The aim of this article is to provide an analysis of the paternalist act and to show how some paternalist acts may be morally justified. Such an analysis should be informative of the nature of the paternalist act and provide the right results with regard to putative cases of paternalist acts. The motivation for providing an analysis of the paternalist act is the same motivation that informs providing analyses of knowledge, causation, the morally good act and so forth; that aside from being of interest in themselves, such analyses are theoretically fruitful in allowing us accurately to deploy these notions across various discourses. I also hope to show that there are paternalist acts that are relatively commonplace and morally permissible. My aim in this article is not to provide a comprehensive review of attempted analyses of paternalism, though I do scaffold my account by way of criticism of Dworkin's analysis and briefly consider other accounts so as to locate my own within the discourse.Footnote ¹

The target of analysis is what is sometimes referred to as hard paternalism, as opposed to soft paternalism. As the name suggests, paternalism generally, that is both the hard and soft variety, is modelled on the father–child relationship. Let's first consider Gerald Dworkin's analysis.Footnote ²

S acts paternalistically towards R by doing (omitting) Z:

1. (C1) Z (or its omission) interferes with the liberty or autonomy of R.

2. (C2) S does so without the consent of R.

3. (C3) S does so just because Z will improve the welfare of R (where this includes preventing his welfare from diminishing), or in some way promote the interests, values, or good of R.Footnote ³

I take it that the basic import of the first two conditions can be grasped without much difficulty; in any case I discuss them in subsequent paragraphs. The third condition, however, warrants immediate attention. The obvious reading of C3, to me at least, is that S's action will, amongst other things, improve the welfare of R.Footnote ⁴ But surely an action being paternalistic does not entail that it will improve R's welfare, never mind S's action being explained by the fact that that action will improve R's welfare. Rather than success, what seems to be required is that S's act be appropriately motivated. Perhaps this is indeed what Dworkin has in mind.Footnote ⁵ If that is so, then my point may be taken to be merely clarificatory in nature.

I’ve written that what is required for an act to be paternalistic is that it be appropriately motivated, but what would make an act appropriately motivated? One option is to say that S acts paternalistically if S does Z just because S is motivated by a belief that doing Z will improve R's welfare.

Consider a case in which a father believes that if his adult son takes vitamin supplement tablets, then his son's health and ultimately welfare will be improved. Knowing that his son hates taking tablets and so is likely to refuse to take the tablets if asked to do so, the father hides the tiny tablets in his son's evening meals and by so doing interferes with the autonomy of his son. Even if it turns out that taking the tablets does not improve the welfare of the son, intuitively the father has acted in a paternalistic way.

However, requiring that S be motivated by a belief, as proposed, is overly demanding. It seems conceivable that someone could act paternalistically and be motivated on the basis of something weaker than belief. Suppose S does Z just because S is motivated by a positive epistemic standing that doing Z may improve R's welfare.Footnote ⁶ Acting on such a basis intuitively doesn't seem to discount the possibility of the act being paternalistic. Epistemic standings that fall short of belief that may also be able to play the same role include acceptance and suspicion. To return to our example, if the father suspects that there is even a small chance that his son taking the vitamin tablet will improve his welfare, and acts on this basis, then the act is paternalistic.

Now let us turn our attention to the first two conditions. A broader rendering of C2 could capture the sense in which paternalism is a threat to liberty or autonomy, arguably making C1 unnecessary. My provisional suggestion is for a condition stating that S does Z irrespective of the consent of R. It seems to me that there is intuitive support for the idea that S could be acting paternalistically even if R did offer consent, so long as S would have acted in the same way regardless of whether R consented or not.

We want our analysis of a paternalist act to allow for the possibility of the happy, let's say Victorian, paternalist, whose husbandry isn't met by resistance from wife and children (and perhaps even his labour force if he has one and is a social reformer of the type typical of his age), but rather is taken by them, being of their time, as legitimate. Even if a charge offers consent to his action on her behalf, it seems intuitive to say that such action could remain paternalist. What makes someone a paternalist plausibly isn't necessarily that he acts in the absence of consent, but rather that the presence of consent or otherwise isn't a factor in his doing what he takes to be for the good of the person on whose behalf he is acting. In other words, someone doesn't stop being a paternalist when their decisions are met with consent, a person stops being a paternalist when their decisions are conditional on there being consent.Footnote ⁷

It may be objected that the revised condition which I have suggested relies on an artificial conception of consent. Saying that R consents to Z suggests that Z was optional for R: it makes no sense to talk of R consenting to Z if Z has already been decided, just as it makes no sense to talk about a child consenting to a particular holiday destination when it has already been decided by his parents, even if the parents would like it if the child was enthusiastic about their choice. This is not a problem for Dworkin, whose condition is simply that there needs to be an absence of consent.

We might think, however, that that one cannot consent to acts of a certain kind, ones in which in fact there isn't an option, is a quirk of the meaning of ‘consent’ and is not something we were trying to capture in our analysis of paternalism. Really the intuition we were trying to capture when we discussed consent and imagined the happy paternalist was that an act could coincide with the wishes of someone and remain paternalistic. These considerations yield the following condition: (C1*) S does so irrespective of what S believes the wishes of R may be.Footnote ⁸

The removal of Dworkin's C1 means that there is no longer any requirement that a paternalist action interferes with liberty or autonomy; what is retained on the alternative analysis, however, implied by C2*, is that paternalist action threatens both liberty and autonomy. To say that an act may be one which does not interfere with liberty or autonomy but still may be a paternalist act does not conflict with the idea that the cases of paternalistic action that have traditionally been of interest are the ones that do in fact interfere with liberty or autonomy.

Nor does it seem inconceivable that an action may be paternalistic without interfering in the liberty or autonomy of the object of the paternalist action. Suppose that a mother is worried about the safety of her son in his new apartment. It's his first time living away from home and his mother knows that there is no smoke alarm in his apartment. She thinks that if she were to suggest that he get one, then he would agree, but knowing him as she does, she doesn't believe that he would actually get one. She knows that he is very proud of his new-found independence, and she thinks that if she were to ask him whether he would like her to buy him one he would say no. She decides to buy him one anyway and, by offering it to him already bought, tries to make his acceptance of it a fait accompli. Does the mother's action interfere with the son's liberty or autonomy? It certainly doesn't interfere with his liberty. His mother has to resort to such tactics because her powers over him are limited. He can simply refuse her gift. What about his autonomy? While his mother's actions certainly don't respect his right to decide things for himself, it's not obvious that they reduce or even hinder his autonomy. Seana Shiffrin offers a similar case in which an agent refuses to help another agent with a task on the grounds that not helping will be good for that other agent; Shiffrin writes that perhaps the other agent is failing to learn skills that he needs.Footnote ⁹ She also draws the conclusion that the violation of autonomy is not a necessary condition for paternalism.

My alternative analysis, which I believe I have shown to be superior to Dworkin’s, runs as follows,

S acts paternalistically towards R by doing (omitting) Z (iff):

1. (C1*) S does so irrespective of what S believes the wishes of R may be.

2. (C2*) S does so just because S has a positive epistemic standing that Z may or will improve the welfare of R (where this includes preventing his welfare from diminishing), or in some way promote the interests, values or good of R.

A consequence of the removal of C1 from the analysis of the paternalist act is that satisfying the requirements of the paternalist act is less demanding than if C1 were not removed. It is therefore likely that a consequence of the analysis is that paternalist acts are more common than would be the case were C1 retained. In fact the removal of C1 from the analysis of the paternalist act results in relatively commonplace acts counting as paternalist acts – thus broadening the category of the paternalist act.Footnote ¹⁰

Why think that the removal of C1 results in relatively commonplace acts counting as paternalist acts? The following acts are relatively commonplace acts, even if they satisfy the requirements of a paternalist act: (a) not buying alcohol for someone who looks like they’ve already had too much to drink; (b) not accepting an expensive gift from a friend who wants to give you that gift but who you know can ill-afford it; (c) not sleeping with someone who wants to sleep with you because one suspects they could easily get hurt as a result of the action; (d) telling the truth to someone one knows would prefer a lie; (e.) reporting a person to the police in the belief that it will be better for that person if they’re stopped now before they’re caught doing something worse; (f) providing support for a friend who has asked not to be helped; (g) insisting that a former partner stay out of one's life because one believes that would be better for the former partner; (h) not allowing someone to run the risk of catching a serious contagious illness from you even though that person wants to be close to you and is prepared to run the risk of catching the illness; (i) prepaying for a doctor's appoint for one's elderly and ill parent because it seems the only way to get that parent to agree to see a doctor – she doesn't like the idea of money being wasted. Some of these acts are more commonplace than others but I take it we’re familiar with such acts, either directly from our personal lives or from our culture – from stories in books, films and so on.Footnote ¹¹

These acts, like the act of not helping described by Shiffrin and the mother getting the smoke alarm for her son, are plausibly paternalist acts that don't interfere with the autonomy of the other agent. Interfering with another's autonomy is surely constituted by an act, including an act of omission; a mere belief that one can improve that other's welfare will not constitute interference in that person's autonomy. Similarly, in none of the cases described is the target's agency manipulated or violated. Nor is the paternalist actor acting in a domain in which the object of the paternalist act's agency is the sole legitimate agency; other agents are involved in each of the cases described. The acts do bear on the autonomy of the other agent: they affect the choices he's faced with and those acts may be acts that run contrary to his will. Still, the legitimacy of the paternalist actor exercising agency in these domains means that the acts don't constitute interference in the autonomy of the other agent.

Given that an implication of the argued for analysis of the paternalist act is a broadening of the category of the paternalist act as the result of the omission of C1, and the denial that these acts interfere with the autonomy of the object of the paternalist act, an attendant question is whether consideration of some of the acts that now qualify as paternalist acts gives us reason to revise the view of paternalist acts as a class as morally wrong acts. Relatedly, what bears on the rightness or wrongness of paternalist acts that don't involve interference in the autonomy of an agent? I take it that some or perhaps all of the paternalist acts just described don't seem morally wrong to most readers.Footnote ¹²

There is, however, an objection to the view that these acts are morally permissible which rests on insisting that these acts do interfere with the object of the paternalist act's autonomy.Footnote ¹³ In all of these cases the act is being carried out by an agent who is acting irrespective of the wishes of the other agent. What's more, the acting agent does so because she has a positive epistemic standing that the act will or may improve the welfare of the other agent. The thought is that it's not the place of the actor to perform that act irrespective of the wishes of the other agent, even if the agent knows that the act will improve the welfare of the other agent.Footnote ¹⁴ So a former partner might insist that even if you’re right that further contact will be bad for her, it's not your concern given her expressed wish to stay in contact. In other words, in matters of the good of an agent, one should defer to the expressed wishes of that agent, unless perhaps that agent is the victim of certain psychological illnesses or incapacities.Footnote ¹⁵ To act in a way that does otherwise is to interfere with the autonomy of that agent. There is some intuitive appeal to such a response.

We know that people have blind spots on certain matters, that they may lack experience and foresight. It would seem perverse then to be satisfied with an answer along the lines of ‘if it's a mistake, it's my mistake to make’ from a former partner when one, say, sees that staying in contact with that partner will be bad for them, or perhaps one sees that it's already bad for them, and that it's one's own action in contacting them that is bad for them.Footnote ¹⁶ My intuition is that the various acts listed are morally permissible. This will probably be unsatisfying to those who don't share the intuition, but for those who do share the intuition and those interested in the view that these acts are morally permissible, identifying the factors that bear on the permissibility of those acts should be of interest.

The particular epistemic standing plausibly has a bearing on our moral evaluations of the described actions. Prima facie, an agent who knows that her action will improve the welfare of the other is in a stronger moral position to act irrespective of the wishes of an agent in comparison to an agent who has some lesser positive epistemic standing that the action will (or even only may) improve the welfare of the other agent also acting irrespective of the wishes of the agent.

It's not just the strength of our epistemic standing that bears on our licence to undertake a paternalist act; also crucial for whether a paternalist act is permissible is the relationship between the actor and the object of the act. While it's one thing for one to break contact with one's former partner irrespective of her wishes motivated by a positive epistemic standing that it will or may improve her welfare, it's another thing again for an outside party to force a break in contact.

The cases described regard interpersonal relationships; they are relationships which we accept will be involved and sometimes messy. There are undoubtedly boundaries beyond which an act will be morally impermissible, though these boundaries will differ from one kind of relationship to another. Our intuitions of moral permissibility or impermissibility in response to the described cases plausibly will in part depend on how we regard the appropriate nature of such relationships. So if one regards the appropriate nature of the mother–adult son relationship as being one in which the mother ‘butts out’ of his life and allows him to make decisions for himself, then presumably one will see her act as morally impermissible.Footnote ¹⁷ Alternatively one could regard the appropriate mother–adult son relationship as a loving, involved relationship in which the mother is concerned for the good of her son, where this includes both his safety and him being able to make decisions for himself – though in some circumstances a concern for his safety might outweigh a concern for him being able to make decisions for himself. Notice that the paternalist needn't be dismissive of the autonomy of the other person. She may recognize that it forms a significant element of the other person's good without holding that it trumps all else in matters pertaining to the good of that person.

Given that the nature of the relevant relationship has a bearing on the permissibility of the paternalist act, I don't take what has been shown here to imply that paternalist acts on the part of the state are morally permissible. We might still think that in societies in which there are diverse conceptions of the good, there are good pragmatic grounds for the state to adhere to neutrality on the good in the absence of an overlapping consensus. The pragmatic grounds might be that state paternalist acts informed by a particular conception of the good could well be alien and unwelcome for agents who do not share the same conception of the good. Those agents would have reason to oppose the order of the state, possibly fostering political instability and conflict.

The degree to which the person's welfare may be affected by the paternalist act also has a bearing on moral evaluations. That we think that the acts described are morally permissible is plausible in part because we understand those acts as potentially making a significant difference to the welfare of the other agent. If the contagious illness to which we would expose the other were a weak cold then it would be less plausible that we should act irrespective of the wishes of the other person than if say the contagious illness to which we would expose the other were Ebola.

What has been said suggests that in such relatively commonplace cases there is a moral cost to acting irrespective of the wishes of the other agent but that the cost is not such that it may not be outweighed by other considerations. Furthermore, our licence to act in such cases varies according to the strength of our epistemic standing regarding the benefit to the other person and the nature of our relationship with that person.Footnote ¹⁸

In what follows I’ll attempt to locate my analysis of the paternalist act in the literature. Given my stated aim I will, rather than repeating the considerations for arriving at the analysis provided above, pay special attention to views that are close to my own view. Daniel Groll helpfully lists a number of prominent accounts of the paternalist act in the literature which are committed to implying a ‘contrary to will criterion’, a criterion to which my own account is not committed.Footnote ¹⁹ Those accounts are as follows:Footnote ²⁰

• • ‘Paternalistic policies are restrictions on a person's liberty which are justified exclusively by consideration for that person's own good or welfare, and which are carried out either against his present will . . . or against his prior commitment.’Footnote ²¹

• • ‘[Paternalism] . . . consists in treating adults as if they were children, or older children as if they were younger children, by forcing them to act or forbear in certain ways . . . “for their own good,” whatever their wishes in the matter.’Footnote ²² As Groll notes, according to Joel Feinberg this defines just one species of paternalism, ‘presumptively blamable benevolent paternalism’.Footnote ²³

• • ‘P behaves paternalistically towards Q [if and only if: . . .] P discounts Q's belief that P's behavior does not promote Q's good.’Footnote ²⁴

• • ‘Paternalism is the intentional overriding of one person's autonomous choices or actions by another person, where the person who overrides justifies the action by appeal to the goal of benefitting or of preventing or mitigating harm to the person whose choices or actions are overridden.’Footnote ²⁵

I don't think it's obvious that Feinberg is committed to the ‘contrary to will criterion’. Perhaps the thought is that if something is ‘forcing’ someone in some way then that something is necessarily going against that someone's will. But this is clearly mistaken. My desire to get the right answer in the complicated maths sum forces me to stop thinking about what I’ll do at the weekend. Similarly, my asking whether the seat is free puts you in a situation in which you’re forced to respond in one way or another, but my asking needn't have been against your will; you might not have even had a will in the matter, or you might have even wanted me to ask. A couple of pages after the passage quoted by Groll from Feinberg, Feinberg claims, and he is considering presumptively blameable benevolent paternalism, that Bernard Gert and Charles Culver have ‘conclusively shown’ that a paternalist act ‘need not be coercive or interfere with anyone's liberty of action’.Footnote ²⁶ In support of this claim he mentions cases of doctors ‘prescribing placebos for anxious but healthy patients’. In such cases we don't have grounds to think, and Feinberg doesn't seem to think it either, that the doctor acts against the will of the patients. Again the object of the paternalist act may not even have a will one way or another on the use of placebos. Alternatively, she may just want to feel better – even if that would mean a doctor giving her a placebo without her knowing it.

Accepting that on Feinberg's account there isn't a ‘contrary to will criterion’, we’re faced with the question of why we should favour the view argued for in this article over Feinberg's view. First, it's not obvious that the two views are in direct competition. Feinberg isn't trying to provide a precise analysis of the paternalist act. If one were to take Feinberg's view, considered here, as an attempt to do so, then fault could be found with the ambiguity of the meaning of ‘treating adults as if they were children, or older children as if they were younger children’ and ‘by forcing them to act or forbear in certain ways’.Footnote ²⁷ For example, what does it mean to treat someone as a child? This is hard to know given that there tend to be differences between how children are treated across times, places and families. How is ‘forcing’ and ‘forbearing in certain ways’ to be spelt out? In contrast, the extension of the two conditions, argued to be necessary and jointly sufficient for the paternalist act, laid out in this article is significantly clearer.

A view that is close to my own comes from Shiffrin.Footnote ²⁸ The conditions she lays out for the paternalist act are as follows:

• [P]aternalism by A toward B may be characterized as behaviour (whether through action or through omission)

1. (a) aimed to have (or to avoid) an effect on B or her sphere of legitimate agency;

2. (b) that involves the substitution of A's judgement or agency for B’s;

3. (c) directed at B's own interests or matters that legitimately lie within B's control;

4. (d) undertaken on the grounds that compared to B's judgement or agency with respect to those interests or other matters, A regards her judgement or agency to be (or as likely to be), in some respect, superior to B’s.

The view is close in that it is consistent with C1* (S does so (acts) irrespective of what S believes the wishes of R may be). Why favour my analysis of the paternalist act over this view? There are a number of reasons. The analysis provided in this article is more elegant; the paternalist act is pinned down with two simple conditions. It's more comprehensive; the analysis put forward provides the necessary and jointly sufficient conditions for knowledge, while Shiffrin's account only says how paternalism of one agent towards another may be characterized.

An area of contrast to which I want to draw the reader's attention, however, is that on Shriffin's view behaviour may be paternalist that is directed at the object's ‘interests or matters that legitimately lie within’ her control in cases in which that behaviour is undertaken on the basis that the actor (or A) ‘regards her judgement or agency to be (or as likely to be), in some respect, superior to B’s’ with regard to those interests or matters. In contrast, on the view advanced in this article, C2*, it is a requirement for an act to be paternalist that the paternalist agent be motivated by the positive epistemic standing that Z may or will improve the welfare of R. In other words, on Shiffrin's view there is no need for a paternalist act to be motivated by the welfare (or interests) of the object of the paternalist act, though such an act may be directed at matters that legitimately lie within the object's area of control. But this seems to yield intuitively very odd results. Consider the following case:

Mindy has clearly acted in an inappropriate way, but intuitively her action hasn't been paternalistic. This seems to be precisely the case because she's not acting out of any concern for Arthur's welfare. This is a straightforward and intuitive counterexample to Shiffrin's account.

Finally let me compare the view advanced in this article with Groll's recently articulated and similar position.Footnote ²⁹ Groll holds that there is a ‘decisive reason criterion’ requirement on the paternalist act. More specifically he holds that ‘A acts paternalistically toward B when A acts for the sake of B's good but does not treat B's will as structurally decisive in determining what she, A, should do.’ By ‘structurally decisive’ Groll's thought, drawing on work from Joseph Raz and Stephen Darwall, is that the will may be treated as authoritative.Footnote ³⁰ More specifically, the fact of someone having a will may be treated as having the power to silence or pre-empt other considerations in certain matters. The thought seems to be that the will of a competent person is such that in matters pertaining to her or his own affairs, that will should be treated as structurally decisive. Furthermore to act in such a way as not to treat it as structurally decisive may lead to an act being paternalistic. Therefore, if there happened to be a coincidence of an act by A to Z and a will on the part of B that A Z, then, as with my account, the result yielded will be that the act may be paternalistic.

The account I’m offering makes clear the motivational component required for an act to be paternalistic; that is, on my analysis A must have a positive epistemic standing that Z may or will improve the welfare of B. Groll acknowledges that the condition he provides may be given an alternative ‘internal’ form such that it's rendered in terms of ‘what A takes to be B's good’.Footnote ³¹ This misses the point, explained earlier, that the motivational component may be such that the paternalist actor may not even believe that her action will improve the welfare (or good) of the target of her action. She may just suspect that it might. Furthermore, the motivational component which I make clear, as previously discussed, has a bearing on the moral permissibility of paternalist acts that don't interfere with the autonomy of the objects of the paternalist acts. The inclusion of this condition, aside from its independent plausibility, is significant in its relevance to normative questions regarding the paternalist act.

In any case, Groll writes that considerations regarding the motivational component are orthogonal to his central concern. Given the concern of this article, however, to offer an analysis of the paternalist act, we have reason to favour the conditions over the decisive reason criterion that Groll offers.Footnote ³²