Addiction is a Brain Disease (and it matters); versus, Addiction is not a Brain Disease (and it matters)

This blog considers alternate view presented by two papers. Leshner's paper, published by Frontiers in Neuroscience: The Science of Substance Abuse; and, Levy's paper published by Frontiers in Psychiatry, Hypothesis and Theory Article.

Papers 14. Papers by Leshnar and Levy

Blog Comments (70)

Stephen Kershnar November 29, 2020 at 3:53 PM


Both Neil Levy and Alan Leshner suggests that whether addiction is a disease has implications for whether an addict should be blamed. Assume blameworthiness is negative moral responsibility.

Here is the argument to the contrary. 

(1) A person is morally responsible for something to the extent that he is reason-responsive with regard to it. 

(2) The extent to which someone is reason-responsive is unrelated to whether he has a disease (or, at best, indirectly related).

(3) If (1) and (2), then whether addiction is a disease is unrelated to whether he should be blamed.

If the state’s decision to incarcerate ought to depend on whether he is to blame, then this debate is irrelevant to incarnation. Ditto for the state’s decision to provide treatment. 

In short, whether addiction is a disease is irrelevant to policy.



Pat D December 1, 2020 at 10:10 AM


I think Levy is making a therapeutic rather than a legal argument. 

Regarding (2), according to Levy, "Mental illness is quite plausibly identified with a defect of rationality of some kind" [p.4]. Delusion due to psychiatric dysfunction (e.g. schizophrenia) entails reason-unresponsiveness - at least around the focus of the delusional thinking. But Levy never says that addiction is a mental illness - rather it is "neuropsychological dysfunction" that happens to be linked to unfavorable "social embeddedness" [p. 6]. He seems to be arguing that changing the social situation can be therapeutic, but that blame and incarceration don't fit the bill as therapeutic social change. He really does side-step the issue whether addiction entails a "defect of rationality." And he never addresses how "neuropsychological dysfunction" relates to "biological dysfunction" [p. 5] or "brain disease." Despite the title of his paper, he does seem to be saying that addiction is a brain disease (= biological dysfunction + social harm) under certain social circumstances.

David H December 4, 2020 at 5:26 PM


It does seem like Levy believes in some scenarios the addict is harmed and perhaps harmed in terms of rationality or capability to pursue a good life. But I think Levy resists describing such addicts as brain diseased because there are accessible environments in which they will not lose control or be unable to pursue a good life. He has that second condition pp 3-4 that "the pathology is sufficient for the person to be suffering from a disease in almost any accessible environment" and on page 4 "if it is the case that there is an accessible environment - where accessibility is is a function not merely of physical possibility, but also of the costs (economic, social, moral) of actually accessing that environment - in which a dysfunction does not cause an impairment, then that dysfunction is not sufficient for a disease."

Stephen Kershnar December 5, 2020 at 10:47 AM


Pat and David:

You make good points. I still do not see why it matters whether addiction is a brain disease. If it neither affects blame nor what policy we should adopt toward addicts, then I do not see why it matters whether addiction is a brain disease.

In short, what is the argument in Levy that it matters as to whether an addict has a brain disease? 


Steve K

Neil Feit December 5, 2020 at 4:20 PM


I think this is a good question and I'm inclined to think it doesn't matter in any important way. It seems to me that according to Levy, the big reason why it matters that addiction is not a brain disease has to do with approaches to treatment: it might be just as efficacious or appropriate to try to change certain social conditions that sustain the addiction, as it would be to treat the addicted person medically, if the addiction is not a brain disease. But it isn't clear to me why this couldn't be just as efficacious or appropriate to do, even if it's a brain disease.

Pat D December 5, 2020 at 5:45 PM

Steve and Neil, I tend to agree with you that Levy falls short in supporting his claim that it matters that addiction is not a brain disease. 

David, I also think that he falls short in making the claim that addiction is not a brain disease. He is not clear whether characterizing the dysfunctional element of addiction as 'biological' and 'neuropsychological' makes a difference. Are pathological 'learning' or pathological 'wanting' disorders of rationality or not? He doesn't say. If defects of rationality are merely part of the harm condition of 'mental' disorders and not what is primarily dysfunctional (pathological), then why isn't addiction simply a brain disease - albeit associated with harm in a narrower range of social conditions than dementia, for instance? Not that I agree with his characterizing mental illness as a 'defect of rationality of some kind'.

Pat D December 5, 2020 at 5:49 PM

David, I see that your comments below make a similar point to what I am saying here.

Bob Kelly December 13, 2020 at 12:14 PM


A couple points. First, I agree that Leshner is a proper target of this argument but disagree that Levy is. Can you point to the place in the text that supports the claim that he thinks disease and blame are conceptually tied? He does think disease is normative since the dysfunction must be sufficient for impairment to be a disease, where determining impairment can invoke norms like whether it is reasonable to demand someone or some group to act so as to actualize particular accessible environments. For instance, for Levy, if we can easily change the environment so addicts are never triggered/cued to act, then we ought to, and hence their dysfunction is not sufficient for impairment. But nothing about this entails anything about blaming them or not. He does discuss the fact that addicts are often not responsible for their environment and how easy (or not) it is to access another better one. But at best this seems to only entail that (sometimes) certain conditions can be sufficient for determining both disease and blame; though, this even seems like a stretch. Moreover, he starts the article by saying that rejecting the brain disease model does not entail moralizing addiction (translation: doesn’t entail introducing blame). I read this as an implicit way of indicating that whether addiction is a brain disease does not affect whether we blame addicts. He thinks we (often) shouldn’t, whether or not addiction is a brain disease—this is precisely why rejecting the disease view does not affect the responsibility status. If he thought they were connected, it seems like he couldn’t do this; he would have to flip his responsibility judgment as well. But he doesn’t. Finally, as you know, Levy is a skeptic, and so thinks no one is blameworthy. As far as I can tell, he is also (at least very close to being) an impossibilist. Hence, the fact that he thinks some addicts can be (and probably are) diseases and some are not should tell us that he doesn’t think disease and blame are conceptually tied. 

Second, I don’t know if (1) is true and I think (2) might be false. At the very least, I’m not sure you have provided sufficient support for either, considering the plausibility of the following. On (1), it seems like other things might be required for responsibility besides RR-ness. RR-ness is a sufficient level of *control* on Fischer’s view, but it is not clear that he thinks this is all that is required. For instance, he might require a certain level of knowledge and/or awareness, and then claim that RR-ness plus those two features is sufficient for responsibility. Or at least someone could reasonably think this. But perhaps an easy modification of (1) would work. On (2), you would have to tell me what you mean by ‘disease’, or at least how you flesh the account out. For instance, Wakefield seems to (pretty straightforwardly, I would argue) claim that faulty RR-ness amounts to a dysfunction. He doesn’t use Fischer terminology, but that is quite precisely the idea. Hence, on a Wakefieldian account of disease, the extent to which someone is RR is directly related to whether they are diseased, since RR-ness constitutes proper functioning and non-RR-ness constitutes a dysfunction. (2) would seem to be false at least in our world, then. 

I think I agree that blame is conceptually unrelated to disease (or, at the very least, to dysfunction). Though, it obviously depends on the account one defends. I am inclined to accept your conclusion that disease and blame come apart conceptually, but I don’t know that this is the argument to get us there.

Bob Kelly December 13, 2020 at 12:25 PM


I think I agree with the sentiment of your claim about treatment and addiction being a brain disease. However, I would just add that if addiction is not a brain disease, then perhaps it might change the way we think about pharmacological treatments that target specific brain processes. Maybe not. Maybe we shouldn't care whether the brain process is pathological when we decide whether we want to create and administer pharmacological treatments since we might only care about, for instance, the behavioral and other effects in terms of harm and so on. But it's just food for thought. Maybe those who favor an internal morality of medicine might think that doctors should not provide pharmacological treatments to addicts if the underlying processes they affect are not pathological. 

Also, I think Levy is interested in the self-motivation worry as well. That is, he (like Marc Lewis) thinks that it can create a kind of acquiescence or lack of self-motivation to call addiction a brain disease. Addicts can take on the patient role and feel like they are doomed to use. Evidence from studies on motivational interviewing suggest that a personal sense of agency can help in treatment success. I think Levy worries about providing addicts a reason to think they can't help it (even if they can). For what it's worth, I don't think he buys into this “disease=compulsion” idea; I just think he knows that many people (including the folk/addicts) buy into it.

Stephen Kershnar December 15, 2020 at 4:47 PM



Great comments. 

Levy says that it matters whether addiction is a brain disease. My claim is that this is false. 

On (1), I agree that it would be better to use the full guidance control conditions (reason-responsiveness + ownership) and, perhaps, specify the absence of an excuse. However, only free will nerds - for example, Kelly and Kershnar - want that specificity. I rely on the claim, and I strongly suspect you agree, that if people are responsible, then it is because they have guidance control. Other theories (agent causation, mesh, rule utilitarian, and sanity) are inferior. 

On (2), the defense is that whether someone has a disorder (harmful dysfunction) matters only if it lessens or eliminates a person's guidance control. But unless an addiction-disorder is filled out in terms of reduced guidance control, a disorder does not matter. 

What is an efficient or just change of triggering conditions and the aptness of blame matter, but they are independent of a disorder.

Plus, Levy needs to start identifying as an impossibilist. 


Steve K

Bob Kelly December 22, 2020 at 12:41 PM


Thanks for clarifying the claims, that was helpful. Very briefly, I would still say that, on (1), we ought to consider other potential responsibility-making/-grounding conditions besides RR-ness, if there are any. I think it’s plausible that there are (e.g. knowledge of what one is doing). Perhaps you meant that such conditions are built into RR-ness. That might be right (though this isn’t super clear to me), and if so, then point taken. On (2), I think I get your response. Though, it still seems like (2) is worded too strongly. And your modification in the reply seems to make it that a disorder that harms someone without reducing their guidance control does not matter—this doesn’t seem right. But point taken about the connection to addiction mattering when addiction is filled out in terms of reduced guidance control (as Wakefield does).      

Stephen Kershnar November 29, 2020 at 3:57 PM


Neil Levy claims that if addiction is a disease, then the following are true. 

(a) There is a dysfunction.

(b) The dysfunction causes impairment (in the right way). 

(c) The impairment causes the person to suffer from a disease in almost every accessible environment.

In short, an addict has a brain disease if and only if he has a harmful biological dysfunction of the brain. Whether a dysfunction is harmful depends on social norms of flourishing. 

On Levy’s theory, if an environment can be avoided without prohibitive cost or injustice, then it is accessible. 

Here are my objections. 

(1) Not a Biological Dysfunction. A mismatch of a biological sub-system and its setting need not involve a biological dysfunction. If addiction is such a mismatch, then it is not a pathology (biological dysfunction). Levy’s arguments suggest this point. 

(2) Extrinsic Property Reductio. On this theory, two addicted doppelgangers who share the same evolutionary history might differ in terms of whether they have a disease because of the environments that are accessible to them. This is counterintuitive. Intuitively, whether an addict has a disease does not depend the price of methadone. 

(3) Economics and Philosophy. Whether an addiction is a disease depends on prohibitive cost and injustice. This makes the issue of disease depend on economic efficiency and justice. This is counterintuitive. Intuitively, we do not need to consult Milton Friedman or Robert Nozick to determine whether a heroin addict has a disease.


David H December 4, 2020 at 6:12 PM


Regarding your third point. I don't think Levy moralizes accessibility. See his footnote 2 in response to Wakefield's query "Suppose for instance that some kind of biological dysfunction causes an impairment only because people were repulsed by it. Would such a condition count as a disease? (I thank Jerome Wakefield for raising this question) I think we should count any dysfunction that causes impairment as a disease when the impairment cannot be (for practical purposes) avoided.

David H December 5, 2020 at 9:17 AM


Well, perhaps the above quote from Levy is still compatible with justice being consideration of accessibility. He does say "accessibility is a function not merely of physical possibility, but also of the costs (economic, social, moral) of actually accessing that environment - in which a dysfunction does not cause an impairment then the dysfunction is not sufficient for the disease." (p.4). So perhaps if an environment can be reached by acts of avoidable injustice then it is not accessible. But if there is no practical way to avoid the injustice, as the "altering (the unjust social conditions) would be prohibitively costly" then it is not inaccessible. That sounds like a terribly unnecessary complication of the harm condition. I like the way you put it, we shouldn't have to consult Friedman or Nozick to determine whether the heroin addict is a disease

Stephen Kershnar December 5, 2020 at 10:56 AM



Great point. 

As you point out, Levy says that a disorder is a function of impairment, which is a function of accessible environment. An accessible environment, on his theory, depends on costs (economic, social, and moral). 

The problem he faces it that (1) he does not want a disorder to depend on whether dysfunction causes suffering due to external oppression but to rule out such oppression (2) he has to moralize the cause of suffering. 

If I remember correctly, one way that you and Neil have suggested, is to identify disorder with dysfunction. Dysfunction might then be filled out in terms of evolution (Wakefield) or statistical (Boorse) based terms. 

In any case, Levy is risking circularity if he accepts the following. 

(A) Disease matters because it informs correct policy.

(B) Disease depends on correct policy. 

I suspect Levy runs into this problem, but perhaps he has a way out.


Steve K

David H December 5, 2020 at 1:59 PM

Could levy avoid circularity because he is talking about different policies. The policies relevant to whether the something is a disease is whether we can make an environment without too great a cost that will not harm the dysfunctional individual and make his addiction an impairment. The policies for treatment recognize what we have to do when the costs were too high or impossible to make the environment not harmful. So, the latter policies mitigate the problems of an addiction that is a disease, the first policies prevent the addiction from being a disease.

Stephen Kershnar December 6, 2020 at 8:08 PM



I think this is a great point. Here is why I think his account is circular.

Levy seems to be committed to the following claims. 

(1) Whether addiction is a disease matters. 

(2) Whether addiction is a disease depends on justice.

(3) Whether something matters depends on justice. 

Hence, the following is true.

(4) Justice depends on justice. 

Levy is explicitly committed to (1). He cannot avoid (2) because both are the results of all-things-considered rather than localized justice. After all, we're trying to decide what is accessible given society's overall wants, needs, etc. (3) is a standard part of justice telling us what issues need to be addressed.

This is precisely why an addiction theorist needs to choose between it mattering justice-wise whether addiction is a disease and it being the case that whether addiction is a disease depends on justice.


Steve K

Bob Kelly December 13, 2020 at 12:42 PM


I again think I agree with most of the sentiment of the objections here. I didn’t like how Levy hung disease on these sorts of factors. However, I wanted to push back on just two things. 

First, a minor point. Aren’t your (b) and (c) just the same thing on Levy’s view? I am not distinguishing your use of ‘impairment’ and ‘suffer’. I thought the intro and abstract showed he had two conditions for disease: (1) x is or involves a dysfunction, and (2), x is sufficient for impairment “in almost any accessible environment.” 

Second, Levy agrees with your first objection, so this is either not a successful objection to his view or there might be some (overlooked) mistake in outlining the view (perhaps by him and not you). Here is the relevant passage: “There are scientific accounts of addiction according to which it does not involve any brain pathology at all. On the theories I have in mind, explaining addiction requires us to postulate non-pathological brain mechanisms. Consider a mismatch account of addiction…” (p. 2). He uses this as an example of a non-disease account of addiction. Maybe I am misunderstanding your point here.

Also, in your response to David just above, I am confused as to how this represents the circularity. As far as I can tell, (1) and (3) entail (2), and (1)-(3) do not entail (4). You would need your English descriptions "whether..." to have biconditionals as the underlying logical structure, but I don't see why this would be so. It seems obvious to me that mattering and justice don't depend on addiction being a brain disease, and so these can't be biconditionals. The underlying logic would be: (1) addiction is a brain disease only if it matters that addiction is a brain disease; (3) X matters only if X is tied to justice in the right way; hence, (2) addiction is a brain disease only if addiction being a brain disease is tied to justice in the right way. (Sorry, the numbers aren’t in order here because I wanted to keep your numbering for the statements.)

Nothing about this is circular as far as I can see. I agree that it is problematic to link addiction being a brain disease to justice in the way he does, but I am not seeing that those claims can plausibly lead to justice depending on justice.

Stephen Kershnar December 15, 2020 at 5:14 PM



Great points, as always.

Here are my interpretive claims about Levy's account. He says that it matters whether addiction is a disorder. 

Let us assume, other things being equal, something matters if and only if it is a consideration of justice 

(1') Justice depends on whether addiction is a disorder (in terms of how to treat users).

Other things matter as well (such as utility, efficiency, and so on) but I don't see this as affecting my argument. Also, think of justice in general terms.

(2') Addiction depends on justice (in terms of accessible conditions viewed from a normative perspective).

See Levy page 4 note 2. 

(3') Hence, justice depends on justice. 

Here is an analogy. 

(1') Whether the criminal justice system gives people what they deserve depends on whether people's get the resource- and opportunity-accessibility they deserve. 

(2) Whether people get the resource- and opportunity-accessibility they deserve, depends on whether the criminal justice system gives people what they deserve. 

(3) Hence, whether the criminal justice system gives people what they deserve depends on whether the criminal justice system gives people what they deserve.

In short, Levy has to choose between two assertions. 

(A) Justice depends on whether someone is an addict.

(B) whether someone is an addict depends on justice. 

Levy should reject both, but (B) is the larger error.


Steve K


Bob Kelly December 22, 2020 at 1:10 PM


Thanks for re-doing the argument like this—it was very helpful. You’ll have to forgive me, though, because I still don’t see the circularity. (To be sure, I see more clearly how you are trying to claim it is circular, thanks to your thorough breakdown.) I think my worry is that there is misstep going on. First, I’ll accept your starting assumption (X matters if and only if X is a consideration of justice) for the sake of argument—though, we should note that this is a big pill to swallow and somewhat stacks the deck in a certain way. Nonetheless, let’s assume that is true. 

I’m just a bit unsure where (1¢) comes from, or perhaps what you actually mean by it. Taking Levy’s claim that “it matters whether addiction is a disorder” together with your starting assumption gives us: 

(1¢¢) Whether addiction is a disorder matters if and only if whether addiction is a disorder is a consideration of justice.

From here (and his claim that addiction being a disorder matters) we can get: 

(2¢¢) Whether addiction is a disorder is a consideration of justice. 

So far, I don’t think we have your (1¢) or your (2¢) yet. My (2¢¢) just says that (on Levy’s view) we have to consider certain matters of justice to know if a case of addiction is a disorder, but this is not the same as your (1¢), “justice depends on whether addiction is a disorder,” which puts the dependence relation in the opposite direction. Perhaps you might think that my (2¢¢) is or entails your (2¢), that addiction depends on justice. I guess I just don’t see how this would be, and it is your (2¢¢) that allows you to get to the circularity of justice depending on justice. But your (2¢) seems very different from my (2¢¢) and it doesn’t seem to be entailed by it either. It suggests that *what addiction is* depends on justice, and my (2¢¢) only says what *whether addiction is a disorder* depends on justice (i.e. is a consideration of justice). It still seems like there is something extra being put in when you jump to your circularity claim in (3) from the other two premises, and it still seems to be an invalid inference from the content given. 

Stephen Kershnar November 29, 2020 at 3:58 PM


Compare these two approaches. 

(1) Incarceration should depend on blame. 

(2) Incarceration should depend on a cost-benefit analysis. 

There are problems with retributivism [seen in (1)]. Here are three of them. Assume that desert is in part or whole a function of blame. 

First, we have no general permission (or duty) to give people what they deserve. Hence, we cannot transfer such a permission (or duty) to the state. 

Second, once we move to a right-forfeiture theory, blame is no longer necessary. Psychotic attackers and short people thrown down wells forfeit rights even if they are not blameworthy (see Fletcher and Nozick). Men watching nephews drown in bathtubs do not forfeit rights even if they are blameworthy (see Rachels). 

Third, if blame is intrinsic in time and space, and it is, then it will not satisfy proportionality or, perhaps, it will not satisfy any sense of proportionality that underlies a contemporary criminal justice system.

Because of these problems, a cost-benefit analysis should guide incarceration. 

On such an analysis, we should incarcerate addicts per se or who commit crimes when it is it is efficient to do so. Accessible environments, disease, and pathology (biological dysfunction) are at most factors that go into the calculations. 

Economics should guide policy, not conceptual analyses of addiction and disease.



Bob Kelly December 13, 2020 at 12:52 PM


Again, we align on some of the ideas here. I don’t like retributivism, as you know. I'm curious about a couple points, though. 

First, why think we have no general permission or duty to give people what they deserve? Do you think no one ever has standing to blame, independently of the impossibilist position you hold? In other words, if people can be morally responsible, do you think that, still, no one would ever have any standing to blame (or praise)? If so, what is the argument for this, since your first problem with retributivism seems to rest on it? If it's based in impossibilism/skepticism about responsibility, I get the argument. If not, I'm curious what grounds the absence of any standing to blame. 

Second, I thought a retributivist could say that the uncle watching his nephew drown deserves punishment *just because* he responsibly committed a wrong (or blameworthy, if you like) action. As you say, even though he forfeits no rights, he is still blameworthy. So, why is this a problem for retributivism if all they need to say is that S deserves punishment just in virtue of being blameworthy for X; that is, just in virtue of responsibly performing X, a blameworthy action? 

What is the reasoning behind the third problem? 

On the last point, I wonder if you think that health insurance should only cover genuine diseases and/or dysfunctions? If so, then disease or dysfunction status would matter to at least a certain type of policy.

Stephen Kershnar December 15, 2020 at 5:27 PM



My points do not depend on impossibilism. It would be interesting to decide whether it makes sense to punish people if it were necessarily true that people are not responsible. This, however, is a discussion for our next phone conversation. 

We have no general permission to give people what they deserve because rights often prevent us from doing so. Imagine a ne'er do well nephew morally and legally inherits his uncle's wealth even though he killed the uncle [Riggs v. Palmer 1889]. We have no general permission to give the nephew what he deserves. The executor has permission to do so, but what explains the permission is his having a right not the nephew's desert.

We agree that the retributivist would say that a person may be punished because, and only because, he deserves it. The problem is that this is false. It matters whether the punisher has a right to punish the wrongdoer. That is, one can be a pure retributivist or one can believe people have rights, but he cannot be both. 

On the third argument, if blame is an intrinsic property, then results - killing or mere attempt - will not matter. On proportionality they matter, however, because proportionality uses harm or right infringement as at least part of what makes a punishment proportional. Hence, if blame justifies punishment, then we lose proportionality, at least any sense of it that resembles the Anglo-American criminal justice system. 


Steve K

Stephen Kershnar November 29, 2020 at 4:03 PM


Rob Kelly suggests that whether someone is addicted depends on a reference class to determine whether he has systematic control. 

(1) Why not follow Neil Levy and make the reference class a normative rather than statistical notion? 

(2) Imagine a person were the first and only human being in the world. That is, there is no reference class. How do we determine whether a drug user lacks enough control to be addicted? 

(3) Is your control-related reference class restricted in time and space? 

(a) Yes. If yes, then why not confine the reference class to situations that the purported addict might face and adjust them by probability? This would reduce control to expected values (outcomes and probability per outcome). It would also eliminate the reference-class theory. 

(b) No. If no, then does the issue of whether someone is addicted to smoking depend on facts about Ancient Egyptians?



Bob Kelly December 13, 2020 at 1:10 PM


These are good (and hard) questions. Working out the details of the reference class is not easy. However, let me just say briefly, by way of motivation, that it's hard to see what else to do if, as I think is incredibly plausible, addiction is a kind of disposition. This is true even if it is not my kind (disposition to systematic loss of control), and perhaps a disposition like David H. suggests (a disposition to have pathological desires). This is because we both need to say (1) something about a threshold (whatever it comes to, it seems implausible that the disposition is determinate—“fires” with a probability of 1 any time it is triggered), and (2) something about certain triggering conditions not counting (since the fact that a bowling ball breaks when we throw it from the Empire State Building does not make it fragile *even if fragility is realized by breaking*). 

In a certain way, I want to think about the normalizing of conditions through a reference class as a way to test or figure out if the disposition is present. In other words, it is like looking to other possible worlds, plopping the individual in the “right” conditions, and seeing what happens. Think of testing whether a person acts from a RR mechanism on JM Fischer's view. But the important point, for non-Humeans like me, is that these possible worlds/conditions do not make it true that a disposition is present; the opposite is the case—the person having (or lacking) the disposition in the actual (and only) world is what makes it true that the individual *would* behave in certain ways *were* they in all of these possible conditions. 

I think appealing to a reference class can help us to figure out what the proper testing conditions are. Maybe this helps us figure out what to say about the first human, though again this is a difficult case. Maybe closely-related species are the reference class. If there are no other beings, maybe they are the only member and we look to conditions that this single individual has a high probability of being in given the history and laws of nature. The idea is that we want the realization of the disposition to be systematic in this sort of way because we want to be able to distinguish one-off or non-genuine cases (like a bowling balls breaking when dropped from the Empire State Building) from cases where we have a genuine disposition to, for instance, break in the right conditions. Wizards masking the breaking of the vase do not make it non-fragile (indeed, its fragility is why the wizard has to work so hard). Similarly, wizards making a bowling ball break when lightly touched in the right spot do not make it fragile (indeed, its non-fragility is precisely why the wizards have to work so hard). Reference classes, while admittedly difficult to get right, help us to start figuring out how to distinguish the right from the wrong conditions for TESTING whether a disposition is present.

I hope that helps some.

Stephen Kershnar November 29, 2020 at 4:05 PM


David Hershenov suggests that moral responsibility depends on health. 

(1) Why not follow Neil Levy and make disease (and, thus, health) depend on facts about prohibitive cost and injustice?

(2) On your theory, blame (via responsibility) depends on health. On Neil Levy’s theory, health depends on blame (via justice and prohibitive cost). If we combine your and Levy’s theory, health depends on health and blame depends on blame. 

Am I right that combining your theories would produce a circular account? 

(3) At best, health indirectly affects reason-responsiveness. Ditto for the ownership-condition for guidance control. Set the point about ownership aside. 

How much violence does your theory do to John Martin Fischer’s theory?

(4) On your theory of medical-professional roles, a medical professional has a duty to protect and promote health and not to protect and promote disease (or, perhaps, pathology). 

Does it matter, then, whether a heroin addict has a disease in determining whether a medical professional – for example, a pharmacist – should sell him methadone?


David H November 30, 2020 at 11:21 AM


I haven't put forth a theory of responsibility in print (or even in conversation.) I just put forth a modified view of Buss's account of autonomy depending upon healthy functioning in the recent or current issue of Religious Studies and in our PANTC/Romanell Center conversations. I tried in the Religious Studies piece to defend a notion of a true self based upon an Animalist account of personal identity in which our nature and persistence conditions are tied to us being essentially metabolizing, self-integrating, self-maintaining entropy resistors or something along those lines. The idea is that we are essentially living beings and continue as long as life processes do, but diseases are defined (in part) as suboptimal contributions to survival. So diseases are contrary to our nature as living beings that persist in virtue of biological processes thus are alien or foreign to our self. The nature of (mental) disease is at odds with our animal nature as such diseases are what make it likely that our persistence is shortened. So autonomy understood as SELF-determination would render actions generated by mental disorders as foreign to the self and not autonomous. 

I don't think responsibility is the same as autonomy. I would think that there can be many scenarios that one can be autonomous without being responsible. For example, there are healthy conditions that mimic the effects of disease on our thoughts and feelings such as non-pathological pains (pregnancy), fatigue, innate fight or flight reactions etc. which may remove responsibility without undermining autonomy. In addition, acting on the basis of some false information may also undermine responsibility without autonomy as I construe the latter in terms of healthy mental functioning. I don't think this makes autonomy unimportant or uninteresting if it doesn't dovetail with responsibility. Admittedly, it may not be as interesting if autonomy was co-extensive with responsibility.

I am sorry to say that I don't have much of a theory of responsibility, not having given it as much thought as autonomy. So I don't have a worked out theory of the relationship between responsibility and autonomy. I just have the idea that autonomy need not be sufficient for responsibility. 

Regarding your threat hijacking question about medical professional roles :), I believe that a pathocentric essence to medicine doesn't require that health care workers limit themselves to combatting pathologies. I only want them to be able to refuse without penalty from inducing pathologies - e.g. abortion, sterilization, physician-assisted suicide etc. I think they should be to give treatments to people who are not suffering pathologies like pain killers to women delivering babies and mood-altering drugs or therapy to people suffering from non-pathological grief, they should prescribe PMS medicines, drugs to help travelers sleep or stay awake etc. I agree with Alan Buchanan that in the future we should enhance people to offset environmental damage and globalization that makes cancers, epidemics, infertility etc. more likely. So we may have to enhance people's immune systems or fertility or thermal regulation or genetic cancer fighting abilities, just to keep them from becoming sick given that we have changed the world by our actions.

Stephen Kershnar December 6, 2020 at 8:16 PM




Great points. 

One response is that in conversation you hold a health-based view of autonomy. That you don't have it fully in print is a good reason to do so. It's an interesting theory. 

Here is a plausible definition of autonomy, 

"Individual autonomy is an idea that is generally understood to refer to the capacity to be one's own person, to live one's life according to reasons and motives that are taken as one's own and not the product of manipulative or distorting external forces, to be in this way independent." [Stanford Encyclopedia on Personal and Political Autonomy]

Here are my assertions.

(1) What unifies these factors into autonomy is their capacity to contribute or actual contribution to moral responsibility.

(2) If (1), then autonomy is either a necessary condition for autonomy or a necessary and sufficient condition for moral responsibility. 

(3) Hence, if the health theory of autonomy is true, then the health theory of moral responsibility (understood as an important necessary condition for it) is true. 

Do you disagree with (1), (2), or (3)? 


Stephen Kershnar December 6, 2020 at 8:21 PM



Here is my counterexample. 


A physician has permission to refuse to induce pathology that goes beyond the permission that loser-philosophers (e.g., Kershnar) have.


A well-educated lawyer decides he is going to induce pathologies when (1) the patient validly consents to it and (2) it makes the world a better place. Consider, for example, sterilization. 

Does he have less of a permission than do physicians? Really?

Does it matter that earlier in his career he left medicine and became a lawyer? How can this matter? 


Stephen Kershnar December 6, 2020 at 8:25 PM



The Delta Force has a physician-member, Morty Wolfowitz, who serves as a medic but also operates the pathology-inducing ray that is used to slowly kill targets of assassination. 

Is it more wrong or bad if Morty shoots the Ray at a legitimate target than his fellow non-physician Delta Force member, Shlomo Rosenberg? I don't see that this is plausible. 


Bob Kelly December 13, 2020 at 1:27 PM

Steve and David,

Super interesting exchange. Steve, one small (yet crucial) point about the SEP definition. Autonomy is not an idea—the idea of autonomy is an idea. Also, I get that this is just an intro to the literature and so on, but the definition also seems somewhat ad hoc ("autonomy is X, and also no funny stuff..."). 

I like (1) since I also think much of this kind of stuff (autonomy, free will, etc.) often comes back to concerns about whether we should hold people morally responsible. (2) has a typo, I think—autonomy is a necessary condition for *responsibility*, not autonomy. I think this seems right, and it seems to me like most people in the free will literature share this intuition. I think it might only be attributionists, or closely-aligned views, that would hold the stronger version of (2) and make (certain understandings of) autonomy sufficient for MR. I doubt that Fischer and other compatibilists would say that they are "merely" securing autonomy, as opposed to some stronger sense of control. That said, Fischer's semi-compatibilism seems to be ok with getting rid of free will and keeping MR (grounded in RR-ness). Maybe if you pushed him, then, he would say RR-ness secures only autonomy (and not a stronger sense of FW), and this is sufficient for MR.

Bob Kelly December 13, 2020 at 1:33 PM


On your first case, why did you switch to the lawyer inducing pathology? The setup was that non-physicians have less permission to REFUSE to induce pathologies that physicians. This confused me when you laid the case out as inducing pathologies. Anyhow, in both cases, why can't David H. say that (1) the lawyer is not acting as a physician, and (2) the physician and Delta Force member, Morty, is not acting as a physician? You ask why this would matter, but I thought the whole idea behind an internal morality of medicine is that it very much matters whether you are acting within the realm of medicine—e.g. *as* a physician. Maybe this is precisely what you are resisting with the cases, but, if so, I worry that it starts to be somewhat question-begging. If we assume IMM, the objection falls flat and we have an out. If we assume ~IMM, then it has force. We seem to need some independent motivations. (For what it’s worth, I am not fond of an IMM and so am sympathetic to your objections.)

Stephen Kershnar December 15, 2020 at 5:38 PM



My point is that IMM fails for two reasons: (1) It has terrible metaphysics and (2) it is strongly counterintuitive. 

The metaphysics claim is that roles - not consent, promise, or forfeiture - change other people's rights. This is bad metaphysics.

The intuition argument is that it is implausible that Wolfowitz and Rosenburg have different moral rights or permissions when they perform acts benefitting or harming non-consenting people. Does this seem right to you? Ditto for past and current doctors doing the same thing. Again, plausible?

Perhaps a role-theorists will say that past doctors acting as doctors are doctors (functional role identification). But this then makes it mysterious what IMM amounts to. Worse, what happened to the rights and privileges of those who doctors treat? 

In short, I do not know whether IMM theorists always hated rights and liberty or whether this is a new thing. 


Steve K

Bob Kelly December 22, 2020 at 1:33 PM


I take your points, and again, am sympathetic to the fact that IMM faces serious challenges. I was seeing if I could throw a bone out there on behalf of the IMM defender. From a sort of parsimony angle, another worry for IMM that your objections raise is that we can explain much of the rights, permissions, and so on with good old fashioned morality, rather than a separate and isolated IMM. 

Jack Freer December 4, 2020 at 9:48 AM

In reading both the Leshner and Levy, as well as a detailed NEJM review from 2016 

(, I found it's easy to get overwhelmed with the neurobiological minutia, and overlook the main question. Is addiction a "brain disease?" To seriously address that question, we might look at what else it might be. 

Leshner sets up his argument for brain disease by contrasting it with a kind of moral failure on the part of the addict. This view of the addicted individual as weak or bad resonates with a segment of the population. This was true in 1997 when the article was published, but remains true now (although probably to a lesser degree). Levy acknowledges that moralization is one alternative to brain disease (but not the only one). He charactertizes addiction as a phenomenon (probably a disease) that entails brain dysfunction most or all of the time. Levy accepts the neurobiological findings, but does not believe that this body of evidence proves addiction to be a brain disease.

Serious academics do not subscribe to the moral failure theory, and most recognize the complex multifactorial nature of addictive behavior. It seems most believe that the growing body of neurobiological evidence is relevant (although they differ considerably in its interpretation). The difference of opinion seems to be the degree to which the brain dysfunction is the cause, or a manifestation of the disease. Since addiction is generally agreeed to be multifactorial with both hereditary and social determinants, it may be a pointless question. We know the etiology is complex. We know behaviors are mediated by the brain. In so doing, the brain would exhibit many of the neurochemical changes seen in both addiction and in other behaviors (pathologic or not). Why is any of this of concern, and why does the moral failure/volitional theory stubbornly persist?

I suspect the answer lies in the subjective sense of agency and autonomy that humans perceive. We have all been tempted by certain foods, erotic thoughts or other substances or experiences. Sometimes we indulge (with or without feelings of guilt). Although we sometimes describe the urge as overwhelming, we still believe we've made a choice one way or the other. We may use the language of "powerlessness" but we rarely believe it literally. Even the most obviously addicted drug user or alcoholic might say: "I can quit anytime." Whether these feelings are illusions, delusions or stone cold reality doesn't really matter. It has a feeling of volition to the individual.



David H December 5, 2020 at 9:46 AM


In line with what you are saying, there may also be a belief that even if addiction is a responsibility removing disease, people were free and responsible to avoid becoming addicts. The idea might be they knew they could become addicted and were free and blameworthy for not avoiding the addiction creating activity. Of course, it may not be useful to addicts to be lectured and condemned after they become addicts. But their earlier responsibility may be a reason for why people are reluctant to throw scarce resources their way. 

It may not just be our own sense of control in the face of temptation that leads to the moralizing approach to addiction, but there may also be evidence for a belief that while ceasing to use is harder for an addict than a non-addict, people still do stop. Some get themselves help even if they can't stop on their own. So there is some basis for blaming addicts for not trying to quit even if they need help to quit - supportive environments, treatments etc.

Addicts also don't use in front of the police so there is again some control. Lay people who moralize about the addict's failings do have some evidence of some addicts limiting use or even quitting or even trying for long times despite relapsing. And I suspect that even addiction therapy involves some moralizing or motivating people to consider what they are doing to loved ones or to show themselves the respect they deserve and that all suggests some control and some role for moralizing in the sense of appealing to moral considerations, though that doesn't mean demonizing the user or blaming them or viewing the addiction as punishment or maintaining that a moral turn around is all that is needed for an addict to become an ex-addict

Stephen Kershnar December 5, 2020 at 11:23 AM


Jack and David:

I'm worried about this quote from Jack, "Serious academics do not subscribe to the moral failure theory, and most recognize the complex multifactorial nature of addictive behavior."

Consider the following facts. 

(1) There are more former smokers than smokers (note, though, that a smoker need not be addicted).

(2) Smoking cigarettes is one of the most addictive drugs.

(3) If (1) and (2), then people can quit using addictive substances (perhaps via many tries and help from professionals).

Why think people's choices not to quit do not satisfy standard responsibility conditions (for example, reason-responsiveness)? 

(a) It might be very hard to do so, but this does not remove it from the realm of moral responsibility (and blame). It might have been very hard for Gary Ridgeway to avoid killing women.

(b) Most people might fail to do so, but this does not remove it from the realm of moral responsibility. A large number of people fail to avoid adultery. I'm not sure that this decisive. 

While blame is backward- or present-focused, as a side benefit blame might affect people's behavior in desirable ways. 

It might be that it is so hard to quit addictive behavior that the user is in effect compelled. Perhaps so, but we need an argument - whether conceptual or empirical - that this is the case. Is there one out there?

Jack - I'm not seeing your argument against moral blame theory. Perhaps I am missing it. 

Side Note: I'm not responsible for being an overeater, not matter what I say above.


Steve K

Jack Freer December 6, 2020 at 9:15 AM


I don't disagree with your description of moral responsibility and blame. By "moral failure" I was referring to the more judgmental or pejorative sense of a bad or weak person, and its associated stigmatization. Personal freedom of choice plays a large part in development of addiction (I would argue this is more pronounced early in the course of the addictive behavior, before behavioral reinforcement sets in). The complexity of addiction as a behavioral, neurological, genetic and yes, volitional phenomenon is discussed in a 2017 Neuroethics issue, see:


Jack Freer December 6, 2020 at 11:47 AM


Our freedom of choice is not binary--either total or non-existent. Even for a single individual, it is variable. Since Steve mentioned his overeating, I'll go into some detail about my food addiction. When I am in a store (or in these plague days, scanning a website), I have more self-control than I do once the object of my desire is in my home. Even though I have a similar rational view of that box of Drakes Devil Dog cakes (bad for me) in both settings, the proximity of them under my roof makes them irresistible to me. I can think of at least one patient (a recovering cocaine addict) who refused opioid analgesics after clearly painful surgery. He believed he could choose to refuse the medication (even if it meant more pain now), but would not be able to resist using later if he accepted the medication now.


Stephen Kershnar December 6, 2020 at 8:32 PM



Great points. Still, I think that we can ask of an addict whether his moral failing is difference in degree or kind from my failing to avoid overeating. I do not see that addiction adds much beyond an empirical claim about the strength of desire -whether want, enjoyment, or changed belief - we are asking someone to not give in to.

In terms of policy, we could ask what is the cost-benefit analysis of the practice of blaming addicts? I do not know the answer to this, but here is a general rule.

(1) If the price of something rises, then, given the same demand curve, the supply purchased and consumed drops. 

(2) Greater blame for drug-and-alcohol use raises the price of it. 

In short, here is my claim. Whether addiction is a dysfunction, disease, or disease of the brain, has no implications - zero - for how we should respond to it. 

That said, I need greater understanding for my overeating. I blame this on Russian Jewish genes and not my weak-willed nature. 


Jack Freer December 7, 2020 at 11:16 AM


Any number of diseases are best seen as extremes along a physiological, anatomical or behavioral spectrum. Not that long ago, one of the criteria for diagnosing diabetes was a fasting blood sugar level of 140 mg/dl or higher. At some later time, the threshold was changed to 126 mg/dl. Less than 100 is normal; 100-126 is prediabetic. In psychiatry, the markers are behavioral, not physiologic. Bipolar disease includes manic individuals who display the typical signs and symptoms, among them:

*Increased activity, energy or agitation.

*Exaggerated sense of well-being and self-confidence (euphoria)

*Decreased need for sleep.

*Unusual talkativeness. 

When these manifestations are severe and interfere with life function, we call it Bipolar I Disorder. Lesser signs/symptoms are sometimes called hypomanic, cyclothymic or borderline personality disorder, depending on the diagnostic criteria.

I see addiction as an analogous situation. There are people who use psychoactive drugs more readily than others. The spectrum probably starts at one end with the guy who just doesn't like taking drugs of any type. As long as he's not doing irreparable damage, he'll rarely take anything--even aspirin for a headache. Sliding down the spectrum, we find more liberal drug taking behavior, including mild opioids for modest pain. Some may even like the way they feel with a couple of Lortabs under their belt. Maybe others actually binge a bit at times. Some get a little nervous about running out and hoard their prescriptions. None of these people I've described so far have any adverse effects (either physically or socially). By any usual standards of disease (for example, Wakefield's harmful dysfunction), this last group does not represent the disease of addiction. But, I see them on the same spectrum.


Bob Kelly December 13, 2020 at 2:01 PM

I think Jack hits the nail on the head in a few places in this exchange. 

Addiction (and the strength of desires and impaired control that come with it) comes in degrees, and these degrees are related to different types of control impairment, capacity vs. exercising, situational factors and their influence, the range of conditions in which someone can exercise the control they do have (and to what degree), and so on. It simply follows straightforwardly from this that pointing out that addicts can sometimes quit and control their behavior doesn't tell us anything new. And pace moral model proponents (and some of the sentiments from Steve’s comments above), it can’t tell us whether addicts have “full” control, whether the individual addict in question or addicts as a group. Moreover, pace anti-brain disease theorists, the presence of some choice cannot tell us that addiction is not a brain disease since, as Wakefield has pointed out, control and disease are not conceptually linked in this way. 

But the fact that some control remains should be incredibly unsurprising to us because, as we all likely agree, addiction (and the desires and control that are involved) come in degrees. To think that a nicotine addict's ability to not smoke on a long flight, or a kleptomaniac's ability to not steal when a cop is standing behind them, is evidence that these conditions do not impair control is to confuse exercise for general capacity. It assumes that “addiction = literal irresistibility” in every case. As far as I can tell, almost no one really thinks this, and rightly so (not even brain disease model proponents would accept this, I believe, despite their sometimes-confusing rhetoric to the contrary). 

Jack is right that the moral model has lost an incredible amount of steam and subscribers, and I don’t think his claim that no reasonable addiction researcher should say that addiction is simply a moral failing is unfounded. Talk of being 'very difficult' is already enough to suggest control impairment, as long as we are being reasonable about control coming in degrees. This is partly why I think "are addicts morally responsible?" and "are addicts free?" are the wrong questions to be asking. Addiction does not completely undermine agential capacity with regard to the object of addiction, let alone undermine it altogether. Focusing on particular actions in particular contexts is what matters. Clearly there are some conditions that might completely eradicate agency and the ability to exercise control or ever be morally responsible. Addiction just does not appear to be like this, given the evidence. But this doesn't mean that it can't impair the capacity for control in the sense that one's actions are less controlled (or fully uncontrolled) in certain instances (like when the right—or wrong!—conditions are met). It would be like asking if dyslexia undermines intellectual ability—not entirely, obviously, and it depends on whether we test that ability with reading or not, and so on. It's just not the right question since it doesn't allow for any nuance and it fails to appreciate the relevance of degrees and conditional factors.

Bob Kelly December 13, 2020 at 2:07 PM

Steve, as a sidenote, the website you sited for (2), that addiction is one of the most addicting drugs, sites two studies and both seem to confuse dependence with addiction. Not to say nicotine isn't addictive. This just seems like a problematic way to operationalize addiction since dependence and addiction come apart. 

We should also be cautious about data on recovery and "quitting without treatment" since (1) most of this data rests on the DSM criteria for inclusion as an addict, and (2) that addicts can (or even often often) quit without treatment tells us nothing about whether their control was impaired. We would need much more precise data to know about their capacity for control (and this is even setting aside the worry in (1) about whether they were truly an addict in the first place, as opposed to a DSM-addict, which almost certainly overdiagnoses addiction and other disorders, as Wakefield has argued). 

We do not need an argument that “addiction = compulsion” (in the sense of literal irresistibility). This is an implausible view, as Hanna Pickard and others have pointed out, and it is unclear whether even people like Nora Volkow would really subscribe to it if pushed. In any case, we don't need literal irresistibility for impaired control.

David H December 4, 2020 at 5:05 PM

Conflating Symptoms of Addiction with the Disease of Addiction: 

Leshner writes “what we now know to be the essence of addiction: comprehensive drug seeking and use, even in the face of negative health and social consequences” (p. 46). Compulsive drug seeking and use, irrationality, Rob Kelly’s Systematic lack of control, strike me as describing symptoms of addiction, not the underlying condition. Symptoms of diseases are often just likely rather than necessary and so we shouldn’t be surprised if there are addicts who don’t have systematic control failures or engage in comprehensive drug seeking. The pathological dependency or craving is what I take addiction to be as that explains and unifies the following cases:

1) babies born addicted who have no compulsive drug seeking and using capabilities or systematic control to lose,

2) injuries that eliminate one’s faculty of control don’t make addicts into non-addicts,

3) the drug dependent baby doesn’t become an addict when developing some ability to seek and use drugs or control that can then fail,

4) addicts who exercise superhuman exertion of control in the face of the constant cravings that preoccupy their minds and keep them from thinking about other things,

5) The addictive drug vanishing forever from the earth so there is nothing to pursue and use or lose control over etc. but the dysfunctional craving remains,

6) Someone with normal control but exceptionally strong cravings that no one normal person could resist is still an addict – such a person is akin to how we view and excuse crimes of passion for none of us could resist such provocations,

7) My account also doesn’t treat someone as an addict if they have normal strength desires but impaired control mechanisms akin to a wanton and so act (often inappropriately and harmfully) on any desire that arises.

(Incidentally, Levy’s comments on Berridge and Holton at p. 3 “the role of dopamine is incentive salience, not learning…addiction is a pathology of incentive salience and not reward prediction. It does not involve pathological learning: rather it involves pathological ‘wanting’” suggests their view may be similar to what I have been pushing. I haven’t read their article so I don’t know if their view is qualified in a way that would rule out addiction in cases 1-6 above.)



Bob KellyDecember 13, 2020 at 2:44 PM


Thanks for raising these issues. I get your concerns, but I think they misunderstand my view at times and can also be overcome in any case.

First, Leshner's characterization of the essence of addiction (and control/compulsion) is not great and I do not subscribe to it. Thus, any objections that are leveled against him are not automatically leveled against my own view. For instance, I don't require harm for addiction, and I don't take "continued use despite negative consequences" to be a satisfactory definition (or component) of ‘addiction’ or ‘impaired control’. 

Second, you say "Rob Kelly’s Systematic lack of control, strike me as describing symptoms of addiction, not the underlying condition. Symptoms of diseases are often just likely rather than necessary and so we shouldn’t be surprised if there are addicts who don’t have systematic control failures or engage in comprehensive drug seeking." I think there is an implicit equivocation here between the disposition (a realizable property) and what realizes it (a process). I identify addiction with the *disposition*, and not the processes (e.g. addictive behaviors, failures of control, etc.) that realize the addiction disposition. Hence, in this sense, I totally agree that you can have an addict who never fails to control their desires (and so, in your terminology, never expresses the symptom of impaired control). But (assuming they are an addict) they would be disposed to (systematically) do so on my view in the right—or wrong!—conditions. This is just like a glass vase being fragile despite bubble wrapping it and locking it in a safe so that the disposition to break is never realized in a breaking process. So, for addiction, the *actual* process of failing to (fully) control some behavior is, as you say, not necessary. Our views align here. 

Third, you go on to say, though, that "the pathological dependency or craving is what I take addiction to be..." As we've discussed before, I think the plausibility of this view will depend on you cashing out dependency and craving a bit more. Dependency as understood in the literature is typically considered to be distinct from addiction (hence the change from DSM-4 to DSM-5 in this regard, non-addicted yet dependent pain treatment patients, addicted but not-yet-dependent individuals, etc.). If by ‘craving’ you mean something like desire, then I don't see how you are going to distinguish addiction from other conditions that involve the disposition to undergo pathological intrusive thoughts that include desires. I also just find it very counterintuitive that someone who has literally zero problem controlling these desires (in the moment, in their ability to get rid of them through indirect means, etc.) is an addict. In my view, the difference between such a person and someone else with the same desires who cannot (sufficiently) control them just is the difference between an addict and a non-addict. 

I address your cases below.

Bob Kelly December 13, 2020 at 2:47 PM

Let me try to say something about each of your cases. 

On (1), babies are born dependent, not addicted. We should not let colloquialisms that push people to describe these babies as "addicted" push our metaphysics around. A really good explanation for why they are merely physically dependent and not addicted is that control is not a factor yet (which seems to align with a view of addiction holding that control is impaired to some degree, like mine). 

On (2), I would want more information about how the injury affects control. If I read it straightforwardly and assume that their *entire* capacity for control (with respect to anything) is destroyed, then dropping the addiction label becomes much less counterintuitive to me. Note that this does not rule out comatose people from being addicted (they might retain the disposition for some time). But once we stipulate that any capacity for control is destroyed, we are back in the baby case—they lose their addiction and become merely physically dependent (disposed to tolerance and withdrawal, understood physiologically). 

(3) is already taken care of by the response above—on my view, yes, they do. 

(4) is also addressed above. Can they act indirectly so as rid themselves of those desires they keep successfully resisting? If not, they do not have full control over them (since even non-superhumans can do that). If so, they just don't seem addicted to me (I am just confused as to why they don't act to rid themselves of those desires yet—maybe they like the challenge since they know they can always defeat it, which seems like a non-addict to me). Also, I would be curious to see how you distinguish this recurring desire they experience (and always defeat) with other conditions like rumination or intrusive thoughts that may not entail addiction. 

(5) seems relatively easy to get around, since it seems to confuse the disposition with its realization again. A fragile vase we bubble wrap and stick in a vault never to be touched again is still fragile. An addiction that doesn't (or can't) manifest is still (or, at least, can still be) an addiction, and my dispositional view helps us capture this. Think of an addict dropped onto a desert island. If we want to say his addiction doesn’t thereby disappear, then the addiction must be dispositional. Hence, the absence of actual uncontrolled behaviors is not inconsistent with the presence of an addiction on my view. 

On (6), I think the case is under-described in a certain sense. I agree that they are not an addict if this is a one-off case. Again, my (systematic) dispositional account helps us capture this. But if they systematically undergo this exceptional craving and lack the capacities to control it (in the moment or indirectly by acting to reduce or eliminate it), I find it hard to believe they have "normal control" (since, again, normally we have the capacity to indirectly work on our desires). I think we need to be careful about the ease with which we can describe people as having "normal" or "superhuman" control alongside "exceptional" cravings. The details of the dispositions are difficult to get right, but they matter.

On (7), I don't see why this is a virtue to not call them an ‘addict’. Do you mean what Frankfurt meant by 'wanton'—they don't have the ability to care? I’m not sure why the *reason* someone fails to control their desires matters. 

Lastly, by 'wanting' Berridge is talking about being moved to act (contrasted with 'liking'). Their view is going to be tightly connected to our control mechanisms since the wanting they are talking about is directly tied to our reward and motivational systems in the brain. I don't think this is in line with what you mean by desire, but if it is then, once again, I don't think you and I are going to disagree as much as you think.

David H December 4, 2020 at 5:06 PM

Mere Semantics? 

Could it be that my opponents and I are just talking past each and describing different phenomena as addiction – I am describing a dependency or pathological craving and the others are describing the harmful pursuit or loss of control etc. Rob Kelly and Jack Freer seem to have taken that view in the previous blog pointing out that I am calling “addiction” what they and addiction researchers call “dependency.” If I understand them correctly, lay usage about addicted babies is inaccurate as they are just dependent and not addicted. Does this mean there is nothing philosophical or scientific to resolve our differences and we should just regulate our language? Perhaps take a poll– or limit the votes to the addiction experts - and settle upon shared usage or accept ambiguity? I think not but suggest further philosophizing should convince those who were calling addiction what I take to be a symptom to use “addiction” as I do for the underlying pathology. I will mention here and explore further three reasons for those who don’t share my linguistic intuitions in the cases above and instead distinguish addiction from dependency/cravings. The first is that my opponents will still have to treat addiction as a symptom of the pathological dependency regardless of their preferred account. The second is that a cure for the dependency or what I am calling addiction will also eliminate the symptoms that opponents are treating as addiction, while the converse is not the case. Third, the dependency/craving explains the symptoms of compulsive search and use or systematic lack of control, not the other away.

David H December 4, 2020 at 5:06 PM


Addiction researchers (we’ve read Levy, Foddy and Savulescu) don’t all believe that addiction is a disease. Some like Rob Kelly may be open to the possibility but seem to be leaning against treating addiction as a disease. If they are distinguishing addiction from dependency and pathological cravings that a newborn can have or someone who is so cognitively impaired that they have no control to lose, they still have to admit then what they are talking about is also a symptom of a disease (the craving or dependency). That may come as a surprise to them and a reason to reconsider their account. They may have assumed that their research into the nature of addiction was aimed at explicating a fundamental condition, not a derivative one. Finding out that what one is describing as addiction is symptom of a disease and not a more fundamental or stand-alone condition may lead one to reconsider what addiction is. That is, they may be assuming that addiction may be or may not be a disease, but it definitely isn’t a symptom. They may be leaning towards addiction not being a disease, but open to the possibility. What they are not open to is that addiction is a mere symptom of a disease. But what those who deny that addiction is a disease are doing strikes me like doctors diagnosing condition x in the following scenario as fatigue. Imagine someone present to doctors who diagnose him with a condition x and offer a definition of condition x as systematic lack of strength and energy in normal activities. If doctors then discover that there is a disease that caused the fatigue, the fatigue is but a symptom of that disease, wouldn’t they reconsider claiming that condition x was the fatigue and instead except that condition x was the underlying disease? Would they really insist they were all along trying to define a symptom? So just as the realization fatigue is a symptom of a disease and not condition x, I think researchers of addiction should resist that the compulsive pursuits and use of drugs or the systematic loss of control is but a symptom of an addictive disease.



Bob Kelly December 13, 2020 at 2:59 PM


A question that came to me in reading your post was: Can a disease cause a disease? If it can, then I don't see why your line of reasoning is at risk of being undermined here. But let us set that aside. My main worry here is similar to that from the exchange a couple posts up, which is that the difference between the disposition and its realization (the process that realizes it) are being conflated. The underlying condition in your case simply seems to be the disposition. That's what I would have been calling the fatigue all along. It doesn't matter if it manifests, just as it doesn’t matter to whether the vase is fragile that it actually falls and breaks. Having the fatigue (or the fragility) is about having the underlying disposition. Similarly, having the disposition is what matters in addiction—and this disposition *must* include the disposition for certain desires on my view, since the disposition is about the ability to control those desires. Again, I don't actually think we disagree about that much here. The problem seems to keep coming to an equivocation on the disposition and its realization. For instance, we could just call the realization of the disposition (i.e. the *actual* uncontrolled behaviors/choices) 'symptoms' of addiction. But if one lacks any capacity whatsoever for this realization, it seems they don't even have the addiction that such uncontrolled behaviors/choices could be a symptom of. It is a good thing we have dependency (disposition to withdrawal and tolerance) to explain the baby cases you appeal to. Another minor worry is that you I’m not sure what exactly you mean by 'dependency', 'craving', or 'desire', and perhaps things would change on particular views of any of these. 

Also, I have strong reservations about addiction being a "natural kind" as opposed to a defined class. I'm not sure how much hangs on this, though.

David H December 4, 2020 at 5:07 PM

Cures and Removals: 

Imagine that an addict can undergo two sorts of interventions. One that leaves the craving and dependency but enables him to exercise systematic control or the other that removes the craving and dependency and consequently enables them to exercise systematic control because they no longer have the overwhelming craving. Which treatment cured their addiction? I would think the cure is the one that got rid of the dependency and its symptoms, not the one that just removed the symptom. (I leave open the possibility that downstream pathologies have left control mechanisms or impulse inhibition damaged so they are impaired even after the dependency/craving is gone. This would be a promising way for loss of control to be a disease itself as well as a symptom of another disease.) According to my opponents, addiction is cured by just restoring control, not restoring control and the removing of the dependency/craving. I believe the underlying cure of the disease and symptom is what people mean by curing or removing addiction. When put in terms of disease and symptoms, does anyone really want to insist that addiction is just a symptom? When you cure an addict, you don’t just remove a symptom of a disease.



Neil FeitDecember 5, 2020 at 4:32 PM


This is really interesting. I find myself inclined to say that both of the treatments that you describe count as cures. Having said that, I don't have any firm views about the nature of addiction and haven't read much except the articles we've been blogging on. But I guess that if I think they're both cures, then I think that the inability to exercise systematic control is not just a symptom.

David H December 6, 2020 at 3:00 PM


There are downstream pathologies - one disease causing another. Doctors could then cure one without curing the other, even if one caused the other. But if loss of systematic control is not a disease, then it can't literally be cured. (Perhaps you won't care about the semantics and are happy to talk about removing rather than curing a condition.) Whether or not it is a disease if it is cured by removing the upstream pathological craving, the fundamental disease, the one that explains the presence of the other, is the pathological dependency/craving. I think that is some reason to call the fundamental, upstream, Ur causal explanatory condition "Addiction", and to call the loss of control a "symptom of addiction but not addiction itself". It is a symptom even if it is a downstream disease produced by the pathological craving. I think those interested in the nature of addiction are ultimately interested in the condition (disease) that explains the others and so usage should reflect that. That is why brain sciences call addiction a brain disease for it explains symptoms such as the loss of control (Kelly) or the compulsive seeking and using (Leshner). On the other hand, perhaps this is too much of a fuss about language and we should be less interested in labels and more interested in just explaining.

Bob Kelly December 13, 2020 at 3:14 PM

I agree with Neil that both cure the addiction (if it is a disease in this case). Again, though, I want to know what you mean by craving and dependency. The latter is typically understood as tolerance and disposition to withdrawal, which again comes apart from addiction according to many diverse views. This is because impaired control unifies almost all views (except rare and outlier strict moral model views). Dependency might influence control (since withdrawal can get really bad), but it's distinct. 

Also, it's quite hard to distinguish the second case from the first. If one "removes the craving and dependency and consequently enables [the addict] to exercise systematic control," I see no reason not to think you have simultaneously restored control. You might say that the first case is still different because it "leaves the craving and dependency," but if you "enable him to exercise systematic control" in that case anyways, you have (at least indirectly) changed the condition so that the craving will now be subject to reduction and elimination. I think craving and control are much more tightly linked that you allow. This is precisely why my dispositional account holds that the disposition which constitutes addiction (involving systematic loss of control) must involve desires. It's also why I find the cases under-described. If you have "full control" but retain these "pathological cravings," I don't know how to make sense of this. How can you have full control, but you can't get rid of these cravings? Normal people with normal ("fallible") control can indirectly adjust their desires by setting up their environment, for instance, as Jack suggested with his food addiction. It seems like the connection between desires and capacity for control is lost in your view, and it may be because I don't know what you have in mind by desires and craving. Adopting the Arpaly and Schroeder view helps, I think, which is why I incorporate this into my account. Desires have to do with representing rewards (in a sort of wanting, not liking, sense, as Robinson and Berridge suggest), where this is about the contribution to reward calculations by our reward and motivation systems and the adjustment of our behavioral dispositions on this basis. I think we should be cautious of letting philosophical desires "float free".

David H December 4, 2020 at 5:08 PM

Explanatory Power: 

Understanding addiction as disease of dependency/craving can explain the symptoms of irrational pursuit, harmful usage, loss of systematic control etc. The converse is not the case. Aren’t addiction researchers interested in explaining more than less with their accounts? If so, they should accept my approach of disease and symptoms. Or if they resist the disease/symptoms approach, they should non-pathologically substitute something of explanatory comprehensiveness of a condition (cravings/dependency) and its consequences. The former is always present, the latter are like symptoms, likely but not essential. Why limit oneself to saying someone whose brain injury removed their control faculty so they couldn’t have failures to control for they had no control at all, is no longer an addict? Control failures are likely and symptomatic but not essential to addiction



Bob Kelly December 13, 2020 at 3:17 PM

I think it is right that *actual* control failures are very likely (maybe we can say 'symptomatic') consequences of having an addiction. Harm is the same way. Still, I don't think one needs to be awake to be an addict. An addict that gets plopped onto a desert island, or suddenly goes into a coma, is not thereby a non-addict. Hence, actual behaviors (manifestations or realizations of an addiction, which is a disposition) need not be present for the disposition to be present. We are agreed here. 

David H December 4, 2020 at 5:10 PM

Responding to a Reductio with a Reductio: 

I always thought it was a reductio of the social harm requirement to disorder/disease/pathology if someone who was say pathologically infertile in one society would cease to be disordered/disease/pathological if moving to say a Shaker society or other society that didn’t want children. But Levy’s response to this absurdity is not to abandon the harm requirement but to absurdly claim that infertility would never be a disease if there are accessible communities where it is not harmful. He didn’t use the example of infertility but instead used homosexuality and peanut allergies. If one’s allergy isn’t a disease because peanut-free environments are accessible, then doesn’t that mean hypothermia not a disease because one can relocate towards the equator, and altitude sickness is not a disease because one can easily live in flat lands, and malaria is not a disease if we can cheaply drain swamps or move from them? If these fail on the ease and cost condition, we can just help ourselves to thought experiments where they aren’t costly – heck, why not free, close fitting portable bubbles for those without immunes systems that are no different in burdens than putting in contact lenses. Or we can perhaps more easily imagine viruses, snake bites and drug overdoses ceasing to be pathological if everyone is easily and cheaply vaccinated or given preventive anti-venom or invent a preventive Narcan or other preventive naloxone-like drugs.

David H December 4, 2020 at 5:10 PM

Plant Diseases? 

I wonder if Levy will accommodate diseases in the plant world by allowing the mindless bushes and grasses to be harmed. I imagine the harm would be the plant being unhealthy and that involves no distress but just a greater probability of not surviving or reproducing. Why not use that criterion for human diseases?



Stephen Kershnar December 5, 2020 at 11:05 AM



Excellent point. Here are my concerns. 

(1) An individual can be harmed by an event only if things go less well for it than they would have but for the event. I just do not see how this can be true for a necessarily mindless entity. 

(2) Survival matters in biological terms only if it leads to reproduction. Yet we can imagine individuals similar to worker bees that are genetically designed to be sterile in order to support the hive. Yet they seem capable of having disorders.

Side note: Worker bees are not genetically infertile, but one can imagine a colony in which they are.

(3) What matters is magnitude of reproduction not reproduction per se in terms of both an individual lineage (with a good probability of continuing) and a species. This suggests that we should look to something like Boorse's biostatistical model. 


David H December 5, 2020 at 1:47 PM


I think the mindless can undergo fluctuations in well-being. They are obviously not fluctuations in desires or pleasure, but I think those are bad theories of well-being. So the theory of well-being is an objective list theory. The only achievement for the mindless is health. When they are healthy, they have high wellbeing and are flourishing, when they are unhealthy, they are doing poorly and have ill-being or low well-being. We can speak NON-metaphorically of things being good for the mindless. I am limiting "good for" to living beings. I have an attributive account of the good where we look to a thing's nature and determine it is good when it is realizing those capacities. 

The costs of denying well-being to the mindless is that you can't explain why it would be a benefit for the mindless to become conscious on a comparative account of benefits and harms. that would involve comparing the well-being of the mindless with the well-being of the minded. But you deny the mindless have well-being. So it won't be bad for fetus to keep it permanently unconscious or impair it so it never becomes conscious and grows into a mindless adult. You can't explain why is is bad for the minded to become mindless when they are mindless. (you can only timelessly compare two lives - one which lapses into unconsciousness and the other that doesn't - and so must give up a time of the badness of being comatose since there is no conscious well-being at that time.) If you recognize interests of the minimally minded in not being killed or tortured that they are then unaware of because of their cognitive immaturity, then consciousness is doing no work as those interests were there before the minimally minded became minded. If you have a desire satisfaction theory of well-being, then you can't explain why it would be wrong to make an infant so impaired that it never regrets it because it is just minimally conscious. There might also be an experience machine reductio for the hedonist who keeps the newborn ecstatic but undeveloped

Oderberg believes that a theory of "good for" an organism is not enough but we need a metaphysics of good per se. He follows the Scholastics in claiming that the good involves perfecting (fulfilling, manifesting) dispositions of a thing's nature and so even inorganic entities can be good. Isn't it good if something does what it is supposed to do given its nature? If so, then a mountain is good at resisting force, producing freshwater runoff, affecting the climate, it may even be a good approximation of a triangle or frictionless plain. So there still resonates with us moderns a sense of good that is found in the inorganic as well as the organic and the sentient world. Oderberg believes organic good for and conscious goods need a deeper metaphysical foundation. I can see your incredulous stare

David H December 5, 2020 at 1:51 PM


reproduction can be construed broadly to include furthering the reproductive success of kin. So the bees designed to be infertile are not malfunctioning if they benefit reproduction of Kin. E.O Wilson once conjectured that homosexuality was not a malfunction on that basis. not a flattering comparison

David H December 5, 2020 at 2:15 PM


I also deny that survival only makes sense in terms of reproduction. I think when they conflict, reproduction determines proper function. For instance, most creatures are designed to go out of existence when they reproduce as most creatures are one-celled organisms. But I think we can speak of proper functioning for entities that don't evolve or reproduce and that involves looking at what they did earlier to maintain themselves in existence. So an organism that didn't have a living ancestor (first and only one of its kind) and was not able to reproduce could still die due to a malfunction if it didn't do what was essential to continue to survive. I don't think reproduction is essential to an organism, but life processes are. So post-menopausal females can still malfunction even though they can't contribute to reproduction or perhaps even reproduction of their offspring's offspring or more distant kin.

Stephen Kershnar December 6, 2020 at 8:39 PM


Here is my argument that survival makes sense only in terms of reproduction. 

(1) If an organism has a basic function - for the purposes of disease -, then it is the end-goal of evolution.

(2) The end-goal of evolution is, and only is, reproduction.

(3) If (1) and (2), then the basic function of an organism is reproduction. 

Other functions (combat, intercourse, and looking good in little black dresses) are mediate goals. 

Consider your intuition about an organism without any ancestor - rather than no living one - and that was infertile. You argue that it might still have a malfunction. I deny this. What was its function? Why choose longevity over happiness or a meaningful life? I can see no reason other than to avoid prioritizing reproduction. 


David HDecember 7, 2020 at 2:26 PM


see my article "A naturalist response to Kingma's naturalism" in the last issue of TMB explaining why survival is a goal of medicine and not happiness and meaning or just reproduction. The very short story is that we are essentially living beings, not happiness seeking or meaning seeking or reproducing beings. Given our nature and persistence conditions, it makes sense that a goal of health is survival, i.e. to continue instantiating our essence. I think reproduction makes sense as a goal as well, especially if one favors etiological accounts of function. Happiness and meaning don't make as much sense of our anatomy as do reproduction and survival.

David H December 4, 2020 at 5:11 PM

Throwing the Normativist a Bone: 

Instead of having social harm do so much work in a theory of disease why not just build its likelihood into the definition of dysfunction by understanding the latter as making death and reproductive failure more likely. In other words, disease is a suboptimal contribution to survival and reproduction. Certainly, early death (and perhaps reproductive failure) is usually a harm, and so I hope that association should be able to accommodate the normativist’s intuition. I think it is this close association of disease and the harm of death and infertility that made it attractive in the first place for the normativist to define disease in terms of harm. My armchair genealogy of normativism may explain the normativist’s mistaken tendency to define rather than just associate harm with disease. These normativist leanings could perhaps be further accommodated with some relativization of environments so say indigenous Americans are not sick if they lack immune responses of the European explorers and settlers, who would be diseased if they lacked them giving their different histories



Stephen Kershnar December 5, 2020 at 11:10 AM



My concern about suboptimal contribution to reproduction is the following. How is magnitude of contribution measured? It could be measured in terms of viable offspring times the probability of these viable offspring themselves reproducing. This will allow us to distinguish between r- and K-selection. 

Also, optimality will have to be judged against the environment-friendliness to survival adjusted by the amount of competition. 

Still I'm not sure what an equation for all this would look like. Thus, I'm not sure that we can judge degree of optimality given that I'm not sure what to put into the denominator. 

Contribution-optimality = 

(number of viable offspring) / (ideal number of viable offspring)


Steve K

David H December 4, 2020 at 5:12 PM

Pathological vs. Disease?

Levy writes “Addiction is not best understood as a brain disease, though it certainly involves pathological neurophysiological dysfunction…I will argue that there is a case for saying that the correlates of addictions are pathological, these correlates are not sufficient for the person to have a disease in some accessible environments.” (p. 1). This makes it sound like “pathology” and “disease” are not synonyms. Or should this be read as implying the correlates are not pathological either in the accessible environments? Or is the idea that diseases are of the whole person, while pathologies are of the parts? I suppose it is the latter as the next sentence is “Regardless of whether the correlates are themselves pathological, the person has a disease only insofar as their functioning as an agent is impaired, and in many environments the correlations of addiction are not sufficient for impairment.” I think if one is going to insist that disease involves impairment and not just dysfunction, then one shouldn’t label the brain dysfunctions as pathological. One should say they are dysfunctions that aren’t pathological because they don’t bring impairments (are impairments necessarily harmful?). At the least, one should be clear how one understands the meaning of pathology and disease since prominent philosophes of medicine use disease, disorder and pathology interchangeably. “Impairment” should also have been clarified. I take it they are loss of capabilities that are harmful in the appropriate environments.



Bob Kelly December 22, 2020 at 2:39 PM


I don’t have much to say besides agreeing that Levy needed to be much clearer with his terminology. I would just add that, as you seem to suggest, I think he has something like a Wakefield-style view of disease in mind, and therefore, is trying to say that the underlying neurobiology might be dysfunctional, even if the person is not diseased (because of a lack of impairment/harm). Though your quote you point to suggests that he might want to say that the underlying neurobiology is indeed diseased while the person of which that neurobiology is a part is not. This seems to bring him much closer to a semantic disagreement—addicts (the person) are not diseased even though, in virtue of having an addiction, they have diseased parts. In short, I agree. His terminological usage was less than ideal.   

David H December 4, 2020 at 5:13 PM

Levy has Too high a standard for disease: 

Why must it be a failure of agency or rationality? He provides examples on page 5 where even heroin addicts with resources to obtain heroin from safe sources need not be diseased. He claims that cravings are not enough for disease because people can avoid the cues that elicit them. But why isn’t avoiding cues for cravings like the hemophiliac avoiding sharp objects? Still, the hemophiliac is diseased. Perhaps Levy will insist that avoiding bruises (internal bleeding) and cuts is not as easy as avoiding cues. But playing around with the probabilities of hypothetical different societies in making cuts and cues for craving rendering them equally likely suggests to me that the dysfunction is sufficient for disease.



Bob Kelly December 22, 2020 at 2:39 PM


This is a nice point. I wonder if Levy is thinking about disease (or addiction) in dispositional terms. This seems to be how we make sense of your point about hemophiliacs. It is because they are disposed in certain ways that they remain hemophiliacs (and diseased) despite never coming into contact with anything sharp. The point is the same whether we think of this as a disposition to bleed abnormally or a normal disposition to clot being masked or otherwise dysfunctional. It comes to one’s being properly (or improperly) disposed, and I think Levy is failing to see disease and addiction this way at times. At the least, this would help him with your worry here. 

David H December 5, 2020 at 1:21 PM

Levy’s Three Inconsistent Claims?

There seems to be an inconsistency or a tension between the following three claims:

A. Addiction is not a brain disease (p. 1 Title of article. Also “Neuroscientists have mistaken some necessary conditions of the disorder with the disorder itself” p. 1)

B. “Addiction is often a disease” p. 1 Abstract, and also on p. 1 “nevertheless, a strong case for saying addiction is often a disease” and again on p. 1 "Addiction is a disorder of the person, embedded in a social context" and p. 6 “Addiction is a pathology…” 

C. “Addiction may always involve brain dysfunction” abstract p. 1 and page 6 “Addiction is a pathology that involves neuropsychological dysfunction”

Claim A is true because there are accessible environments where the brain dysfunction doesn’t produce impairments in rationality and agency or pursuit of the good. pp. 4-5

If Claim B is true that must be because there are sufficient accessible environments where addiction leads to disruptions of rationality or agency

However, if there are sufficient environments for addiction to be a disease and Claim C is true that addiction involves brain dysfunction, then there should be sufficient environments for brain dysfunction to meet the second condition (impairment/harm) for disease



Stephen Kershnar December 6, 2020 at 8:50 PM



I think you are being too kind. Levy argues that the following is true. 

(1) Addiction is a neuropsychological dysfunction (that is, a pathology).

(2) A disease is a dysfunction that causes suffering in almost every accessible environment. 

I think it intuitively correct that an addiction to heroin, meth, and crack causes harm in almost every accessible environment. This might only consist of the cost of getting professional help or having others angry or ashamed of you. 

So why isn't he committed to the following. 

(3) Most stereotypical addictions are diseases. 

I also do not see his argument as to why it matters whether addiction is a brain disease. 

Rob Kelly - Am I missing his argument? 

Side note: I think Levy is mistaken about (1). The sub-systems are doing what they are designed to do, they are just operating on a novel substance.


David H December 7, 2020 at 10:31 AM


I think all dysfunctions are diseases even if not harmful. But if harm is a condition for disease, what do you think about caffeine addictions? Wakefield thought they were dysfunctions but not harmful or is it not harmful enough to be a disorder (disease)? That said, I do agree with you that it is hard to imagine the standard drugs that are listed as drug addictions (heroin, cocaine, etc.) not being harmful. Even high functioning addicts and alcoholics have accessible worlds where their addiction is harmful

Bob Kelly December 22, 2020 at 2:39 PM


I again think you make a good point about Levy being too unclear in his terminology. Let me try to help with him out with these three claims, though. 


Assume a Wakefield-style account of disease, where dysfunction is insufficient for disease since disease requires dysfunction + harm/impairment. 


Addiction (probably) entails some dysfunction in the brain (a part, a system/process, etc.). In essence, let’s assume Nora Volkow and other brain disease proponents get something right about the neurobiology being dysfunctional (though, see Marc Lewis’s “The Biology of Desire” for a counter argument). This, then, is claim C: addiction always involves brain dysfunction. 

However, given our starting assumption, that dysfunction is itself not enough for disease, and moreover it is not true that all addictions (and their dysfunctions) entail harm/impairment. I think you and I agree that harmless addictions are possible. If that is true, given the account of disease we are assuming, addiction is not (by definition, or according to its nature) a disease. This is his claim A: despite the presence of dysfunction, addiction is not essentially a disease (since it is not essentially harmful). 

However, it remains consistent with the foregoing that some (perhaps many) instances of addiction (roughly, for levy, a certain sort of dysfunction involving desires and control impairment) are also cases where the dysfunction involves harm/impairment. This is his claim B: sometimes, addiction involves both the dysfunctional underlying neurobiology as well as the harm/impairment. On his view, both harmless and harmful addiction count as addictions (in virtue of something else, such as impaired control over desires or some such thing). But since disease entails harm, they cannot both be diseases. Hence, addiction is sometimes but not necessarily (or by definition) a disease. 


I take him to hold a Wakefield-style view, and then arguing that despite dysfunction (maybe) being essential to addiction, disease is not. Hence, if this view of disease is right, then addition is not essentially a disease. If this distinction doesn’t matter or isn’t worth taking about, it seems like the same criticism could be directly applied to Wakefield’s view. However, as you point out, Wakefield has the virtue of being exceedingly clear and careful. 

Bob Kelly December 22, 2020 at 3:01 PM


On Levy’s argument for why it matters that addiction is a disease, see my response above to Neil. 

On the point about him being committed to your (3), that most stereotypical addictions are diseases, I would make two points. First, most is still not all. His point (which is similar to one I have made in our Romanell workshops) is that one can consistently hold that addiction is not defined by (or essentially a) disease yet many instances of addiction can be instances of a disease. See me reply to David above. Second, we should be cautious about our intuitions about “stereotypical addictions” since, well, there really is no such thing. As Rasmus has pointed out about psychopathy, much (most?) of the data we have on addiction comes from either treatment or jail/prison. This should give us pause, and at the least we should be careful with intuitions about addictions being harmful, especially if they stem from what we think of as “stereotypical” addictions. Consider this point from Nick Heather (though he was making a slightly differently point, it still applies, especially the second half):

“When reading literature on addiction, both in the popular media and in scientific publications, it is surprising how seldom authors actually tell us what they mean by the term, whether by a formal definition, a rough characterization of how they see addiction, or what they consider to be the “hallmark” (Skog 1999, p. 173) of addictive behavior and experience. It is as though authors simply assume that, when speaking of addiction, everybody will know what they mean. This might have been understandable in the early days of addiction studies in the 1950s or 1960s, when the word conjured up images in most people’s minds of a disheveled individual injecting heroin into a vein or desperately seeking the means to do so. Most people would agree that such stereotypes should have been abandoned long ago but there nevertheless remains a tendency for writers on addiction to assume, for whatever reason, that the term needs no clarification. (Heather, 2017b, p. 3)

On your last point about (1), Marc Lewis makes the same argument in his “Biology of Desire.” He argues, roughly, that (1) the neuroscientists are right about all of the evidence they provide about the systematic changes in the brains of addicts, yet (2) none of this entails these changes are pathological, because (3) the brain (reward system) is doing exactly what it evolved to do—learn via rewards and carve out neural pathways on this basis. That said, see Wakefield’s exchange with Lewis (on the Brains Blog and in Neuroethics, I think) where he argues that the brain changes are dysfunctional precisely because they are operating on objects they were not designed to operate on (like when a duckling imprints on a fox).

David H December 7, 2020 at 10:16 AM

Levy's Misleading Title: 

If some addictions are diseases, all addictions involve brain dysfunction, then some brain dysfunctions are diseases. So it would have been helpful if Levy had written “some brain dysfunction are diseases and some aren’t.” He never says that. Instead his title is “Addiction is not a brain disease” when he himself speculates there probably is not a single addiction that also qualifies as disease because it is harmful that isn’t also a brain dysfunction, ergo a brain disease! He should have titled his paper “All Addictions that are Diseases are Brain Diseases”. He writes addiction is not a brain disease when he believes it is likely that for any addiction that is a disease is also a brain disease! That is misleading. Why couldn’t he write “not all brain dysfunctions are diseases. Not all addictions are diseases. Those addictions that are diseases are also brain diseases”? A disease involves dysfunction and harm. Some brain dysfunctions are harmful. Ergo, some brain dysfunctions are diseases. If some addictions are diseases but not all, then some brain dysfunctions are diseases but not all. It would have been a lot less confusing if he said some brain dysfunctions are diseases since he says some addictions are diseases. I find the following sentence of the abstract very misleading “Hence addiction is not a brain disease, though it is often a disease, and may always involve brain dysfunction.”