘until we know why freedom matters.’ I do not wish to disagree with that view, so my only defense of the method I use is to say that I believe at least some progress can be made by conceptual analysis alone. The definition I shall offer may be only a tentative one, subject to revision once all the argument for and against paternalism are sorted out. It is helpful to have at least some idea of what paternalism is in order to address the arguments surrounding it.
A simple definition of paternalism is that it is the restriction of a person’s freedom, usually against his or her will, for his or her own good. This definition captures many cases that are often thought to be paternalistic, such as motorcycle helmet and car seatbelt legislation, preventing a person from unknowingly walking across a dangerous bridge, a doctor refusing to inform his patient of her terminal illness, requiring people to undergo blood transfusion when needed despite their religious objections, and laws against drug use, certain sexual activities, and gambling. The definition also explains the motivation for paternalism – to benefit the paternalised – and suggests why it may be morally troubling – because it involves a restriction of freedom, usually against the paternalised’s will.
The simple definition, however, does not capture all that we mean by paternalism. A more precise definition is the following:
X behaves paternalistically towards Y only if X aims to close an option that would otherwise be open to Y or X chooses for Y in the event that Y is unable to choose for himself; and to the extent that X does so in other to promote Y’s good.
According to this definition, only behavior that aims to close an option or that aims to choose for a person can be paternalistic, and such behavior is paternalistic insofar as it is for the target’s own good.
Paternalism is therefore concerned with the good of the person subjected to the intervention. Restricting a person’s freedom solely to prevent him from harming others, or society in general, or to promote some non-person-affecting good (if there is such a thing) is not paternalistic. Many acts, however, may involve multiple rationales. Someone may intervene with my action both to protect others and prevent me from harming myself. The intervention is paternalistic with respect to the latter motivation. That is the meaning of the phrase ‘to the extent that …’ in the second definition above. Moreover, paternalistic motivation may be a major or a minor reason for acting. A wife who tries to stop her husband from drinking primarily because she believes abstention is for his own good may do so also because she is aware that he becomes violent towards her when drunk. A person’s main reason for interfering with another may be to prevent harm to others, but she may also do it to help the person being interfered with. Both these cases involve paternalism to a greater or lesser extent.
In the above definition of paternalism, X and Y may range over groups or individuals. X may or may not be the state since paternalism is not limited to laws and public policy. A friend acts paternalistically when she hides my cigarettes from me, a firm acts paternalistically when it (unforced by the state) makes its employees put part of their earnings into a pension plan, and parents often act paternalistically towards their children. If (and only if) the intervener is the state, then the intervention is a case of state paternalism, and if (and only if) the state uses the law to intervene, then it is legal paternalism
Paternalism is characterized above as a form of behavior. The meaning of this is that it includes both acts and omissions. X withholding from Y that Y’s spouse has died to prevent Y from experiencing distress when about to undergo surgery is rightfully described as an act of paternalism. And so is a doctor’s withholding information from a patient for what he sees as the patient’s own good. Thus an accurate definition of paternalism covers omissions as well as acts, so long as the omissions are intended to benefit the person.
The paternalist aims to close an option of the paternalised, but may fail to do so. Some people may obtain recreational drugs despite legal prohibitions on those drugs. Such prohibitions are nevertheless paternalistic insofar as they are intended for the good even of those who dodge them. Another way that paternalism may fail is that the paternalist may aim but fail to bring about the good of his target. Laws against prostitution may be intended to benefit those who would otherwise be prostitutes. But they may in fact end up worse off under those laws. Such legislation would be unsuccessful paternalism but paternalism nevertheless.
Since paternalism involves attempting to benefit another person, it is a type of benevolence. But not all benevolent behavior is paternalistic. Giving food to a hungry person is not normally paternalistic. The above definition explains why, since such an act does not close an option. The hungry person may always give away or abandon the food. No opportunity is closed that would otherwise be open to the person. We can imagine circumstances where an option is closed. A person may not only give food to a hungry man, but also forces him to eat it. Such an action seems paternalistic, and my definition explains why.
There are competing accounts of what makes benevolent behavior paternalistic. Some scholars define paternalism as ‘interference with a person’s liberty of action’ for the good of ‘the person being coerced’. However as several commentators have pointed out, paternalism need not be coercive nor interfere with liberty of action. Consider a case where Peter tells a mother on her deathbed that her son is doing well when Peter knows the son has just been killed attempting to escape from prison after being found guilty of multiple rape and murder. Peter is acting paternalistically towards the mother, but no coercion is involved, he is not restricting her liberty of action, and is not even attempting to control her behavior. The case is one of paternalism on my definition. The mother cannot choose what to hear. She is at the mercy of what Peter chooses to tell her. Peter chooses for her, choosing to lie, for what he sees as her own good.
Another case of paternalism that does not restrict liberty of action or involve coercion is giving a blood transfusion to an injured, unconscious person whom we know to be opposed to blood transfusions. Such a case is accounted for and is what motivates the second part of the clause of my definition. It is not covered by the first part. The options of the patient are already closed in virtue of the fact that she is unconscious. Her unconsciousness prevents her from having the options to choose whether or not to have a transfusion. Hence the paternalist cannot close the patient’s options in this respect. But he can choose for the patient since she is unable to choose for herself. In doing so he would be acting paternalistically. (It would also be paternalism if the unconscious patient is not opposed to transfusion, although not as morally interesting a case as when she is.)
Some scholars argue that in light of examples of paternalism that do not restrict liberty of action, paternalism should be thought of as covering the restriction of any freedom, including but not limited to restriction of liberty of action. Paternalism, may abrogate other freedoms such as the freedom of thought and expression, and the freedom to be left alone. They suggest that not all examples of paternalism restrict freedom. For example, my refusal to play tennis with someone because that person gets upset when he loses is a case of paternalism that does not restrict freedom in any way. The plausibility of a definition that makes restriction of freedom necessary for paternalism depends upon what conception of freedom to consist in the absence of impositions on a person. This leads to the proposition of the following definition of paternalism:
X acts paternalistically in regard to Y to the extent that X, in order to secure Y’s good, as an end, imposes upon Y.
It is clear from the above definition that an imposition is taken to be the same as restriction of freedom. This is a mistake. If someone asks me the time of the day, he is imposing (his will) upon me but not restricting my freedom in any way. It seems to make the mistake of thinking that since all restriction of freedom are impositions, all impositions must be restrictions of freedom. This has the result of making the definition of paternalism too broad. If I remind a friend of the time of day so that she is not late for an appointment, this seems hardly an instance of paternalism, yet it fits the above definition since my action would be an imposition.
It also considers two other conceptions of freedom. One is the view that freedom consists in the absence of constraints on doing what one desires. This, however, cannot be the sense in which freedom is restricted by paternalism since as will be suggested below, some cases of paternalism do not conflict with the target’s desire. The second conception holds that freedom is ‘the non-restriction of options’. This I suggest, is the sense of freedom in which paternalism almost always restrict freedom. The resulting definition of paternalism is almost equivalent to my own. But my definition avoids taking a stand on the controversial question of what freedom is. Even if freedom is not, or not only, the non-restriction of options, what paternalism involves is the restriction of options. This is often but not always the case, since as was seen in the blood transfusion and mother-on-her-death-bed cases above, paternalism sometimes does not restrict options. In those cases, the subject is unable to choose for him or herself and the paternalist makes the decision.
There are several other ways that some authors have attempted to define paternalism so that it embraces more than restrictions on liberty of action but without making it so wide as to cover all benevolent behavior. One is to hold that paternalistic acts or omissions are those that either are, or are believed by the paternalist to be, contrary to the target’s will. One author holds some such view, since part of his definition of paternalism is that X believes that his behavior is contrary to Y’s ‘operative preference, intention, or disposition at the time’ X performs the action or omission. Another author makes it a necessary condition of paternalism that the intervention be ‘carried out either against [the paternalised’s] present will … or against his prior commitment’.
Both definitions are however too narrow. Laws requiring and passengers to wear seatbelt while in motor vehicles may not be contrary to the operatives preferences of many of those drivers and passengers. Similarly, laws requiring the wearing of helmets while riding on motorcycles may not be against the will of some of those affected. But requiring people to wear seatbelts and helmets are for their own good are, I suggest instances of paternalism even if many of those people will voluntarily choose to do so anyway. What is important is not that the act or intervention is contrary to the person’s operative preferences but that the option is closed, whether the paternalised wants it to be closed or not. Whether they want to or not, they are made to do something for their own good. Laws against prostitution and pornography intended for the good of those whose freedom is restricted are paternalistic even towards those who have no desire to engage in the prohibited activities. The mistake these authors make, I suggest, is to concentrate on only the more interesting or morally problematic cases of paternalism. Whilst instances of paternalism that are not against the target’s will may strike us as less interesting or morally troubling than others, this should not lead us to believe mistakenly that they are not paternalism at all.
Similar remarks apply to another proposed definition which claims that ‘an essential feature of paternalistic behavior towards a person is the violation of moral rules (or doing that which will require such violations)’. Building into this feature into the definition of paternalistic behavior does not mean that they think it always unjustified. Someone could have an adequate justification for violating a moral rule. But this account is inaccurate since there are cases of paternalism that do not involve the violation of any moral rule. For example, a husband who hides his sleeping pills from his suicidal wife violates no moral rule. They are his pills and he may put them wherever he wishes. Again, the more interesting cases of paternalism may be those that violate a moral rule. But this is not a reason to make it part of the definition of paternalism and exclude as cases of paternalism those which do not violate moral rules.
My proposed definition also contrasts with another recently put forward definition which limits of paternalism to cases of coercion and interference with liberty of action only. A more recent proposal is that as act is paternalistic only if it is a limitation on Y’s autonomy. By this it means that there must be a usurpation of decision-making, either by preventing people from doing what they have decided or by interfering with the way in which they arrive at their decisions.
As an aside, I think it is a mistake to hold that all usurpation limit autonomy. According to one view, autonomy requires general freedom from coercion and manipulation, an adequate range of worthwhile and morally acceptable options, and identification with and loyalty to one’s choices. A particular usurpation of decision-making that is neither coercive nor manipulative, therefore, may not violate autonomy. But it may be paternalistic, which suggests that not all paternalism violates autonomy.
Putting this aside, the definition is still too narrow because paternalism does not always usurp decision-making. A blood transfusion to an unconscious patient does not prevent that patient doing anything he’s decided (since he has not decided anything, being unconscious), nor interfere with the way he makes his decisions. As I suggested above, laws against prostitution are paternalistic even towards those who do not or would not choose to prostitute themselves, nor need those law interfere with the way in which people make decisions. In other cases, the paternalised may have no capacity for decision-making for the paternalist to usurp; this could partly be what motivates the paternalist.
Moreover, an author argues by restricting paternalism to behavior that limits autonomy as a non sequitur. He considers the case of a father (a lawyer) who wants his daughter to be a lawyer. The daughter believes she would be such a good lawyer that she would be more successful than her father and that her success would make her father unhappy. She therefore decides not to become a lawyer. Other parts of the definition – that it be contrary to the person’s will and that it be for his own good – are satisfied in this case. He claims that since it is not a case of paternalism an additional part is needed in the definition. He argues that it is not an instance of paternalism because the daughter does nothing to interfere with the self-determination of the father. From this he concludes that paternalism involves a violation of a person’s autonomy. However, his discussion could simply support something like part 1 of my definition (that the paternalist closes the option of, or chooses for, the paternalised) without the need to claim that this involves a limitation on autonomy. The daughter does not close any option that was previously open to the father. It is not his option what his daughter does, but hers (although he has a preference for how she should choose amongst her options). Since my definition can account for excluding the case as one of paternalism, the condition is not necessary. And since the definition seems too narrow – some paternalism does not usurp decision-making – then there is reason to prefer my definition to the author’s.
To reiterate, I define paternalism in the following way:
X behaves paternalistically toward Y: (1) only if X aims to close and option that would otherwise be open to Y or X chooses for Y in the event that Y is unable to choose for himself; and (2) to the extent that X does so in order to promote Y’s good.
The above definition is morally neutral in the sense that it allows the possibility of both justified (whether simply permissible or also required) and unjustified paternalism. The definition allows us to point to examples of paternalism without already judging their moral permissibility. Descriptive considerations are separated from evaluative ones. But the definition also suggests why is often thought to be wrong. There are various methods by which a person may close another’s options that may be at least presumptively wrong. These could involve the use of force, coercive threat, manipulation, physically impairing, causing pain, promise-breaking and so on. Paternalism is also sometimes thought to be wrong (either presumptively or absolutely) in itself, regardless of whatever means are used by the paternalist. Various suggestions have been made regarding paternalism’s permissibility. Another suggestion is that paternalism is acceptable only if the target has consented to the intervention or would do so if she were aware of all the relevant circumstances. A third possibility is that paternalistic intervention is permissible only when the conduct involved is substantially non-voluntary or when temporary intervention is needed to establish whether it is voluntary or not. My concern in this article , however, has not been with the evaluation of paternalism. Rather, I have merely tried to describe it.
