Show Summary Details

Page of

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, POLITICS ( (c) Oxford University Press USA, 2016. All Rights Reserved. Personal use only; commercial use is strictly prohibited. Please see applicable Privacy Policy and Legal Notice (for details see Privacy Policy).

date: 26 March 2017


Summary and Keywords

In the standard view, A acts paternalistically toward B if and only if: (i) A restricts B’s liberty, (ii) A acts against B’s will, (iii) A acts for B’s own good. For example, the state may tax or prohibit smoking in the interest of citizens’ health in circumstances in which such measures are resisted by them or some of them. Telling counterexamples have been produced to each of these conditions. In the revised view, A acts paternalistically toward B if and only if: (i) A acts so as to influence B by the use of means other than rational persuasion; (ii) A does not regard B’s will as structurally decisive (i.e., A takes the prevention of voluntary self-regarding harm to constitute a reason for influencing B); (iii) A does so for B’s good or to affect matters within B’s legitimate sphere of control; (iv) A’s act cannot be justified without counting its beneficial effects on B in its favor. The wrongness of paternalism lies in the way in which a paternalistic act by A toward B infringes B’s autonomy: A does not consider B’s will authoritative in determining how A should treat B in B’s self-regarding matters―A subjects B’s will to his in this sense. Hard paternalism as thus understood should be distinguished from soft paternalism or anti-paternalism. According to the latter, the prevention of voluntary self-regarding harm is never a good reason for interference. The latter is justifiable only to prevent involuntary self-regarding harm―harm pertaining to acts that are not his or do not represent his values or preferences. Hard paternalism may, pace what soft paternalism or anti-paternalism claims, sometimes be justifiable. This is particularly so when the voluntary self-regarding harm involved is significant and the infringement of liberty required to prevent it limited or acceptable given the harm at stake. The question of when a good or an advantage is profound and when an infringement of liberty is limited is, however, difficult and worthy of further investigation. Paternalistic justifications should be distinguished from other liberty-limiting principles. That is, they should, first, be distinguished from moral paternalism focusing on improving the person’s moral character and hence his moral well-being or on making the person better (as opposed to the improvement of the person’s physical and psychological condition focused on by ordinary or welfare paternalism). Second, it should be distinguished from legal moralism concerned with barring conduct that is intrinsically morally bad (that is, bad for reasons independent of how it affects people’s character or their physical or psychological condition).

Keywords: Paternalism, hard paternalism, soft paternalism, autonomy, liberal reconciliation strategy, moral paternalism, legal moralism


In the standard view, A acts paternalistically toward B if and only if: (i) A restricts B’s liberty, (ii) A acts against B’s will, (iii) A acts for B’s own good. For example, the state may tax or prohibit smoking in the interest of citizens’ health in circumstances in which such measures are resisted by them or some of them.1 Liberals, including John Stuart Mill in his On Liberty, have traditionally resisted such policies (Mill, 1985). Part of their reason for finding such policies wrong is that a paternalistic act by A toward B may be seen to infringe B’s autonomy: A does not consider B’s will authoritative in determining how A should treat B in B’s self-regarding matters―A subjects B’s will to his in this sense. Paternalistic policies need not, however, be wrong, all things considered. Often the good they promote may be considerable and the infringement of autonomy required to do so limited or acceptable given the good at stake.2 This seems true, for example, in the paradigmatic cases of crash helmets and seat belt regulations. Moreover, the emergence of so-called libertarian paternalist policies that may be held to infringe people’s autonomy in a minimal sense, if at all, while benefiting people profoundly has heightened the interest in investigating further the nature and justification of paternalist policies. A critical overview of the existing literature is provided and areas worthy of further investigation are suggested. The section What is Paternalism? confronts the thorny question of the appropriate definition of paternalism, including the distinction between hard and soft paternalism. The section Justification of Hard Paternalism considers the extent to which hard paternalism is justifiable and points to underexplored avenues bearing on this issue. The section Other Liberty-Limiting Principles situates paternalistic justification within the frame of alternative liberty-limiting principles.

What Is Paternalism?

The Standard View

According to Mill’s “one very simple principle”:

The sole end for which mankind are warranted, individually or collectively, in interfering with the liberty of action of any of their number is self-protection.… [T]he only purpose for which power can be rightfully exercised over any member of a civilized community, against his will, is to prevent harm to others. His own good, either physical or moral, is not a sufficient warrant.

(Mill, 1985, p. 68)

This principle makes acts that are harmful to others eligible to be interfered with, and acts that are merely self-regarding harmful not so (Jones, 1985, pp. 136–137).

In Richard J. Arneson’s influential reformulation of Mill’s “antipaternalist principle”:

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 (when his present will is not explicitly overridden by his own prior commitment) or against his prior commitment (when his present will is explicitly overridden by his own prior commitment). Mill’s principle states that paternalistic policies so defined are always wrong.

(Arneson, 1980, p. 471)

In his seminal 1971 article, Dworkin relies on a similar understanding of paternalism:

By paternalism I shall understand roughly the interference with a person’s liberty of action justified by reasons referring exclusively to the welfare, good, happiness, needs, interests or values of the person being coerced.

(Dworkin, 1971, p. 108)3

According to the Mill-Arneson-Dworkin definition of paternalism, A acts paternalistically toward B if and only if: (i) A restricts B’s liberty; (ii) A acts against B’s will; (iii) A acts for B’s own good. Each component of this definition has been subjected to fierce criticism. However, some contemporary authors stand by―with certain refinements and additions―the standard view. Peter de Marneffe is one of them. A “government policy is paternalistic toward A,” according to him, “if and only if

  1. (a) it limits A’s choices by deterring A from choosing to perform an action or by making it more difficult for A to perform it;

  2. (b) A prefers A’s own situation when A’s choices are not limited in this way;

  3. (c) the government has this policy only because those in the relevant political process believe or once believed that this policy will benefit A in some way; and

  4. (d) this policy cannot be fully justified without counting its benefits to A in its favor” [i.e., consideration for the well-being or good of the person whose liberty is restricted, is a necessary condition for the justification of the policy in question4] (De Marneffe, 2006, p. 74. Notes excluded).

On the “Restriction of Liberty” Condition

In their 1976 article Bernard Gert and Charles M. Culver produce telling counterexamples to this condition. For example, a doctor who administrates a blood transfusion to an unconscious member of a religious sect, who does not believe in blood transfusions, acts paternalistically, if anyone does. However, it is unclear that she coerces the patient or otherwise restricts his liberty―she is arguably not even trying to affect or control the patient’s behavior (Gert and Culver, 1976, p. 46).

In a similar vein, Seana V. Shiffrin points out that paternalism need not diminish freedom and may take the form of an omission. For example, A may decline to assist B in building a set of shelves, because doing so would in her opinion promote B’s good in various ways (B should for his own sake learn to become more self-reliant and so on) (Shiffrin, 2000, p. 213).

Moreover, A may act paternalistically by enhancing B’s freedom, against B’s will, A acting on the belief that it will benefit B to cope with the further options in question (Shiffrin, 2000, p. 214). Governmental policies such as those providing incentives for citizens to engage in certain activities, for example visiting art museums or going to the opera, may also increase people’s options while at the same time appearing paternalistic (Le Grand & New, 2015, p. 12).

Incidentally, Mill’s and Dworkin’s focus on “interferences,” as opposed to Arneson’s focus on “restrictions of liberty,” may allow for the indicated more expansive notion of what might count as a paternalistic act or policy, or at least for aspects of this (cf. VanDeVeer, 1986, p. 18).

The need for a more encompassing account of the kind of acts (or omissions) that may involve paternalism than the one suggested by Arneson’s focus on “restrictions on a person’s liberty” or Dworkin’s definition if we emphasize that paternalism according to it involves a person “being coerced” has also become obvious by the emergence of the immensely influential so-called libertarian paternalism or nudge policy spearheaded by Richard H. Thaler and Cass R. Sunstein (2008). While priding itself on not involving coercive measures or restrictions on liberty, it still, by its own admission, influences people in ways conducive to their own good, and does so in a way that is not so innocuous as not to deserve the label of paternalism (this term arguably suggesting that a certain, at least prima facie, wrongness pertains to the acts or policies it characterizes). Hausman and Welch suggest that we should conceive of such acts or policies as some that “shape” people’s choices, this implying “the use of flaws in human decision-making to get individuals to choose one alternative rather than another” (Hausman & Welch, 2010, p. 128).

Before considering an appropriate account of what may characterize the acts or policies involved in paternalism in terms of how they influence their targets, we should note that some authors resist expanding the set of acts (or omissions) in our interest beyond coercive or legally coercive acts or they argue that focusing especially on the latter class of acts is justified in that the question of the justification of the state coercing citizens for their own good takes center stage in our concern with paternalism (De Marneffe, 2006, p. 76; cf. Feinberg, 1986, pp. 4, 7–8). However, as we have seen, governmental acts or policies, including incentives and shaping, aimed at averting harm to the agent him- or herself or intending to promote his or her good, need not involve coercion or restrictions of liberty, and still they seem paternalistic. Moreover, even if paternalism in law or with respect to policies is often associated with interferences in liberty, this, as aptly pointed out by Gert and Culver, “is due to the nature of law, not to the nature of paternalism” (Gert & Culver, 1976, p. 45).

How should we characterize the influence feature of paternalistic acts or policies? A recent, apparently apt, proposal stresses the involvement of means other than those of rational persuasion (Hausman & Welch, 2010, pp. 128–129; Scoccia, 2008, p. 353). This includes, inter alia, coercion as well as incentives and shaping.

On the “Against B’s Will” Condition

The idea that a necessary feature of a paternalistic act or omission is that it is adopted “against the will” of its intended beneficiary has, as has the influence feature discussed previously, been subjected to a number of forceful objections or counterexamples. Daniel Groll recently provided a useful account of some of the central objections to this criterion and added an ingenious one of his own. Together they justify a serious revision of the criterion or the jettisoning of it altogether.

The “against B’s will” criterion faces first the “No Will” problem (Groll, 2012, p. 698). The problem concerns cases in which B has no will on the matter or has not as of yet formed one. For example, A may prevent B from receiving a credit card offer (intercepting the offer that arrives in the mail and ripping it up)―an offer that would not be advisable for B to accept, but that he is all the same strongly inclined to accept (Shiffrin, 2000, p. 214). B, say, has not reflected on the matter (i.e., on whether he would want A to act in such ways as to protect B’s interests) or formed an intention with regard to it. Accordingly, A would not seem to be acting against B’s will, but still, it seems plausible to say, be acting paternalistically (Groll, 2012, p. 698; Shiffrin, 2000, p. 214).

Second, the criterion under consideration has to confront the “Accidental Concordance” problem (Groll, 2012, p. 698). To explain: in some cases A may act so as to promote B’s good or to avert harm to him in, as it happens, perfect consistency with B’s will, but without paying attention to B’s will on the matter. The fact that A’s act squares with B’s will is coincidental or a matter of luck (Cohen, 1994, p. 120 and following; Groll, 2012, p. 698). For example, Sabina decides to pay off Ernest’s debt―something that as a matter of fact sits well with Ernest’s wishes―without Sabina giving regard to Ernest’s will on the matter: she pays off his debt “without any thought to whether this is what Ernest would want her to do” (Groll, 2012, p. 698). Sabina, it seems clear, “evinces a paternalistic attitude” (Groll, 2012, p. 699). She does not let herself be guided by what B’s will on the matter is, denying control to B in this sense or appropriating control over matters that are properly within B’s control (cf. Cohen, 1994, p. 120 and following).

Third, consider what may be called the “Relevant Reason” problem―Groll’s own ingenious objection. A may regard B’s will as being what Groll calls substantively decisive in determining what A should do. This implies, roughly, that B’s will always outweighs competing considerations, including B’s good. Still, the latter enters A’s calculations as a consideration to be balanced against what B’s will on the matter is, although it is not necessarily decisive; indeed, insofar as A regards B’s will as substantively decisive, it in fact never outweighs the consideration pertaining to B’s will.

For example, if A, a doctor, regards a patient B’s will as being substantively decisive with respect to whether or not B is to undergo surgery, this means that if B is against undergoing the treatment in question this outweighs whatever competing consideration there is, including the consideration that the treatment is highly likely to benefit B profoundly with low or no non-trivial risks involved. Likewise, consider the case of B who voluntarily chooses to incur a significant risk of harm―for example, the risk of a debilitating or even lethal fall to the ground, involved in trying to jump from one balcony of a tall building to another. A takes B’s will to be substantively decisive if A, in his considerations of what he ought to do, regards the fact that B freely chooses to absorb the harm or risk of harm in question to outweigh the reason A has to try to avert the self-regarding harm to B in question (A may still, of course, have other-regarding reasons to try to stop A, including the protection of others from being hit by B’s falling body in the case B misses a jump, and from eventual psychic harm from witnessing B hit the ground).

In these cases A deliberately acts in accordance with B’s will. Moreover, it may seem as if A treating B’s will in this way suffices as a bulwark against paternalism, or, more relevant here, that such treatment would suggest that A did not act paternalistically. After all, an unwanted treatment is never forced on a patient, and an agent is never barred from voluntarily shouldering a significant risk of harm.

However, and this is Groll’s apt observation, there may be something disrespectful and paternalistic in A allowing considerations other than what B’s will is to enter into A’s determination of what she should do―in A treating B’s will as substantively decisive. B’s will, it might be said, is appropriately sovereign or authoritative in what is B’s self-regarding matters. That is, B’s will rightly silences or excludes whatever competing considerations there might be, including considerations of what is in B’s interest or for his good. At least this is true as long as B’s will is intact or he is competent. Only if A regards B’s will in this way―regards it, in Groll’s terms, as structurally decisive―does A not treat B paternalistically.

On a first view, Groll’s suggestion that when A treats B’s will (merely) as substantively decisive, A treats B paternalistically might seem implausible. However, it squares with Feinberg’s formulation of legal or hard paternalism―“the illiberal principle so emphatically rejected by Mill” (Feinberg, 1986, p. 4. cf. p. 12):5

It is always a good and relevant (though not necessarily decisive) reason in support of a criminal prohibition that it will prevent harm (physical, psychological, or economic) to the actor himself.

(Feinberg, 1986, p. 4)

According to this principle, A counting the prospect of benefiting, or averting harm, to B as a relevant reason in determining what to do, although this reason may always be outweighed by the fact that the harm in question is one that the agent voluntarily chooses to absorb, implies that A treats B paternalistically.6 If A takes the reason provided by B’s good sometimes to outweigh the fact that B consents to the harm involved or incurs it voluntarily, or, a fortiori, takes this to be the case often or always, A in a stronger sense treats B paternalistically, but arguably not in a way that is categorically different. In both cases A treats B paternalistically or in accordance with the doctrine of hard or legal paternalism.

A necessary feature of a paternalistic act by A toward B is then not that A acts “against B’s will” but that A does not treat B’s will as structurally decisive. Combining this feature with the influence feature discussed previously, it may be said that a paternalistic act by A toward B is characterized by at least the following two necessary features: A acting to influence B by the use of means other than rational persuasion and A not treating B’s will as structurally decisive in determining how A should act. Consideration will now be given to the motivational component integral to the standard view, namely that A acts as he does “for B’s good.”

On the “For B’s Good” Condition

According to the “for B’s good” condition, a paternalistic act by A toward B is undertaken by A for B’s good. The good in question is B’s interests or welfare: “these interests [the interests catered to by A] are defined in terms of the things that make a person’s life go better; in particular their physical and psychological condition” (Dworkin, 2016).7 B’s good or welfare may be understood either subjectively, referring to B’s own conception of his good, or objectively in terms of what in A’s view would cater to B’s good independently of B’s view on the matter. The sort of legal or welfare paternalism vehemently resisted by Feinberg, for example, is of the former, subjective, kind (Arneson, 1989, 2005; Midtgaard, 2015a; Scoccia, 2008, p. 361). Is A catering to B’s good in either of these senses a necessary condition for saying that A treats B paternalistically? Of the three conditions stemming from the standard view this is the least controversial, that is, the claim that it is a necessary condition is almost universally accepted.

However, Shiffrin, for one, takes issue with the “for B’s good” condition. She does so based on cases such as the following. A park ranger may prevent a hiker from embarking on a steep, dangerous mountain path, not out of concern for the safety or well-being of the hiker (in which case, we all agree, the ranger would be acting paternalistically), but because of how it would adversely affect the hiker’s spouse if an accident befalls him (Shiffrin, 2000, p. 217). If the park ranger does so, Shiffrin argues, her act has an important feature in common with the case in which she bars the hiker from embarking on the path for his own sake. This feature is what Shiffrin takes to be crucial to characterizing paternalism by A toward B, namely, roughly, that it involves A substituting her judgement or agency for B, A catering to B’s interests or matters legitimately within B’s control, on the grounds that A regards her judgement or agency to be superior to B’s with respect to those interests or matters:

“Paternalism by A toward B may be characterized as behavior (whether through action or through omission)

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

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

  3. (c) directed at B’s own interests or matters that legitimately lie within B’s control8

  4. (d) undertaken on the grounds that compared to B’s judgment or agency with respect to those interests or other matters, A regards her judgment or agency to be (or as likely to be), in some respect, superior to B’s” (Shiffrin, 2000, p. 218).

Sharing this feature, the two cases also, according to Shiffrin, “involve the same sort of intrusion into and insult to a person’s range of agency” (Shiffrin, 2000, p. 218).

Notice first that paternalism as characterized by Shiffrin appears to bear similarity to Groll’s account in its emphasis on how A treats B’s will or his judgement or agency in matters related to B’s interests or legitimate sphere of control. A, in Shiffrin’s account, certainly does not regard B’s will as authoritative with respect to these matters. A substitutes her own will for that of B.

The question is whether paternalism, as Shiffrin suggests, appropriately encompasses A treating B’s will in this way in certain other-regarding matters in addition to A’s similar treatment of B in B’s self-regarding matters. With Mill, we may believe that once the conduct interfered with has significant other-regarding effects, “the case is taken out of the self-regarding class” and becomes eligible or amenable to be coercively interfered with (Mill, 1985, p. 148). For example, an individual may indulge in conduct such as gambling and heavy drinking, which is inadvisable in terms of his self-regarding interests. If that were all, others would not be entitled to interfere with him for his own good. By doing so they would contravene the anti-paternalistic principle. However, by engaging in the conduct in question the person may come to violate duties to assignable others including his wife and children. If this happen, then, as Mill stresses, his in the first place self-regarding harmful conduct becomes eligible to external interference precisely because it enters the class of other-regarding acts, which is appropriately thus eligible. Similarly, we may say, if the ranger interferes with the hiker on the ground that his conduct is likely to involve him violating certain duties of his to his spouse, the ranger interferes on grounds that are appropriately conceived as other-regarding, the interference in question being a legitimate one, but not a paternalist one.

Shiffrin’s controversial suggestion to include as a form of paternalism A’s attempt to influence matters legitimately within B’s control (A substituting her judgement for B’s regarding how these matters are best handled) for the sake of another party C is not as easily dismissible as the reaction of some philosophers would suggest (De Marneffe, 2006, p. 74 n. 17; Le Grand & New, 2015, p. 16 n. 15). The plausibility of counting such instances as paternalism depends, it seems, partly on whether B confronted with A’s behavior may plausibly express paradigmatic anti-paternalist indignation of the form “this is for me to decide” or “whose life is it anyhow?” In the case of A acting so as to affect B for the sake of C, C being B’s child, this rebuttal in the mouth of B is often, as aptly pointed out by Mill, lame (Mill, 1985, pp. 175–178). Many aspects of a child’s life and many of his or her interests, for example in an appropriate education, is not “for B to decide” or left to his discretion. Many aspects of how B chooses to conduct his relationship with C, C being B’s spouse or partner, including the extent to which he chooses to incur risks knowing that he may cause her grief, are, by contrast, up to B to decide, ideally in accordance with a common understanding with C regarding the terms of their relationship. Consider the case of Andrew McAuley, who drowned trying to kayak from Australia to New Zealand (Snelling, 2014, pp. 149, 151). His wife, left with their small child, supported his project all the way. If a ranger had blocked McAuley’s adventure for the sake of his wife this would, apparently, have been a paternalist act (albeit perhaps a justifiable one) both with regard to McAuley (intruding in how he chose to discharge his duties to his wife and child) and with respect to his wife (intruding in her assessment of the kind of the risk it was acceptable for her husband to incur).

Justified by Reference to B’s Good

The claim that a paternalistic act by A toward B is motivated by B’s good has been discussed. As pointed out by De Marneffe, there may be reasons for adding a justificatory condition to this account of paternalism (De Marneffe, 2006). Following John Kleinig (1983, p. 11), justification may be understood as a “success-word,” suggesting that the reasons for the act in question defeat the reasons against (cf. Raz, 1986, p. 152). Motivation, it might be said, refers to the reasons upon which the agent acts irrespective of the fact that the reasons in question may in fact be defeated by others. (The “rationale” of an act may refer either to its justification or to how it is motivated.) For example, crash helmet legislation may be motivated by the so-called “flying rock” rationale (to wit, that flying rocks from the wheels of other motorists may hit a helmetless motorcyclist, thus making her lose control of her bike, jeopardizing the safety of other motorists), although the reason invoked to support the legislation in question is insufficiently weighty (perhaps the risk quoted is diminutive) to defeat the reasons against (including the infringement of the liberty of motorcyclists preferring to ride without a helmet).

On a purely motivational understanding of paternalism, all it takes to show that a policy is not necessarily paternalistic is to cite a potential non-paternalistic motive such as the “flying rock” rationale irrespective of the success of this reason toward justifying the policy in question. However, when consideration is given to the various attempts by liberal anti-paternalists in the literature to account for apparently paternalistic policies―policies that appear sound and that we would be loath to repeal―what results is attempts to justify the policies in question (e.g., Anderson, 1999; Bou-Habib, 2006; Shiffrin, 2000). They appear, in other words, to assume that what is needed to defeat a policy’s status as paternalistic is to produce a plausible non-paternalistic justification for it. To accommodate this important strategy in the literature we should include in our understanding of paternalism the condition that a paternalistic policy is one that cannot be justified without reference to its beneficial effects on the person whose liberty gets restricted or must be justified exclusively by reference to such effects.


To sum up the discussion of the conditions for A’s conduct toward B to count as paternalistic, in the revised view:9

A acts paternalistically toward B if, and only if:

  1. (a) A acts so as to influence B by the use of means other than rational persuasion

  2. (b) A does not regard B’s will as structurally decisive (i.e., A takes the prevention of voluntary self-regarding harm to constitute a reason for influencing B)

  3. (c) A does so for B’s good or to affect matters within B’s legitimate sphere of control

  4. (d) A’s act cannot be justified without counting its beneficial effect on B in its favor (cf. De Marneffe, 2006, p. 74).

On the Wrongness of Paternalism

The definition of paternalism reached is arguably of a normative kind capturing something that is prima facie wrong (cf. Grill, 2009, 2013). When A, by the use of means other than rational persuasion, aims to influence B for B’s own good or with regard to matters legitimately within B’s control and in doing so does not take B’s will as authoritative in determining what to do, A in effect subjects B’s will to his, or substitutes his will for B’s. In brief, A infringes B’s self-direction or autonomy (cf. Blake, 2002, sec. II; Raz, 1986, pp. 148–157, chap. V). This, we may hold, is prima facie wrong in that it infringes the right of a person of “ripe years” with an intact will to be sovereign in self-regarding matters. The autonomy-related account of the prima facie wrongness of paternalism seems closely related to the notion that A when acting paternalistically imposes his conception of the good on B (Arneson, 2005, p. 266; Dworkin, 1971, p. 122). Infringements of autonomy may also generate the kind of reactive attitudes that are pivotal to the prima facie wrongness of paternalism according to some conceptions, including Shiffrin’s mentioned previously. Paternalist behavior, Shiffrin suggests, “manifests an attitude of disrespect toward highly salient qualities of the autonomous agent. The essential motive behind a paternalist act evinces a failure to respect either the capacity of the agent to judge, the capacity of the agent to act, or the propriety of the agent’s exerting control over a sphere that is legitimately her domain” (Shiffrin, 2000, p. 220). Similarly, Elizabeth Anderson suggests that if the state adopts various measures on paternalistic grounds, for example a scheme of compulsory insurance, it is “effectively telling citizens that they are too stupid to run their lives, so Big Brother will have to tell them what to do,” in this way showing outright disrespect for citizens (Anderson, 1999, p. 301).10

Others try to define paternalism in non-normative terms. VanDeVeer, for example, defines a paternalistic act roughly “as an act in which one party interferes with another with the aim of promoting his or her own good” (VanDeVeer, 1986, p. 20). “Interferences,” VanDeVeer points out, are not necessarily even presumptively wrong. For example, A may without using means other than rational persuasion, but against B’s will, try to influence B for B’s sake (for example, A may try to make B quit smoking by drawing B’s attention to the risk of harm involved). So paternalism is not necessarily presumptively wrong although it often, VanDeVeer admits, involves acts that are presumptively wrong (Grill, 2013, pp. 31–32).

There are reasons for and against adopting a non-normative definition of paternalism (Bullock, 2014; Grill, 2009, 2013). One advantage of the latter is that if we can get clear on the descriptive meaning of a term, this may make normative disagreements more transparent and tractable (Grill, 2013, p. 32). On the other hand, one might consider it an advantage of a normative definition that it reflects why paternalism matters to us, for example that it reflects a morally problematic insult of the person exposed to it, or involves the state’s use of problematic means in its dealings with citizens, such as coercion (De Marneffe, 2006, p. 75; Grill, 2009, p. 14; Shiffrin, 2000, p. 212).

Hard and Soft Paternalism

What has been outlined previously is an account of hard or legal (or genuine, if you like) paternalism. Soft paternalism or anti-paternalism denies hard paternalism. Specifically, it holds that the prevention of voluntary self-regarding harm is never a good reason for acting so as to try to influence B. In Mill’s forceful and eloquent phrases:

As soon as any part of a person’s conduct affect prejudicially the interests of others, society has jurisdiction over it, and the question whether the general welfare will or will not be promoted by interfering with it becomes open to discussion. But there is no room for entertaining any such question when a person’s conduct affects the interests of no persons besides himself, or needs not affect them unless they like (all the persons concerned being of full age and the ordinary amount of understanding). In all such cases, there should be perfect freedom, legal and social, to do the action and stand the consequences.

(Mill, 1985, pp. 141–142)

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.

(Mill, 1985, p. 142)

The anti-paternalist or soft paternalist norm, as Arneson points out, “protects the agent’s freedom to choose now to waive or alienate some or all of her future freedom to act or even personal sovereignty itself” (Arneson, 2005, p. 265). It is not a doctrine of prudence (this, as indicated, would make a sort of subjectivist hard paternalism).

Positively, soft paternalism enjoins the restriction of people’s liberty or influencing them by means other than rational persuasion “to give effect to [or to block actions that do not reflect] their current evaluations of where their own good lies and their own present will as to how far it should be pursued” (Arneson, 2005, p. 266).11 Acts that are not reflective of the person’s own conception of his or her good are not his or her acts. Interfering with them does not infringe the person’s autonomy. Paradigmatic examples of acts that are not the agent’s own in this sense include people acting in self-regarding harmful ways, or with great risk of such harm, on false beliefs or because they are subjected to coercion or duress. For example, if the balcony jumper mentioned previously embarks on his quest believing that he will flow gently from one balcony to another (cf. Dworkin, 1971, p. 122) or he engages in the activity against the background of a credible threat issued by his girlfriend that she will dump him if he does not undertake a series of jumps,12 then soft paternalistic interferences or attempts at influencing the conduct of the balcony jumper are in place.

Deciding on whether individuals who act in self-regarding harmful ways do so in a sufficiently autonomous way to be appropriately immune to external interventions is not easy. The most comprehensive analysis of this question is due to Feinberg. He approaches the question in the form of a development of a standard of voluntariness. A perfectly voluntary choice, according to Feinberg, is one in which: the chooser is competent; he does not choose under coercion or duress; he does not choose because of more subtle manipulation, he does not choose because of ignorance or mistaken belief; and he does not choose in circumstances that are temporarily distorting (Feinberg, 1986, p. 115). Few if any actual choices satisfy this standard completely. A given self-regarding harmful choice may, however, according to Feinberg, be sufficiently voluntary to be protected by the anti-paternalist norm. Whether or not it is so depends, inter alia, on how risky and irrevocable the harm involved is (the more risky and irrevocable the harm in question is, the higher the appropriate standard of voluntariness is).

Soft Paternalism and Justification

Soft paternalism is usually considered less controversial than hard paternalism, if controversial at all. However, a number of concerns regarding the justification of soft paternalism have been raised. First, soft paternalism, at least on Feinberg’s influential account with a relatively high threshold for what is to count as a voluntary choice, might be difficult to distinguish from a version of hard paternalism in that it includes features such as calmness and deliberateness similar to those involved in a prudent or rational choice (Arneson, 1989). Second, as argued by Julian Le Grand and Bill New in their recent book (2015), even though the conduct interfered with by soft paternalists is non-autonomous, such interferences partly infringe autonomy or at least perceived autonomy (perceived autonomy or the fact that the agent regards himself as being capable of directing him- or herself, on their view, constituting an integral part of autonomy) (Le Grand & New, 2015, chap. 6). Similarly, Kalle Grill and Kristin Voigt argue that we do at least have some reason to respect even some choices the voluntariness of which is vitiated (Grill & Voigt, 2016, p. 4). Third, Jason Hanna has recently argued that the anti-paternalistic or soft paternalistic tenet that ill-informed self-regarding harmful choices may be interfered with while informed such choices are protected by autonomy is indefensible (Hanna, 2012). In the course of his argument he stresses the point that culpably ignorant self-regarding harmful choices or risks voluntarily assumed should sometimes not be protected by autonomy. For example, a reckless hiker may refrain from making use of an option to familiarize himself with the conditions of the bridges in the park in which he goes hiking. Along his course he encounters a bridge that unbeknownst to him is unsafe. He decides to cross it simply to complete his hike (not because he relishes the risk or wishes to kill himself). A bystander capable of intervening so as to avert the self-regarding harm in question should, Hanna submits, intervene. That is, the choice of the hiker is not protected by autonomy.13

Justification of Hard Paternalism

Soft paternalism or anti-paternalism is, as has been shown, the view that voluntary self-regarding harm is never a good reason for criminal prohibitions or for trying to influence people by the use of means apart from rational persuasion. It is, as already indicated, rooted in a general idea of personal sovereignty or autonomy (cf. Arneson, 2005, pp. 261–263). Apart from appeal to this ideal Feinberg makes an intuitive case for anti-paternalism. That is, he refers to strong and arguably widely shared intuitions regarding our right to, roughly, make our own choices in self-regarding matters. A right that we would very much prefer be observed by others in their dealings with us―and if they don’t we try to rebuff their interventions using language such as “That’s my business, not yours,” or “That’s no concern of anyone else” (Feinberg, 1986, p. 98)―and a right that we ought to observe as well in our dealings with others.14

A third argument is related to what could be considered a problem for or challenge to the hard paternalist. In counting self-regarding harm as a reason for intervention it is incumbent upon the hard paternalist to specify when important or fundamental interests are at stake and when the liberties that get infringed in order to protect these interests are limited or the infringements acceptable in light of the good at stake. Undertaking such balancing in a non-arbitrary way is, Feinberg submits, an unsurmountable challenge (Shafer-Landau, 2005, pp. 184–185). Soft paternalism, by contrast, relies on a clear distinction between voluntary and non-voluntary self-regarding harm (although, as mentioned previously, it might in fact be difficult to determine the correct threshold for a sufficiently voluntary choice and to do so in a way that renders the criterion for a voluntary choice distinct from what counts as a rational or prudent choice).

A crucial thought behind hard paternalism is that it seems implausible to hold that the prevention of self-regarding harm, irrespective of its magnitude and irrespective of the potentially limited infringements of liberty required to prevent the harm in question, is never a good reason for intervention. A mild form of paternalism is, recall, constituted by the view that takes a person’s will as substantively decisive, that is, as always outweighing other concerns, including the person’s own good (the latter considerations entering the scales, so to say, but never actually being weighty enough to justify blocking or restraining people’s self-regarding voluntary choices). Paternalism of a kind that would actually justify paternalistic policies denies that a person’s will is always substantively decisive, that it holds that sometimes, at least, considerations pertaining to the person’s good outweigh our reasons to respect his self-regarding harmful choices or his willingness to absorb the harm in question. When, if ever, might this be the case?

There is an obvious answer to this that is given by a number of authors including even liberal anti-paternalists inclined to think that hard paternalism is (almost) always wrong (e.g., Anderson, 1999, pp. 301–302; Barry, 1995, p. 87). It goes roughly like this. Some self-regarding acts involve profound harm or risk of such harm. The risk in question may be slight, but the outcome truly catastrophic if it materializes. Paradigmatic cases include not fastening one’s seat belt when driving or not wearing a crash helmet when riding a motorcycle. In many cases the driver or the biker will arrive unharmed. They have good option luck, so to say.15 At the same time, the risk of the harm in question can apparently be averted by the use of means that are not very costly in terms of the infringement of liberty or autonomy they involve, or the cost in question is acceptable given the magnitude of the risk of harm they prevent. In the cases mentioned, the relevant such means would be to make the use of seat belts when driving and of a crash helmet when operating a motorcycle obligatory. So what we have here is arguably cases in which the good or interest at stake (or the harm that might be prevented) is profound whereas the costs in terms of infringement of liberty is if not trivial then of a limited nature and acceptable in light of the good at stake. Insisting that the person’s good does not even in this case count as a reason for interference or that it may count as a reason, but never one that is actually weighty enough to outweigh our reason to respect people’s self-regarding choices, seems fanatical.

As indicated, this rationale for certain hard paternalistic policies is endorsed even by liberal anti-paternalists and it is taken to justify the “tried and true” policies of seat belt and crash helmet regulations. It constitutes also, however, the backbone in the arguments of a number of authors defending a wider range of hard paternalistic policies (e.g., Arneson, 2005; Conly, 2013; Grill & Voigt, 2016; Scoccia, 2008, 2013; Shafer-Landau, 2005). While this may be suggestive of the strength of the argument, it is also, and perhaps more likely, indicative of its indeterminacy or vagueness―the weakness of hard paternalism that Feinberg believes to be detrimental to it. The latter pertains especially to the question of what constitutes a trivial or limited interference with a person’s liberty and when a person’s profound good or interest is at stake (and of course when an infringement of liberty is acceptable given the good or interest at stake). In Anderson’s influential criticism of choice-sensitive conceptions of distributive justice, for example, she accepts crash helmet and seat belt regulations by reference to an argument of the kind just sketched, but then moves to deny the justifiability of various schemes of compulsory insurance on the ground that the infringement of liberty involved is too serious to be justifiable paternalistically or in a way that lets the good to the person whose liberty is restricted count in its favor (Anderson, 1999, pp. 301–302). This obviously begs the question of how to distinguish in a more systematic way between serious and non-serious infringements of liberty. But Anderson is not alone in relying on tacit or intuitive notions of the kind of infringements of liberty involved in various cases (and the good at stake). De Marneffe, remarkably, in his important case for hard paternalism not always being wrong presumes that there is a distinction to be made between fundamental and non-fundamental infringements of liberty, this distinction being of importance to the kind of hard paternalism he endorses and even more clearly to the circumscribed liberal principles of anti-paternalism he also endorses, but does not set out himself to provide such a distinction (De Marneffe, 2006, p. 69; see, however, De Marneffe, 2010, chap. 3). And indeed this may not be surprising given that the questions of what constitutes people’s crucial interests or what counts as advantages for them and when people’s liberty is infringed in a serious way and when they are infringed in a less serious way are indeed philosophically deep and controversial. However, getting clearer on these questions seems crucial to making progress on understanding hard paternalism and the degree to which it is justifiable. This is not to say that the existing literature does not include material addressing these important topics. Especially in areas such as, for example, smoking and prostitution there are many valuable contributions involving attempts at balancing the concrete goods and harms at stake and assessing the seriousness of the infringements of liberty involved (De Marneffe, 2010; Grill & Voigt, 2016). And perhaps addressing the tricky balancing on a case by case basis is the best, or most realistic, strategy. Before moving on to another way of justifying hard paternalism, or to deflect anti-paternalistic justifications of certain policies that we would be loath to give up, it bears mentioning that although the strategy sketched here is broadly consequentialist and the problems raised may seem to pertain especially to such a view, contractualist justifications of hard paternalistic measures―that is, paternalistic measures justified by reference to what parties in a certain situation and with certain characteristics, for example being “fully rational,” would accept as reasonable restrictions on the liberty of citizens in an ongoing society (Dworkin, 1971; Rawls, 1999, pp. 218–220; De Marneffe, 2010)―are subject to the same problems or challenges. They are so in that they must, when characterizing the contractualist situation, specify the kind of interests and their relative importance on the basis of which parties may accept certain paternalistic restrictions.

The justification of hard paternalism may proceed in a more indirect manner. That is, it may consist in an attack on the so-called liberal strategy of trying to justify apparently paternalistic policies non-paternalistically, also known as the liberal strategy of reconciliation (cf. Midtgaard, 2005b). It seeks to reconcile the liberal repugnance against hard paternalism with the fact that a number of seemingly sound policies appear to be paternalistic. Arneson attacks this strategy in an ingenious way. His attack works by excluding by stipulation various non-paternalistic reasons for interfering in a certain activity. He then argues that even in the absence of such reasons for interference in the activity in question, there appear to be compelling reasons for interference. The latter are even undiminished in strength after the removal of the non-paternalistic reasons. This would seem to show that the compelling and salient reasons in play are paternalistic. Consider his case of a group of daredevils entering the forest during hunting season dressed as shooting targets. We presume, first, that they do so on an entirely voluntary ground or on a ground that is sufficiently voluntary to disallow interfering with their activity on soft paternalistic ground.16 Second, to exclude harm to others we assume that hunters generally constitute “a prudent, culturally conservative bunch” who do not lose an hour of sleep in case they accidentally hit one of the daredevils (Arneson, 2005, p. 273). So there is no “psychic harm” involved―a harm that may justify interferences, liberal theorists of reconciliation suggest (e.g., Feinberg, 1986, pp. 134–142).

Having excluded various “harm to others” grounds for restricting the conduct of the risk takers in question and grounds referring to vitiated voluntariness (the grounds exemplified are of course not exhaustive, but additional grounds may be excluded by stipulation in ways similar to the exclusion of the grounds mentioned previously), Arneson claims that our reasons for barring the youngsters from entering the forest dressed as shooting targets remain as intuitively strong as ever, which goes to show that the salient reason at play for restricting the conduct of the darevevils is to prevent the self-regarding harm or risk of harm they voluntarily incur. Preventing severe harm of even a voluntary self-regarding kind arguably has a strong intuitive appeal. It is important to prevent people from making bad decisions (Voorhoeve, 2008, pp. 198–199).

Although the self-regarding harm pointed to by Arneson appears to provide us with strong reasons for preventing it, it might seem as if he overstates its importance, or, specifically, that his claim that our reasons for intervention remain as strong as ever, even informed that the conduct is completely or sufficiently voluntary, is implausible.17 It seems indeed that our reason for preventing self-regarding harm is reduced in strength if the harm in question is voluntarily incurred or one that the person it befalls consents to absorb. But this is consistent with the basically paternalistic stance that the prevention of self-regarding voluntary harm at least offers a reason for preventive measures. How compelling the reason in question is depends of course in part on the magnitude of the harm involved (Voorhoeve, 2008, pp. 198–199). This brings us back to the previous considerations about the necessity of trying to achieve a better understanding of what constitutes profound and non-profound harm.

Returning briefly to the question of the degree to which people’s liberty gets infringed in attempts to promote their good or to prevent harm from befalling them, it is notable that libertarian paternalism is partly motivated on the ground that the interferences involved are limited in nature to the extent they constitute interferences at all. At the same time, the degree to which people’s self-regarding decisions may be improved by these arguably innocuous means is often held to be considerable. However, we should keep in mind that the means by which libertarian policies affect people’s decisions―by shaping their choices, that is, using “flaws in human decision-making to get individuals to choose one alternative rather than another” (Hausman & Welch, 2010, p. 128)―share with other paternalistic means of influencing people’s choices the characteristic that they do not apply rational persuasion.18 What is more, they often, as has been pointed out by Hausman and Welch, affect, in a way that is more insidious than more traditional coercive means, people’s control over their own evaluations and convictions. Accordingly, it is at best an open question whether the means of influence proposed by libertarian paternalists are less problematic in terms of the infringement of liberty they involve than are more traditional coercive means. From the perspective of the proposed analysis of the wrongness of paternalism, to wit, that it infringes autonomy, shaping may appear at least as problematic as more traditional coercive means. This is not, of course, to say that libertarian paternalist policies can never be justified, irrespective of the good that can be achieved by their application, rather it is to say that such policies are not of such an uncontroversial nature that using them is justifiable as soon as some considerable good can be promoted by doing so. Hence libertarian paternalists are in the same boat as more traditional paternalists in having to say something relatively determinate regarding what kind of goods may make certain infringements of autonomy justifiable.

Other Liberty-Limiting Principles19

Hard paternalism or welfare paternalism is one among other liberty-restricting principles or principles justifying trying to influence people’s conduct by the use of means other than rational persuasion. Paternalism or welfare paternalism, recall, is concerned with B’s interests or welfare, where “interests are defined in terms of the things that make a person’s life go better; in particular their physical and psychological condition” (Dworkin, 2016). Moral paternalism is concerned with the moral welfare of the person. In one version of moral paternalism, the paternalist is concerned with the person’s conduct because this would improve the person’s welfare or a certain aspect of this associated with his moral character. On another, the paternalist is concerned with making the person in question better without this necessarily being better for the person (i.e., without necessarily improving his well-being or a certain aspect of this related to his moral character). Legal moralism prevents inherently immoral conduct, that is, conduct that is immoral or wrong for reasons independent of harm to the person’s welfare or moral character. Finally, the doctrine of inalienable rights extends the ordinary understanding of the harm principle. On the latter, harm consented to or willingly absorbed by others does not constitute wrongs or wrongful setbacks of interests and hence should not be interfered with on an appropriate understanding of the harm principle (cf. Feinberg, 1984, pp. 32–36). According to the doctrine of inalienable rights, by contrast, there are things we are not permitted to do to others even with their consent (for example, enslaving them). This would amount to infringing their rights (Feinberg, 1971, p.118; Dworkin, 2005, 2016, 2012).


For very useful comments to earlier versions of this entry I am grateful to the participants in a seminar in Oslo on State Paternalism organized by the Center for the Study of Mind in Nature (CSMN), January 25–26, 2016. I am also indebted to the members of the section for political theory at Department of Political Science, Aarhus University. For detailed comments in writing I am grateful to Rasmus Sommer Hansen, Kasper Lippert-Rasmussen, and Jens Damgaard Thaysen. I am also grateful to the two anonymous referees of this entry for useful and constructive comments.


Anderson, E. S. (1999). What is the point of equality? Ethics, 109(2), 287–337.Find this resource:

Arneson, R. J. (1980). Mill versus paternalism. Ethics, 90(4), 470–489.Find this resource:

Arneson, R. J. (1989). Paternalism, utility, and fairness. Revue Internationale de Philosophie, 43(170) (3), 409–437.Find this resource:

Arneson, R. J. (2005). Joel Feinberg and the justification of hard paternalism. Legal Theory, 11, 259–284.Find this resource:

Barry, B. (1995). Justice as impartiality. Oxford: Clarendon.Find this resource:

Blake, M. (2002). Distributive justice, state coercion, and autonomy. Philosophy & Public Affairs, 30(3), 257–296.Find this resource:

Bou-Habib, P. (2006). Compulsory insurance without paternalism. Utilitas, 18(3), 243–263.Find this resource:

Bullock, E. C. (2014). A normatively neutral definition of paternalism. Philosophical Quarterly, 65(258), 1–21.Find this resource:

Cohen, G. A. (1994). Amartya Sen’s unequal world. New Left Review, I(203), 117–129.Find this resource:

Conly, S. (2013). Against autonomy: Justifying coercive paternalism. Cambridge, U.K.: Cambridge University Press.Find this resource:

De Marneffe, P. (2006). Avoiding paternalism. Philosophy & Public Affairs, 34(1), 68–94.Find this resource:

De Marneffe, P. (2010). Liberalism and prostitution. Oxford: Oxford University Press.Find this resource:

Dworkin, G. (1971). Paternalism. In R. A. Wasserstrom (Ed.), Morality and the law (pp. 107–126). Belmont, CA: Wadsworth.Find this resource:

Dworkin, G. (1988). The theory and practice of autonomy. Cambridge, U.K.: Cambridge University Press.Find this resource:

Dworkin, R. (1989). Liberal community. California Law Review, 77(3), 479–504.Find this resource:

Dworkin, R. (2000). Sovereign virtue: The theory and practice of equality. Cambridge, MA: Harvard University Press.Find this resource:

Dworkin, G. (2005). Moral paternalism. Law and Philosophy, 24(3), 305–319.Find this resource:

Dworkin, G. (2012). Harm and the Volenti Principle. Social Philosophy and Policy, 29(1), 309–321.Find this resource:

Dworkin, G. (Summer 2016). Paternalism. In E. N. Zalta (Ed.), Stanford Encyclopedia of Philosophy. Stanford, CA: Stanford University.Find this resource:

Feinberg, J. (1971). Legal paternalism. Canadian Journal of Philosophy, 1(1), 105–124.Find this resource:

Feinberg, J. (1984). Harm to others. Oxford: Oxford University Press.Find this resource:

Feinberg, J. (1986). Harm to self. Oxford: Oxford University Press.Find this resource:

Gert, B., & Culver, C. M. (1976). Paternalistic behavior. Philosophy & Public Affairs, 6(1), 45–57.Find this resource:

Grill, K. (2009). Anti-paternalism and public health policy. (PhD thesis). Stockholm, Sweden: KTH.Find this resource:

Grill, K. (2013). Normative and non-normative concepts: Paternalism and libertarian paternalism. In D. Strech, I. Hirschberg, & G. Marckmann (Eds.), Ethics in Public Health and Health Policy: Concepts, Methods, Case Studies (pp. 27–46). Dordrecht, The Netherlands: Springer.Find this resource:

Grill, K., & Voigt, K. (2016). The case for banning cigarettes. Journal of Medical Ethics, 42(5), 293–301.Find this resource:

Groll, D. (2012). Paternalism, respect, and the will. Ethics, 122(4), 692–720.Find this resource:

Hanna, J. (2012). Paternalism and the ill-informed agent. Journal of Ethics, 16, 421–439.Find this resource:

Hanna, J. (2015). Libertarian paternalism, manipulation, and the shaping of preferences. Social Theory and Practice, 41(4), 618–643.Find this resource:

Hausman, D. M., & Welch, B. (2010). Debate: To nudge or not to nudge. Journal of Political Philosophy, 18(1), 123–136.Find this resource:

Jones, P. (1985). Toleration, harm and moral effect. In J. Horton & S. Mendus (Eds.), Aspects of Toleration: Philosophical Studies (pp. 136–157). London: Methuen.Find this resource:

Kleinig, J. (1983). Paternalism. Totowa, NJ: Rowman & Allanheld.Find this resource:

Le Grand, J., & New, B. (2015). Government paternalism: Nanny state or helpful friend? Princeton, NJ: Princeton University Press.Find this resource:

Midtgaard, S. F. (2015a). Paternalism, public health ethics, and equality. World Political Science, 11(2), 405–420.Find this resource:

Midtgaard, S. F. (2015b). Non-renounceable rights, paternalism, and autonomy. Utilitas, 27(3), 347–364.Find this resource:

Mill, J. S. (1985) On liberty. Harmondsworth, UK: Penguin Books. Originally published 1859.Find this resource:

Oshana, M. A. L. (1998). Personal autonomy and society. Journal of Social Philosophy, 29(1), 81–102.Find this resource:

Rawls, J. (1999). A theory of justice (rev. ed.). Oxford: Oxford University Press.Find this resource:

Raz J. (1986). The morality of freedom. Oxford: Clarendon.Find this resource:

Regan, D. H. (1983). Paternalism, freedom, identity, and commitment. In R. Sartorius (Ed.), Paternalism (pp. 113–138) Minneapolis: University of Minnesota Press.Find this resource:

Shafer-Landau, R. (2005). Liberalism and paternalism. Legal Theory, 11, 160–191.Find this resource:

Scoccia, D. (2008). In defence of hard paternalism. Law and Philosophy, 27(4), 351–381.Find this resource:

Scoccia, D. (2013). The right to autonomy and the justification of hard paternalism. In C. Coons & M. Weber (Eds.), Paternalism: Theory and practice (pp. 74–92). Cambridge, U.K.: Cambridge University Press.Find this resource:

Shiffrin, S. V. (2000). Paternalism, unconscionability doctrine, and accommodation. Philosophy & Public Affairs, 29(3), 205–250.Find this resource:

Snelling, P. C. (2014). What’s wrong with tombstoning and what does this tell us about responsibility for health? Public Health Ethics, 7(2), 144–157.Find this resource:

Thaler, R. H., & Sunstein, C. R. (2008). Nudge: Improving decisions about health, wealth, and happiness. New Haven, CT: Yale University Press.Find this resource:

Thomson, J. J. (1990). Realm of rights. Cambridge, MA: Harvard University Press.Find this resource:

VanDeVeer, D. (1986). Paternalistic intervention: The moral bounds on benevolence. Princeton, NJ: Princeton University Press.Find this resource:

Voorhoeve, A. (2008). Scanlon on substantive responsibility. Journal of Political Philosophy, 16(2), 184–200.Find this resource:


(1.) Although citizens are arguably concerned with their own health they may resist state intervention because they believe that the value of smoking justifies shouldering the risk to health involved.

(2.) Infringing autonomy or an autonomy right means doing what the right disallows. The term leaves it an open question whether or not doing so is morally wrong, all things considered (Arneson, 2005, p. 264, n. 10; Thomson, 1990).

(3.) That a paternalistic act of A against B to Dworkin, just as it is to Mill and Arneson, is one that is undertaken against B’s will is suggested by his discussion of cases in which people willingly submit to coercive measures reflecting their own view or judgement of what best promotes their interests (Dworkin’s example is restrictions on hours of work. Another example, attributable to Feinberg, is a curfew rule) (Dworkin, 1971, p. 112; Feinberg, 1971, p. 123; De Marneffe, 2006, p. 73 n. 15). In such cases, Dworkin submits, the restrictions in question are not paternalistic.

(4.) Apparently, according to this condition, a policy, in order for it to be paternalistic, need not, pace the Mill-Arneson-Dworkin conception, be justified exclusively by reference to the good it does for the person’s whose liberty gets restricted. It is just that this reference is necessary to the justification of the policy in question. The import of this deviation from the standard view is unclear (besides that De Marneffe’s conception appears to allow that policies with mixed rationales [paternalistic and non-paternalistic] may count as paternalistic). The liberal strategy of reconciliation that De Marneffe aims to accommodate by his condition (d) aims to show that a given, apparently paternalistic, and seemingly sound, policy can be justified exclusively by reference to non-paternalistic reasons (De Marneffe, 2006, pp. 68, 69–70; Feinberg, 1986, p. 25). If successful, this strategy would, on any of the proposed conceptions of what a paternalistic policy is, go to show that the policy in question is non-paternalistic. See, further, the discussion regarding justification versus motivation and the liberal strategy of reconciliation.

(5.) Feinberg’s illiberal principle is in fact not unequivocally rejected by Mill’s anti-paternalistic principle quoted previously. We may take Mill’s claim that a person’s own good “is not a sufficient warrant” for the rightful exercise of power to mean that catering to the person’s own good never outweighs countervailing considerations (pertaining to the fact that the harm is one the agent incurs voluntarily) (cf. Dworkin, 1971, p. 107). If this is the correct reading then it is clear that Mill’s anti-paternalism does not reject Feinberg’s illiberal principle. However, we may take Mill to be saying that a person’s own good fails to constitute a legitimate “purpose for which power can rightfully be exercised” (emphasis added), that is, it fails to provide a reason at all for preventive measures. If this is the correct reading Mill may indeed be taken to reject the illiberal principle as formulated by Feinberg.

(6.) It is assumed here that the principle of hard or legal paternalism is mutatis mutandis applicable to interpersonal conduct although Feinberg focuses on criminal prohibitions (cf. Arneson, 2005, pp. 262–263). That the reason for interference emphasized by hard paternalism is held by this doctrine to persist when the self-regarding harm in question is of a voluntary kind or that whether or not it is voluntary is of no import for our reason to interfere is suggested by supplementary claims Feinberg makes about this doctrine (Feinberg, 1986, pp. 11–12).

(7.) Mill’s anti-paternalist principle quoted previously is not crystal clear in delineating paternalism in this way (Dworkin, 2005, p. 306). See, further, the distinctions made between welfare or interest paternalism outlined here and so-called moral paternalism and legal moralism.

(8.) On the liberal view that people have legitimate domain or control over their interests the first clause of (c), referring to interests, may, as Shiffrin points out, be seen as a prominent subset of the second clause’s “appeal to matters that legitimately lie within B’s control” (Shiffrin, 2000, p. 219, n. 24; cf. Grill, 2013, p. 34); (c) means that a full account of paternalism will depend on an account of the interests and matters legitimately within an agent’s control. Shiffrin does not provide such as account. One of the issues she does not address is the case in which B has handed over his legitimate control of a certain matter to A. Say that the hiker in the case mentioned has authorized A to decide on the kind of risks it is acceptable he incurs given his obligations to his spouse. Given that this is authorized by B this would quite clearly not count as a case of paternalism. It would constitute a form of prior commitment (Arneson, 1980, p. 471; Gert & Culver, 1976, pp. 49–50. See, however, Regan, 1983). To reflect this (c) should be followed by a parenthesis with a rider of the following kind: “where B has not delegated control over these matters to A.”

(9.) Conditions (a)–(c) may be understood either internally (or subjectively) or externally (objectively) (cf. Grill, 2009, p. 6; Groll, 2012, p. 707). That is, they might concern what A takes to be “means other than rational persuasion,” what he takes B’s will to be, and what he believes is for B’s good, or, alternatively, they might concern the actual nature of the means employed, what B’s will is, and what is for B’s good. In that (a)–(c) are arguably motivational or aiming to capture motivational features of paternalism the subjectivist reading is perhaps the most appropriate. By contrast, (d) concerns, as pointed out previously, whether or not the policy in question is justifiable by reference to B’s good or cannot be justified without reference to his good.

(10.) One might consider the objection to paternalism that it involves disrespect a separate ground for objecting to it (independent of its infringement of autonomy). However, the disrespect in question seems closely related to a distrust of the person’s capacity for self-direction. Similar questions arise with respect to other objections to paternalism, for example objections inspired by the so-called endorsement constraint according to which, roughly, people’s lives cannot be improved by components that they do not endorse or do not have an appropriate pro-attitude towards (Dworkin, 1989, 2000, pp. 268–274). Again, however, this objection seems closely related to the idea that a life valuable to the person is one that is lived from within. This is not, of course, to say that any objection to paternalism boils down to the fact (if it is a fact) that it offends in some way the value of autonomy, but autonomy may be the primary value paternalism infringes (cf. Oshana, 1998, p. 82).

(11.) Should this core of soft paternalism―to wit to give effect to the person’s authentic preference or not to give effect to those of the person’s preferences that are not authentically his or her―be understood subjectively or objectively (cf. n. 9)? It seems that it could be understood either way depending on the context of our particular interest. That is, we may focus on the motivation of the intervener holding that he acts paternalistically, or is at least paternalistically motivated, when he knocks the salt out of my hand acting on a belief that I am about to sprinkle my eggs with arsenic. However, we may also sometimes want to characterize an act in terms of whether it in fact bars us from acting on desires that we do not have (for example, to sprinkle our eggs with arsenic and to suffer the likely consequences if we go on to consume the eggs).

(12.) It seems plausible that while it is a least in part up to the individuals in a relationship to decide upon the kind of risk it is reasonable for each of them to shoulder it is not up to the couple to decide that their relationship includes permission to issue threats of this kind.

(13.) For a contrary view see Arneson (1980, pp. 475, 482–483, 485). Pace Hanna (Hanna, 2012, p. 424 n. 6) it is not clear that Arneson has abandoned this view in his 2005 piece. Although Arneson here endorses a form of hard paternalism that countenances infringements of people’s autonomy or personal sovereignty it is not obvious that he has abandoned his 1980 Millian view that autonomy protects the choice in question: “The root idea of autonomy is that in making a voluntary choice a person takes on responsibility for all the foreseeable consequences to himself that flow from his voluntary choice,” where “consequences of an act are forseeable by the agent … when at some time prior to the act he has the opportunity and power to ascertain these consequences by taking normal and reasonable means” (Arneson, 1980, pp. 475, 482–483. Emphasis in original).

(14.) The intuitive argument sketched here is related to Feinberg’s so-called argumentum ad hominem methodology where “we seek to persuade an interlocutor starting from premises she accepts or can be brought to accept by considerations that have intuitive appeal” (Arneson, 2005, p. 260 n. 3; Feinberg, 1984, pp. 16–19).

(15.) Option luck “is a matter of how deliberate and calculated gambles turn out―whether someone gains or loses through accepting an isolated risk he or she should have anticipated and might have declined” (Dworkin, 2000, p. 73). The driver refusing to fasten his or her seat belt or the biker not wearing a helmet risks a catastrophic outcome in the form of an accident not cushioned by efficient safety measures to avoid the inconvenience of driving with seat belts or to enjoy the wind in his or her hair. If he or she arrives safely he or she has good option luck; if he or she experiences an accident, he or she has bad option luck. Option luck is contrasted with “brute luck.” The latter is “a matter of how risks fall out that are not in that sense deliberate gambles” (Dworkin, 2000, p. 73). A driver or a biker who experiences an accident due to a bridge collapsing where its unsafe condition was unknown and not reasonably foreseeable to the driver or the biker is a case of bad brute luck.

(16.) Note that although the behavior of the daredevils is imprudent, even manifestly so (involving an extreme discounting of the value of future goods), this does not necessarily render it involuntary. The soft or anti-paternalist norm, as we have noted, protects the choices of the current self even in the case that they are manifestly imprudent (Arneson, 2005, pp. 265–266).

(17.) And some may even make the stronger point that once we are assured that a self-regarding harm is completely or substantively voluntary we have no reasons at all to intervene to try to avert it (cf. Dworkin, 1988, p. 126). Indeed this is the uncompromising anti-paternalist or soft paternalistic stance.

(18.) Still, they may, as Jason Hanna interestingly points out (2015), function so as to make people responsive to good reasons, in this way being non-manipulative and perhaps acceptable tools for improving people’s decisions for their own sake.

(19.) “Liberty-limiting principles” is, of course, given what has been said previously slightly misleading (in that paternalism, and perhaps moral paternalism and legal moralism, is not restricted in its writ to behavior that limits a person’s liberty). However, I follow standard usage here originating from Feinberg (Feinberg, 1986, pp. xvi–xviii).