Ten years ago, it seemed that paternalism was among the major problems with which citizens of liberal states had to deal. These times are gone. Still, the debate about paternalism and its limits reveals basic dimensions of what it means to value autonomy, freedom, equality and respect for persons.

1 Liberalism as Anti-Paternalism

Individual autonomy is an idea that is generally understood to refer not only to the capacity but also to the claim “to be one’s own person, to live one’s life according to reasons and motives that are taken as one’s own and not the product of manipulative or distorting external forces” (Christman 2020, Intr.). Liberalism as a political, moral and legal theory is centered on the notion of an egalitarian normative individualism. Normative individualism protects the conditions of autonomy as a set of individual rights expressive of one’s sovereignty over oneself and is an egalitarian notion, as everybody has these rights equally. Historically, the liberal tradition in legal and political philosophy starts with a strong anti-paternalist momentum, turning against the dominant, extremely paternalistic and perfectionist legal and political theory in 17th century central Europe, especially the perfectionist enlightened absolutism upheld by Christian Wolff and his successors (Gutmann 2005). Liberalism’s beginnings show that a liberalism of rights is necessarily a non-perfectionist concept and that it had to be an anti-paternalistic concept from the start.

According to Immanuel Kant, a rights-based liberal, the individual’s right to develop and act on one’s own conception of the good is essential to one’s dignity as a person. To override this right was to treat the individual with ultimate disrespect:

“A government might be established on the principle of benevolence towards the people, like that of a father towards his children. Under such a paternal government (imperium paternale), the subjects, as immature children who cannot distinguish what is truly useful or harmful to themselves, would be obliged to behave purely passively and to rely upon the judgment of the head of state as to how they ought to be happy, and upon his kindness in willing their happiness at all. Such a government is the greatest conceivable despotism, i.e. a constitution which suspends the entire freedom of its subjects, who thenceforth have no right whatsoever.” (Kant 2004 [1793], 74).

Isaiah Berlin does not relate to Mill’s strong but utilitarian liberalism and anti-paternalism (Mill 1991 [1859]), but rather to a Kantian notion of entitlement and recognition:

“Paternalism is despotic, not because it is more oppressive than naked, brutal, unenlightened tyranny, nor merely because it ignores the transcendental reason embodied in me, but because it is an insult to my conception of myself as a human being, determined to make my own life in accordance with my own (not necessarily rational or benevolent) purposes, and, above all, entitled to be recognized as such by others. For if I am not so recognized, then I may fail to recognize, I may doubt, my own claim to be a fully independent human being.” (Berlin 2002, 203).

One can argue whether this “classic” critique covers the entire spectrum of paternalistic interventions, including every form of a benevolent structuring of choice environments (as in so called libertarian paternalism, see e.g. Sunstein 2016 vs. White 2013); Kant’s and Berlin’s conceptions certainly referred to forms of “hard” legal paternalism. Insofar, anti-paternalism is rooted in a theory of individual (moral and legal) rights, and the very idea of individual rights is anti-paternalistic. As the German Constitutional Court put this in 2011: “Constitutionally protected individual liberty rights incorporate the right to make use of them in a way which – in the eyes of others – is detrimental to the right-holders’ best interests.” (Bundesverfassungsgericht 2011: n. 48).

2 Concepts of Autonomy

Within the theory of paternalism (cf. Gutmann 2014; Fateh-Moghadam and Gutmann 2014; Grill and Hanna 2020), there are strong reasons to draw one important distinction more seriously: the differentiation between (a) “personal autonomy” designating a family of ideal, gradual and hence at least moderately perfectionist conceptions about what an autonomous person is and (b) “autonomy as right” (Feinberg 1986, 47). These are not different dimensions of one concept but independent concepts with disparate, even conflicting normative functions and should be kept clearly separate from each other. Autonomy—the question of whether a person is autonomous or acts autonomously or has the authority to act as she likes—has different functions in particular normative settings. One main notion of autonomy, personal autonomy, which is a concept of what it is to lead a good, valuable and flourishing life, appears in different conceptions (cf. Gutmann 2021). In this perspective, an autonomous person is authentic in the sense that she is able to reflectively endorse (or at least not be alienated from) her basic organizing values, motivations and commitments in the light of her diachronic practical identity, i.e. the historical processes that have given rise to these characteristics (Christman 2009, 133 et seq., 143). An autonomous person can evaluate the formation of her desires in a process of sustained critical reflection (Christman 2009, 152), exercising the repertory of skills necessary for autonomous self-discovery and self-definition (Meyers 1989, 49). An autonomous person in this sense is able to fully develop the affective attitudes—attitudes of self-respect, self-trust and self-esteem—necessary for a view of herself as the legitimate source of reasons for acting (Mackenzie 2008, 524-7). As an ideal, the notion of autonomy used to qualify the structure of a person’s life belongs to the realm of value-centered reasons (Benn 1988, 10). In almost all the varieties of its philosophical use, such a substantial concept of personal autonomy designates an ideal that real persons can, at best, only approximate. Autonomy in this sense is always a matter of degree—a person can be more or less autonomous (Mackenzie and Stoljar 2000, 18). As a goal to be attained, “it may well be enjoyed by very few if any individuals” (Christman 2020, 1.1.). In this development-orientated and thus broadly perfectionist sense, however, autonomy cannot be an essential condition for guaranteeing persons areas of liberty for self-responsible decision-making.

The other basic function of autonomy is related to a person’s claim (her moral or legal right) to be respected as an agent entitled to choose her own path and follow it. “Autonomy as a right” is what is meant when the “principle of respect for autonomy” is invoked in the way Beauchamp and Childress use the term in their Principles of Biomedical Ethics in regard to “individuals’ decision-making”: “To respect autonomous agents is to acknowledge their right to hold views, to make choices, and to take actions based on their values and beliefs” (Beauchamp and Childress 2013, 101 et seq., 106; Beauchamp 2005). A right is a legally (or morally) “respected choice” (Hart 1982, 189). Autonomy in this sense of having a set of rights expressive of one’s sovereignty over oneself, i.e. of having a protected space in which one may choose freely between options for action (cf. von Savigny 1867, 6; Feinberg 1986, 28; Wellman 1985, 95 and Steiner 1994, 90) is necessarily a formal notion. This realm of rights (Jarvis Thomson 1990) to self-determination is a central feature of our concept of normativity and deeply embedded in our moral and legal practices. Rights to self-determination and personal liberties are not only preconditions for self-determination but its main form of realization. People become agents of their lives by making choices that are protected and respected. In a sense, respect for persons may simply be respect for their rights and their capacity to assert claims (Feinberg 1970). In this respect, rights-based liberalism has to be a structurally anti-paternalist concept because paternalism undermines the very concept of persons living their lives by making decisions within their guaranteed spheres of liberty.

3 The argument from justice

3.1 The argument

The conception of the autonomous person, however, also serves as the model of the person whose perspective is used to formulate and justify political principles, as in social contract models of principles of justice. And here we might encounter a problem because, as for all policies, paternalistic and anti-paternalistic policies have distributive effects of their own. Accordingly, paternalism might be a matter of distributive fairness. Richard J. Arneson was the first to express this argument clearly:

“[I]t is easy to see that imputing to individuals a strict lexical preference never to suffer paternalistic interference […] will predictably work to the advantage of the haves and to the disadvantage of the have-nots. Good choosers will predictably fare better under a regime of soft anti-paternalism than bad choosers, because the imposed value of sovereignty will be more likely to constrain the state or other would-be interferers from carrying out paternalistic acts that really will work to the benefit of the latter. The distributive dimension of the paternalism issue, hitherto largely ignored, is significant. Once acknowledged, it cuts against the advocacy of any form of strict anti-paternalism.” (Arneson 1989a, 437).

Joel Anderson’s critique is similar to Arneson’s critical analysis that the focus on autonomy as equal rights fails to protect less-competent groups in society because

“the central fact about individual autonomy is how limited the competence of many individuals is and how vulnerable groups are significantly disadvantaged by arrangements in which benefits are distributed in part on the basis of the autonomy-competence that individuals develop ‘naturally.’” (Anderson 2014, 365).

3.2 Critique

There are three options to refute this claim:

  1. a)

    The first reply to Arneson’s claim is that the personal ability to make use of individual spaces of protected autonomy is not part of the realm of distributive justice at all. In other words, we would then have a mission creep of justice talk. There is nothing to be done about the fact that, even in a just society, there always will be good choosers and bad choosers. Insofar as Arneson himself advocates a luck-egalitarian concept of equal opportunity for welfare (Arneson 1989b), he should be committed to this position. Arneson’s argument from justice is self-contradictory.

  2. b)

    The second option is simply to stick to Rawls’s thesis that we have (and should have) a “higher-order interest” in legal safeguards enabling us to advance our determinate conception of the good or rational plan of life (Rawls 1999, 491, 493). So we have (and should have) a strict lexical preference not to suffer paternalistic interference (at least not in critical life decisions). This means that following the principle of respecting persons will not enable us to maximize the well-being of bad choosers. But we are willing to pay this price, as there is a structural trade-off between respecting persons and maximizing utility anyway (and this seems to be what is at issue in the whole family of liberal theories of justice starting with Rawls, cf. Rawls 1999, xi): “Justice denies that the loss of freedom for some is made right by a greater good shared by others. The reasoning which balances the gains and losses of different persons as if they were one person is excluded. Therefore in a just society the basic liberties are taken for granted and the rights secured by justice are not subject to political bargaining or to the calculus of social interests.” (Rawls 1999, 25). In this perspective, paternalism’s “argument from justice” is a non-starter because autonomy—understood as autonomy as a right, as a person’s claim to be respected as an agent entitled to choose her own path and follow it—cannot be sacrificed for a consequentialist notion of aggregating or maximizing autonomy as an ideal. Consequentialist notions of self-determination are inconsistent, because they negate the very idea of the separateness of persons the concept of individual self-determination is built upon (cf. Rawls 1999, 24 et seq.).

  3. c)

    Moreover, the idea of curtailing the realm of individual liberty rights by a regime of paternalistic interventions working to the benefit of “bad choosers” is not convincing, because in a pluralist society there is no inter-personal consensus of what is meant by a “good” or a “bad” choice. Furthermore, most critical life decisions depend on action reasons and values that are incommensurable (Raz 1986, 321 et seq.). It is regularly impossible to compare in value two states or two possible careers of one person. Therefore, the paternalistic state has no sound criteria to decide whether a freedom-curtailing intervention really works to the “benefit” of people. Important personal decisions cannot be made from a third person perspective. At any rate, a person’s life cannot be improved by forcing her into some act or abstinence she thinks valueless (Dworkin 2000, 269).

  4. d)

    Even given that we might have special justice-based obligations towards bad choosers, if we value autonomy at all, it would be an illegitimate policy to curtail the good choosers’ right to self-determination in order to make life easier for bad choosers. There cannot be any leveling down in autonomy rights, because this would be incompatible with the very idea of autonomous persons. So Joel Anderson’s alternative to a liberal regime of autonomy, a “solidaristic regime” that is “an opposition to policies that open up room for those with more autonomy-competencies to fare well but that leave those with less competency more vulnerable” (Anderson 2014, 363) is not justifiable.

3.3 The dialectics of protection and tutelage

Given that leveling down in autonomy rights cannot be justified, we need to look for other options to fulfill possible justice-based obligations towards bad choosers (if there are any). It might be argued that we would have to enable bad choosers to become better choosers. Joel Anderson suggests a moderately “perfectionist regime” that focuses on an expansion of the skills needed to handle the expanded choice. For perfectionists, valuing autonomy involves facilitating improvements in individuals’ autonomy-competency in general. From the perspective of autonomy rights, there is nothing wrong with helping people to become better choosers, e.g. by offering better education. As human beings can only become autonomous in any meaningful sense through socialization (Habermas 1992), educating people per se is neither perfectionist nor paternalistic (although certain forms of education can be justified paternalistically as well as based on perfectionist rationales). And there seems nothing wrong with the idea of a regime of autonomy that would focus on optimizing the context of decision-making for everyone, the resources available and the opportunities for strengthening one’s appreciation for the reasons that one has. A good point can be made that such measures, which do not add up to hard paternalism against the good choosers (or anyone), are a matter of social justice. Moreover, if Joseph Raz is right in claiming that a really autonomous choice must be one between objectively good, meaningful options (Raz 1986, 369 et seq.), there is a link between an adequate theory of autonomy and a theory of well-ordered (or just) societies. So a lot can be done to create better decision environments for bad choosers (and for better choosers as well). But what do we do in the meantime, as long as we still have a lot of bad choosers? Anderson suggests that, in order to protect basically competent, but vulnerable persons against their own bad choices, they “have to be licensed to make certain moves in the practice” (Anderson 2014, 358). People will be allowed to make choices after they have passed an adequate specific competency test and have obtained their license.

For a society based on equal freedom, this argument is self-refuting. Autonomy-enhancing approaches like Anderson’s blend out the dialectics of protection and tutelage. If there is a ‘they’ who need to be taught how to make choices, there is a ‘we’ that teaches them. So if we are to take decision-making perfectionism for the (adult) have-nots seriously, we will get a class society divided along autonomy lines, with autonomy alphas and betas, or more exactly: we will have another form of premodern social organization with a stratified system of license holders having unequal access to the use of their autonomy rights. The idea of equal freedom for all will be refeudalized; autonomy will be cut back from an equal status for all (basically competent) persons to a system of graded privileges. In a Habermasian sense, the autonomy perfectionism as conceived by Anderson, if taken seriously, undermines the precondition of a society of equals.

As Anderson concedes, the “predicate ‘autonomous’ is a powerful element within the social practices in which statuses, entitlements, immunities, liberties, and the like are attributed, withheld, and contested” (Anderson 2014, 356). In an autonomy-orientated society, classification as incompetent or as still not competent enough means fundamental exclusion. By contrast, having the license to make use of your rights must be the default position for adults. In a regime with liberal characteristics, full decision-making authority (and responsibility) is accorded on the basis of meeting a low threshold of competence. It is only through such a low-threshold concept of autonomy that the fundamentally egalitarian as well as the inclusive character of the concept (or regime) of autonomy can be ensured. It is clear from the start that not everybody will maximize her well-being this way. The low threshold requirement, however, is necessary for reasons of equality because a system of individual rights has to be inclusive and must apply to at least almost every adult. In order to achieve normative inclusion and equality, the ontic status of autonomy as a right—having rights to self-determination and the authority to use them—must include at least most adults. Ascribing rights to persons (whether liberty rights or rights to democratic participation) has to be an extremely inclusive concept. In this perspective, autonomy has to define a default position. It is the very sense of ascribing rights within a liberal society based on rights (or in a notion of morality based on equal respect) for this to be as general a concept as possible. Otherwise, it would be impossible to treat persons “as equals” or “with equal concern and respect” (Dworkin 1977, 181 et seq., 273). This is the dialectics of protection and tutelage. Protecting people from making wrong choices by invalidating their choices is incompatible with recognizing them as equals. This seems to be strongest refutation of paternalism’s “argument from justice”.