J. S. Mill on Harm Prevention

Abstract

According to J. S. Mill’s liberty principle, the only legitimate justification for restricting the freedom of competent adults is to prevent harm to others. However, this is ambiguous between two interpretations. The harm causation version (Brown, 1972) has it that only conduct that is itself harmful is liable to interference. In contrast, the general prevention of harm version (Lyons, 1979) allows interference with conduct that does not itself cause harm, such as refusals to assist others, so long as this interference prevents harm from occurring.

Mark Tunick (2024) has recently offered new arguments for the harm causation interpretation, suggesting that only this can explain Mill’s resistance to legal interference with prostitutes. This paper challenges Tunick’s arguments. First, I show that Mill does not clearly restrict interference to the proximate causes of harm. While he prefers interference to focus on the clients, rather than singling out the prostitutes, he is prepared to countenance interference with the prostitutes as well. Further, his preference for focusing on the clients is explicable, even if not required by the liberty principle.

1. Introduction

The central thesis of On Liberty is that ‘the only purpose for which power can be rightfully exercised over any [competent adult] member of a civilized community, against his will, is to prevent harm to others’ (CW18: 223).¹ Call this Mill’s Liberty Principle (henceforth: MLP).² It protects individual freedom by limiting justifications for coercive interference.³ However, there are many ambiguities and interpretive puzzles.⁴ This paper focuses on the connection between interference and harm prevention. More specifically, it concerns whether legitimate interference is limited to harmful conduct or whether MLP allows interference with harmless conduct, provided some harm is prevented.

Mill holds that ‘there is no room for entertaining’ interference ‘when a person’s conduct affects the interests of no persons besides himself’ (CW18: 276). On this basis, some interpreters (Brown, 1972; LaSelva, 1988; Tunick, 2024) think that interference should be restricted to actions that cause harm to others. Let us call this the harm causation view (HC).

HC: A competent adult’s conduct is liable to coercive interference iff that conduct causes harm to others.

Some finessing is required here. Mill allows for interference where there is merely a risk of harm, rather than actual harm (e.g., CW18: 282, 292). Further, at least when it comes to legal interference, it may be necessary to prohibit all actions of a type, even if some tokens may be harmless. Thus, a suitably refined version of HC might permit interference with some actions that are themselves harmless, though they must at least be of a type that tends to cause harm. This explains why self-regarding actions cannot be interfered with; if an action does not affect anyone other than the agent then a fortiori it cannot harm anyone else.

However, this HC interpretation faces difficulty explaining those cases where Mill says that people can be compelled to perform positive acts, such as testifying in court or performing easy rescues (CW18: 225). Mill suggests that interference is justified in these cases because one causes harm by inaction, but the idea that one can cause harm by omission is controversial. We might ordinarily say that an agent who does nothing merely allows harm to occur. In light of such cases, some interpreters (Lyons, 1979, 1982; Threet, 2018, p. 544; Waldron, 2007, p. 18) hold that MLP permits interference with actions that do not themselves cause harm, provided that this interference somehow prevents harm. Let us call this the general prevention of harm reading (GPH).

GPH: A competent adult’s conduct is liable to coercive interference iff that interference prevents harm to others.

On this interpretation, the harm that is prevented need not be caused by the agent or action interfered with. Thus, interference is permissible with conduct (actions or omissions) that causes or fails to prevent harm (Lyons, 1982, p. 56). This would allow the enforcement of duties to rescue, because this saves people from harms, such as drowning, that would occur were they not rescued.

While this interpretive dispute goes back to the 1970s, Mark Tunick (2024) has recently offered new arguments in favour of HC. According to Tunick’s interpretation of Mill, it is illegitimate to interfere with prostitutes, rather than their clients. Tunick holds that this restriction is only explicable if MLP restricts interference to the proximate causes of harm. However, I find his interpretation unpersuasive on both counts. First, Mill does not clearly limit interference to proximate causes in either of the cases that Tunick discusses. Rather, in these passages and elsewhere, he seems willing to countenance interference with non-proximate causes. Second, while it is true that Mill sometimes prefers interference to target the proximate causes, that does not commit him to HC. This preference is compatible with GPH. Thus, Tunick’s case for HC is inconclusive.

2. Harms and Rights

Before addressing the causation of harm, it will be helpful to clarify what Mill means by harm. Tunick (2024, p. 6) considers harm to be a moralised notion. That is, he takes harm to consist not simply in bad consequences, but to involve violation of some right.⁵ Tunick is by no means the only interpreter to read Mill this way (e.g., Brink, 1992, p. 85; Fuchs, 2006, p. 150), but I believe this is a mistake.⁶ To be sure, Mill does consider rights violations to constitute harms of the sort that might justify interference. However, this does not commit him to the view that all harms involve rights violations. A rights violation, in my view, is sufficient but not necessary for harm.

One passage that Tunick (2024, pp. 5–6) cites is that where Mill says that unsuccessful competitors have no right to be protected from loss (CW18: 293). Tunick takes this to indicate that there is no harm here.⁷ However, I believe this misconstrues the purpose of these remarks. Mill introduces this case to warn against the mistaken assumption that ‘because damage, or probability of damage, to the interests of others, can alone justify the interference of society, that therefore it always does justify such interference’ (CW18: 292). Mill’s point is that harm, while necessary, is not sufficient, to justify interference. This requires that losses from competition are harms. Mill does not say that society is forbidden from interfering here by MLP, but only that it is not ‘called on to interfere’ because it is ‘better for the general interest of mankind’ not to (CW18: 293). I take this to be a matter of expediency, rather than principle.⁸ In other words, I consider the case for free competition to be like that for free trade, which Mill explicitly distinguishes from MLP (CW18: 293).

Tunick (2024, p. 6, n. 10) also references Mill’s remark that the conduct which each can be bound to observe ‘consists first, in not injuring the interests of one another: or rather certain interests, which […] ought to be considered as rights’ (CW18: 276, emphasis added). However, Tunick’s quotation omits the ‘first’, thereby giving the impression that this is the whole of Mill’s view. In fact, it is only the beginning of a list.⁹ The passage quoted continues, adding ‘secondly, in each person’s bearing his share […] of the labours and sacrifices incurred for defending the society or its members from injury’ (CW18: 276). It is this second provision, rather than the first, that seems to justify requirements to testify in court or to perform easy rescues. So, there is no need to subsume these cases under rights-violations.

Further, Mill goes on to add that ‘an individual may be hurtful to others, or wanting in due consideration for their welfare, without going the length of violating any of their constituted rights. The offender may then be justly punished by opinion, though not by law’ (CW18: 276). Ironically, this passage has sometimes been cited to support the view that MLP only concerns rights-violations (Kogelmann and Carroll, 2024, p. 3). The reasoning seems to be that, if law cannot interfere with these hurts, then they cannot constitute harms (or, at least, not the kind of harm relevant to MLP). But, while some modern interpreters are concerned only or primarily with legal interference, this is not Mill’s only concern. MLP is introduced to govern ‘compulsion and control, whether […] in the form of legal penalties, or the moral coercion of public opinion’ (CW18: 223).¹⁰ Hence, saying that certain things are not appropriate for legal interference does not imply that they are not liable to other forms of social interference, much less that they are not harms at all. If people can be punished for these hurts, even only by opinion (or other non-legal forms of social control), this means that these hurts must constitute harm in the relevant sense.

3. Prostitution

So far, I have suggested that Tunick misinterprets Mill’s notion of harm, but this need not undermine his arguments about harm causation, to which I now turn. Tunick’s first example is drawn from evidence that Mill gave to an 1870 Royal Commission, concerning the Contagious Diseases Acts of 1866 and 1869 (CW21: 349–71).¹¹ These acts sought to limit the spread of sexually transmitted diseases by placing restrictions on prostitutes. Mill objects that the measures involved are ‘opposed to one of the greatest principles of legislation, the security of personal liberty’ (CW21: 351). He is particularly critical of the fact that the interference licensed by these acts targets the female prostitutes, rather than the male clients, who are the ones directly culpable for transmitting infection to innocent, non-consenting parties (CW21: 354, 362). Tunick (2024, pp. 7–9) contends that Mill’s objection to punishing the prostitutes, rather than clients, makes sense only if MLP is interpreted as HC.

However, while Mill clearly objects to punishing the prostitutes rather than the clients, he does not rule out punishing both (as he should, if only the proximate cause of harm can be punished). Mill suggests that, if anyone is to be subjected to personal examination, ‘it should be applied to men as well as women, or if not to both, rather to men than to women’ (CW21: 356, emphases added).¹² He repeats this point later: ‘the woman should not be singled out to be subject to examination, but the men should be subjected to it also, or even if the women were not subjected the men might be, but if the one is, certainly I should say both’ (CW21: 363). These passages do not rule out examining the women, as one might expect if – as Tunick suggests – interference ought to be limited to the proximate cause of harm. What they object to is only singling out the women, while doing nothing to the men. In fact, since they allow that prostitutes can be subjected to examinations, so long as clients are too, they appear to contradict Tunick’s claim that interference should be limited to the proximate causes of harm.

One might still respond that GPH cannot explain Mill’s preference for interfering with the clients, rather than the prostitutes. This might be true. But this is not something that needs to be explained by MLP itself. While GPH would equally permit interference with either party, we need not think it a matter of indifference who is interfered with. Once the MLP test is passed, the form that interference should take is determined by other considerations. In deciding where to interfere, and with whom, we should be sensitive to factors such as the costs of the interference itself and its expressive effects (CW18: 225). These considerations can explain Mill’s preference for interfering with clients, even if the prostitutes are also liable to interference. Moreover, there is evidence that such considerations influenced Mill’s comments. For instance, one reason he prefers measures targeting the male clients is because he regards personal examinations as ‘exceedingly degrading to the women’ but ‘not in the same degree to men’ (CW21: 356). Further, Mill worries that state examination of prostitutes, like licensing, might be seen as officially condoning prostitution (CW21: 357). In contrast, imposing a penalty on soldiers or sailors who are found to be diseased would clearly express disapprobation of those who visit prostitutes (CW21: 360). Thus, Mill’s preference for targeting clients, instead of prostitutes, is fully explicable, even if both are liable to interference.

4. Other Cases of Non-Proximate Causation

Tunick’s second example concerns freedom of expression. Again, this is intended to demonstrate that interference is limited to the proximate causes of harm. Tunick (2024, p. 11) suggests that one should be permitted to publish dangerous opinions, even ones that might encourage harmful acts such as tyrannicide or lynching, because it is generally the perpetrators – not the speaker – who should be held morally responsible for any resulting harm. This restriction on interference supports HC since, according to GPH, it would be permissible to interfere with the speakers, if that was an effective way to prevent harm.

Once more though, careful attention to Mill’s text does not entirely support Tunick’s reading. For sure, Mill does think that expression – even encouragement to immoral acts – should generally be permitted. But, as in the previous case, he does not clearly limit interference to the proximate causes of harm. In the footnote near the start of chapter two of On Liberty, Mill says there ‘ought to exist the fullest liberty of professing and discussing, as a matter of ethical conviction, any doctrine’, even the morality of tyrannicide (CW18: 228n). The reference to ethical conviction here presumably limits this ‘fullest liberty’ to cases of mere discussion, rather than action. However, Mill goes on to add that ‘the instigation […] may be a proper subject of punishment, but only if an overt act has followed, and at least a probable connexion can be established’ (CW18: 228n). Thus, punishment is not limited to the perpetrators of the act alone; it can also include the instigators, where they are a (non-proximate) part of the causal chain.¹³ Again, this contradicts Tunick’s claim that interference should be limited to proximate causes only.

So, Mill does not clearly restrict interference to the proximate causes of harm in either of Tunick’s examples. Moreover, there are other cases, in On Liberty and elsewhere, that seem to involve interference with those who are not proximate causes of harms. For instance, Mill suggests that ‘a soldier or a policeman should be punished for being drunk on duty’ (CW18: 282). Their drunkenness might allow others to commit crimes without impediment, but it is presumably the criminals who are proximately responsible here. Further, in an 1868 letter [#1361] to James Beal, Mill writes that:

[T]here sh[oul]d be a great increase of efforts to root out the receivers of stolen goods […] without them a criminal class, as a class, could not exist. If there were no receivers there could be no professional or habitual thieves […]. I am not in a condition to say what means should be adopted for making receivers of stolen goods more amenable to justice […] but I am satisfied that this is a direction in which the law requires either to be strengthened or to be more vigorously enforced. (CW16: 1523–24).

Maybe Tunick could show that these cases do somehow involve proximate harms, as he argues in other cases (Tunick, 2024, pp. 12–13).¹⁴ But, at first sight, these cases involve interference with non-proximate causes.

Even if Mill did generally favour interference targeting the proximate causes of harm, as Tunick claims, this is not strong evidence for HC. It might be that there are reasons to focus on the proximate causes, even when others are also liable to interference. In contrast, evidence that Mill would, even sometimes, allow interference with non-proximate causes refutes Tunick’s argument that interference must be limited to the proximate causes of harm. However, this does not necessarily support GPH. Even if interference is not restricted to proximate causes of harm, it might still be limited to those whose conduct causally contributes to harm in a more extended sense. For instance, LaSelva (1988, p. 494) suggests that one need only be ‘part of the sequence of events that results in harm’ in order to be liable to interference or punishment. Unlike Tunick’s proximate version of HC, this extended HC would permit interference with prostitutes and instigators.

5. Conclusion

Given Mill’s looseness of language, and the various formulations of MLP that he offers, it is difficult to find any interpretation that is consistent with everything that he says, even in On Liberty.¹⁵ Nonetheless, Tunick’s recent arguments for HC conflict with various things that Mill says, even in the passages that Tunick cites for support. While it is true that Mill often favours interference targeting the proximate causes of harm, and therefore criticises the Contagious Diseases Acts for focusing on the prostitutes rather than the clients, he does not say that interference must be restricted only to the clients. In fact, he seems to allow the possibility of interfering with both clients and prostitutes (and, similarly, with instigators as well as the perpetrators of harmful acts). Again, this observation need not support GPH, since it is still consistent with thinking that interference should be restricted to those who are part of the extended causal chain producing harm. Nonetheless, I suggest that Mill did not hold that interference must be limited to the proximate cause of harm.¹⁶