Solidarity and Responsibility in Health Care

This article argues that health care systems that are grounded in solidarity have the right to penalise some users who are responsible for their poor health.

Paper 4. Solidarity and Responsibility in Health Care by Ben Davies and Julian Savulescu


Jack Freer June 7, 2020 at 10:12 AM

I see a number of flaws in the arguments put forth by Davies and Savulescu (D&S) in regard to personal responsibility in solidarity-based healthcare systems.

1. UK's NHS can be characterized in a number of ways. Yes, it is an example of a system grounded in solidarity, which (in theory) would impose obligations on participants, as well as providing benefits. It also can be characterized as a means to provide universal healthcare (at a predefined level) to all citizens.

The former description *could* allow for penalties, diminished services, or other types of modifications for those who continue to demonstrate indifference (or even hostility) to healthy choices. Reacting to these "bad choices" necessarily requires an elaborate system of rewards and benefits, based on normative assumptions made by the system. These determinations are potentially arbitrary and capricious, as well as prone to "anti-fat" or "anti-sedentary" bias.

The latter description (universal coverage), on the other hand, is straightforward and simple to administer: provide the preset minimum of healthcare to all, depending only on medical indication. There is nothing necessarily built into the NHS requiring that we accept the first and not the second description.

2. D&S oversimplify the behavioral elements of healthy lifestyle. Smoking, excessive drinking, and overeating are compulsive behaviors that involve elements of addiction. Many well-meaning adults are desperate to end these self-abusive behaviors and are repeatedly thwarted. There is an illusion of total freedom of choice, in which those whose personalities lend themselves to more disciplined behaviors view the rest of us as slovenly, lazy and undisciplined by choice.

3. D&S make unfounded assumptions about causation in behavioral choices. I think it is clear that failure to wear a seatbelt increases vehicular morbidity and mortality, as would driving without headlights at night. On the other hand, moderate obesity has a much more tenuous relationship with optimal health.

4. I am unclear about how D&S view "golden opportunities." The description in the article implies that these are further indicators that the patient has squandered a clear, highly effective means to modify behaviors. I am awaiting delivery of the 2018 article that describes such opportunities, but I remain skeptical about their effectiveness.


Stephen Kershnar June 7, 2020 at 1:57 PM


You state the following.

"D&S oversimplify the behavioral elements of healthy lifestyle. Smoking, excessive drinking, and overeating are compulsive behaviors that involve elements of addiction. Many well-meaning adults are desperate to end these self-abusive behaviors and are repeatedly thwarted. There is an illusion of total freedom of choice, in which those whose personalities lend themselves to more disciplined behaviors view the rest of us as slovenly, lazy and undisciplined by choice."

Do you think that some people who drink too much and eats too much are blameworthy or that none are blameworthy? If it is the former, then we could in theory investigate who is blameworthy for these things and penalize only them.

This investigation might fail a cost-benefit analysis, but in theory there is nothing wrong with it.


Steve K

Stephen Kershnar June 7, 2020 at 2:02 PM


You say the following.

"D&S make unfounded assumptions about causation in behavioral choices. I think it is clear that failure to wear a seatbelt increases vehicular morbidity and mortality, as would driving without headlights at night."

At least one economist thinks that safety belts do not make us more safe.

If safety belts were to make us less safe, do you think we should penalize those who wear safety belts? Perhaps we could ticket them or hit them with an additional tax (e.g., Pigou tax).


Steve K

Jack Freer June 7, 2020 at 6:47 PM


I'm intrigued by the notion of "blame" (reminds me of that T-shirt: "I didn't say it was your fault; I said I was going to blame you"). I think the real issue is the consequences of blameworthiness. If there is a real penalty (even stigmatization), one would hope that it would modify the behavior in the future. I don't think it is possible to fairly assign blame to a degree that would distinguish one person from another with similar behaviors.

When you suggest that safety belts might make us less safe, I'm assuming that you are proposing a situation in which newer research demonstrated conclusively that belts caused more injury/death than unrestrained occupants. I think the onus would fall on automakers to discontinue, remove and/or replace with safe alternatives. If someone continued to use them for sentimental reasons, it does not seem like the kind of thing we'd penalize.


Bob Kelly June 8, 2020 at 12:27 AM


On (1), I'm not sure I see the real problem. I think they use that institution as the example since it's really close. If it is, in fact, not a solidarity system, then we can imagine it as one, or think of another one. It was just a useful example, I think, to illustrate the theory. Moreover, you're right that normative assumptions have to be made in order for the system to function (I don't know if they are made "by the system," as you say). But why is this a problem? It's a normative theory, so we'll need to have some normative assumptions. For instance, they assume that there is something like a standing obligation to not abandon others, out of which some solidarity obligations arise. I don't see why that is a problem. Maybe it's false. But the method doesn't seem suspect.

On (2), I agree with Steve. We just need it to be true that some people who engage in addicted behaviors are responsible for doing so (perhaps only indirectly). What's more, I think D&S agree with the sentiment of your comment. This is why SMOKING is not a case of full autonomy (they even think pressure from social norms is enough to undermine control), and also why they discuss the role of genetics and so on in undermining full autonomy (and so responsibility). It seemed to me like they are completely open to many addicts (drug, food, or otherwise) having compromised autonomy. Interestingly, though, Savulescu famously thinks addiction does not necessarily involve impaired control (but mostly because he thinks we can't know that). In any case, perhaps out of an attempt to be more open with the account, he and Davies seem to allow for a lot of wiggle room in terms of certain stigmatized behaviors like addictions.

On (3), I again think this point is inconsequential to their argument. All they need is for it to be plausible that some of these behaviors/choices have significant causal effect on the individual's health, such that they need to take resources from the HC system and so impose costs on others. Surely it is plausible that some cases of obesity work this way (keep in mind this doesn't mean obesity has to be the SOLE cause of their ill health). For seat belts, the thought was just that their case was not one where an autonomous, responsible choice was made with a Golden Opportunity. This could happen with seat belts. But their case didn't seem that way. If you think the mere fact that the decision not to wear the seat belt was impulsive is insufficient to undermine autonomy, then that seems like a reasonable objection. It is even more problematic when we combine that claim with their reliance on negligence, which is precisely indifferent to whether or not one's decision was impulsive. Negligence doesn't care about one's current state (mostly), but only whether it arose from a previous failure to do what you should have done. If they allow negligence as much as they seem to, then SEATBELT might need to count as a responsible choice on their view.

I basically agree with (4).

Stephen Kershnar June 8, 2020 at 8:45 AM

Jack and Bob:

Consider these two things regarding an activity X (overeating or drinking too much).

(1) A is blameworthy for X.

(2) The states treats A as if he is blameworthy for X.

First, assume that (2) depends on (1). Jack's response is that we can't know (1). If so, then, in some sense, it is objectionable to do (2). Still, if we can know people are at fault in the criminal law and our personal lives, I don't see why in principle we can't know this with regard to unhealthy behavior.

Second, assume that (2) is independent of (1). For example, we might have J. J. C. Smart's forward-looking theory of responsibility. If this is correct, then, if it were the case that seat belts make us less safe, I don't see why we would ticket, (Pigou) tax, or otherwise discourage seat belt use.

Your response is that (a) we need conclusive proof and (b) the onus should fall on automakers.

Proposition (a) is odd. We don't require conclusive proof with regard to many policies (consider, for example, pandemic lockdowns, non-algorithmic medical school admissions, or statutory rape laws). There is no reason to require it here.

Proposition (b) strikes me as mistaken. If we're interested in modifying behavior in the healthiest direction (or, perhaps, in the economically efficient direction), then we want to apply incentives to the party who will best respond (least cost avoider). I don't understand why you've chosen to target the automakers rather than drivers.

Bob - I'm wondering if you think the Golden Opportunity is shorthand for autonomy-ensuring condition or whether it is something else. Also, I'm wondering if you think the state should make false statements (A is blameworthy when he is not) if doing so is overall efficient.

Thanks for your thoughts,

Steve K

Jack Freer June 8, 2020 at 11:41 AM

Quick response to Steve. I’ll have to give Rob’s reply more time and thought.

I would want conclusive evidence for seat belt harm because it would run counter to a large body of existing evidence.

Onus on automakers because they are the ones placing the belts in the cars. Unlike new evidence that, for example, driving with a cup of hot coffee in your lap was safer than no coffee, the driver can not circumvent the no-belt guidelines without someone else having first installed the belt.

Stephen Kershnar June 9, 2020 at 9:52 AM



Quick reply. Here are two groups we could put the burden of proper seat belt policy on.

(1) The burden should be put on whomever makes the car.

(2) The burden should be put on whomever uses the car.

The decision between (1) and (2) should be based on efficiency. That is, who produces more benefits (health or money) at a cheaper cost. You would either have to know what is efficient a priori or not sufficiently value efficiency.

By analogy, liability for misuse of guns is put on the individual people and not the gun-makers. Ditto for alcohol.

One might think that safety belt decisions should be left to the discretion of car makers and buyers because this is a free country. Sadly freedom's out of fashion.

Thanks for the note and I hope life is treating you well,

Steve K

Bob Kelly June 9, 2020 at 1:21 PM


Regarding the initial comment to Jack, you said:

"(1) A is blameworthy for X.

(2) The states treats A as if he is blameworthy for X.

First, assume that (2) depends on (1). Jack's response is that we can't know (1)..."

I don't want to speak for Jack, but I don't see where he said that. He said it is hard to determine the causal relationship between obesity and optimal health (and perhaps HC costs generally). The point being, I think, that obesity may not qualify as a case where treatment is deprioritized (since we can't tell if or what actions caused unhealth or HC costs, and so can't tell is a solidary obligation is violated). I don't think he said we can't know if they are blameworthy. In fact, his previous point seemed to suggest they may not be (due to compulsion). This line of thought confused me.

Second, to me you said:

"Bob - I'm wondering if you think the Golden Opportunity is shorthand for autonomy-ensuring condition or whether it is something else. Also, I'm wondering if you think the state should make false statements (A is blameworthy when he is not) if doing so is overall efficient."

On the first question, yes, something like that. Maybe even worse. It seems circular. S is MR for A only if they unreasonably refused a GO. And a GO requires that it was given in appropriate conditions, where they say this is conditions that are "conducive to responsible choice." Hmmmm.

Apologies, I am not sure I am following the motivation for the second question. I can't tell what I said above that implied an answer either way to that idea. Also, I don't know what you mean by 'efficient', so I don't know how to answer it. If you're asking whether I think a consequentialist view of MR is correct, then I would say no. But if the question is more like, would it be ok to utilize a consequentialist conception of MR if we found out that MR ascriptions never referred, then I am less certain that the answer is no. Skeptics like Pereboom and others propose responsibility systems without MR (though, his is rights-based). But the point is that, assuming skepticism, we would still need some kind of responsibility and/or punishment system, and so (again, depending on what 'efficient' means) the answer to the second question might have to be yes.

Stephen Kershnar June 14, 2020 at 10:30 AM



This is a really good point. I am also worried that the golden opportunity notion is circular. I wonder, though, if it could be defined in terms of something which ensures sufficient knowledge or voluntariness, where these can be defined independent of responsibility.

The more general concern is this.

(1) If a person is morally responsible, then he has a sufficient amount of each of the following: (a) capacity, (b) knowledge, and (c) voluntariness.

I wonder if (a) through (c) – either the concepts in this context or the needed amount - can be specified in a way that does not beg the question.

If they cannot be specified in a way that does not beg the question, neither does the notion of a golden opportunity.


Steve K

Bob Kelly June 14, 2020 at 2:38 PM


I think you're right here. The move for D&S to make is to say that they are NOT providing an account or definition of the nature of MR in step one ("S is MR for A only if she unreasonably refused a GO"). Instead, they are providing a particular qualifying condition for S's A-ing to be subject to penalization. Once we get to step 2 ("S unreasonably refused a GO only if S did so in MR-conducive circumstances"), then we can say more about what it means to be MR. This is where they would introduce your conditions (knowledge, capacity, etc.).

So, it should look more like this. Suppose A-ing externalizes HC costs. Then, S is liable to penalization for A-ing only if S was MR when she refused a GO to not-A. S was MR for refusing a GO to not-A only if her refusal was done with sufficient capacity, knowledge, and voluntariness. That doesn't seem circular anymore (assuming, as you point out, that we can fill out the last three terms in a non-circular way). I suspect that almost everyone in the FW/MR literature would say we can fill them out non-circularly. It seems like they must say this, or else that literature is in big, big trouble.

Stephen Kershnar June 7, 2020 at 1:46 PM


Davies and Savulescu (D & S) argue that solidarity can generate obligations and that failure to meet these obligations may be penalized. They assert that one prominent contemporary use is to “invoke ‘emotionally and normatively motivated readiness for mutual support’. This includes a willingness to promote others interests, or the interests of the group, even at personal cost.” [See p. 134]

Solidarity does not fit into one of the two most plausible justifications for state legitimacy or an obligation to obey the law.

(1) Consent.

The authors note that solidarity-based obligations can exist even when a person cannot practically or even ethically opt out of. [See p. 139] This suggests it is not consent-based.

(2) Duty of Fair Play.

D & S do not state that solidarity-based obligations rest on the active receipt of benefit from a just cooperative scheme. Nor do they cite the usual suspects. Consider, for example, H. L. A. Hart and John Rawls. This suggests it is not a version of the duty of fair play.

(3) Communitarian and Role-Based Duty.

D & S assert that “Solidarity obligations arise because of relationships and similarities that already exist between individual. … [G]roup membership is the basis of solidarity obligations … .” [See p. 139]

Perhaps they view such obligations as the result of social roles or relationships, that is, communitarian or role-based duties.

Why would a relationship, similarity, or group membership ground a duty? If a person is forced into a relationship (or group) against his will and further forced to accept the benefits of the relationship (or group), no relationship-based duties result. For example, he is not obligated to support the relationship or group. This is true even if, setting aside coerced membership, the relationship or group is otherwise just.

In short, if this interpretation is correct, solidarity does not ground obligations.

Objection #1: Consent.

If we assume that the healthcare system is justified by people’s consent (this is more plausible), then the system may have whatever penalties consenters want.

Objection #2: Duty of Fair Play.

If we assume that the healthcare system is justified by the duty of fair play, then the system may have whatever penalties are just (or fair). In such a case the justification of the system (duty of fair play) plays no role other than to point to the role of justice.


Steve K


Bob Kelly June 8, 2020 at 12:35 AM


On (3), I also think this is closest to what they have in mind. But rather than relationships, aren't they trying to ground the solidary obligations in other duties? For instance, in the part where you put the ellipses in the quote, they say, "It is

because we cannot abandon people that they in turn derive obligations to play their part by not overly burdening the system we share: it would be unreasonable of them to burden that system, since it is a system the rest of us cannot ethically—and perhaps even practically—opt out of." It sounds like "we cannot abandon people" and "cannot ethically...opt out of" are meant to invoke other duties we have. I'm not sure if this helps them escape your objection here, but I thought it worth mentioning that this seemed to be what they had in mind. I agree, though, that it's unclear where the ground of solidarity is supposed to be.

Stephen Kershnar June 8, 2020 at 8:59 AM



Excellent point.

(1) Instrumental Theory. If D & S think that solidarity is an instrumental reason depending on other reasons, we need to know what those are before we can decide whether these sort of penalties. For example, if a criminal justice system is justified by, and only by, utilitarian reasons.

D & S couldn't accept that and then cite solidarity reasons to explain why the particular practices in that system are governed by retributionist norms (contra Rawls).

(2) Non-Instrumental Theory. If, instead, solidarity has a communitarian basis, then the particular community values will control the policy on penalties. This will be community-relative. If solidarity is role-based, then we need to know the relevant roles (healthcare distribution agent, physician, etc.). They can't skip this step.

In short, on either interpretation solidarity will not do the work they want. Hence, I think their argument does not work.


Steve K

Stephen Kershnar June 8, 2020 at 9:10 AM


In the part you quote, D & S's argument is as follows.

(P1) We should not abandon people.

(P2) If (P1), the a recipient may not overly burden the system.

(P3) If a recipient may not overly burden the system, then he may not make an unreasonable choice free of penalty.

There are a couple of problems here. First, many people want to dissent from being wards of others by opting out of Medicaid, Medicare, Social Security, etc. and purchasing their own insurance. If D & S respect autonomy, dignity, etc., then it is not clear they can trample on people's rights merely by citing a duty not to abandon people. This is even more clear if the system is prohibitively expensive or works poorly. Hence, (P1) is either conditioned on consent or false.

Second, even if the argument is true, I don't see their argument for (P3). Here is an analogous argument.

In the part you quote, D & S's argument is as follows.

(P1) We should not abandon people.

(P2) If (P1), the a recipient may not overly burden the system.

(P3) If a recipient may not overly burden the system, then a poor woman may not have a child out of wedlock that she cannot pay free of penalty.

This strikes me as a parallel argument. I wonder if D & S support this conclusion. Jack - What say you?


Steve K

Bob Kelly June 9, 2020 at 1:44 PM


I know D&S addressed the issue of wealthy people dissenting from a solidarity system by getting private insurance they (but not many others) can afford. Is this the idea you had in mind? They end up saying, "Moreover, if private patients are not fully internalising the costs of their choices, then the same obligations of solidarity would apply." I took this to mean that it would need to be clear that dissenting comes along with internalizing all of one's HC costs, and I guess they aren't confident this would happen. I think they also hang a lot on the claim that reasonable people understand the duty not to abandon others (and so, in a sense, understand the solidary nature of the HC system). If so, then dissenting might be knowingly skirting a duty, knowingly imposing the burden of fulfilling that duty on others, and perhaps also not internalizing all of one's HC costs (free-riding).

Also, if there really is a duty not to abandon (whatever that means), then this would presumably compete with people's rights. If so, then I don't see why they can't cite this duty and also respect people's autonomy.

On the (P3) point, I just think there isn't enough information in your example. We have to start with the assumption that this is plausible, and not the opposite. Doing the latter makes a single case seem to have more weight than it has or might have. We need to spell out the content of the premises with their view a bit more. Suppose (P1) is true, and that because of this it is wrong in some sense to externalize your HC costs (P2). But (P2) *must* include the qualification that it is wrong only when externalizing your HC costs meets certain conditions (you had a GO for help, you refused it autonomously, you had reasonable foresight of the externalized costs, and so on). Suppose this is all true, and add to this the requirement that someone else is clearly and straightforwardly competing for the required HC resources due to a condition completely outside of their control. It now sounds less crazy to say that, in some sense, it is wrong to have the child in those conditions. Maybe not all things considered. We need to further add that they do not advocate banning low SES people from having children (even when they refuse a GO). They advocate differential disbursement of HC resources to some unspecified degree, where things other than the immediate facts about refusing a GO and so on can count. It just seems like there is much more nuance to their view that (P1)-(P3) suggest, making the conclusion much less absurd than your argument implies.

Stephen Kershnar June 14, 2020 at 10:39 AM


Excellent point. I gather your response is that given a GO and D & S’s reasoning the conclusion of the second argument is not so absurd.

(C1) [Because a recipient may not overly burden the system] A poor woman may not have a child out of wedlock that she cannot pay free of penalty.

One thing of interest here is that D & S use penalties solely in terms of less support. From a Law & Economics perspective, though, we also want to consider other methods of regulation, such as fines, incarceration, and higher taxes. Here (C1) sounds a little more troubling. I wonder what D & S could say in response to these other penalty-types.


Steve K

Bob Kelly June 14, 2020 at 2:29 PM


I see your worry, and I think they would certainly need to address it to fill out the view. My point was just that it's not clear from their account what the response to (C1) would be given the further information we would need to know. We would have to know what the circumstances were like so as to know we had a GO on our hands, whether it was refused unreasonably, whether "help" or "support" was offered, what the other person's/people's (competing for the HC resources) circumstances were like, and, most importantly, if there are any other variables that might override the GO refusal or perhaps make it reasonable. To your point, it ends up being a criticism of their view that we don't have any of this figured out. So even if I am right and we don't know what their view entails about your (C1), this itself is also a problem.

Stephen Kershnar June 7, 2020 at 1:50 PM


D & S argue that, “[A] violation of solidarity obligations requires more than avoidability: it requires both that the risk taken is unreasonable, and that it was made under conditions that are conducive to autonomous choice, as embodied as being offered a Golden Opportunity.” [141] They say, “Those who choose to impose unreasonable burdens on others – or choose unreasonably failing to consider the burdens on others – have failed to show this reciprocal concern [for others].” [136]

What is an unreasonable risk?

(1) Inefficient Risk

An unreasonable risk is not an inefficient risk. D & S do not appear to accept this because they think that whether a risk is reasonable depends on duties to others, broader social context, and conditions of choice. [136] These do not sound like a cost-benefit analysis.

(2) Blameworthy Risk

An unreasonable risk is not a risk that one is blameworthy for taking. D & S do not appear to accept this because they reject luck egalitarianism [135-136] and distinguish unreasonableness from the conditions of autonomous choice.

(3) Risk Not Supported by the Balance of Reasons

Perhaps an unreasonable risk is a risk that does not reflect the balance or reasons or, perhaps, moral reasons. If this is correct, though, then the theory says that solidarity justifies the system penalizing those who (a) take risks (b) they ought not to take and (c) for which they were blameworthy.

Reciprocity is independent of risks members ought not to take. Some risks are not worth taking, but do not lessen reciprocity. Some risks lessen reciprocity but are worth taking.

This appears, then, to be a way of making us do the right thing or, perhaps, improving us. But this is not the government’s (or healthcare system’s) job. I do not want the government to penalize people for adultery, divorce, obesity, promiscuity, racism, or Scientology membership because it is the wrong thing to do or because avoiding these things would improve us.

Perhaps the unreasonableness standard does not reflect the balance or reasons (or, perhaps, moral reasons). In this case, it is mysterious.

Hence, the unreasonableness standard is unreasonable.


Steve K


Neil Feit June 9, 2020 at 12:21 PM

Steve, this seems right to me and I also find the D & S notion of an unreasonable risk to be mysterious. Their fundamental obligation of solidarity seems to be, roughly, "don't risk externalizing costs unreasonably." At the very least, the notion of unreasonableness here is so under-developed that it makes the general perspective too vague to evaluate. (This is even granting that the HC system, like the NHS, is grounded in solidarity and that this is what generates obligations.) Perhaps D & S could go with something like your (3) above, though I don't see why (c) [blameworthiness] should be included. If this could be done, then the penalizing issue would not worry me so much. This is because (I think) one is penalized for behavior only when it does impose unreasonable burdens on others by externalizing costs and the suggested penalties -- like lower priority on waiting lists or getting only partial covering of HC costs -- seem proportional.

Bob Kelly June 9, 2020 at 1:54 PM

Neil, my guess is that (c) blameworthiness is included for two reasons. First, it seems to be exegetically correct. D&S explicitly say that a GO requires conditions that are conducive to responsible choice. I, like Steve, don't know how to understand responsible choice here other than a choice that would make one blameworthy were the choice wrong. Second, I think that blameworthiness is a stand-in here for a choice made free and autonomously, or is meant to introduce a fairness standard, or both. So it wouldn't be enough to perform a risky behavior that one shouldn't. Perhaps they acted under duress or were otherwise not free/blameworthy. The choice might still be risky and ought not to be performed (externalizes costs). It would only be fair/appropriate to penalize, though, if the risky, wrong behavior was done freely (where this means it was done in such a way that we can appropriately blame the person for doing it).

I may be missing something in your point, though.

Neil Feit June 9, 2020 at 3:16 PM

Bob, I get that and I might be missing something. I guess I was thinking that in the quote Steve starts with, D & S seem to give two necessary conditions for a decision to violate a solidarity obligation: (i) it involves an unreasonable risk, and (ii) it involves refusing a Golden Opportunity (roughly). I was thinking blameworthiness goes to (ii), that is, refusing the GO, but is not needed for (i), that is, the risk counting as unreasonable. Maybe I was wrong about that.

Bob Kelly June 9, 2020 at 3:41 PM

Oh, I see. In that case, I think I would say two things. D&S might be putting into (i) as well since they seem to talk about the reasonableness standard in a way that suggests risk-taking is only unreasonable when we can hold the person responsible for it. For instance, they say that a single mother not working two jobs is not taking an *unreasonable* risk by not cooking fresh meals for her kids because we couldn't hold her MR for making that trade-off. She needs to sacrifice the time it would take in order to provide for her kids, so the risk is not unreasonable at least partly because her decision is not blameworthy.

If that is incorrect, the second thing I would say is that they may be slipping it into (i) in this way implicitly. I think a reasonableness standard, while intuitive, is partly problematic because we need to then ask what determines the threshold, and answering this question can sometimes run in a circle. For instance, if you think responsibility gets cashed out in terms of what it is reasonable (e.g. reasonableness of having reactive attitudes, requiring that one couldn't reasonably have been asked to avoid the act, etc.), then you cannot fill out the reasonableness standard in terms of responsibility. However, as Steve has argued in conversation before, one worry is that the only plausible way to determine which kinds of factors count as reasonable-making or not are those which tell us whether we would blame you (or hold you responsible) or not. For instance, S is MR for A only if it was reasonable to ask them to avoid A-ing, and circumstances C make it reasonable to ask them to avoid A-ing only if we could blame them for A-ing in C. Maybe D&S are making this error?

Stephen Kershnar June 14, 2020 at 10:53 AM


Neil and Bob:

Great points. Along with you guys, I think there are three concerns with the family of concepts.

(1) Autonomous choice

(2) Avoidable choice

(3) Reasonable choice

(4) Disproportionate externalizing of costs

First, as you guys point out, there is a real concern that these concepts are circular. Not just in the way we specify them, but in terms of their truth-maker.

Second, factors such as externalizing of costs and reasonable choice can be specified by a cost-benefit analysis or via brute intuition. I realize this needs more argumentation. I will just assert it here. As you guys and I have discussed before, the former is distinct from blameworthiness in part because it depends on the efficient rule which in turn depends on facts about others.

The second puts a lot of weight on intuition. Perhaps our intuitions can bear some of this weight at least in extreme cases. I just do not know.

The third is the need of so many in the literature to use ‘autonomous choice’ rather than ‘morally responsible choice’. Until we know whether the author thinks these are the same, I see no reason to use the former.


Steve K

Bob Kelly June 7, 2020 at 11:28 PM

I thought that D&S made a decent contribution to the discussion surrounding responsibility in health care. I am not a Savulescuan about the topic now, since I also think there are several issues with the account, but nonetheless I think their discussion was interesting and novel. Moreover, a number of the points where the theory was weak were so because they had underdeveloped areas of their account. This leaves room for development, and I think despite some of the problems there is something to the backbone structure of their theory that is appealing. Perhaps a better dealing with some of the central concepts could do better justice to the core of the view they want to defend. But again, I still think that they put enough out there that was novel and that made interesting enough moves (and had substantial consideration of practical questions) to consider the piece worth including in the responsibility and health care literature. Ok, that's my defense of including this as one of the two options for this round. As D&S note, it is possible to make responsible (and I will add: free) choices even when some of the conditions under which the choice was made were out of our control (like someone else choosing which options you get to pick from).

I outline some of my worries in a separate post below.

Bob Kelly June 7, 2020 at 11:51 PM


1. This theory applies straightforwardly when there is a clear conflict for resources (e.g. when they are scarce). But it seems like there might be some (many?) cases where resources aren't scarce, and there isn't any other obvious way in which treating a particular individual will take from others. Perhaps someone needs a relatively quick and easy treatment due to activity that would qualify as a morally responsible violation of a solidary obligation. If such cases exist to any relevant extent, then the question becomes whether we (the system?) must make some extensive effort to redirect those resources somewhere else--e.g. search for someone else we can redirect them to, calculate the cost of using them to treat and use that money elsewhere, etc. If we should do this, then that may start to seem pretty counterintuitive. If not, thenit seems like we might get two dopplegangers who differ with respect to their treatment rights/prioritization due to seemingly irrelevant factors (e.g. someone else happened to get in an accident that evening and needs the medical resources).

2. I wanted more discussion of the role of consent to or foreknowledge about being a part of such a solidarity system. Do we have to know we're a part of such a system for our responsible behavior to undermine our right to treatment? They seem to sort of address this on p. 139 when they say solidarity (and its obligations) might be grounded in other obligations to "not abandon" others. On one reading, this suggests that we don't need to know. There are simply standing obligations towards other people, and some of these are what give rise to a solidarity system (and its obligations). This reading is suggested by their response to the objection that wealthy people might escape the system. If this reading is right, I can't tell if solidarity is supposed to be happening many times over for each individual (perhaps according to how many groups one identifies with, or has sufficient similarity with, or would identify with, or whatever else). If so, then lots of conflicting obligations might arise. If you don't have to know, intentionally be solidary, explicitly consent to the system or what have you, then this will seem to raise more problems for the responsibility side of things. One might be responsible for a behavior without being responsible for that behavior violating the solidary obligation (since they were ignorant about the latter). This needed more focus.

3. Negligence seemed to do a lot of work for them since, as they suggest, many failures to uphold solidary obligations will be failures of intention (i.e. failing to intend what they ought to have intended). This needs much more discussion and defense than they gave it. To be fair, most people are pretty dismissive of anything that isn't an uncritical acceptance of responsibility for negligence. But a common (the main?) reply to people like Levy and Rosen (who argue that we can't ever be justified in responsibility judgments for negligence and that responsibility cannot apply to negligence, respectively) is something like this, "But then lots of people we thought were responsible wouldn't be. Then what?" This is not a good reply to their arguments. All D&S do is cite Raz and give two cases they think supply the intuition that a negligent person is responsible. That is insufficient.


Stephen Kershnar June 8, 2020 at 9:42 AM



Excellent points. Let us consider the following question.

Question: Must one consent in order to be subject to community-based or role-based duties? Perhaps the consent would be to be part of the community or to occupy a role.

Horn #1: Yes. If so, then, in the community case, it intuitively seems that the system depends at least in part on that to which people have consented. In the role-based case, it similarly seems that aggregate consent can change the role. This is true even if roles have Platonic essences and ground duties.

Horn #2: No. If one need not consent, then there is no reason that D & S should be so anxious about autonomy (and golden opportunities that enable it). That is, why focus on autonomy at the level of individual health decisions, but not membership in the particular healthcare system.

Perhaps D & S think that consent cannot effectively operate at the system-wide level (see A. John Simmons' arguments), but can at the act level (Should I eat a box of Entenmann's chocolate donuts for breakfast?). Fair enough.

I then return to whether we should be able to penalize people who overly burden government spending via cash welfare and free or subsidized food, housing, medicine, and school for their children. Why limit the penalties to healthcare? Perhaps D & S think that people are less blameworthy for work and family-structure decisions than they are for alcohol and food decisions.


Steve K

Bob Kelly June 8, 2020 at 12:09 AM


4. D&S also needed to say more about free will (or autonomy) and moral responsibility. They distinguish their INACTIVITY case from the SMOKING and SEATBELT cases on the basis of the former (but not the latter) failing to meet conditions of freedom/autonomy. They risk running the two together, but only actually use "freely" once and stick with "autonomy" mostly. Still, their Frankfurt-style description of autonomy conditions and the like are very much in the realm of free choice, especially when the context is discussing moral responsibility for actions. Lots of people in the FW/MR literature would (and probably do) scoff at the idea that autonomy or acting in character is sufficient for free choice (where free choice is necessary for moral responsibility). If they want to be attributionists, that's fine (not really, but dialectically fine). They just need to be up front about this (mistake) and get clearer on the conditions for autonomy, free choice, and moral responsibility. Moreover, the notion of "Golden Opportunity" required that the conditions of choice be "conducive to responsible choice." They never spelled this out. Perhaps they thought they did with their autonomy discussion (e.g. second-order desire, choice comes from reflection, etc.). But the notions mentioned there are not sufficiently worked out. Is second-order desire necessary or sufficient? Both? Is coming from a process of reflection necessary for free or autonomous choice? Libertarians are likely to deny the former. In any case, these notions weren't treated with enough care. Perhaps if they were, the account could get some legs under it.

5. I wanted more discussion of the problem of working out what it means to be offered help for one's lifestyle change (or healthy choice-making). For a behavior to be morally responsible (and so capable of violating a solidary obligation and undermining one's claim to treatment), they require that a patient refuse a Golden Opportunity. But one condition of a GO being present and refused is that help is offered to support the patient in their lifestyle change or healthy choice. The thought is that this is often very hard to do, and there are competing values and varying resources, and so forth. So, an offer of support is needed to make the refusal unreasonable (justifying the claim that they did not show reasonable or due care for others in the solidarity system). But it's unclear how often this ever happens, or what a sufficient offer of help looks like. For instance, consider a functioning alcoholic who is told by their doctor that they need to quit or drastically reduce their drinking or else they may face some health problems in the future. To support them, they get them in touch with a local 12-step rehab and AA group. Is this enough? Consider the fact that AA and 12-step programs are no better than chance at helping alcoholics quit drinking (see "The Sober Truth" What kind of help is this to offer? Just because most people think AA is really helpful does not make it so, and hence does it make it a good support option for the patient. If he refuses, should we say he was even offered sufficient help? What if he refuses because he doesn't think they work? Again, I just think this condition of offering support for the GO to be realized needed much more fleshing out.

Jack Freer June 9, 2020 at 7:44 PM

Rob, Steve,

I apologize for not addressing your posts/questions yet, but I was distracted by 2 observations.

1. Solidarity. I believe that there are two different kinds of solidarity. The variety mentioned in the article (with UK NHS as the example) represent solidarity by design. The other type is a more organic/naturally evolving solidarity. What I have in mind is the way in which the Amish community pays for the health care of its members. The Amish traditionally do not obtain health insurance (they consider insurance "gambling with God"). They do, however, sometimes become ill and run up big bills. They consider it a community responsibility and they generate the cash via craft auctions (to which everyone in the community contributes). I'm pretty sure participants do not analyze their system, but I also suspect there is negligible resistance to the individual's responsibility to the community. It is unlikely that systems with designed solidarity work so smoothly.

2. Golden opportunities. If I understand GOs correctly, Savulescu, (Savulescu J. J Med Ethics 2018;44:59–61.), gives the example of not choosing ("safer") e-cigarettes instead of smoking. My favorite example is the diabetic who continues to drink sweetened Coca Cola. My GO is usually to suggest Diet Coke (much better risk/benefit profile than e-cigarettes). Over the years, many patients have told me, "but I don't like diet soft drinks" and continue to drink the sweet stuff. A fair number of patients have had dangerously high blood sugars (requiring hospital care) after a cola and candy binge. Is that patient blameworthy? Sure. can we justly penalize that patient? I don't think so because human behavior is so complex that I don't believe any of us can tease out the volitional elements that are purely derived from the individual's free will. There is so much else involved (genetics, parenting, conditioning etc) that it is presumptuous to believe that everything that *feels* autonomous actually is.


Neil Feit June 11, 2020 at 11:19 AM


This is really interesting and your points on GOs also relate to Bob's earlier questions about what counts as adequate support by the provider. (I know someone in a similar situation who gets migraines from the sweeteners in diet colas...) I agree with you about your patients: at least some are blameworthy. And I'm also sympathetic with your claims about penalizing them, but on the other hand perhaps it depends on the nature of the penalty. If we suppose the HC system is a UK-style one, then D&S might plausibly say it's appropriate to penalize such a patient by requiring them partially to cover the costs of their treatment. (Having said that, I'm skeptical that NHS is grounded in solidarity and I'm inclined to agree with Steve's argument that even if it is, this doesn't generate the relevant obligations.)

Jack Freer June 12, 2020 at 9:23 PM

Neil, I agree NHS is probably not grounded in solidarity, and in any case, does not obligate anyone to shoulder a larger share of responsibility to the system because of their unhealthy behaviors. Indeed, if (as I believe) the system is designed to provide medical treatment to all equally, no such penalties are even possible.

Bob Kelly June 12, 2020 at 11:23 PM


Interesting points you make here. I have a few thoughts on these.


1. ONLY ONE TYPE: I think you're distinction between two types of solidarity may be a distinction without a (relevant) difference. First, why should we think that "naturally" occurring solidarity systems like those displayed by the Amish are not designed in the relevant sense? It seems like there is probably something of a social contract in that case, which seems somewhat "designy" and analogous to the contract-like feature of the NHS type. It's also hard to see how it would be implicitly consented to in such a small community with such explicit rules/norms.

2. DISTINCT BUT NOT DIFFERENT: Second, even if it were implicit (not explicitly designed) in the Amish case, what relevance would that difference have to D&S's arguments? That is, what relevance would "by design vs. organic" (whatever these end up meaning) have to whether solidary obligations give rise to duties not to unreasonably externalize system costs, or to the role of responsibility in determining resource allocation? I see the plausibility of *some* difference, but not the relevance.

3. DIFFERENT BUT RELEVANT: Third, suppose you offered a reason to think the design/organic distinction was relevant. Why not think that their arguments just work for designed solidarity systems and not organic ones? Towards the end, you suggest that it is unlikely that designed systems will work as smoothly? Can you say why (i) we should think this is so, and (ii) why we should think the organic ones work so smoothly?


1. EQUIVOCATION: It seems like you are equivocating on 'responsibility' and/or 'blameworthiness'. On the one hand, you seem to think it's pretty clear that (i) some patients are BW for their unhealthy conditions (and need for treatment). You are saying "some patients are BW." On the other, you suggest that (ii) this doesn't mean we can penalize (blame) them because they did not act autonomously due to influencing factors beyond their control. You are saying that "a patient being blameworthy doesn't entail punishment due to undermined autonomy." I don't know how to make sense of both (i) and (ii) unless 'blameworthy' means something different in each. If S is BW, they are an appropriate target of blame (where this might be minor punishment, or being the proper target of blaming attitudes like resentment). That seems analytic. So if they are BW, of course we can justly penalize them. The question is just how much. This is true even if we decide to forgive them. Also, if S is BW, then they acted freely (or at the very least, autonomously). So if there are genetic factors, or upbringing, or other things that would make us withhold blaming them due to antecedent uncontrolled factors compromising their autonomy, then they are, in fact, not blameworthy. So I was confused by the two points being put together.

2. EPISTEMIC PROBLEM: It seems like your BW-ness worry is an epistemic one. You're concerned that we can't really ever know for certain the "springs of one's actions" as Frankfurt might put it. Perhaps. But perhaps there is a standard for knowing being invoked here that is unfair (i.e. not required in other areas of inquiry). In any case, why not think there is still a substantive point being made here? The claim is that responsibility matters in HC decisions if there is a background solidarity system. Whether we can know who is responsible and when would not affect that argument. You might object that without knowing that, the argument doesn't matter. I would disagree. It wouldn't be able to be put into practice until we figured out how to know who was responsible and when. But the theory being true would still matter--i.e. it would ground an obligation for us to figure out how to incorporate it.

Bob Kelly June 12, 2020 at 11:31 PM

Also, in your last reply to Neil you said, "Indeed, if (as I believe) the system is designed to provide medical treatment to all equally, no such penalties are even possible." I don't think this is a fair treatment of D&S's project. The starting assumption, like some previous articles we've blogged about, is that sometimes we have to make decisions about prioritizing resources. When we do, we want to know what the rules are. D&S argue that, in certain contexts (a b/g of solidarity), responsibility for needing care in the first place should count. Others think that QALYs matter. Some think still other factors matter. But either way, these are situations where someone is going to be "penalized" in a certain sense. We just have to pick who, and we do so based on what we think counts. Hence, we take the penalization to be, relative to the circumstances of allocating scarce HC resources, justified. It seems a bit like you want to say that we should always provide equal medical treatment to everyone. This either denies the possibility of situations with scarce resources, or it entails that we do something like not treat anyone when resources are scarce, or split up the resources we do have equally (even if this mean no one gets enough to survive). These options all seem unreasonable. Once we grant that we're in a situation where we are allocating scarce resources, and trying to figure out the principles for doing so justly, I don't see why (i) we should always go with "provide medical treatment to all equally" or, (ii) why we should think that "no such penalties are even possible."

Stephen Kershnar June 14, 2020 at 11:09 AM


Neil, Jack, and Bob:

Great points. Bob makes two strong responses. I want to explore a possible implication of them.

(1) ONLY ONE TYPE. The distinction between two types of solidarity may be a distinction without a (relevant) difference.

(2) DISTINCT BUT NOT DIFFERENT. Even if the two types of solidarity are distinct, they either both justify policy in the same way (likely via consent) or do not justify policies at all.

Part of the problem here is that we don’t know if D & S have a communitarian or role-based theory of government legitimacy and a duty to obey the law (or analogous features of a system. Part of the problem is that they do not even footnote a theory of communitarian or role-based legitimacy and obligation so that the reader can know how they are getting around the usual consent-is-required reasoning.

My question here is a political one. If one agreed with D & S, this would be a good reason (although perhaps not sufficient) to have homogenous populations and local healthcare systems. This is because solidarity-like feelings are probably enhanced by smaller and more homogenous populations and systems. See, for example, Robert Putnam’s work. If this is correct, I wonder if this is a reason to break up the British National Healthcare Service into much smaller parts, for example, town or locality systems. It is also a reason to restrict dissimilar immigration, disfavor integration, and support in-group dating, friendship, and marriage. Anyway, just a thought.


Steve K