Paternalism and Liberty/Autonomy as Dialectically Related Concepts

Abstract

Following a seminal definition by Gerald Dworkin, paternalism comprises interventions which interfere with the liberty or autonomy of a person, lack the concerned person’s consent and are conducted with benevolent intention. In a paper titled “The Concept of Paternalism” Dominik Düber advises abstaining from applying the notions of liberty and autonomy in a conceptual analysis of paternalism in order to avoid conceptual confusion and moral preconception regarding paternalism. In the present paper, I will argue to the contrary that both the concept and justification of paternalism depend dialectically on the notions of liberty and autonomy, respectively. The concept of paternalism alters its structure and meaning according to these notions. Likewise, the legal and ethical evaluation of particular instances of paternalism depend on the related concepts of liberty and autonomy.

1 Introduction

Paternalism is an enduring controversial issue, especially in medical ethics (Schöne-Seifert 2009; Beauchamp and Childress 2013; Schramme 2015a) and in the context of public health policy (Conly 2014; del Savio 2015; Dieterle 2020). Moreover, and rarely noticed in the philosophical discussion, the phenomenon of paternalism is prevalent in private contexts (Kühler 2017). Recently, discussions of possible AI-paternalism have entered the field of ethics of technology (Kühler 2022; Diaz Milian and Bhattacharyya 2023). A notorious example for quite heated debates on paternalistic interventions in Germany are so-called ‘veggie days’ in public canteens. More recent examples for political and public indignation include the discussion of a possible ban of meat-based food options at fairs and folk festivals. Interference with people’s dietary choices by public or institutional authorities is frequently criticized as ‘paternalistic’, roughly meaning ‘unduly patronizing’. In public discourse, the terms ‘paternalism’ and ‘paternalistic’ are often used with a negative connotation from the start. However, neither are all instances of allegedly (liberal) paternalistic interventions in fact paternalistic (Düber 2016a), nor does paternalism imply self-evident moral or legal wrongness. In order to determine and morally evaluate instances of paternalism, one has to take into account complex conceptual and (meta-)ethical preconditions. Discussing the moral and legal status of situations with regard to paternalism requires philosophical effort, in the first place. In the words of Thomas Schramme: “[B]efore we take a closer look at the issue of justification of paternalism, it might be helpful to sort out, in more detail, what paternalism is and which forms it might take” (Schramme 2015b, 2).

In the following, I will be concerned with sorting out what paternalism is and which forms it might take. To this effect, I will scrutinize a conceptual suggestion by Dominik Düber (Düber 2015) who advises abstaining from applying the notions of liberty and autonomy in a conceptual analysis of paternalism in order to avoid conceptual confusion and moral preconception regarding paternalism. I will argue to the contrary that both the concept and justification of paternalism depend dialectically on the notions of liberty and autonomy, respectively. The concept of paternalism alters its structure and meaning according to these notions. Likewise, the legal and ethical evaluation of particular instances of paternalism depend on the related concepts of liberty and autonomy. The aim of the paper is to draw a first sketch of conceptual and normative conditions for the determination and ethical and/or legal evaluation of various instances of paternalism.

2 The Concept of Paternalism

2.1 The Definition of Paternalism

According to a seminal definition by Gerald Dworkin, paternalism comprises interventions (acts or omissions) which interfere with a person’s liberty or personal autonomy, without their explicit or implicit consent, and for their own good. The latest version of Dworkin’s definition in the Stanford Encyclopedia of Philosophy conceives paternalism as follows:

“I suggest the following conditions as an analysis of X acts paternalistic towards Y by doing (omitting) Z:

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

2. X does so without the consent of Y.

3. X does so only because X believes Z will improve the welfare of Y (where this includes preventing his welfare from diminishing), or in some way promote the interests, values, or good of Y.” (Dworkin 2020, chap. 2)¹

Following Dworkin’s definition, interference with the liberty or autonomy of the paternalized person, lack of consent by the paternalized person as well as an exclusively benevolent intention on the paternalizing agent’s part² can be considered as necessary – and together sufficient – conditions of paternalistic interventions.³ Accordingly, if a certain intervention does not meet one of the three conditions it is not paternalistic by definition.

In a paper titled “The Concept of Paternalism” (Düber 2015) Dominik Düber provides some refinements of the three conditions proposed by Dworkin.⁴ I will adopt his suggestions by and large and present them in reversed order, the reason for which will become clear in a moment.

Düber terms Dworkin’s third condition the “benevolence condition” (Düber 2015, 36). I agree with his assumption that in the above-mentioned interpretation (exclusively benevolent intention), the third condition appears too strong. Pace Dworkin, it appears sufficient that the intention on the paternalizing agent’s part is at least partly or perhaps primarily benevolent, without excluding other coexisting intentions (e.g., self-interest, indirect benefit of other persons or the realization of some intrinsic value). Although a proper instance of paternalism necessarily includes the aim to somehow benefit the paternalized person or spare them harm, this “does not exclude mixed rationales that aim at the good of the affected person and of third parties. Such an action is paternalistic insofar as it aims at the good of the paternalized person” (ibid., 37). If one took Dworkin’s suggestion for granted that the intervention has to aim exclusively at the good of the paternalized person, proper instances of paternalism would arguably be very rare. In many cases of paternalistic interventions, for instance in the context of public health, it appears plausible that “mixed rationales” regarding the motivation on the paternalizing agent’s part prevail. Moreover, Düber emphasizes that the assessment of an intervention as paternalistic cannot depend on its factual success in promoting the good of the concerned person. The paternalizing agent can certainly be mistaken about the paternalized person’s good – independent of settling the meta-ethical question whether there is some objective good for the person or their good depends on (inter-)subjective evaluation. However, if the intervention takes place on grounds of partly or primarily benevolent intention, it is to be classified as paternalistic (ibid., 36 f.).

Regarding the second condition which Düber terms the “consent condition” (ibid., 34), he proposes that the relevant notion of consent should be “actual consent” (ibid.) in contrast to hypothetical, prospective or retrospective consent. Thus, the scope of paternalistic interventions comprises “two types of cases” (ibid.):

“Cases in which an action is carried out against the known will of the paternalized person and cases in which the – known or unknown – will of the paternalized person is declared irrelevant. This reading clearly excludes cases of actual consent to a benevolent coercive measure. Thereby, we gain a clearer understanding according to which an action is paternalistic. The action is paternalistic insofar as it aims at the good of those who do not consent.” (ibid., 34 f.)

Such an understanding rules out cases in which there is explicit consent by the concerned person. If, for example, a psychologist interferes with the autonomy of a patient who gives actual consent to a certain medical procedure, say, hypnosis therapy for the sake of quitting smoking, the psychologist may have benevolent intentions towards their patient, and the patient’s autonomy may be constrained under hypnosis, still, this intervention does not count as paternalistic due to the patient’s actual consent. Assuming hypothetical or retrospective consent of a person whose autonomy is (temporarily) interfered with may be relevant for the ethical evaluation and justification of an intervention, for example, in cases in which the person’s actual will cannot be ascertained due to their medical condition and there is no valid advance directive at hand. However, it does not play a role for the conceptualization of the intervention as paternalistic. Following Düber’s interpretation, paternalistic interventions are simply carried out despite a lack of consent or against the concerned person’s actual will.

Regarding the first condition which Düber terms the “coercion condition” (ibid., 33) or “interference condition” (ibid., 34), he proposes that “the use of multifaceted terms like liberty and autonomy should be avoided, if possible. The reason for this is not only that there is a wide range of conceptions of liberty and autonomy, but also the danger of neglecting […] a morally neutral definition of paternalism” (ibid.).⁵ I fully agree that it is advisable to clarify conceptual questions in the first place, before engaging in a normative argumentation in favour or against paternalism. Despite my support for Düber’s claim regarding a morally neutral definition of paternalism, Dworkin’s first condition is where our interpretations of Dworkin’s definition part. In contrast to Düber’s assumption, I suggest that it is worthwhile to keep the concepts of liberty and autonomy in the first condition as proposed by Dworkin, because their application has conceptual implications for the very notion of paternalism.

My conceptual suggestion is as follows: Assuming that it is a necessary condition for an intervention to interfere with the liberty or autonomy of a person in order to instantiate a case of paternalism, paternalism can be considered as an ambiguous concept which alters its structure and meaning according to the interrelated concepts of liberty and autonomy, respectively.

For the sake of the following line of argument, it is worthwhile to start with a rough differentiation between a (legal) binary concept of liberty and an (ethical) gradual concept of personal autonomy. Following Bijan Fateh-Moghadam and Thomas Gutmann, one can differentiate between

“liberty of action (as underlying concepts of basic legal rights and moral rights) and personal autonomy (as used in many moral-philosophical conceptions). For legal and legal-philosophical theories of paternalism, it is advisable to follow up on intervention in liberty of action, which every person is fundamentally entitled to regardless of her competence.” (Fateh-Moghadam and Gutmann 2014, 388)

On the basis of this differentiation, I take liberty to be a thin concept which is frequently applied in the context of law and legal philosophy. Regarding liberty as a thin concept means I assume that the notion of liberty does not refer to substantive autonomy-warranting personal capacities. Furthermore, I assume that liberty – in terms of positive freedom of action, negative freedom from interference or liberal rights (cf. Paulo 2015) – is usually attributed to persons that exceed a certain threshold. For example, any person who reaches the age of consent is attributed legal contractual capacity and is, therefore, free to conclude legally binding contracts by default. Any person who has not yet reached the age of consent is not attributed legal contractual capacity and is, therefore, not free to conclude legally binding contracts. Accordingly, liberty (for example, to conclude legally binding contracts) is not some capacity with which persons are endued ‘by nature’, but an assigned (social and legal) “deontic status” (Anderson 2014, 358) which entitles its bearers to certain rights (and obligations): it “licenses certain further moves in the relevant practice” (ibid.). The attribution of such a status does not depend on substantive autonomy-warranting personal capacities, but only on exceeding the relevant threshold.⁶ A person who exceeds the threshold is “taken to deserve to be treated in certain ways, and these normative implications relate, for example, to rights of non-intervention” (ibid., 359). Therefore, I take liberty to be a threshold or rather binary concept.

In contrast, I take personal autonomy to be a thick concept which is frequently applied in the context of normative ethics. Regarding autonomy as a thick concept means I assume that personal autonomy is usually attributed to persons on the basis of certain substantive autonomy-warranting personal capacities.⁷ Regardless of how these capacities are spelled out in detail,⁸ I assume that they can be more or less developed. Accordingly, a person can be more or less autonomous, according to the degree they realize the relevant autonomy-warranting personal capacities. For example, young children or cognitively incapacitated persons are commonly considered to be less autonomous than average adult persons, for instance, in terms of being less competent to engage in rational (self-)reflection and make authentic decisions on the basis of their own values and beliefs regarding a good life. A person can be more or less autonomous compared to other persons and realize varying degrees of autonomy over the course of their lifetime. Therefore, I take personal autonomy to be a gradual concept.

This differentiation granted, if one presumes that liberty as a binary concept applies in the first condition of the above-mentioned definition of paternalism, an intervention which is conducted without the concerned person’s consent (second condition) and on grounds of benevolent intentions (third condition) can be either paternalistic or not depending on the attribution of liberty or the lack thereof to the concerned person. If interference with a person’s liberty is a necessary condition for a paternalistic intervention, only persons who exceed a relevant threshold and, therefore, are attributed liberty can be treated paternalistically. In this case, paternalism amounts to a threshold or rather binary concept, too.

In contrast, if one presumes that personal autonomy as a gradual concept applies in the first condition of the above-mentioned definition of paternalism, an intervention which is conducted without the concerned person’s consent (second condition) and on grounds of benevolent intentions (third condition) can be more or less paternalistic depending on the person’s respective degree of autonomy. This means, for example, that the scope of possible paternalism towards children increases with their growing autonomy.⁹ Inversely, the scope of possible paternalism towards persons suffering from dementia or otherwise (temporarily) cognitively impaired persons decreases along with their autonomous capacities. If interference with a person’s autonomy is a necessary condition for a paternalistic intervention and persons can be more or less autonomous, paternalism amounts to a gradual concept as well.¹⁰

These conceptual differentiations antecede the ethical evaluation of any particular instance of paternalism and, therefore, can very well preserve the required neutrality of the definition in moral terms.

2.2 Types of Paternalism

Dworkin (2020) provides a differentiation of several types of paternalism which I will address briefly in the following. At first, it is reasonable to differentiate between a “narrow” and a “broad” understanding of paternalism (ibid., chap. 2.2) which refers to the question of who or what is the paternalizing agent. In a narrow sense, paternalism only comprises legal or state paternalism. In a broad sense, paternalism also includes paternalistic acts by institutions, related persons in private contexts (cf. Kühler 2017) and perhaps even artificial agents (cf. Kühler 2022).

Furthermore, Dworkin differentiates between “soft” and “hard” paternalism (Dworkin 2020, chap. 2.1). Soft or autonomy-oriented paternalistic interventions aim at determining the autonomy of the concerned person’s choice or action in a given situation. As a precaution, the paternalizing agent may temporarily interfere with the paternalized person’s choice or course of action, however, once it has become clear that the respective choice or action is actually autonomous, the paternalized person is free to proceed as they wish. In contrast, hard paternalistic interventions foreclose possible courses of action for the paternalized person by simply overriding their will and autonomous decision. The paternalizing agent supposes to know better than the paternalized person what is good for them and leaves them no alternative option for action.

Another differentiation is the one between “weak” and “strong” paternalism (ibid., chap. 2.3). Weak paternalistic interventions include criticism of or interference with the means to some autonomously chosen end while refraining from being concerned with the end in question. To this effect, the paternalizing agent “interfere[s] with the means that agents choose to achieve their ends, if those means are likely to defeat those ends” (ibid.). In contrast, strong paternalistic interventions include criticism of or interference with the very ends a person autonomously chooses. In extreme cases, the paternalizing agent calls into question “the entire life plan [of the paternalized person], i.e., their concept of a good and flourishing life as a whole” (Kühler 2017, 74; own translation).

The above-mentioned types of paternalism can certainly be combined and there are blurring lines between different types of paternalism. Imagine, for example, a case of overly concerned parents who honestly want the best for their adult son and constantly criticize his lifestyle, choice of field of study, subsequent profession, and so forth, in a strong paternalistic way. Once they undertake effective measures to prevent him from, for example, taking up a particular career, they transgress the line to hard paternalism.

Dworkin provides two further conceptual differentiations. The first one is between “pure” and “impure” paternalism (Dworkin 2020, chap. 2.4). Pure paternalistic interventions aim directly at the good of the person interfered with. Impure paternalistic interventions, in contrast, interfere with a third party’s conduct in order to indirectly prevent the paternalized person(s) from harm or promote their good. An example would be legal regulation of sugar content in foods for the purpose of protecting consumers from adverse health effects.

The last conceptual differentiation is between “welfare” and “moral” paternalism (ibid., chap. 2.5). All above-mentioned types of paternalism fall into the category of welfare paternalism, since the respective aim is to promote the non-moral good of the paternalized person – roughly in the sense of their (objective) welfare and/or (subjective) well-being. Moral paternalism, in contrast, includes the (classical virtue ethical) assumption that a person’s good also comprises their moral constitution – their being a virtuous person, so to speak. Paternalistic interventions in this sense aim to make the paternalized person, their decisions and actions morally better, not for the benefit of third parties, but for the sake of their own good life. This type of paternalism is not to be confused with “legal moralism” (ibid.). The latter “does not justify [limitations on liberty] with the welfare of the party concerned, but on the grounds that certain self-harming dispositions violate socio-moral values, lead to moral disintegration in society (i.e. damage a collective good), or that they are simply intrinsically bad (malum in se)” (Fateh-Moghadam and Gutmann 2014, 385).

So much for the different types of paternalism. In the following, I will address some implications of the presented interpretation of Dworkin’s definition of paternalism for the ethical and legal evaluation of paternalistic interventions. In doing so, I will primarily be concerned with structural as against substantial presumptions for the justification of particular instances of paternalism.

3 Ethical and Legal Evaluation of Paternalism

Notwithstanding the moral neutrality of the above-mentioned definition of paternalism, different concepts of liberty and autonomy certainly have an impact on structural and substantial criteria for the legal and ethical evaluation of (alleged) paternalistic interventions.

For example, it is commonly assumed by both proponents and opponents of paternalism that soft or weak paternalistic interventions can be ethically justified, if they are only conducted in order to determine the actual state of autonomy of the concerned person (soft paternalism), aim only at suitable means for some autonomously chosen end (weak paternalism), or if the concerned person lacks the required capacities to estimate what is good for them as, for example, young children, persons suffering from dementia or otherwise (temporarily) cognitively impaired persons. However, one can question whether interference with the choices and conduct of the latter group of persons amounts to paternalism, in the first place. Fateh-Moghadam and Gutmann (2014, 389) give an example for such an (alleged) paternalistic intervention. According to their estimation, the

“assignment of legal autonomy [i.e., liberty] itself […] entails risks of paternalism: lack of this ‘basic competence’ usually means that a person is considered as not being able to state her will self-consistently or to give consent effectively, and others have to decide in her place. ‘Incapacitation’ or legal tutelage represents the most intensive form of paternalistic intervention, even when, without any doubt, it is a matter of autonomy-orientated, soft paternalism.”

In contrast, I would argue that it is dubious whether the example actually instantiates a case of (soft) paternalism. At least as far as narrow (i.e., legal or state) paternalism is concerned, strictly speaking, a person who completely lacks the “basic competence” required for the assignment of legal autonomy in terms of liberty cannot be treated paternalistically by definition, since the first necessary condition of paternalism (interference with liberty) does not apply.¹¹

It is a further question beyond the conceptual determination of the intervention as paternalistic whether such benevolent treatment without the concerned person’s consent is legally or ethically justifiable. However, if young children, persons suffering from dementia or otherwise (temporarily) cognitively impaired persons lack a sufficient degree of autonomy or fall short of a certain threshold of liberty in the above-mentioned sense, they cannot be paternalized, since the first necessary condition for paternalism (interference with liberty or autonomy) does not apply. They can be treated with benevolent intention and without their actual consent,¹² and such treatment certainly requires ethical and legal justification, however, this justification is independent of the assessment of the particular case as an instance of paternalism.

To the contrary, interference with the choices and conduct of persons who are assigned legal autonomy in terms of liberty and possess a sufficient degree of personal autonomy without their consent and for their own good clearly amounts to an instance of paternalism. Such interventions – if they are not considered plainly morally outrageous (by firm anti-paternalists) – usually require legal and/or ethical justification. If an intervention requires ethical justification, it can be considered as morally problematic. The term “morally problematic” as I use it here indicates that the intervention in question is morally relevant and, thus, deserves ethical consideration (cf. Ach and Pollmann 2017). It does not mean that the intervention is morally wrong, in the first place. Instead, particular moral problems can be considered as specific practical problems, namely ones that are appropriately treated from the viewpoint of normative ethics – in contrast to, for example, the perspective of law, religion or mere social etiquette. The evaluation of a morally problematic intervention as morally right or wrong has to be justified by ethical argumentation. If and to what extent a paternalistic intervention is morally right or wrong depends on the underlying ethical theory and the according axiological order of liberty/autonomy and welfare/well-being (as, for example, in some varieties of deontology¹³) or rather – if the theoretical basis allows for balancing values – the weighing up of liberty/autonomy against welfare/well-being (as, for example, in consequentialism¹⁴ or principlism¹⁵).

That said, I aim to argue in the following that the ethical considerability and moral relevance of instances of paternalism, i.e., whether and to what extent they pose a particular moral problem – in contrast, for example, to a (merely) legal problem – are likewise interconnected with the related concepts of autonomy and liberty, respectively.

To this effect, paternalistic interventions can, on the one hand, be considered as binary morally problematic concerning a person’s liberty (in terms of legal autonomy, as explicated above), i.e., they are of particular moral relevance or not. This conceptual relation obtains in cases of narrow, i.e., legal or state paternalism. For example, if a legal seatbelt requirement interferes with the liberty of persons for their own good and despite a lack of their actual consent, this amounts to a case of state paternalism. In this example, positive law or rather legal theory is the appropriate normative basis for the normative assessment of the paternalistic intervention. Assuming that (today) there is no need for an additional ethical consideration of a seatbelt requirement, the intervention in question poses no moral problem, only a legal one, for example, regarding proportionality (cf. Paulo 2015, 129 f.)¹⁶. If, however, for some reason legal considerations are insufficient to normatively evaluate an intervention which interferes with the liberty of persons for their own good and despite a lack of their actual consent – take as an example compulsory vaccination – and it, therefore, requires additional ethical consideration, the intervention indeed poses a moral problem, not only a legal one.¹⁷ Again, the actual evaluation of a particular paternalistic intervention depends on underlying theoretical and normative presumptions. For example, one can argue that hard state paternalism can in no way be legally (or ethically) justified, since – presuming a (deontological) rights theory – it is impossible to balance basic liberal rights of one and the same person, and interference with such rights is only justifiable in cases of infringement with the rights of third persons (cf. Paulo 2015, 139).

On the other hand, paternalistic interventions can be considered as gradually morally problematic contingent on the degree of the concerned person’s autonomy, i.e., the ethical justification of a particular intervention can be more or less demanding. Roughly speaking, the more autonomous a person is, the harder the justification for any paternalistic interference becomes. This conceptual relation obtains in cases of broad paternalism, i.e., in institutional or private contexts. The demandingness of the justification for a particular instance of paternalistic interference with personal autonomy depends on the actual degree of autonomy of the person interfered with – this assumption reflects the above-mentioned widespread intuition that (alleged) paternalistic interferences with very young children and incapacitated persons are ethically justifiable. Accordingly, the ethical evaluation has to take into account the affected person’s autonomy-warranting personal capacities¹⁸ – and the type of paternalism as explicated above.

This means that, on the one hand, minor interferences with personal autonomy, as in cases of soft or weak paternalism or in cases of interference with persons who possess comparatively low autonomous capacities, arguably pose a comparatively light burden of justification for the paternalizing agent. Again, the actual justification depends on underlying theoretical and normative presumptions. However, since the burden of justification is comparatively light, such instances might be morally exaggerated by firm anti-paternalists, so to speak.¹⁹ This misjudgement might, in turn, lead to inept moralism in the ethical evaluation of the paternalistic intervention in question (cf. Ach and Pollmann 2017, 55 ff.). For example, although my autonomous capacities are comparatively high under current circumstances, and therefore, any infringement with my autonomy has to be well-justified, it might be inept to accuse my colleague of morally wrong interference with my personal autonomy by advising me – against my conviction – to engage in social media for the sake of staying in touch with remote friends. This amounts to a case of weak paternalism which arguably is of minor moral relevance.

On the other hand, grave interferences with personal autonomy, as in the case of hard or strong paternalism or in cases of interference with sufficiently autonomous persons arguably pose a comparatively heavy burden of justification for the paternalizing agent. Again, the actual justification depends on underlying theoretical and normative presumptions. However, since the burden of justification is comparatively heavy, such instances might be morally underestimated by advocates of paternalism, so to speak. This misestimation might, in turn, lead to inept ignorance of moral significance in the ethical evaluation of the paternalistic intervention in question (cf. Ach and Pollmann 2017, 58 f.). For example, it might be seriously morally wrong to treat an elderly person whose autonomous capacities are beginning to decline as if the person were actually unable to care for their own good life. To install, for example, ambient intelligence technologies at the person’s home for the purpose of preventing accidents and facilitating their daily routines despite a lack of their consent may be well-intentioned and even actually promote the person’s welfare. However, instead of being a morally unsuspicious autonomy-enhancing means, as is often taken at face value by techno-optimists, this measure, due to its impact on the affected person’s whole conduct of life, amounts to both strong and hard paternalism and arguably is of considerable moral relevance, thus deserving a strong ethical justification.

Of course, in order to apply this matrix to particular instances of paternalistic interventions, the respective concepts of liberty and autonomy, the criteria for autonomy-warranting personal capacities and, not least, the axiological relation between the good of a person and their autonomy have to be spelled out in detail. However, my aim so far was to illustrate how the very concept of paternalism depends dialectically on the concepts of liberty and autonomy. Depending on whether the first condition of the definition of paternalism refers to a binary concept of liberty or a gradual concept of personal autonomy, the concept of paternalism alters its state as a binary or gradual concept, too. Moreover, this conceptual dialectic has an impact on structural and substantive criteria for the legal and ethical evaluation of paternalistic interventions, since, again depending on whether the first condition of the definition of paternalism refers to liberty or personal autonomy the normative framework for the evaluation of the intervention in question differs. In the following paragraph, I will draw some recapitulatory suggestions from the given examples.

4 Summary of Conceptual and Normative Conditions for the Assessment of Paternalism

In the light of the presented considerations, I propose the following conditions with regard to the conceptual and – ethical and/or legal – normative assessment of paternalism:

In accordance with the positions of Dworkin and Düber, only interventions which meet all three conditions in the above-mentioned definition of paternalism (interference with the liberty or autonomy of the paternalized person, lack of consent by the paternalized person, benevolent intention on the paternalizing agent’s part) are paternalistic. The conceptualization of a particular intervention as paternalistic does not predetermine the ethical and/or legal evaluation of the intervention in question.

Pace Düber, Dworkin’s first condition for paternalism should be maintained, because the concept of paternalism depends dialectically on the interrelated notions of liberty and personal autonomy, respectively. This means, paternalism can account for a binary concept (in cases of interference with legal liberty) or a gradual concept (in cases of interference with personal autonomy). In cases of narrow, i.e., legal or state paternalism, the first condition of the above-mentioned definition of paternalism refers to a person’s liberty (in terms of a thin concept of legal autonomy). In cases of broad paternalism in both institutional and private contexts, the first condition of the above-mentioned definition of paternalism refers to a person’s autonomy (in terms of a thick concept of individualistic or relational personal autonomy).

The justification of particular instances of narrow, i.e., legal or state paternalism depends on the assignment of liberty to the person interfered with and the presumed legal or legal-philosophical theory. The justification of particular instances of broad paternalism, i.e., paternalism in institutional or private contexts, depends on the actual degree of autonomy of the person interfered with and the presumed ethical theory.

Only persons who are assigned a legal status of liberty and/or possess a sufficient degree of personal autonomy can be paternalized by definition. Persons who lack a legal status of liberty and/or a sufficient degree of personal autonomy²⁰ can be treated in benevolent ways and without their actual consent. However, this does not imply a paternalistic intervention, since the first necessary condition for paternalistic interventions (interference with liberty or autonomy) does not apply.

A paternalistic intervention can, but does not necessarily, pose a particular moral problem which requires according ethical deliberation. The demandingness of the ethical justification of paternalistic interventions varies according to the degree of autonomy of the paternalized person and to the type of paternalism.

5 Conclusion

In the present paper, I have scrutinized a conceptual suggestion by Dominik Düber who proposes some refinements regarding Dworkin’s seminal definition of paternalism. Contrary to Düber’s assumption, I have argued that the conceptual differentiation of paternalism along the lines of liberty and autonomy does not anticipate moral appraisal of paternalism. Instead, it reveals dialectic conceptual relations between the concept of paternalism and the concepts of liberty and autonomy, respectively.

Depending on the presumption of paternalistic interventions as interfering with (legal) liberty or (ethical) autonomy, paternalism amounts to a binary concept or a gradual concept. Likewise, depending on the concepts of liberty and autonomy, the need for legal and/or ethical justification as well as the weight of the burden of justification for paternalistic interventions varies. Instances of narrow state paternalism arguably pose a legal (as opposed to a moral) problem, in the first place, which has to be considered from the standpoint of legal or legal-philosophical theory. Instances of broad paternalism pose a moral problem which has to be considered from the standpoint of ethical theory. The legal and ethical evaluation of particular instances of paternalism requires a case-by-case analysis on the basis of explicit conceptual and normative premises regarding the underlying legal and/or ethical theory, the respective concept of liberty and/or autonomy, criteria for autonomy-warranting personal capacities and assumptions about the axiological relation of the autonomy and the good of a person.

The aim of the paper was to draw a first sketch of conceptual and normative conditions for the determination and ethical and/or legal evaluation of various instances of paternalism. Discussing the moral and legal status of situations with regard to paternalism requires a great amount of philosophical effort. As the above analysis indicates, the conceptualization and justification of paternalism requires complex conceptual and (meta-)ethical considerations on different levels of argumentation. However, this is no specific of paternalism, but a regular characteristic of ethical deliberation.²¹