Welcome to our NDPR Forum on Andrew Jason Cohen’s Toleration and Freedom from Harm: Liberalism Reconceived (Routledge 2018), which was recently reviewed by Peter de Marneffe in NDPR. Please feel free to join in on the discussion, about the book, the review, or related issues. Andrew will be joining in soon.

From the book jacket: “Toleration matters to us all. It contributes both to individuals leading good lives and to societies that are simultaneously efficient and just. There are personal and social matters that would be improved by taking toleration to be a fundamental value. This book develops and defends a full account of toleration―what it is, why and when it matters, and how it should be manifested in a just society. Cohen defends a normative principle of toleration grounded in a new conception of freedom as freedom from harm. He goes on to argue that the moral limits of toleration have been reached only when freedom from harm is impinged. These arguments provide support for extensive toleration of a wide range of individual, familial, religious, cultural, and market activities.”

From de Marneffe’s review: “According to Cohen, a better formulation of [Mill’s] harm principle is this.

“HP1A*: The only purpose for which power can be rightfully exercised over an agent, against her will, is to prevent that agent from harming others or to set minimal policies, including those involving the minimal necessary taxation needed to reliably prevent all harms to others, no matter who would do the harm. (106)

“By harm Cohen means a ‘wrongful setback of interests’ (50). If a setback of interests is not wrongful, it does not count as harm.

“A predictable question is, when is a setback of interests wrongful? Cohen does not tell us. He might have said it is wrongful when it violates a person’s rights, but Cohen does not endorse this view. Early on he writes, ‘I will say very little about rights in this book’ (12). He remains true to his word.

“Another predictable question is, what constitutes a ‘setback of interests’? Cohen suggests it is a decrease in well-being, where he uses the phrase ‘one’s interest, understood as well-being’ (50). So perhaps he means by harm a wrongful reduction of well-being. If so, making this explicit would have been helpful.

“Then there is the issue of paternalism. Understood in one way, paternalism refers to the following view:

The fact that a person will be better off when his liberty is limited in some way is a good reason for the government to limit his liberty in this way, even if he opposes this policy. Although this reason might be decisively outweighed, reasons of this kind should be counted and in some circumstances will be sufficient to justify the limitation in question.

“Mill’s harm principle is a rejection of this view. It says that the fact that a person will be better off when his liberty is limited is never a good reason for the government to limit his liberty when he opposes this policy. Reasons of this kind should not be counted, and there are no circumstances in which reasons of this kind are sufficient to justify the government in limiting a person’s liberty against his will. Mill’s harm principle is therefore a principle of antipaternalism. So is Cohen’s HP1A*. What reason, then, is there to think that principles of antipaternalism like Mill’s and Cohen’s are valid?

“No reason, I would say. Cohen gives some broadly utilitarian arguments for toleration (68-82), but they don’t explain why HP1A* is valid. This is partly because they don’t explain why every alternative principle of liberty is invalid, including those that recognize and protect rights to basic liberties but allow a limited range of paternalistic policies. In Chapter Seven, Cohen briefly considers some arguments against the harm principle, but does not address any of the recent attacks on antipaternalism by Danny Scoccia, Sarah Conly, Jason Hanna, and others. “I will not extensively discuss paternalism,” he announces with notable understatement. “I don’t find any of the newer work seeking to defend it persuasive even where such work has improved discussion” (134). Why does he find this recent work unpersuasive? This, too, he leaves unexplained.”


8 Replies to “NDPR Forum on Andrew J. Cohen’s Toleration and Freedom from Harm

  1. First, let me officially say “thank you!” to David Shoemaker for the opportunity to respond and to PeaSoup for hosting this forum! Now to jump in:

    We philosophers are, of course, odd creatures in that we thrive on criticism. I am thus grateful that Professor de Marneffe took the time to read and comment on my book. That a thinker whose work I have read and appreciated over the years would read mine is gratifying. De Marneffe raises 6 worries about the book, most unfortunately accurate. Still, I’ll take this opportunity to respond. The first 2 I will be very quick with: I don’t explain what “minimal policies” are and I fail to address paternalism. I’ll then talk more about the other four: (1) my definition of harm is not clear enough, (2) I don’t address risks of harm that are not themselves harms, (3) I don’t have an account of what have been called “cumulative harms,” and (4) I don’t offer a full defense of my interpretation of the harm principle (HP1A*).

    Regarding the first 2 worries: De Marneffe is simply right that what I say about “minimal policies” leaves that notion indeterminate. I say as much on page 110. While I offer some indications of how it should be understood, it’s reasonable to want a more definitive clarity than I offer. Similarly, it’s reasonable to want someone with my sort of view to say more about paternalism. Fortunately, as I note (126), other anti-paternalists have offered significant and substantive rebuttals to paternalist thinking. My friend and sometime co-author, Bill Glod, for example, has done excellent work on this topic. I admit I was torn about whether to write more against paternalism, but as always, one must pick and choose what to delve into in a book. Neither pages nor time are free, after all. This may be a future project.

    Now for the issues I do say more about.

    1. First, a key part of the definition of harm I offer in the book is as follows: to undergo a harm is to be the subject of an event wherein one’s interests are wrongfully set back (56; the rest of the definition makes clear mine is an event-based account rather than the more common state-based account). De Marneffe does not approve of this because it leaves open what counts as wrongful and because he thinks I am unclear about what counts as a setback of interests, noting that I suggest “it is a decrease in well-being, where [I] use[] the phrase ‘one’s interest, understood as well-being’ (50). Perhaps I should have been clearer when I noted, following Feinberg (1984, 34), that “One’s ‘interest,’ understood as well-being, requires that one’s ‘interests,’ understood as those things one has an interest in, are not thwarted.” My intended use of the term “interests” in my definition is the plural term not the mass noun. That is, it is a setback to one’s interests, not one’s ‘interest’ that I am concerned with when I speak of harms. “’To have an interest’ is to ‘have a kind of stake’”—this is not the same as one’s overall interest. That said, no doubt the terms wrongfulness and setting back of interests are both indeterminate—and I fully understand that some find this unsatisfying. Nonetheless, as I say in the book,

    I believe that indeterminacy is precisely what pushes us to do the right sort of normative work that is required before determining if interference with another is permissible. Given the existence of uncontroversial tokens of harm (and lacks of harm), my Feinbergian account nicely tells us to use those to help in reasoning analogically (as one would expect in juridical situations) about cases that are controversial. It allows us to separate easy and hard cases and pushes us to do the difficult normative work in the latter before concluding that a harm is present and interference permissible (51-52).

    In short, I prefer to let determinations of harm be done in a quasi-jurisprudential manner. Moreover, of course, providing a full account of wrongfulness would require another book. Perhaps I will attempt that in the future.

    2. De Marneffe’s claim that I don’t say enough about risks of injury that are not actually harmful seems to ignore chapter 8, wherein I argue (with Hugh Lafollette) for parental licensing as a way to minimize risks of harm to children. It’s true, of course, that I don’t myself do the social scientific work that would be needed to give a full account of when interference is permissible. I say, for example,

    What we should say, I think, is that only when an activity is statistically likely to be harmful in the sense of wrongfully setting back interests, is regulation warranted. Driving a car is like that—it’s not that driving merely potentially causes significant harms, but that it does so in a statistically significant number of cases. …Endorsing HP1A* allows one to defend state policies and activity meant to prevent statistically likely harm (146).

    I add, in a footnote (158, n 24), that “I leave open how statistically likely the harm must be. I do not mean merely that the odds of a harm occurring on any given occasion of the practice in question are more than zero. They must be non-negligible. Surely, though, the odds need not be 9 to 1 either.” I also note that “less than 1% of the population [is] injured in a given year” by car accidents and that most think interference (to license drivers) is warranted there. I’m not committed to endorsing that, but I think it reasonable.

    3. De Marneffe also worries that I do not say enough about what might be called “cumulative harms.” I do note the worry (65, n 12) and point to one solution (provided by Andrew Kernohan). One simple view is this: while it’s true, as de Marneffe notes, that “the amount of pollution that any individual fire would add is not enough to constitute a wrongful reduction of anyone’s well-being,” after a certain number of these fires have been set, additional fires contribute the pollution that pushes us past a threshold to a level such that all of us would face reduced well-being due to the pollution. At that point, wrongfully lighting the fire would be wrongfully setting back people’s interests, so at that point interference would be allowed by HP1A*. Of course, one might wonder why we should not interfere before we reached that threshold. I take no stance on this matter, but recognize it may be reasonable to interfere at the earlier stage. If so, such a policy would presumably be part of the minimal policies allowed in my form of liberalism.

    4. Finally, de Marneffe is right that I don’t offer a definitive defense of HP1A*. I begin the book by noting that I will be schematic in laying out my liberalism (1). There are clearly things I could—and now suspect, should—have gone further into. That said, I should indicate that I do provide three arguments in favor of the broad view I offer, 2 of which are made explicit and 1 that is left implicit.

    The first is really a set of arguments from economics. The biggest of these is indebted to David Ricardo and supports the view that extensive toleration is valuable. “In discussing that argument, I also invoke John Stuart Mill and contemporary social science regarding the value of diversity” (12). This argument is consequentialist in nature and the one on which de Marneffe concentrates. Its presented in chapter 4, but as I note at the end of that chapter, I think we have a duty to tolerate others far more often than is typically thought—and I provide my primary argument for that in the next chapter.

    In chapter 5, I explicate a conception of “freedom understood as freedom from harm—a narrow moralized form of negative freedom” (13). Even as I signed off on the book, I was under no delusion that the argument here was going to persuade many—but I wanted to try. De Marneffe has shown me that I should have tried harder. Given more time and space, I would have more argumentation seeking to show that freedom from harm was conceptually more in line with what we value than freedom as non-domination, freedom as independence, and other forms of freedom—as well as a more robust argument showing how its value means we must protect it and how this means we must have extensive toleration (that is, tolerate in accord with HP1A*). Once again, I owe de Marneffe a debt for showing where further work on this is needed.

    Finally, there is what might be considered an argument from demonstration. The second half of the book is “meant to target certain areas [showing what should and should not be tolerated therein] so that we have the tools to consider all other possible objects” of toleration (13). If the limits of toleration I argue for are attractive, that is some reason to think the general view correct. That said, I realize my view is not in the mainstream of current political philosophy, so I am not surprised that de Marneffe would not find it attractive. My hope is that readers who value extensive liberty but who also are concerned about the welfare of all will find it attractive and help defend the view going forward.

    Thanks again to Professor de Marneffe for reading my work and offering helpful indications of things I might want to work on moving forward. It’s appreciated.

  2. I’m glad to see the harm principle discussed on PEA Soup! I’m wondering about the response to de Marneffe’s concern about “risky” behavior. (Although I’ve read the review, I have not yet read the book.) The question here is this: how risky does some form of activity have to be in order for it to be subject to restrictive regulation? Cohen suggests that the risk must be non-negligible. I wonder whether the concern could be met in a different way. It might be that some form of activity should be restrictively regulated only if the risk of harm it poses is sufficiently severe *when weighed against the interests people have in participating in the relevant form of activity.* My driving a car (responsibly, when sober, etc.) imposes on others a slight risk of serious harm. Suppose, for the sake of argument, that my choice to open an outdoor archery range in my suburban backyard (or to keep a pet lion, or to use an outdoor fire pit during a drought) would impose on others precisely the same small risk of harm as my choice to drive. It may well be permissible for the government to prohibit me from opening the archery range (keeping the pet lion, using the fire pit), even if it would not be permissible for the government to prohibit me from driving. And this might be because it would be much more costly for me to refrain from driving than it would be for me to refrain from opening the archery range (etc.). The difference here doesn’t have to do with the magnitude of the risk I would be imposing on others, but instead with the costs that a restrictive policy would impose on me. That sort of response seems promising to me. I wasn’t sure whether it was what Cohen had in mind, or whether it is consistent with his way of addressing risk.
    I’m also wondering whether the line of response I’ve suggested might fuel de Marneffe’s concern about paternalism. Here’s what I have in mind. Consider a policy that restricts some form of activity that imposes moderate risks of harm on others. Suppose that someone who wants to participate in this activity sincerely claims that he is greatly burdened by the policy. Most of the rest of us disagree: we think that the policy would impose only slight costs on him, or no costs, or that it would even benefit him on balance. And suppose that the disagreement does not concern empirical matters; instead, the disagreement might concern how important or beneficial the relevant form of activity is. (Here might be an example: An opponent of gun control might concede that an assault weapons ban or some such measures would reduce gun-related deaths. But he might claim that these measures are still unjustified because he benefits greatly from owning assault weapons. Supporters of gun control laws might deny that these laws would impose huge costs on him, but agree that these laws would be unjustified if they did impose huge costs on him or other would-be gun owners.) What should we do about this sort of disagreement? Here, it may seem implausible to say that we should defer to the policy’s opponent and conclude that the policy really does impose major costs on him (or should be regarded as though it imposes major costs on him) *simply* because he sincerely says that it does.
    But then what should we say about paternalism? There might be a certain tension between (a) the claim that we should sometimes be prepared to override or disregard a person’s judgment about the magnitude of the costs that a policy would impose on him, when this policy reduces the risk of harm to others, and (b) the claim that we should never be prepared to override or disregard a person’s judgment about the costs that a policy would impose on him, when this policy reduces the risk of harm to him alone. This tension wouldn’t arise if we reject (a), but I wasn’t quite sure what Cohen would say about it (again, having not read the book). Regardless, I enjoyed reading Professor Cohen’s post and thinking about these issues.

  3. Jason Hanna wrote: “But then what should we say about paternalism? There might be a certain tension between (a) the claim that we should sometimes be prepared to override or disregard a person’s judgment about the magnitude of the costs that a policy would impose on him, when this policy reduces the risk of harm to others, and (b) the claim that we should never be prepared to override or disregard a person’s judgment about the costs that a policy would impose on him, when this policy reduces the risk of harm to him alone. This tension wouldn’t arise if we reject (a), but I wasn’t quite sure what Cohen would say about it (again, having not read the book).”

    Great comments, and my initial two cents if I’m interpreting you correctly: I wouldn’t reject (a), but I’m not clear what the tension is because there’s an asymmetry between (a) and (b). One reason for not disregarding/overriding P’s judgment about the costs a policy imposes on her alone is precisely that her (mis)judgment does not impose costs on others. That seems like a paradigm instance of the Harm Principle as Mill, and, I believe, Cohen defend. P isn’t “imposing” costs on her future self in the way she would if others had no choice but to suffer (the risk of) harm from P’s decisions. In contrast, P has the choice not to bear the costs of her (mis)judgment by, well, not acting on it. But, for whatever reason good or bad, she decides to perform the self-harmful actions, and an anti-paternalist can argue that we should let her perform those-self-harmful actions, even though they might clash with her own beliefs and values. Whether P is responsible for her (mis)judgment may be a separate matter, of course, but that seems to call for the (Feinbergian) soft/hard paternalism distinction and an investigation into why she is making self-harmful decisions.

  4. Thanks for the thoughtful comment Jason Hanna! “The question here is this: how risky does some form of activity have to be in order for it to be subject to restrictive regulation?” My answer thus far, “when an activity is statistically likely to be harmful in the sense of wrongfully setting back interests” is, no doubt, incomplete and I admit, I am not yet sure how to make it complete.

    On one view, taking a risky action would be permissible (and so presumably not harmful) if (and only if) the benefit to the actor of taking the risk outweighs the burdens imposed. This has a certain plausibility to it, but I doubt it helps that much and I am uncertain how it fits into my overall view. The basic view is that interference with A is permissible when A’s action would cause a wrongful setback of interests to B–which is best determined in a quasi-jurisprudential manner (to determine both if A wrongs B and if B’s interests are set back). When we are talking about risks, matters are more complicated.

    When does a risk itself count as a harm? Of course, some risks do so directly. If you force me to play Russian roulette with the gun aimed at me and no physical harm occurs (the sole bullet in the gun is not triggered), you cause me a risk that (I think we all agree) sets back my interests independently of the harm that would occur if the bullet tears through my flesh. In this sort of example, there is clear psychological distress that is (I assume) unjustified and hence wrongful. In this sort of case, the risk itself initiates a harm event. Other risks, of course, don’t involve harm at all. They are themselves non-injurious.

    Most people don’t, so far as I can tell, suffer psychological distress at the thought that Average Joe’s driving puts them at risk. In these sorts of cases, I think there are (at least) two options: (1) do nothing unless an actual injury occurs or (2) institute some form of advance legal restriction to prevent the harms made possible by risks from occurring. This does not mean, of course, instituting a complete legal prohibition. We can license people to perform the activity instead. We do this with driving–requiring that those who wish to drive pass an exam to show that the risk they impose is small (in that the likelihood that they will actualize a risk is low). In chapter 8 of the book I argue for parental licensing on these grounds.

    There are difficulties with instituting a parental licensing scheme–indeed, any licensing scheme–and our social science is not, so far as I can tell, sophisticated enough yet to institute such. But we know that roughly 3% of biological parents mistreat their offspring and if we could institute a licensing scheme that prevented some significant portion of that 3% from having children to mistreat while not preventing others from having children, we would likely reduce harms significantly (not just to the children saved from abuse). The trick is to avoid false positives (and hopefully false negatives).

    We might ask now whether the fact that roughly 3% of biological parents mistreat their offspring is sufficient to show that any response is warranted. This is, I think, Jason’s question. My view does not provide a good answer, I am afraid. The only thing I can say definitively is that most people seem OK with licensing drivers, which is designed to prevent a much smaller chance of harm, so consistency suggests they should be OK with licensing parents. But, of course, one response is to say “I had no idea the risk of driving is as low as it is! we shouldn’t be bothered licensing drivers!” (This, of course, is confused because the risk of driving that we know of is a risk of driving in a system with licenses; the risk may be significantly higher if we did not have a licensing program.) Another response is to argue that the magnitude of harms that could result from risks are vastly different. This is true, but its not clear which way it would push us.

  5. Thanks for the replies! They gave me more to think about than I can probably get to in a short post, but here goes.

    Bill Glod writes: “One reason for not disregarding/overriding P’s judgment about the costs a policy imposes on her alone is precisely that her (mis)judgment does not impose costs on others.” There could be an important asymmetry between behavior that only harms oneself and behavior that harms others. But I was concerned that *one* defense of this asymmetry might be weakened by the way in which we deal with cases involving risk. I think my concern might best be illustrated with an example. Some people defend motorcycle helmet laws on the grounds that helmetless motorcyclists pose a risk to others on the road because, e.g., they’re more likely to get hit by rocks, get bugs in their eyes etc., and crash and injure other drivers. This strikes many people as an unconvincing argument, but let’s just stipulate that it is in fact true that riding without a helmet slightly (but only very slightly) increases the risk that a motorcyclist will crash into someone else. This fact might seem to provide some reason, though perhaps only a very weak reason, to require helmets, even if we accept the harm principle. But suppose that motorcyclists respond to this argument by pointing out how wonderful it is to feel the wind in their hair and claim that the laws would be really costly to them. How should we respond to this argument, if we think that in fact helmet laws would not leave them worse off (and might even make them better off)? And how should our response affect our views on straightforwardly paternalistic arguments for helmet laws? I suppose the concern is that *if* we oppose paternalism by arguing that straightforwardly paternalistic helmet laws “impose values” on people by privileging certain views about just how valuable it is to feel the wind in one’s hair (etc.), then perhaps we should also think that we impose values on people by privileging a similar view when our aim is only to reduce risks of harm to others.

    Andrew Jason Cohen’s example of the Russian roulette case is interesting. I wonder about variants on the example. Suppose someone places a single bullet in a million-chambered gun and plays Russian roulette with an innocent (and unconsenting) bystander. And suppose that neither the bystander nor anyone else finds out about this, so there’s no distress. Should this sort of behavior be prohibited under the harm principle? I agree that the answer is “yes,” though it’s not so clear to me that the behavior in this case is more harmful or more risky than other kinds of behavior that probably are not eligible for restriction under the harm principle. Does the difference have to do with the agent’s intentions? I’m not sure. Maybe this is the kind of case where the quasi-jurisprudential method is the best we can do.

  6. Jason-More great questions. I’ll stick with the last part. I agree, of course, that if someone places a single bullet in a million-chambered gun and plays Russian roulette with an innocent (and unconsenting) bystander, is a bad thing. That is, I think the actor does a wrong. Perhaps they do a wrong to the bystander–though I am less certain about that. Do they set back the bystander’s interests? I am less certain still (given the bystander has no idea about it and is not actually shot). I think, though, that these are precisely the sort of questions that should be discussed before a decision is made regarding whether to interfere. I admit that my prima facie response is “interference is permissible,” but I don’t have a great deal of confidence in that response. This may be because of the small chance of anything actually happening.

  7. Jason Hanna:

    I take one of your main points to be that there are some liberties we want to have even though they impose risks on others, but we tend to differentiate between certain liberties as more permissible despite imposing equivalent risks of harm on others. One reason, of some weight, for restricting certain such liberties is because we are overriding/denying the person’s judgment of the benefit of the activity *for that person* all else equal. That may be true, but I sense at least three other considerations are in play when we restrict some activities but not other equally risky ones, and all seem to be subsumed under the Harm Principle without needing to evoke paternalism: (1) the severity and identifiability of the source of that harm, (2) the social benefits as weighed against risks to others attendant on some activity, (3) the social costs of harms that may eventuate from activities that are mostly of personal not social benefit.

    Then there’s a fourth: straightforward paternalism cases where there are unclear benefits to anyone but perhaps costs to the actor if we do not intervene. Having numbered things, I’m probably underestimating the subtlety of your observations. Nonetheless, I’ll proceed.

    Let’s say we have good actuarial numbers on the harm data and interpret this data the same way. In light of risks, I don’t see this as a paternalism matter so much as a tort matter or some situation that could call for restricting liberty based on weighing overall costs and benefits. Typically, we can’t identify my germs causing you (threatening you with) severe harm, but we can identify my careless motorcycling or Russian roulette playing. OTOH, if we could identify me as a major influenza vector, perhaps I should be quarantined from going out in public. Focusing on identifiably caused severe harms traceable to specific sources might be a relevant matter, where jurisprudential matters like torts etc come in. Not sure philosophers can or should say a whole lot on the nitty-gritty, but the framework makes sense.

    Moreover, we protect certain risky-to-others activities such as the freedom to drive the speed limit or go out in public with a mild cold when these activities are often necessary in the ordinary conduct of one’s life and generally beneficial to others (e.g., driving is usually more efficient than walking, or we still often need people to show up for work even if they’re a mild contagion risk). OTOH, popping wheelies or being “volunteered” for million-to-one-odds Russian roulette are not activities generally beneficial to others, who had no choice in being subject to the risks, even if the primary actors get a major thrill from them. So even if these latter activities impose equivalent (or even lesser) all-things-considered risks on others than the former activities, the “benefits” side of ledger only involves benefits (if any) to the individual imposing the risks.

    Then there are social costs with regard to activities that mainly only bring personal and not social benefits. As things stand, rock climbers and footballers are free to engage in their endeavors even if their injuries are more costly to taxpayers than social costs imposed by unhealthy consumption patterns (yet numerous studies suggest that smokers and obese people are net benefactors to the health care budget given their shorter lifespans). The discussion moves to singling out which liberties “matter more” regardless of the risks they might impose on others (or the actor). But then it’s no longer a discussion about (equivalent) risks imposed on others that the Harm Principle can cover. Certain liberties are being picked out for scrutiny, perhaps rightly so, but that’s another discussion.

    The fourth case is someone who passes the above tests, who is not imposing any clear costs on others. They are mainly costs to herself.

Comments are closed.