Fundamentals

Liberalism vs. Paternalism: Autonomy, Harm, and the Limits of Power

By Daniel Sardá · Published on September 10, 2026

7 min read1,500 words

The disagreement between liberalism and paternalism is not about denying risks; it is about who may impose a response, and within what limits.

A policy can arise from sincere concern and still exceed the legitimate limits of power. That is the tension liberalism identifies in paternalism: it asks not only whether an individual decision is prudent, but who should have the authority to override it.

In the public-policy sense, paternalism occurs when someone’s choice is interfered with in order to protect that person’s own welfare, even when they do not agree with the intervention. The debate is therefore not about whether suffering should be avoided or whether everyone chooses well. It is about whether a competent adult may be compelled to live according to the judgment of officials, experts, or majorities.

The liberal tradition does not offer absolute immunity from every rule. It begins with a presumption of liberty: people should be able to direct their private lives and bear the consequences of their choices unless a sufficiently strong public reason justifies restricting them. That presumption is among the core commitments of classical liberalism, and it explains liberalism’s distrust of power presented as protection.

Key idea: The fact that a choice appears unwise does not, by itself, show that the state should prevent it.

The starting point: autonomy, not infallibility

Defending autonomy does not mean claiming that individuals are always rational, informed, or farsighted. It recognizes something different: each person occupies a special position in deciding which ends are worth pursuing in their own life. They may make mistakes, change their mind, or take risks that others would not take. None of that automatically makes their choice an administrative matter.

In On Liberty, John Stuart Mill offered an influential account of this limit: for adults with mature faculties, coercion is characteristically justified to prevent harm to other people, not to preserve someone from themselves. This is neither a formula that settles every case nor an identical expression of every liberal current, but it supplies a decisive question: does the intervention protect an identifiable third party, or does it replace the judgment of the person who chiefly bears the risk?

A simple example helps. Someone may spend time and money on an activity their family considers unwise. If the harm falls mainly on that person, others’ disapproval is not enough to prohibit it. But if their conduct concretely exposes another person to an unconsented risk, the analysis changes: the issue is no longer only the agent’s welfare, but possible harm to someone else.

This distinction avoids two common shortcuts. The first is calling every regulation paternalistic. The second is calling every social consequence that is inconvenient, costly, or contrary to a majority preference “harm to others.” Neither regulation nor an externality should be assumed: it must be shown which right, interest, or person is affected, and how.

Self-regarding risk and harm to others are not the same

The distinction is central because the kind of justification changes. A paternalistic prohibition seeks to prevent someone from harming themselves; a rule against fraud, violence, or pollution seeks to protect people who bear consequences they did not choose. Both rationales may coexist in one policy, but they are not interchangeable.

Preventing a business from withholding decisive information from customers, for example, is not primarily a way to save them from their own preferences. It protects their ability to consent on an informed basis and reduces deception. Likewise, an activity whose risk can be shown to be shifted onto neighbours or passers-by raises a question of responsibility to third parties, not merely of self-care.

Useful distinction: Liberalism does not require every decision to be correct; it requires coercion to rest on a reason other than an official preference for a “better” life.

References to collective costs deserve particular caution. Sometimes private conduct does create real effects for others; sometimes the phrase merely conceals moral disapproval or a highly diffuse burden. Before restricting a choice, it is worth specifying the causal connection, who is affected, and why a less intrusive measure would not suffice. Invoking general welfare without that examination leaves too much room for any governmental preference to become a command.

When a choice ceases to be fully autonomous

The liberal presumption protects people’s own decisions, not fraud or coercion disguised as choice. For a decision to merit respect, it matters that it be sufficiently voluntary, informed, and made by someone able to understand what is relevant.

This is where the terms soft paternalism and hard paternalism are often used. The former is, in principle, limited to checking whether there was genuine consent: it may intervene in cases of deception, coercion, ignorance of a decisive fact, or incapacity. The latter restricts a choice even when the person is thought to understand what they are doing and to decide freely, because an authority judges the outcome to be bad for them.

The boundary is not always sharp. Information can be complex, pressure difficult to detect, and some people may be in vulnerable circumstances. But acknowledging these difficulties does not authorize treating all adults as incapable by default. The response should fit the problem: requiring clear information is not the same as cancelling an option; investigating fraud is not the same as imposing a life plan.

Moreover, capacity should not become a label available for silencing unpopular decisions. A free society makes room for choices an observer disapproves of. Private life includes precisely that sphere in which a diversity of ends and ways of life need not seek political permission.

Prohibiting, informing, and influencing: different degrees of intervention

Not all public measures affect liberty with the same intensity. A prohibition removes an option and may be backed by sanctions. An information campaign adds material for deliberation. A default setting organizes a choice in a way that may steer conduct while preserving an exit for those who wish to depart from it.

This latter possibility is known as libertarian paternalism, a term associated with Richard H. Thaler and Cass R. Sunstein. Their proposal seeks to influence decisions through choice architecture without formally removing alternatives. A form that presents a preselected option but allows it to be changed illustrates the idea.

The term is contested for good reasons. Keeping an alternative available does not mean the influence is irrelevant. A design can exploit inertia, hide the exit, or make it unnecessarily costly. It is therefore not enough to say that no prohibition is involved: the mechanism’s transparency, the genuine ease of rejecting the default, and who decides which conduct to steer also matter.

Caution: A nudge may be less restrictive than a prohibition, but preserving formal options does not exempt it from scrutiny.

From a liberal perspective, providing comprehensible information and combating deception will often respect agency more than replacing a person’s decision. Yet official information is not always neutral, and recommendations are not free from bias. The institutional question returns: what authority is granted, what controls govern it, and how may its use be challenged?

A standard for evaluating public intervention

The contrast between liberalism and paternalism is not settled by asking whether a measure pursues a good end. Almost any restriction can be presented as protection. The more demanding test is to examine the means, the evidence, and the power normalized by applying it.

When an intervention limits a private choice, it is worth asking:

• Is there concrete harm to others, fraud, coercion, or a decisive lack of information—or only a risk the person assumes for themselves?
• Can the person decide with sufficient understanding and freedom?
• Is the measure necessary and proportionate to the identified problem?
• Is there a less restrictive alternative, such as information, transparency, or liability for harm caused?
• Who oversees the authority, and what real opportunity does the citizen have to challenge the decision?

These questions do not yield an automatic answer. They do prevent moral disagreement from becoming a license to command. They also fit the idea of limited political power: public authority may protect rights and address demonstrable harms, but it must justify with particular rigor its entry into decisions that belong to the person.

Liberty as a rule of responsibility

Liberalism’s response to paternalism does not idealize human conduct or promise that every choice will turn out well. Its contribution is more modest: a responsible life requires room to decide, learn, correct oneself, and assume one’s own risks. Where there is deception, incapacity, or identifiable harm to others, there may be grounds for intervention; where there is only a choice an authority considers mistaken, the presumption should favour the person who lives with its consequences.

The limit matters not because welfare is irrelevant, but because granting power the authority to define it for everyone carries an institutional cost. Keeping that difference visible protects autonomy and requires public protection, when necessary, to remain limited, intelligible, and reviewable.

Sources consulted

• John Stuart Mill, On Liberty, ch. I.
• Stanford Encyclopedia of Philosophy, “Paternalism”.
• Richard H. Thaler and Cass R. Sunstein, “Libertarian Paternalism” (2003).