Savulescu and Foddy: A Liberal Account of Addiction

Philosophers and psychologists have been attracted to two differing accounts of addictive motivation. In this paper, we investigate these two accounts and challenge their mutual claim that addictions compromise a person’s self-control. 

Paper 13. Savulescu and Foddy: A Liberal Account of Addiction


Stephen Kershnar October 25, 2020 at 8:53 AM


On page 15 para. 6, the Foddy-Savulescu liberal theory of addiction (LAA) makes three assertions.

1. Appetitive Desires. Addictive desires are just strong, regular appetitive desires.

2. Valuation. We do not know whether an addict values anything more than the satisfaction of his addictive desire.

3. Autonomy. We do not know whether an addict behaves autonomously they he uses drugs.

Here is a competing theory: The Kershnar-Kelly Theory of Addiction (others might also have put forth this theory).

An individual has an addiction to a thing if and only if the following are true.

1. Desire. He has an unhealthy desire.

2. Control. He cannot systematically control his use of the thing.

3. Explanation. (a) explains (c).

The view has two other features.

4. Compulsion. This is a subjective compulsion.

5. No Substance. The objection of addiction need not be a substance (consider, for example, sex or OCD).

Like LAA it is neutral on claims about valuation and autonomy. Unlike, LAA, it asserts the following.

A. Normal Activities. A person cannot be addicted to normal levels of healthy activities such as breathing, food, or water. Nor does it make a 19-year-old male with a normal sex drive with his girlfriend addicted to sex.

B. Control and Health. It gets closer to our intuitions that addiction is inextricably linked to control and unhealthiness.


Robert Kelly October 26, 2020 at 12:37 PM


First, I have made some considerable changes to the way I think about addiction since we spoke about and came up with this theory you lay out. First, I don’t think ‘addiction’ should be defined in terms of disease. I agree that desires are necessary, but they need not be unhealthy. At least, I don’t think this should be part of the definition of ‘addiction’. One motivation is that an alternate history could change whether a given condition is a disease or not, without changing anything intrinsic to the condition. If addiction were a disease by definition, then two doppelgangers (in terms of control, desires, brains, etc.) could differ in terms of having an addiction. I think we should avoid this result if we can. Thankfully, we can: just don’t put disease (or “unhealthiness”) in the definition. This would be consistent with every instance of addiction being unhealthy—it would just be an empirical fact about our world and our (evolutionary) history. It’s sort of a nit-picky point, but why build in something to the definition that will close off metaphysical possibilities and produce counterintuitive results if you don’t need to?

Second, I agree that systematic loss of control is necessary, though I think it is better to describe this as a loss of control over one’s desires to engage in the relevant behavior, rather than over their use. One reason is that you can fail to control your desires before use even comes into question by, for instance, failing to implement your cognitive resources that would allow you to suppress the desire, or modify it, or distract yourself from it, or what have you. Alternatively, you might fail to secure or strengthen those cognitive resources, rather than failing to implement them. In any case, you might say that this ultimately leads to failing to control your behavior that is eventually produced (in part) by those desires. Fair enough, but failing to control the desires is more direct and primary. Were you to be able to control your desires, which drive your addicted behavior on my view, then you would control that behavior. Since you don’t, you don’t. So I think lack of control over one’s desires is a more apt description.

Third, I don’t talk about the desire to phi explaining the failure to control phi-ing. This may be true. However, I instead focus more on systematicity, and in particular, the systematic nature of the disposition that is addiction. One has a disposition to fail to control their desires to engage in some type of behavior, and that disposition is systematic, meaning that it is, roughly, sufficiently strong (likely to realize when triggered) in the right conditions (where ‘right conditions’ needs careful handling, which I am trying to work out). The point is that I shift the focus from desires explaining impaired control to having a systematic disposition which is realized in failures to control one’s desires. The explanation component I have now is just that the disposition explains those failures, but this is just part of what dispositions do—they explain (in a making true way) what their bearers can do.

Fourth, I don’t include ‘compulsion’ anywhere. This is mostly rhetorical. Brain disease model components have ruined that word. Everyone now thinks it means the person literally cannot do otherwise, and sometimes they think it means they always literally cannot do otherwise (at least regarding their object of addiction). There is no nuance or room for choice once compulsion comes in (according to its now normal use in the literature). I think this is a confusing way to think about control in addiction, so I stay away from ‘compulsion’.

Fifth, I agree with your sentiment at the end that people are not addicted to breathing. However, I want to be careful. I certainly think people can be addicted to normal, healthy behaviors like jogging, eating, and even drinking water (Hutcheon & Bevilacqua, 2010; Edelstein, 1973). But of course, there needs to be a way to distinguish the appetitive urge to breathe from the desires involved in addiction. I appeal to Arpaly & Schroeder’s reward theory of desire to do this.

Lastly, I disagree that people have intuitions that addiction is unhealthy. They have intuitions that addiction is bad, wrong, and harmful, and my guess is that this is what would make people think it is unhealthy. But we know that none of those are sufficient for unhealth, and so we should also be careful about people’s intuitions about addiction being unhealthy. For what it’s worth, I would also want to see the data on such intuitions first. I have not come across any.

Phil Reed October 29, 2020 at 12:51 PM

Steve - what are (a) and (c) in the Kershnar-Kelly Theory of Addiction (or maybe this is just the Kershnar theory, now that Kelly has disavowed it)?

Phil Reed October 29, 2020 at 12:57 PM

Isn't the sense in which the addictive desires are "unhealthy" a broad sense? One advantage of the Kershnar theory is that it includes behavioral addictions: gambling, say, or smart phones. These do not seem to be unhealthy (at least in the Boorsean sense), but they are unhealthy in a broader sense - in that they are bad for the person, or harmful.

I think, and Rob says he agrees, that people do have intuitions that addictions are harmful (unhealthy in the broad sense) and that would be a point against the Savulescu and Foddy view.

People do sometimes use the phrase "healthy addiction," but I think that is a metaphor more than a coherent concept.


Stephen Kershnar October 30, 2020 at 10:30 AM



Good point. I did not fill out the conditions correctly. It should say the following.


An individual has an addiction to a thing if and only if the following are true.

1. Desire. He has an unhealthy desire.

2. Control. He cannot systematically control his use of the thing.

3. Explanation. 1 explains 2.


Stephen Kershnar October 30, 2020 at 10:31 AM



I think this is a really good point.

“People do sometimes use the phrase "healthy addiction," but I think that is a metaphor more than a coherent concept.”

I cannot think of a healthy addiction, except in unusual cases where unhealthy habits are lifesavers. For example, a heroin addict fails to show up to work on the day that someone goes postal there.


Stephen KershnarOctober 30, 2020 at 10:31 AM



Great comments. Quite interesting. Below are my responses.

You write the following.

“First, I don’t think ‘addiction’ should be defined in terms of disease. I agree that desires are necessary, but they need not be unhealthy. At least, I do not think this should be part of the definition of ‘addiction’. One motivation is that an alternate history could change whether a given condition is a disease or not, without changing anything intrinsic to the condition. If addiction were a disease by definition, then two doppelgangers (in terms of control, desires, brains, etc.) could differ in terms of having an addiction. I think we should avoid this result if we can.”

Two doppelgangers could differ in disease only if they had different evolutionary histories or were different bio-statistically to their reference class. This does not strike me as problematic. A very slow human might have a disorder and a chimpanzee does not have it even if they were thought-for-thought identical.

I think that unhealthiness is central to our thinking about a disease. Consider, a super healthy and flourishing person, who has normal desires – but also equally frequent and intense as paradigmatic compulsive desires – to breath, drink water, and sleep. We would not want to say that he is addicted to these things even if his desire and lack of control were relevant similar in intensity and effect to that of a heroin addict. Here then is my argument

1. A frequent, strong, and control-undercutting desire to take heroin is part of an addiction [Assumption].

2. A desire to drink water, eat food, or sleep that is similar in frequency, strength, and control-undercutting is not an addiction [Intuition].

3. If 1 and 2, then an addictive desire is unhealthy [This is the most salient difference between these desires].


Stephen Kershnar October 30, 2020 at 10:34 AM


Rob: You also argue the following.

“Second, I agree that systematic loss of control is necessary, though I think it is better to describe this as a loss of control over one’s desires to engage in the relevant behavior, rather than over their use. One reason is that you can fail to control your desires before use even comes into question by, for instance, failing to implement your cognitive resources that would allow you to suppress the desire, or modify it, or distract yourself from it, or what have you. Alternatively, you might fail to secure or strengthen those cognitive resources, rather than failing to implement them.”

The problem with this is that one normally cannot control his desire content, strength, or even its prevalence.

Case #1: 18-Year-Old

Consider, for example, the average heterosexual 18-year-old males desire for sex. While the male cannot control his desire, he can control acting on it. In some cases – whether actual or hypothetical – this distinguishes those who are and are not sex addicts.

Here is the argument.

1. In non-addiction cases, a person lacks systematic control over his desire, specifically, the frequency, intensity, and control-undermining force (or lack thereof) of his desire.

2. If 1, then addiction does involve a lack of control over a desire.

Case #2: Super Spartan

A highly disciplined Spartan has a desire for heroin as strong as that of an out-of-control junky in San Francisco. He cannot control his desire. However, his greater discipline allows him to never use heroin, even though he desperately wants to do so. He is not addicted.

1. The Super Spartan is not addicted to heroin.

2. If addiction focused on control over a desire, then the Super Spartan would be addicted.

If one were to think the Super Spartan is addicted – my intuition is not as clear as I would prefer – then he is addicted because given his desire strength, a person with an ordinary capacity for control would be addicted to heroin. Thus, control operates through heroin use, albeit counterfactually, and only in part via desire.

Stephen Kershnar October 30, 2020 at 10:35 AM



You make the following argument, “The point is that I shift the focus from desires explaining impaired control to having a systematic disposition which is realized in failures to control one’s desires. The explanation component I have now is just that the disposition explains those failures, but this is just part of what dispositions do—they explain (in a making true way) what their bearers can do.”

As argued above, I do not think addiction involves failure to control one’s desires. This is true for many things to which we are not addicted.

The general disposition condition is mistaken. Here is why.

Case: Evil Demon

A person is addicted to meth but not heroin. However, every time he goes to shoot up meth, an evil demon causes the meth to transform into heroin before it enters his body. He is not addicted to heroin but has a disposition to take it.

You might respond that the disposition is not a feature of him, but rather of his situation. Fair enough. But the same might be true for a meth addict that is compelled when, and only when, people use meth around him.

We need to specify what sort of disposition. This is when desire-impairing-control specifies the particular disposition.

Stephen Kershnar October 30, 2020 at 11:17 AM


Rob, excellent point.

You write the following, “I don’t include ‘compulsion’ anywhere. This is mostly rhetorical. Brain disease model components have ruined that word. Everyone now thinks it means the person literally cannot do otherwise, and sometimes they think it means they always literally cannot do otherwise (at least regarding their object of addiction). There is no nuance or room for choice once compulsion comes in (according to its now normal use in the literature). I think this is a confusing way to think about control in addiction, so I stay away from ‘compulsion’.”

Let us define compulsion as follows.

(1) A person is compelled by a desire if and only if he lacks systematic control over acting on that desire even if he intended or judged that he ought not to do so all things considered.

Here is an alternative version.

(2) A person is compelled by a desire if and only if, were he a reasonable person, he would lack systematic control for acting on that desire, even if he intended or judged that he ought not to do so all things considered.

Account (1) is at the heart of how we think of addiction because it is the sort of lack of control on which addiction focuses.

Here, then, is my argument.

1. Addiction focuses on a sufficiently strong loss of control (compulsion).

2. If 1, then addiction should have a compulsion condition.

It is an interesting issue as to whether it should be tied to the person or to a reasonable person. Also, this is what distinguishes addiction and akrasia.

Robert Kelly October 30, 2020 at 3:53 PM


I agree that capturing behavioral addictions is a good thing. Did you mean to imply that F&S do not capture this? I think they do. Their definition as a ‘strong appetite’ allows behavioral addictions. Their claim is just that addiction is a, roughly, more severe case of akratic action. It’s less clear what makes it more severe, but I think the most plausible interpretation is that there is (i.e. the addiction just is) a stronger proneness or tendency for the appetite to occur and to be effective. Possibly they also mean the actual desires are typically stronger in addiction as opposed to akratic actions. But this is all consistent with them capturing behavioral addictions.

Also, why would the intuition that addictions are harmful be a point against F&S? If you mean that this would allow addictions to be diseased/unhealthy, in your broad sense, then I don’t see why they can’t just agree to this. When they deny addiction is a disease or unhealthy, they are certainly not denying that it is typically harmful. If you broaden the notion of ‘disease’ to your view of being bad or harmful, then nothing about that conception of disease is inconsistent with their view. Obviously many addictions are harmful. If my explanation of their distinction between addiction and akratic actions is correct (namely, proneness/frequency and strength), then they could even easily use this to explain why addictions are typically harmful whereas akratic actions (maybe) are not.

Why is “healthy addiction” incoherent if, on your broad view, this just means “harmless addiction”? I actually think it is quite obvious that harmless addictions are possible. I think there is also good reason to think that harmless addictions are actual. And at the very least, there is a seriously difficult burden to bear in showing that addictions are necessarily harmful. Given the obviousness of the possibility claim, the plausibility of the actuality claim, and the heavy burden of requiring harm as a necessary condition, I don’t know why we would think that addiction is necessarily harmful. If, as you suggest, people saying “healthy addiction” is a metaphor, the metaphor likely refers to your broad sense of ‘disease’. In other words, it is only a metaphor in the sense that the mean ‘harmless’ by ‘healthy’, and that is metaphorical because ‘harmful’ is not a good definition of ‘disease’. It feels like there may be a bit of equivocating in your responses between health/disease in a more robust sense and health/disease understood as just harmful or not.


I respond to your other comments below, but wanted to address your “I can’t think of a healthy addiction” comment here. I think you just aren’t being imaginative enough, or perhaps are being selectively imaginative. You have dirty, down-trodden, wasted away heroin addicts with needles hanging from their arms as your go-to picture of addiction. This is a naïve picture of addiction. Moreover, you seem to be, like Phil, thinking of health and disease in terms of harm (given your going postal example). You also seem to have dysfunctional desires just plugged into what it means to be addicted, and so I think this contributes to your difficulty in imagining a healthy addiction. Wakefield himself, who requires harm for disease, agrees that addiction and addictive disorder (harmful addiction) are different. This is because he understands that addiction is fundamentally about control over your desires, and that it would be a really hard claim to defend that addiction is necessarily harmful, given the evidence.


Robert Kelly October 30, 2020 at 3:55 PM


You say “Two doppelgangers could differ in disease only if they had different evolutionary histories or were different bio-statistically to their reference class. This does not strike me as problematic.”

Right, that is not what is supposed to be problematic. These cases seem possible, which is what introduces the problematic entailment: two qualitatively identical conditions would differ in their counting as addictions. I am not saying their being different in terms of disease is a problem—indeed, my objection requires this. The point is that this should not entail that one is an addiction and the other isn’t. They are both addictions; it’s just that one is a disease and the other isn’t due to the different evolutionary histories or biostatistical differences. What addiction is shouldn’t change in this way, since it is just about the ability to control your desires to engage in certain behaviors. Both people systematically lack control over their desires to phi. One is a disease because, given that world’s history and laws, such a condition is a dysfunction (where this is cashed out in evolutionary terms in the example).

You also say, “A very slow human might have a disorder and a chimpanzee does not have it even if they were thought-for-thought identical.”

These are not doppelgangers, and so the analogy doesn’t hold as far as I can tell.

You also say, “I think that unhealthiness is central to our thinking about a disease.”

Agreed. This seems trivially true. Maybe you meant ‘addiction’ at the end, since you go on to say, “Consider, a super healthy and flourishing person, who has normal desires – but also equally frequent and intense as paradigmatic compulsive desires – to breath, drink water, and sleep. We would not want to say that he is addicted to these things even if his desire and lack of control were relevant similar in intensity and effect to that of a heroin addict.”

A couple things. First, base appetites for breathing and drinking water are not desires in any way that people who talk about desires in addiction understand those desires. This is probably why conversations about this are relatively rare in the literature. No one is really that worried about distinguishing addictive desires from impulses to breath produced by the autonomic nervous system. They are not the same thing. Of course, we can form “ordinary” desires about breathing and sleeping and eating and so on, but these are different that the base impulses that are there anyways and that you are (and need to be) referring to in your argument. This difference is also why I can say that people can be addicted to water and eating, for instance. The base impulses that are there from the autonomic system are not sufficient – but through repetition and reward learning, you can develop desires for water that would count (so long as one systematically fails to control them).

Robert Kelly October 30, 2020 at 3:56 PM


Two things might be happening here, I think. One is that I don’t think we actually disagree about what I said in the quote you are addressing. The other is that you are misinterpreting (and possibly straw-manning) my view.

In short, I agree that the Spartan is not addicted, and it is because he can control his desires. By systematically ailing to control one’s desires, I just mean that they systematically fail to successfully resist them, reduce them, modify them, etc. when they try. We can do all of these things. Of course I don’t mean control over desires in the sense that you would have to will the desire to arise in order for it to arise. Desires simply arise, just as thoughts do, when they are triggered. But I also don’t think of desires as these “felt urges” you probably have in mind. I think these are the result of having certain desires and being in certain conditions that trigger them. Desires are representations of states of affairs as rewards, where this means those representations contribute positively to the overall reward calculation. Thus, you can desire something even though the overall reward calculation ends up being negative (“that was worse than expected!”). Controlling our desires is complex and often indirect—though we can try to simply resist them (or the felt urges they produce) when they occur, but we’re just usually not very good at that. We can utilize our cognitive resources to anticipate desires, to distract ourselves from them, to formulate desire-combatting habits (which may be other desires), we can set up our environment to make triggers less prevalent, and on, and on. You are restricting the claim about “controlling desires” to either (i) directly resisting felt urges, or (ii) having robust free will in the moment over when and where your desires arise. Neither of those is my view (nor should it be anyone’s). Controlling whether you act on your desires is to control your desires (since, to anthropomorphize them, what they want is for you to act on them). Also, given the indirect control we have over them and what they are (on Arpaly & Schroeder’s reward theory of desire that I endorse), it is still true to say that we can control their strength, frequency, and so on. You know you desire to eat sugary sweets late at night and rarely resist those desires when they come up (and sugary sweets are on hand). So, you stop buying sugary sweets and accepting them as gifts. It is totally plausible (and normal) for desires regarding a certain object or behavior to vary in strength in different situations or contexts (such as a desire to eat them now when they are available vs. a desire to buy them when at the store vs. a desire to accept an offer to take a bag of candy).

I think the most plausible explanation here is that I didn’t explain my view well enough and you misinterpreted it too narrowly, and so we don’t actually disagree about what addicts lack control over.

Robert Kelly October 30, 2020 at 3:57 PM


Maybe I should have left this point alone. I again don’t think we really disagree here. There is a perfectly fine sense in which you can explain the systematic lack of control by appeal to the desires. My point was about my account now emphasizing the dispositional nature of addiction; and doing so would bring the desires (explaining control loss) for free. Maybe I should leave it at that (but I won’t).

To be fair, I don’t know that I fully understand your evil demon objection and the preceding and following claims. But this is probably because, again, I failed to explain this component of my view well. You say, “We need to specify what sort of disposition.” Right, and I do, but probably didn’t here. The systematic disposition is for (realized in) failures to control one’s desires to engage in certain types of behavior. In your case, the meth addict would have desires to engage in meth-related behaviors (ingesting it, obtaining it, etc.). Nothing about your case entails that he is not addicted to meth, nor that he is addicted to heroin, on my view. It is underdescribed, but I will just say that, if they are a meth addict, then they have a systematic disposition to fail to control their desires to (for instance) take meth. This is still true and the evil demon changing it to heroin doesn’t change this (though it probably will eventually change this, given how desires and reward work). The presence of the disposition is not explained (made true) by actual realizations of it—the opposite is true. So, the fact that we have one case where someone accidentally takes heroin does not show he has a disposition to take heroin (the easiest conclusion you could draw), and it certainly does not show he has a systematic disposition to take heroin (which wouldn’t count as an addiction on my view), and it certainly doesn’t show he has a systematic disposition to fail to control desires to take heroin (indeed he doesn’t even have such desires yet).

Robert Kelly October 30, 2020 at 3:58 PM


I basically agree with everything you say here. The point I was making though, and so the only disagreement I have, is that we shouldn’t use the word ‘compulsion’. If by ‘compulsion’ you just mean the systematic lack of control over (acting on) one’s desires to engage in certain types of behavior, then yeah, I think addiction requires compulsion. I just speak in terms of ‘systematic lack of control’ because (i) it emphasizes the dispositional nature of addiction, (ii) gets everything we want about the middle ground between brain disease views using (implausible) conceptions of compulsion as literal irresistibility and moral models that say addicts have full control, and (iii) it does this without using a word that everyone confuses and misunderstands, and that is embedded in confused and stalled debates.

Again, I agree. I would just strongly advise against using that word, given the state of the addiction literature surrounding it.


Stephen Kershnar October 25, 2020 at 8:54 AM


I do not see why addiction is not a disease if it is a harmful dysfunction. Harm is an empirical matter. Still, under the counterfactual comparative theory of harm, many addictions are going to be harmful. This might be affected by the way society is or should be structured, but in many cases, this will be met. In addition, the appetitive desire or the person’s pursuit of it are dysfunction in the sense that the strength of desire or focus on it are not in line with evolutionary design.

Here is Chris Boorse’s biostatistical theory (from Steven Tresker)

1. The reference class is a natural class of organisms of uniform functional design; specifically, age and sex of a species.

2. A normal function of a part or process within members of the reference class is a statistically typical contribution by it to their individual survival or reproduction.

3. Health in a member of the reference class is normal functional ability: the readiness of each internal part to perform all its normal functions on typical occasions with at least typical efficiency.

4. A disease is a type of internal state which impairs health. That is, it reduces one or more functional abilities below typical efficiency.

Many cases of addiction satisfy Boorse’s theory.

If these are the two most plausible theories of disease (or disorder or pathological condition), then many addictions are diseases.

I do not think anything follows from this. Health-based theories of the good, the right, moral rights, and responsibility do not succeed. Still, many addictions are diseases.


Robert Kelly October 26, 2020 at 12:39 PM


First, as you point out, you don’t capture every (at least possible) case of addiction. It’s like you don’t capture all actual cases either, since there are almost certainly lots of “functioning addicts” out there, as they are sometimes called. Pickard, Sinnott-Armstrong, and I think some others have tried to say that the loss/impairment of control itself is a harm, and so addiction is necessarily harmful in this sense. I think this is cheating, since they really mean (if you read them closely) that control loss is a risk of harm. If that’s what they mean by harm, then fine, it’s harmful. But I don’t think risk of harm is a harm. If Limbaugh says that is damage, then fine, addiction is damaging. Though I doubt Limbaugh would say this since, based on our conversations, he also agrees that addiction is not necessarily harmful. It’s possible to have a harmless addiction. This just seems like a pretty solid intuition. But more than that, there is absolutely no evidence that addiction is always harmful. If anything, we should be skeptical of the claims that addiction is necessarily harmful because of the evidence we have on addiction—it all basically comes from treatment centers, rehabs, etc., or prisons, or studies that use the DSM for the inclusion criteria (which is littered with harm requirements). Of course it is going to look from the data like addiction is always harmful. People who are not being harmed in virtue of their addiction are pretty unlikely to go to jail, be in rehab or treatment, or take part in a clinical study on addiction.

Second, even Wakefield admits that addiction is not necessarily harmful. This is why he distinguishes between addiction (no harm required—just his theory of the dysfunctional impairment in deliberative capacities) and addictive disorder (that plus harm). Hence, even if every actual instance of addiction is an addictive disorder (i.e. a disease) on Wakefield’s view, this would not make addiction necessarily harmful.

Third, Marc Lewis has a book-length treatment of why the desires are not dysfunctional. Wakefield responds, arguing that addiction can be a disease even if Lewis’s treatment of the neuroscience is correct. He thinks the capacity for deliberating and reasoning over our desires is dysfunctional in addiction—however, nothing about his view requires that this is because of their strength or our focus on them. He accepts Lewis’s comparison of the “brain on love” and the “brain on drugs,” and says that addiction (but not love) can still be a disease because of the external cause of the brain states/processes that develop in addiction. The brain systems/processes are caused to be how they are by something that was not part of the design of those systems/processes. It’s the same reason he thinks the duckling’s imprinting mechanisms is dysfunctional if it imprints on a fox, even if it looks identical to a duckling’s imprinting system that imprints on its mother. It wasn’t designed to have that object as its cause. The point is that the desire need not be abnormally strong or in focus in order to be unhealthy, even according to Wakefield himself.

In the end, I agree with your last claims. Every instance of addiction may very well be a disease (though I doubt it). This is possible. But if we are Wakefieldians, it’s almost certainly false since there are harmless addictions (and he sees this and avoids it). If we are Boorseians, I don’t know how we are going to deal with Lewis’s brain on love analogy other than to deny its possibility (of an addicted brain and a loving brain being qualitatively identical in the relevant ways).


David H October 30, 2020 at 2:44 PM

I agree with Steve that the disease component is critical to avoid addictions becoming used promiscuously to involve any strong desire that we can’t control, assuming we wanted to. Our most important difference is that my preferred account doesn’t make loss of control a defining feature. See my stand-alone comments below. I would also have a quibble with Steve in that I take it he means by “unhealthy desire” a pathological excess of the desire rather than a unhealthy pathological lack of the desire. I would also think “desire” is too narrow as it tends to be too intellectualized as one can’t desire X without being able to conceptualize X. I want the pathologically excessive drive to be less cognitive. Neonates can be born addicted but they don’t desire the drug as they have no conception of it and they don’t desire its cessation as they have no conception of their persistence across time. But they are addicted as they have an unhealthy excessive, non-desired need for the drug. Perhaps most importantly, they don’t have any control mechanisms that are failing to operate. So, control is not necessary to addiction. A more extensive defense of these claims is given in a stand-alone comment below.


Robert Kelly November 4, 2020 at 10:06 PM


First, I don’t have such an intellectualized account of desires. I endorse Arpaly & Schroeder’s reward theory of desire, and this roughly makes desires representations of states of affairs as rewards. To desire that p is to represent p as a reward, where this means that representation contributes positively to overall reward calculations. Neonates’ reward (dopaminergic) systems are certainly up and running, and so have desires in this sense. I think they still probably don’t have any control that could be impaired, though, and so this leads to a second point.

Second, on my view, the neonates are born dependent, not addicted. This need not even be about desires or control, really. This is just based on the fact that the neonates you have in mind only suffer from physical dependence, as far as we know. They have had drugs in their system, and so they develop tolerance and disposition to withdrawal (mostly the latter). That is physical dependence. See:


David H November 4, 2020 at 10:08 PM


Representing Rewards: I haven’t read the Arpaly and Schroeder’s account of desires as representations of rewards but it still seems an intellectualized account of desires as newborns won’t be able to represent themselves being rewarded in the future. It seems plausible to me that they are living in the moment or nearly so. They can’t represent to themselves obtaining a drug or being without pain. They are just in pain, not cognizant of it as pain or able to want to be without the pain. We ascribe them interests in pain relief but not desires. That is, it would be good for them to be free of pain or to have a reward but they don’t have an interest in i.e., desire either. Their reward systems are not up and running in the sense they represent future rewarding states. Perhaps I don’t understand what sort of representation you have in mind.

Addiction or Dependence: My knowledge of the addiction literature is limited to the articles we have read over the years in PANTC, Blameless Buffalo, and Romanell Center working dinners and this S and F paper. So I could very well not be using addiction the way the professionals are. I suspect that Jack Freer is right and I am just unduly influenced by the layperson’s usage of kids being born addicted and lab animals rendered addicts. Well, that makes my thesis more of an uphill battle but doesn’t lead me to abandon it, though I should not be confident given that I am a bit of a tourist and not a local steeped in the literature. Fodor once joked in the preface of his book Psychological Explanation (?) that he always believed that if he had a free weekend that he could fix up psychology. I have the same suspicion about addiction but unlike Fodor, who was knee deep in the psychological literature, I have just put the tip of my toe in the addiction literature. So I am far more likely to make a fool of myself. Fortunately, no one reads our blog but the bloggers. We are engaged in reflective equilibrium as were Wakefield and Boorse in their debate over whether diseases should be defined in terms of harm as a necessary condition. Boorse stressed the usage of pathologists and Wakefield helped himself to clinicians and laypeople. Boorse was especially willing to break with medicine on whether some conditions were pathologies but wanted to capture the paradigm cases of disease. I too want to capture the paradigm cases of addiction, but I am willing to extend addiction to cover the very young, cognitively impaired, and non-human animals who don’t seem to be failures of control. So I don’t think I mean something else by addiction, but believe (hope) I have the better conception of the concept ‘addiction’ and can convince researchers that they are wrong about the extension of addiction. My strategy is to appeal to philosophy of medicine and distinguish diseases from symptoms and claim that the disease addiction can explain the symptom of loss of control. There is no reason that empirical considerations and philosophical reflection can’t reach a reflective equilibrium where fetuses and neonates are addicted and control failures are symptoms of addiction caused and explained by addiction rather than define addiction. I try to flesh this out in later posts


Robert Kelly November 4, 2020 at 10:11 PM


On rewards, no, Arpaly and Schroeder’s account of representation does not over-intellectualize desires. Newborns have a working reward system, and they represent states of affairs in the basic sense that their perceptual system, for instance, represents states of the world (to their brain, or the rest of their brain, so-to-speak, which is taken in as information to be processed). An infant could feel cold, or taste something bad, or see a dark figure, or smell her mother, or what have you. These are all they mean by representing some state of affairs. When that representation contributes positively to the overall reward calculation—that is, when it increases the chances that a positive learning signal will be produced (which is basically a message in the brain saying, “Hey, the world is better than expected”)—it is a desire. When it contributes negatively (decreasing the chance of a negative learning signal (“worse than expected”), it is an aversion. Typically experience and learning is how desires are developed (from very early on), but some are also innate. Certain states of the world will be represented as rewards simply in virtue of our nature thanks to evolution. Most others will be learned through experience (reward learning). This does not require infants to have any sophisticated mentality going on—nor animals. Both have desires in this sense. I am not sure about the comatose—they don’t discuss this. It is an interesting question, though. My account seems to be able to handle this, though, even if they can’t currently represent the world (and so can’t currently undergo a desiring). Similarly with control. As long as they maintain the appropriate dispositions—to represent certain states of affairs as rewards, to fail to control those desires, etc.—then they would be an addict. The easy case is an addict who gets knocked on the head and goes into a coma. They (obviously, I think) do not thereby immediately fail to be an addict, just as an addict who is kidnapped and dropped on a desert island doesn’t immediately fail to be an addict just because his desires and impaired control won’t be manifested (as we can stipulate). We need dispositions to make sense of this, and luckily I have a dispositional account. Eventually the comatose and desert island prisoner will lose their dispositions, and thereby lose their addiction.

On the rest, again, I don’t think you have to abandon the impaired control component to capture non-human animals. Other mammals control their behaviors in various ways. These ways get disrupted when they are (forced in animal studies to become) addicted. Again, Waller’s and Dennett’s accounts of free will seem to be able to accommodate this, and even Fischer’s framework can be watered down to allow for a kind of reason-responsiveness of some mammals. Of course, they wouldn’t recognize reasons as such, at least not in the way we do. But a minimal RR of animals does not seem incoherent to me. I think I sent Steve a paper I found on this a while back when he was writing something on animals and control. I also wonder why you are dead-set on making newborns addicted. This is far from a paradigm case of addiction. I whole-heartedly agree with your approach, though, and one of my main criticism of existing accounts is that they almost always appeal to symptoms or typical effects (and causes) of addiction when defining it. Harm is the most prevalent example, and I know you and I agree that harmless addictions are possible (and very likely actual). I just don’t think impaired control is this way, once properly understood as a disposition to systematically fail to control your desires in various kinds of ways, some of which are more or less robust. Mostly I care about the framework, and not the particular account of control that goes into it.


Stephen Kershnar October 25, 2020 at 8:56 AM


Foddy and Savulescu assert that on the basis of their liberal theory of addiction (LAA), criminalization of recreational use of drugs (drug enforcement) should not criminalize autonomous drug use (see p. 20 para. 1) and should address only those problems that are intrinsic to the drugs themselves (see p. 19 para. 5). They also assert that treatment programs should (at least in part) aim at control their usage (see p. 18 para. 6).

Consider legalization. Autonomy is

1. a matter of degree,

2. not the only value, and

3. contingently related to liberty.

Hence, unless one is a liberty freak, criminalization should probably aim at a cost-benefit analysis.

If this is correct, then in some cases – none of which might be actual - the state may criminalize autonomous drug use and decriminalize non-autonomous drug use.

Here is the argument.

(P1) The state should pursue all, and only, efficiency.

(P2) If 1, then the state should criminalize drug only if it is efficient to do so.

(C1) Hence, the state should criminalize drug use only if it is efficient to do so. [(P1), (P2)]

The efficiency calculation should consider the value of users’ and others’ pleasure.

If, instead, the state should focus on, and only on, protecting freedom – whether as a side-constraint or through maximization – then drug use is none of the government’s business. Such a view would rest on the strong trumping effect of rights (see, for example, Trolley-and-Fat-man-type cases).

Instead, it should stick to its main function, which is directly protecting against force, fraud, and theft. Here is the argument.

(P1) The state should focus on, and only on, protecting freedom.

(P2) If the state should focus on, and only on, protecting freedom, then the state should not criminalize recreational drug use.

(C1) Hence, the state should not criminalize recreational drug use.


Robert Kelly October 26, 2020 at 12:39 PM


I don’t have much to say about legalization (or doctor’s and others’ duties towards addicts, as in your below questions to Reed, Hershenov, and Timmerman). All I want to say is that I think it is relevant to keep in mind the following point. There are likely tons and tons of people in treatment, rehab, jail/prison, and so on that (i) are there because they (and others) take themselves to be addicts, and (ii) are not addicts. This is because treatment of addiction seems to have made the need (or even desire) for help concerning reward-inducing behaviors synonymous with addiction. If I drink alcohol and it is causing problems in my life, and I can’t seem to get myself together so I seek treatment, then I am pretty much done qualifying as an addict in most clinical/medical contexts. It doesn’t matter whether it is true that I have lost control over my drinking. Just that I drink, am harmed by it, and seek help seems to be enough. Sometimes, as in the jail/prison cases, just the behavior and the harm is enough without actually seeking treatment.

I note this because it seems like it makes the questions about policy really, really hard. And more important for my purposes, it seems to make them not about addiction. Legalization and so on seems to be about harmful use, at least in large part. This is simply not addiction. So, I guess I think the question is ill-formed if it is “should we criminalize addiction?” or “is Phil Reed or the state obligated to prevent addiction?” or “Are Hershenov or clinicians obligated to enforce harm-reduction policies for helping addicts, but not necessarily eliminating addictions?” I think they are ill-formed because they conflate harmful use with addiction—or they risk doing so.


Phil Reed October 29, 2020 at 1:04 PM

Rob - it is hard for me to imagine someone being harmed by drinking in this way but not being addicted to it. If the drinking is causing the problems and the person "can't seem to get himself together" and is seeking treatment, in what sense can he systematically control his drinking?


Stephen Kershnar October 30, 2020 at 10:50 AM


Rob and Phil: Great points. Still, I have to disagree.

Rob writes the following, “So, I guess I think the question is ill-formed if it is “should we criminalize addiction?” or “is Phil Reed or the state obligated to prevent addiction?” or “Are Hershenov or clinicians obligated to enforce harm-reduction policies for helping addicts, but not necessarily eliminating addictions?” I think they are ill-formed because they conflate harmful use with addiction—or they risk doing so.”

Here is one reason to think that addiction matters – although this is not available for responsibility-impossibilists or fans of cost-benefit analyses (that is, philosophy weirdos).

1. If a person has a right to be left alone by the state, then he is sufficiently morally responsible for what he does.

2. [In relevant cases], addicts (or addiction beyond a threshold) are not sufficiently morally responsible for what they do.

The argument might also focus more narrowly on when coercive treatment is permissible.

The problem with focusing on harm is that given the best theory of it – the counterfactual comparative account (CCA) – harm is not a good reason, whether other things being equal or all things considered, for a duty in third parties or the state. This is because many right-respecting acts are harmful in the sense that a person would be better off if he were not to do it. Consider, for example, refraining from exercising, dropping out of college, or dating a psychopath.

This is a different approach than Phil’s as he thinks harm connects to addiction. He says the following, “Rob - it is hard for me to imagine someone being harmed by drinking in this way but not being addicted to it. If the drinking is causing the problems and the person "can't seem to get himself together" and is seeking treatment, in what sense can he systematically control his drinking?” Phil’s use might relate to our everyday concern about harmful drinking rather than CCA-type accounts.

Perhaps the state should focus on right-infringement, but this will reintroduce responsibility-undermining addiction.

Once we drop the focus on harm, responsibility, and right-infringement, we are left with a cost-benefit analysis.


Robert Kelly October 30, 2020 at 3:49 PM


I hesitate to engage because I am not very competent on the legal/policy stuff. But let me take a stab at responding to your argument. You say:

“1. If a person has a right to be left alone by the state, then he is sufficiently morally responsible for what he does.

2. [In relevant cases], addicts (or addiction beyond a threshold) are not sufficiently morally responsible for what they do.”

(1) seems false, or at least I am having trouble seeing the connection between S having a right to be left alone by the state and S being MR for what they do. How does their right to be left alone by the state have anything to do with determining whether they are MR for phi-ing? MR for phi-ing seems to be about properties of S (control, knowledge, etc.), and these don’t seem to be necessarily affected by whether S has a state-related right. This allows that sometimes their state-related right might affect whether the MR properties are present (maybe their right affects their control in some cases?). But why would think these are necessarily connected in a way that makes (1) true?

(2) is, I think, again ill-formed. We simply cannot make such general, capacity-based claims about addicts’ MR, especially when the target of their MR is as broad as “for what they do”? Addicts are not automatons in virtue of their addiction. They lack control in usually the very same ways we lack control over our desires—they just do so systematically, and with respect to certain types of behaviors. One thing that has promoted confusion over these questions in the literature is people focusing on such borad, capacity-based claims about addicts “lacking free choice” or “not being responsible”. Addicts can make free and responsible choices the same as anyone else, sometimes. This may be off topic some from what your argument is getting at, but (2) is ill-formed because of it and so whatever conclusion or point (2) is supporting will suffer because of this.


Robert Kelly October 30, 2020 at 3:51 PM


It is important (and maybe sneaky, but unintentionally so) that I said he can’t “seem” to get himself together. Maybe he is being purposefully obstinate because he really doesn’t believe it is a problem (he might be in denial or truly doesn’t see either (i) that it is harmful, or (ii) that it is the drinking that is doing the harm). Pickard talks about such cases a lot when discussing the role of denial and knowledge in addiction. But this kind of case seems possible with someone who does not lack control over their drinking. He may just be seeking treatment because his family has encouraged him to do so, and he loves them, so he goes reluctantly. But maybe after a couple weeks, low and behold he makes the connection between his drinking and the problems, finds new reasons to see that the other things he was blaming are not responsible, and also is able to see that he has his own motivations for quitting that he was ignoring (perhaps because he didn’t think it was doing much harm right now). Suppose further that he could have noticed all of this if he had just tried a little harder, but he hates ruminating about all this stuff and so intentionally distracted himself from doing so, etc.

The point is just that we can seem to have cases where someone is still not to a point where they systematically lack control over their desires to drink, but where their drinking is causing problems. The puzzle would be to ask why they are still drinking, but I hope the points about denial, ignorance to the causes, and so on help. There might be lots of reasons why the person doesn’t see the harm as sufficient to stop (even though they could), or as even related to their drinking. They may also just have false beliefs about their own self-control and seek treatment for that reason. All of this, as far as I know, is not unheard of in actual clinical or treatment settings. At the very least, it seems possible. Hence, problematic drinking does not entail addiction, if addiction requires (systematic) control loss.


Stephen Kershnar October 25, 2020 at 8:59 AM


If I understand his theory correctly, and I might not, Phil Reed argues that an individual has a duty to prevent suicide even among the fully autonomous and transfers the duty to the state.

I find this implausible.

If the state has such a duty, then a suicidal person has a pre-institutional, positive, and non-waivable right to have another prevent his suicide, without his owning the body or labor of the person against whom he has this right.

Still, let us assume there is such a right.

I wonder if Dr. Reed thinks the state has a duty to prevent addiction-caused death and harm.

If I understand his theory correctly, and I might not, David Hershenov argues that physicians have rights, duties, and permissions that are justified, at least in part, by their role or the value of health.

I find this implausible because not only would such moral factors not be grounded in autonomy or interest – on the most plausible theories of interest – but they would have to trump natural rights and rights derived from them.

Still, let us assume such moral factors exist and that addiction is unhealthy.

I wonder if Dr. Hershenov thinks that physicians and pharmacists have role-based permission to avoid policies that lessen the effects of addiction, rather than eliminating them. For example, paying addicts to use fewer drugs but not stop using them or introducing one addiction to eliminate another.

Travis Timmerman argues that there is a duty to prevent unavoidable and undeserved harm. In the context of public confederate monuments, he uses this to argue that we should take down such monuments.

If there were such a duty, and I think the problems here are similar to those faced by Hershenov, I wonder if Dr. Timmerman thinks that both individuals and the state should support criminalization and forced treatment of addicts if it were to eliminate or lessen unavoidable and undeserved harm.


Phil Reed October 29, 2020 at 1:07 PM

Yes, I think individuals and society have a duty to prevent addiction-caused death and harm.

Whether this duty should be transferred to the state is up for debate. I'm inclined to think there are better ways to execute this duty than state intervention.

Stephen Kershnar October 30, 2020 at 11:01 AM



Thank you for your response.

Here is my argument, “Phil Reed argues that an individual has a duty to prevent suicide even among the fully autonomous and transfers the duty to the state. … If the state has such a duty, then a suicidal person has a pre-institutional, positive, and non-waivable right to have another prevent his suicide, without his owning the body or labor of the person against whom he has this right.”

Phil responds as follows, “Yes, I think individuals and society have a duty to prevent addiction-caused death and harm. Whether this duty should be transferred to the state is up for debate. I'm inclined to think there are better ways to execute this duty than state intervention.”

Here are my replies.

1. If A owes a duty to B to prevent B from being having addiction-caused harm or death, then B has a right (claim) against that A does so.

2. If this were true, then (a) there would be natural positive rights, (b) B would not own the right (in the sense of having the power to waive it), and (c) the right – if relevant – would trump B’s right over his body.

3. Conditions (a) through (c) are false.

4. Hence, A does not owe B a duty to prevent B from having addiction-caused death or harm.

First, this is an implausible account of natural positive rights. If B has a natural right against A that A do certain things for him, then B owns A or A’s labor. This is false.

Second, if B cannot waive his right, then either no one can or someone else can. It is odd that a person can own a right and not be able to waive it. If someone else can waive it, then that person owns B as if he were a slave.

Third, there is no good reason to think that were the purported right to exist, it would trump B’s right to his body. Body-rights are very strong. This is why raping the comatose is so wrong.


David H October 30, 2020 at 2:47 PM

Steve asks “I wonder if Dr. Hershenov thinks that physicians and pharmacists have role-based permission to avoid policies that lessen the effects of addiction, rather than eliminating them. For example, paying addicts to use fewer drugs but not stop using them or introducing one addiction to eliminate another.” After engaging in conceptual analysis via reflective equilibrium (see my Christian Bioethics paper “Conscientious Objection or an Internal Morality of Medicine” or the earlier and less developed “Pathocentric Medicine and a Moderate Internal Morality of Medicine.” 2020, Journal of Medicine and Philosophy. 45: 1, 16–27. They should both be on my web page), I concluded what is essential to being a health care professional is to fight pathologies - understood as not just curing but preventing and mitigating the symptoms of disease. If they don’t do that, but use their physiological knowledge to just interrogate, execute, euthanize, sterilize, abort etc. then they are not doctors. They are doctors if they fight pathologies in their practice even if they don’t politically advocate for policies that promote the fight against disease. So I don’t view the nature of medicine as requiring its practitioners to be politically active in ways that benefit improving societal health care I also think doctors can refrain from inducing pathologies even if those pathologies produce good results such as prevent more pathologies of the same. So maybe doctors can refuse to participate in policies that will induce addictions (pathologies) in some patients to prevent worse pathologies in others. I am not opposed to doctors inducing a pathology in service of preventing a worse pathology as when surgeons cut open someone and destroy healthy tissue as a means to some pathology fighting ultimate end of the operation. So maybe switching pathologies from the more harmful to the less harmful should be viewed in the same way as cutting off a gangrenous or cancerous organ or limb to save other organs or limbs


Stephen Kershnar October 31, 2020 at 11:41 AM


Excellent point. Still, I do not think this works.

Here is what you say, "I am not opposed to doctors inducing a pathology in service of preventing a worse pathology as when surgeons cut open someone and destroy healthy tissue as a means to some pathology fighting ultimate end of the operation. So maybe switching pathologies from the more harmful to the less harmful should be viewed in the same way as cutting off a gangrenous or cancerous organ or limb to save other organs or limbs."

(1) If this were true, then abortion, euthanization, sterilization, etc. would a way in which one pathology is introduced to prevent or lessen another. Leaving aside how the balance turns out, this conflicts with your earlier claim that in doing these things, a person is not a doctor..

Side note: If the introduction of these things need only be properly motivated, then the people introducing these things are acting as doctors even if they get it wrong.

2. If instead there is a pathology-based side-constraint against introducing pathologies (abortion, euthanasia, sterilization, etc.), then a physician could not provide methadone (it is addictive). This is implausible.

Hence, leaving aside the issue of what would ground this duty, it appears to either be too permissive (if the duty is a maximizing type notion) or too restrictive (if the duty is a side-constraint).


David H October 31, 2020 at 2:07 PM

Abortion is producing a pathology in a healthy person to prevent a disease in another (anyway, mental diseases due to pregnancy are often dubious but more likely problems of living). So abortion is not like surgeon's cutting skin to get at a diseased organ. Surgical cuts produce smaller pathologies to prevent greater pathologies. Euthanasia doesn't cure anyone of any pathologies though it can eliminate pathologies by eliminating the person. What diseases does sterilization cure? Even if it did eliminate another disease, it is likely not a necessary means to cure a disease. Perhaps you mean it will prevent a disease.

Stephen Kershnar November 2, 2020 at 9:40 AM


Consider cases in which a physician introduces a pathology in order to prevent what the physician perceives is a larger pathology (for example, sterilizes a hemophiliac to prevent a life-endangering pregnancy-delivery).

This is a pathology designed to prevent another pathology.

Is your account of a physician a success term or a intention term?

(1) If it is a success term, then we can say that someone acted as a physician only when we say that they did in fact prevent a lesser pathology. Hence, a seeming-physician who fails to cure someone (introduces chemo but does not stop the cancer) is not a real physician.

(2) If it is an intention term, then so long as the physicians who perform prophylactic abortions, sterilizations, etc. and have reasonable grounds to do so act as a physician.

As a result, we are back to the previous conclusion: Hence, leaving aside the issue of what would ground this duty, it appears to either be too permissive (if the duty is a maximizing type notion) or too restrictive (if the duty is a side-constraint).



Phil Reed November 4, 2020 at 4:37 PM

Steve, your argument against me gives this first premise.

"1. If A owes a duty to B to prevent B from being having addiction-caused harm or death, then B has a right (claim) against that A does so."

I reject this premise. Some duties do not entail corresponding rights. Here are some examples: the duty to prevent others from committing suicide, the duty to prevent others from engaging in drug-inflicted death and harm, the duty to save a child from drowning.

You think that duties without corresponding rights make no sense. That's because you are assuming, as you always do, that normative ethics has to be reduced to rights, which is false.

David H November 4, 2020 at 5:19 PM


abortion is not a prophylactic for the aborted fetus. It may be good for the health of the woman. but a pathocentric account of medicine doesn't allow killing a patient to help another. Sterilization is not a means to curing a disease like cutting skin to remove a burst appendix. If you find or invent a disease where it is part of the cure, then I would treat it like surgical cuts or cutting off a cancerous leg to prevent spread



Stephen Kershnar October 25, 2020 at 9:20 AM


The Constitution does not permit the federal government to criminalize drug use (see Article I Section 8). Sadly, ship has sailed on the view that Congress has few and defined powers.

There is an interesting issue as to whether, the Fifth and Fourteenth Amendments in the United States permit the federal government or the states to criminalize drug use.

Here is part of the Planned Parenthood v. Casey, 505 U.S. 833 (1992) majority opinion (co-written by Anthony Kennedy, Sandra O’Connor, and David Souter).

“These matters, involving the most intimate and personal choices a person may make in a lifetime, choices central to personal dignity and autonomy, are central to the liberty protected by the Fourteenth Amendment. At the heart of liberty is the right to define one’s own concept of existence, of meaning, of the universe, and of the mystery of human life. Beliefs about these matters could not define the attributes of personhood were they formed under compulsion of the State.”

If drug use is a significant intimate-and-personal choice, central to a user’s personal dignity and autonomy, or an important part of a user’s defining the concept of his own existence, then at the very least drug criminalization should be subject to strict scrutiny.



Jack Freer October 26, 2020 at 11:36 AM


Foddy and Savulescu's description of the Liberal View contains several references to the pleasurable aspect of drug use (and the autonomous choice to partake). My sense is that this is a gross oversimplification of the subjective experience of the addicted person. While this may be an accurate description of the early exposure to many addictive stimuli, it seems to evolve over time. I am mainly referring to drug use (as opposed to gambling and shopping), and specifically opioid use (as opposed to cocaine and meth), but the differences may only have to do with the degree of physical dependence (high in opioids, low in gambling), so it may be true of addiction in general.

The pattern I'm describing is an intense pleasure early in the course of the addiction, but a steady change toward use for *relief* rather than pleasure. Indeed opioid dependent individuals often describe taking their medicine when they feel "sick." Those addicted to various drugs (I've heard the description most often with cocaine) sometimes speak longingly about the first time they used (and a quest to achieve that pleasurable experience again). Heroin addicts have been known to taper/discontinue the drug in order to achieve that opioid-naive pleasure again.

Although one might equate pleasure and relief for the purpose of the autonomy argument, it seems the absolute value of the subjective experience is fundamentally different because of the direction of deviation from baseline (inducing an unpleasant experience for the sake of subsequent relief is not what most people mean by pleasure).


Robert Kelly October 26, 2020 at 5:24 PM


A la Wakefield, addiction could still be a mental disease even if their Lewis-style argument against the neurobiological evidence is successful (p. 4).


Robert Kelly October 26, 2020 at 5:26 PM


Ambivalence Unnecessary: It is not necessary that ambivalence is present in addiction, even if it is typical, and need only be, at most, a disposition towards ambivalence. Still further, we need only see addiction itself as a disposition, and understand ambivalence as either (i) a possible (part of a) manifestation of that disposition, or (ii) being present in the sense that were the addict to wish to abstain, then she wouldn’t be able to resist temptation were she to try. Hence, I can think addiction requires loss of control without thinking that addiction requires ambivalence in the sense they understand the Willpower View to require it.


Robert Kelly October 26, 2020 at 5:29 PM


Their argument that withdrawal isn’t that bad assumes that withdrawal must be a central reason for loss of control (e.g. motivating compulsion), but this is false.

a) Complex Control: Factors independent of withdrawal are relevant to control, and it is only one possible cause of control loss.

b) Withdrawal-less Addiction: It is possible to be addicted without withdrawal, even if addiction involves loss of control, since one might be an addict yet take a withdrawal-elimination pill, or die before withdrawal sets in.


Phil Reed October 29, 2020 at 1:10 PM


There are a lot of funny lines in this article. Here are some of my favorites:

“At least one case of milk addiction has been observed.” (5)

“Everybody who eats becomes somewhat addicted to the chemicals that comprise food.” (6)

“It is very hard to get addicted to water, because you must force down liters of it every day.” (8)

“Being addicted to breathing, for example, will not produce imprudent actions except in extreme cases.” (20)



Robert Kelly October 30, 2020 at 3:47 PM


These are pretty funny claims to think about. However, I think they can be useful for pointing out something that F&S and I agree on: one can probably become addicted to anything. Above, I discussed addiction to breathing and drinking water in response to Steve. There are important differences we have to be aware of – desires in addiction are not the same as “desires” or base impulses or whatever you want to call them in breathing or becoming thirsty when deprived of water for too long. However, and this is the point in thinking about these funny claims, it is still possible to become addicted to them. I agree with F&S that this would be really hard to do, and that is why such cases are extremely rare (and probably non-existent for breathing). But this doesn’t mean they couldn’t happen. As long as you can develop desires towards these things (again, ordinary desires of the reward-theory kind), and as long as you could then systematically lack control over them, then you can be addicted to them. The fact that they are so rare suggests that it is just really hard to develop ordinary desires about breathing, and this further suggests (or rather supports) the fact that such desires would be completely different than the base impulses to breath that are already produced by the autonomic nervous system. There is simply no need to desire breathing—it is done for us. But if we did (or could), I see no reason to think we couldn’t become addicted to it. The fact that this would be a rare and very weird case (or world) shouldn’t dictate whether we think it is truly possible.


Phil Reed October 29, 2020 at 1:11 PM


Foddy and Savulescu claim several times that their Liberal view is the view that addictions are “just” strong desires. This cannot be the right view of addiction. Whenever I debate Kershnar, I have the strong desire to strangle him, but I do not have an addiction to strangle Kershnar.

What is more reasonable, and perhaps what they mean but say less frequently (p. 15 mentions the desires being “regular”), is that addictions are strong and stable (persistent over time) desires. This view is much more plausible.


Neil Feit October 30, 2020 at 11:02 AM

I agree that this is more plausible, Phil. I take "stable" to indicate something like a disposition to form an explicit desire. But don't they also say that addiction is strong *appetitive* desire. It might be tricky to make this precise enough to evaluate all cases, but your desire to strangle Kershnar is arguably not an appetitive desire.

Robert Kelly October 30, 2020 at 3:46 PM

Phil and Neil,

I agree that they are not clear about this, and I agree with Neil that they have something dispositional in mind. Here is their definition of ‘appetite’ from their reply to critics (and from the notes I emailed around):

“An appetite is a disposition that generates desires that are urgent, oriented toward some rewarding behavior, periodically recurring, often in predictable circumstances, sated temporarily by their fulfillment, and generally provide pleasure.”

From this it is pretty obvious that by “strong appetite” they mean (i) a disposition to desires for rewarding behaviors (‘disposition’ and ‘desires’ captures the points about predictable circumstance, generally pleasurable, etc.), and (ii) that this disposition is strong. They say basically nothing about what it means for an addict’s appetite regarding X to be strong, as compared with an akratic’s appetite regarding X that is not strong (and so only akratic). This is a good but underdeveloped part of their account. My account develops this dispositional component further (and adds impaired control).

Neil Feit October 30, 2020 at 5:52 PM

Thanks, that makes sense. So, Phil is right that without going with stable or dispositional desires, he'd count as being addicted.

Robert Kelly November 4, 2020 at 10:13 PM


Yeah, I think he is right. And this point is important because it brings out a sort of hole in these kinds of “no difference” arguments. F&S are not the only people to argue that addiction is no different in kind from akratic actions. However, they all do something similar in introducing some kind of stability, or frequency, or degree condition, and they usually sound very dispositional. It seems like they have to in order to make the claim plausible (obviously there is *some* sort of difference between one-off akratic acts and an addict). I exploit this feature of their argument, and I think it gives their game away. The difference that they try to gloss over in terms of stability, degree, or what have you is, on my view, exactly what matters. This is the systematicity (dispositional) component of my view

Phil Reed October 29, 2020 at 1:12 PM


Rob - can you give any insight on whether the Disease View and the Willpower view, as described by Foddy and Savulescu, are accurately portrayed and representative of the main views?

I’m open to the view that addicts suffer from weakness of will. I would think that this would count as some version of the Willpower View. However, they characterize this view (and the disease view) very strongly: the addicts’ actions are non-voluntary, that addicts lack control over their actions, that they act like robots, and that they are “unwilling, disordered slaves” (13). I accept none of those characterizations and the idea that addicts suffer from weakness of will does not require them.

When a more nuanced view of willpower is considered (p. 12), the authors claim it is incompatible with how they characterize the Willpower view as one where “addictions are necessarily cases of lost self-control.”

I’m worried that the two targets of the article are straw men.


Phil Reed October 29, 2020 at 1:53 PM

Oops sorry Rob. Just read your comments on the paper and saw that you made this point.

Robert Kelly October 30, 2020 at 3:45 PM


Yeah, in short, you are right that they are straw-manned too much. For instance, I would fall under the “willpower view” but I don’t hold many of the claims they attribute to it (like ambivalence or literally irresistibility being required). However, I will just add that it is only slightly straw-manned because of the literature being such a mess. There are indeed reasonable ways to come away from papers defending the brain disease view, for instance, thinking that the author said things that F&S attribute to that view. The same goes for choice models (or “willpower views”). In other words, I don’t think it is entirely their fault that they misrepresent these views, or at least the plausible interpretations of them. The literature facilitates that.


Phil Reed October 29, 2020 at 1:30 PM


The authors are so concerned to fight the cultural taboo against drugs that they flirt with the idea that addiction is not really a problem or the idea that we have no reason to be more concerned about strong, stable desires for heroin than we do about such desires for coffee or food or water. And this is just silly, in my opinion. Drug addiction is not just a problem because it devalues health (as the authors seem to assume, e.g. pp. 8-10) but because it devalues so many other goods that make life worthwhile.

They do mention (p. 8) that drugs are “uniquely dangerous” but then go on to say that “the addictive process is, from a biological standpoint, identical whether it is achieved with sugar or with cocaine.” If this is true, then the biological standpoint is the wrong standpoint from which to understand addiction.

Suppose they are right that the Liberal view shows addicts do not necessarily lack autonomy. That does not mean that their addictions are non-problematic. I don’t see how the Liberal View changes the status quo in the way they say. The Liberal view doesn’t entail that we should have safe injecting rooms and needle exchanges. The Liberal view doesn’t entail that we should avoid laws against drug use. Even if the desires are autonomously chosen, they might still be all things considered harmful to the person who has them and there would be reason for society to intervene in some fashion. Even libertarians might believe this.

The authors write: “To get at the truth about the nature of addiction, we need to allow each person to hold his own set of desires and values” (20). In fact, what they identify as the nature of addiction has nothing to do with whether we should allow each person to hold his own set of desires and values. Their political commitments seem to be driving what they take to be the implications of their view of addiction.


Stephen Kershnar October 31, 2020 at 11:50 AM


This is a good point. Here, perhaps, is an underlying argument for their notion. Let us set aside catastrophic overrides (because they have to compete against an individual right not aggregate rights).

1. If state coercion is permissible, then it should not infringe people's moral rights.

2. Drug-related state coercion infringes people's moral rights (because they retain autonomy and have neither lost or waived their rights).

3. Hence, drug-related state coercion is wrong.

You might respond that harm, not right-infringement, is a good reason for state coercion. This would rule out fast food, dropping out of school, and dating or marrying people with various disorders (e.g., borderline personality and psychopathy). This is true whether we look at prima facie or ultima facie harm.

So, while you're right, their political commitments are driving their view of addiction, I suggest that these are political commitments that we should find quite plausible.


David H October 30, 2020 at 2:48 PM

Autonomy is tied to health - Pace Foddy and Savulescu, neither Procedural or Substantial.

Inspired by Buss, I link autonomy to healthy mental functioning – such a passive approach avoids regress of idealized deliberative and endorsement models. Buss’s view in not purely a procedural or substantial account as understood by Savulescu and Foddy “Procedural accounts of autonomy claim that our autonomy depends only on our capacity to process information and make choices in accordance with our preferences, whatever these preferences are. Substantive accounts of autonomy claim that, to be autonomous, we must possess certain normatively rational preferences.” (p. 9) Her account doesn’t just judge autonomy on internal coherence as does procedural account nor makes the substantial claim that anything unreasonable is non-autonomous. The idea is unhealthy cognitive and affective processes (those that statistically suboptimal or below evolutionary selected contributions to survival and reproduction) are not autonomous. So, choosing to risk your life to save your child’s life from danger is not heteronomous though a threat to health, but depression, eating disorders, phobias etc. and the motivations and reasons they produce will not be autonomous. Choosing to risk your healthy to save your kids will not be unhealthy and autonomy impinging as there that might be selected concern or statically normal levels of concern. Addictions will be threats to survival (and sometimes reproductive success” perhaps if we construe the latter as raising and not just producing kids) that involve malfunctioning (not selected or statistically suboptimal) cognitive and affective systems. The disease view, properly understood, doesn’t require that one can’t abstain from drugs, that one is not reasons response, that one is engaged in irrational reflexive behavior, nor that there is brain damage (oddly understood by F and S as inability to undergo further change p. 15 but a lot of brain structures harden but are not damaged). More about what hijacking entails below


Stephen Kershnar October 31, 2020 at 11:58 AM


Here is my response to the health theory of moral responsibility (autonomy is too ambiguous to be helpful in relation to F & S.

1. If something affects responsibility, then it affects a person's ability to shape his psychology or act on the basis of his psychology.

2. If something affects a person's ability to shape his psychology or act on his psychology, then it directly interferes with his mental states (see, for example, ignorance and manipulation), reasoning (see, for example, reason-responsiveness), or options (see, for example, threats).

3. Hence, if something affects someone's moral responsibility, then it directly affects his mental states, reasoning, or options.

4. Heath does not directly affect someone's mental states, reasoning, or options.

5. Hence, health does not (directly) affect someone's moral responsibility.

Health causally affects it, but so do lots of factors extrinsic to moral responsibility.



David H October 31, 2020 at 2:13 PM

Responsibility is not the same as autonomy. You may not be responsible for acting on false information but that doesn't mean the action wasn't autonomous - under your control. autonomy is concerned with actions originating from you rather than being foreign in some sense. Mental dysfunction will often mean the disease generates the reasons for acting.


Stephen Kershnar November 2, 2020 at 9:27 AM


You state the following.

(1) An person's act is autonomous if and only if it originated from him in the relevant way.

The question is the following: Is the relevant way moral-responsibility-making or a necessary condition for the moral-responsibility-maker?

Let us pose the following dilemma.

(1) Autonomy is closely connected to responsibility (via identity or as an essential condition).

(2) If autonomy is closely connected to responsibility, then the above argument can be restated in terms of autonomy and the health theory of autonomy is false.

(3) If autonomy is not closely connected to responsibility, then the healthy theory of autonomy loses its intuitive appeal (for example, it loses its connection to normatively rational preferences).

(4) Hence, the health theory of autonomy is either false or there is no intuitive argument for it.



Robert Kelly November 4, 2020 at 10:15 PM


I have some independent reservations about the Buss view. Still, I guess my first thought here would be that this view is compatible with my account. If autonomy is linked to health, then, on my view, some addicts might be autonomous, and some might not. Moreover, as I mentioned in some above comments, it is consistent with my view that every actual instance of addiction is an instance of disease. I doubt that this is true. Maybe it is true, though, that the systematic lack of control over one’s desires is a dysfunction. I am not that opposed to addiction ending up being classified as a disease. My reservations have to do with possible worlds or histories that make qualitatively identical people addicted and not addicted on the grounds that, according to one world or history, their physical makeup constitutes a dysfunction and, according to the other, it doesn’t. But if nothing changes about the person and their physical makeup, then I want to say that we have two addictions, one of which is diseased. This is because they would both have the systematic disposition to fail to control their desires for X. It’s just that, because of their history, for instance, this is dysfunction for one of them.


David H November 4, 2020 at 10:16 PM


I don’t think people in the functions literature are or should be bothered by your healthy and unhealthy doppelgangers. Consider two artifacts that are qualitative duplicates but one is a functioning X and the other is a dysfunctional Y because of their histories and the intentions of their creators. Imagine the creator of the dysfunctional Y was trying to make a functional Y and it just came out doing what an X does rather than a Y. Or imagine two moles millennia apart. One lives above ground and is blind and dysfunctional and the other is born centuries later after moles have moved underground and their eyes have become vestigial. Their eyes could be duplicates.

I think a better case to make your point might be someone who is fatigued but healthy and who fails to do something just like someone who is not fatigued but dysfunctional. There are a lot of cases where the healthy are sleepy, pregnant, fatigued and can’t do certain things just like those who are unhealthy


Robert Kelly November 4, 2020 at 10:24 PM


Maybe I was unclear above. The fact that one person is diseased and the other isn’t is not the problem. The two moles differ with respect to having a disease. That is fine. But they are both still blind. Being blind is not *defined* in terms of disease. Both moles are blind, but one of them is diseased and the other isn’t. If blindness was defined in terms of disease, the non-diseased mole would not just be non-diseased; it would not be blind! That is my problem with addiction being defined in terms of disease. I don’t care if the two doppelgangers differ *with respect to having a disease*. That’s fine; history and environment matter to a condition counting as a disease. But they are both still addicted; it’s just that one addiction is a disease and the other addiction is not. If addiction is defined as a disease, then we don’t just lose the disease status in the alternate world, we lose the fact that this person even has an addiction. Whatever the same type of condition is that gets instantiated in both cases—that is what I am calling an addiction. This can’t be disease because, by stipulation, one is diseased and the other isn’t. As you said, this is not a problem because disease status can depend on facts external to the agent. But whatever the condition is that depends on the agent and their physical makeup (their dispositions, in particular)—i.e. being disposed to have certain desires and being disposed to systematically fail to control them—that is the same in both cases (again, by stipulation). And that is the addiction. If this is true, then addiction is not a disease by definition.

Also, control seems to come apart from autonomy anyways, but I’m not sure how that changes things. For instance, I (and probably many incompatibilists) think that determinism undermines free will but perhaps not autonomy. The Buss view could explain this, but so could other accounts. The point is just that appealing to a Buss-style view of autonomy does not yet tell us whether addicts have or lack a normal capacity for control, since they might have impaired control yet still be autonomous.


David H November 4, 2020 at 10:24 PM

My point was about autonomy not free will so why does it matter that determinism undermines free will but perhaps not autonomy? Other accounts of autonomy can make addicts non-autonomous but they don’t explain why addiction is foreign to the self. They can’t provide an account of a true self or nature of the self. For example, there is no reason to think we ESSENTIALLY have Franfkurtian second-order volitions since we were once mindless fetuses and may become mindless adults. I am interested in taking autonomy literally as self-determination so I need an account of a true self and operations that are not expressive of that true self. My modification of Buss’s account that we are essentially living beings and disease is defined, in part, as making it more likely we will die sooner than the healthy. So that makes it possible to see disease as contrary to our nature, foreign to our self. Buss claims all the idealized deliberative or endorsement intellectualized accounts of autonomy suffer a regress as one can’t deliberate or endorse everything.

My point is a small one directed at F and S that is paternalism is harder to justify when the person is autonomous than not. They will have addicts who may be autonomous. I will have no autonomous addicts – at least none who are acting on reasons generated by their addiction. So I have guaranteed that that addicts are then not autonomous. If one is interested in paternalism – then I have made the paternalists job a little easier. I suspect libertarians like Steve are right about legalization but that is another issue.

You write that my Buss-inspired account can’t tell us whether addicts have a normal capacity for control since they might have impaired control yet still be autonomous. My account doesn’t make control essential to addiction so there can be addicts with normal control. I imagined addicts with superhuman control. I would disagree with you that they are autonomous. So this critique perhaps begs the question against my account


Robert Kelly November 4, 2020 at 10:25 PM


I may have misunderstood the original point. Your clarification was helpful. My question was about what exactly is diseased in addiction and what exactly is not part of the individual’s true self. I worry that, once we take into account what addicted desires are, you are going to make bungee jumpers, cave divers, race car drivers, and so on diseased. But you address this below, so I will shift to that comment and respond there.


David H November 5, 2020 at 12:28 PM



I don’t know the responsibility literature. But I have a faint memory of a PANTC or Blameless Buffalo? discussion about saying someone is responsible if they are praiseworthy or blameworthy. On the idealized agent accounts of autonomy, someone who is weak willed is not autonomous but intuitively blameworthy for eating the last piece of pie. So autonomy is not a necessary condition for responsibility on such account. I suspect it is a sufficient condition for responsibility

Buss rejects idealized agent accounts, in part because they suffer a regress as one can’t deliberate or endorse all of one’s deliberations or endorsements, so she claims autonomy needs a passive mode – healthy mental functioning. Autonomy is self-determination and so it is determinations by the true self that are autonomous. I try to make her talk of a true self more ontologically respectable with an animalist account of our nature and persistence and a definition of disease that that makes death more likely which leave diseases foreign and contrary (in a sense) to our nature and persistence conditions.

Healthy mental functioning seems intuitively compatible with someone being blameless for their misconduct due to misinformation. Falsehoods are not foreign to one’s self. (For all I know, Freud was correct that some mental operations (repression) are functioning correctly when they produce false beliefs!) If one can be autonomous with false beliefs that render one blameless, then autonomy is not sufficient for responsibility. I also think there are conditions that mimic disease – fatigue from thinking too hard for too long, sleep deprivation that leads to poor thinking, to the non-pathological pain that leads to unreasonable thoughts. If I don’t extend the loss of autonomy to conditions that mimic disease, then there will be healthy autonomous actions that one is not responsible for. So if such conditions are autonomous but undermine responsibility, then autonomy is not sufficient for responsibility

Is autonomy necessary for responsibility. I don’t know. One reason is my uncertainty about an asymmetry in responsibility – we aren’t blamed for doing certain things but if we didn’t do them we would be praised. If someone has a mental disease but they resist the action that the disease “promotes” through some superhuman will power, I think they are praiseworthy and so responsible for the act and deserve credit. But if they failure to resist the urges or voices of their mental disease, I wouldn’t blame them. So it seems someone might be responsible without being autonomous. But I haven’t thought much about these matters

Is autonomy uninteresting or useless if it is neither necessary nor sufficient for responsibility? I don’t think so as it will often be the case that if an act is not autonomous that one won’t be responsible for it. I am not confident that autonomy isn’t a necessary condition for responsibility as my counterexample is pretty under-described and farfetched. Anyway, I think it is valuable to know what is expressive of your true self and what is not


David H October 30, 2020 at 2:48 PM

The Absence of Autonomy and Justifying Paternalism:

The authors liberal account is open to addiction being an autonomous pursuit of pleasure given their ranking of such pleasures above health and other goods. So paternalism, which they don’t rule out, still cannot be justified in part on the grounds that the addict is not autonomous. But on the Buss-inspired view of addiction that I favor, the addict will not be autonomous. So insomuch as paternalism is easier to justify for the non-autonomous than autonomous, that consideration is still available for the paternalist


David H October 30, 2020 at 4:38 PM

More carefully, some of the addict's reasons and desires won't be autonomous. Having a mental dysfunction doesn't render heteronomous all of one's reasons and actions

Stephen Kershnar October 31, 2020 at 11:59 AM

I do not see why a mental dysfunction does not render heteronomous reasons and actions if unhealthiness does so. The two would seem to have the same sort of effect.

David H October 31, 2020 at 2:01 PM

there are many mental functions and so a dysfunction of one doesn't render all other reasons heteronomous. a person who is paranoid may still have autonomous reason to do certain things that are not produced by the paranoia

Stephen Kershnar November 2, 2020 at 9:32 AM

Good point, but we are talking about reasons and actions that flow from the mental dysfunction. Not ones separate from it.

You need the following claim.

(1) Addiction produces heteronomous desire in a way different from mental dysfunction.

I am not sure I see why we should think this is true. This is especially true if in people an addiction is a type of mental dysfunction.


David H November 5, 2020 at 11:51 AM

Whatcha talkin' bout Willis?


David H October 30, 2020 at 2:48 PM

The need for autonomy to be based upon a true self and foreign disease:

The authors find themselves in an epistemic quandary when they doubt that we can know the addict’s ranking of desires and so can’t say whether they are acting contrary against their strongest desires – Joes just may value the pleasure more than ten million dollar (p. 17). They offer what amounts to a futuristic scientific Hail Mary that there will be a neurological breakthrough and we will be able to measure desires or something. “It may be that neuroscience can tell us whether addictive behavior is autonomous. But empirical strategies for answering this question would depend upon technology that is not available.” (p. 17) But this inability just shows the need for an account of autonomy like Buss’s. Autonomy is self-control and so if there is a true self and diseases that are contrary to that true self, then we have a good framework for judging addiction as autonomous or not. So I offer (see my paper entitled “Health as the key to autonomy in a divinely determined world” the current issue of Religious Studies June 2020 vol 56 issue 2) that we are essentially living organisms, diseases are defined, in part, as making survival less likely, so they are contrary to our nature as living beings. Diseases are not foreign to our essence like being a car or squareness is contrary to the nature of a circle, but it still fits the bill as accounting for how disease can be alien to ourselves, contrary to our true self and nature (essence). So mental diseases can be autonomy undermining as they lead to motives and reasons that stem not from our true self but are in, the appropriate sense, foreign to our selves as so when the ‘disease is producing reasons to act”, we are not expressing our true self in such heteronomous actions.


Robert Kelly November 4, 2020 at 10:26 PM


I guess I am having some trouble with the view. What exactly is the disease supposed to be in addiction? Pathological desires? What does that mean? Is at described at the psychological level, such as being too strong? Is it at the behavioral level, such as being too frequent? Is it at the neurobiological level, such as having too many or few dopamine receptors, or something like this? Is it pathological reasons? What does that mean

David H November 4, 2020 at 10:29 PM


Pathologically strong need is my first stab at addiction. I am assuming there is a healthy level of certain needs for X or health involves no need for X. I prefer some etiological account of function and dysfunction. I argued it didn’t require being too frequent as the person with super control who resists levels of needs the rest of us would give into is an addict. I don’t know what the neurological level looks like. It is just a functional account. I suspect on rare occasions there is little neurological difference as Foddy and Savulescu claim between the addict and non-addict but the difference is in the object craved as Wakefield argued in his exchange with Lewis.

Robert Kelly November 4, 2020 at 10:31 PM


I still don’t understand what “pathologically strong need” means. What is a need? What makes it pathological? Wakefield argues that the object *can* make it pathological. Though he is not just talking about a need. He involves the whole desire-and-deliberation process, including its disruption (which is easily understood as an impairment in control). He only introduces the object of addiction/desire because he is entertaining Lewis’s claim that we can get a qualitatively identical brain, desire, and deliberation process for someone in intense love. The object can distinguish love from heroin addiction in the way he wants. But this only gets one so far. Now we need a story about evolving to have only certain contents of desire (or objects of deliberation), as opposed to evolving to have certain desiring and deliberation processes or systems, working in particular ways (such as strengthening dispositions when we represent states of affairs as issuing in more reward than expected, regardless of the rewarding objects/processes). I don’t recall his response to the authors that argued we did actually evolve to have rewarding plant chemicals as objects of our desire (i.e. as objects/inputs of/to our reward and motivation systems). Setting that aside, I still want to know what a pathological need is. If you don’t require frequency, strength, number of dopamine receptors, etc., then you are only left with pathological need being “a need (?) that has as its object something which is outside of our design.” We did not evolve to desire to go to the movies—is this pathological? I don’t see how you break the analogy to desiring heroin, or gambling, or what have you, if what is supposed to make them pathological is that we weren’t evolved to desire such things. Harm (or being likely to reduce lifespan or some such thing) also won’t help you, since, first, you agree that there are harmless addictions, and second, this would make the desires of bungee jumpers and bull riders diseased (if these aren’t good examples, we can surely find an example that works—I don’t have the data handy on risk of harm/death for various risky activities people engage in every day).

I think it would be quite difficult, given the extremely varied way in which addiction can be instantiated and manifested, to pinpoint the specifics in answering those questions. In Wakefield’s exchange with Lewis, he ultimately falls back on the object of the desire being something for which we were not designed to desire (in the way it is desire in addiction—so strongly maybe?). Part of the reason is that Lewis, a neuroscientist, gives some compelling reasons to think that addicts’ brains are often doing quite normal things in terms of the reward and motivation system. I’m not saying he is totally right. The point is just that he seems to be right about the brain changes, while systematic, being of the same types of changes we see in lots of reward learning. So we’d have to speak in terms of degrees of something—strength? Frequency? Dopamine levels?—to pinpoint the disease/dysfunction. What is this on your view?

Also, how do you pick which desires, or strengths of desires, or brain systems, or brain functionings, or whatever are your true self and which aren’t? Is this a metaphorical sense of “true self” or is the addict’s dopaminergic system literally like an implant or foreign object when it starts operating with the drug of choice as its object? What about when that same reward system is operating later that day with respect to food, or a fight with the spouse, or while watching football? Does it become a part of them again? Is it about the specific neurons that are talking to each other as a part of that system or the system as a whole? I’m just having some trouble getting my head around what the view of addiction is, where and what exactly the pathology or dysfunction is, which anatomical parts of not the true self, and whether that is metaphorical or literal.

David H November 4, 2020 at 10:40 PM


Not all diseases involve foreign bodies. But diseases are foreign in the sense they are contrary or at odds with our nature as essentially living beings, self-maintaining, integrating entropy resisters or how ever we define organisms. That is they are defined, in part, as making it less likely we stay alive. That is, diseased parts make suboptimal contributions to survival undermine our persistence conditions. Since our nature determines our persistence conditions (to some extent), diseases that increase the likelihood we don’t persist are contrary to our nature. They are not contrary in the way being a table is. We can’t be a human being and a table but we can be a human being and terminally ill.

Robert Kelly November 4, 2020 at 10:43 PM


See my comments above on risky or harmful desires. Also, again, this seems inconsistent with your previously professed commitment to the view that harmless addictions are possible.

David H October 30, 2020 at 2:50 PM

Disease model prevents too liberal or promiscuous spread of addiction:

As Steve has already noted, if addiction is just a strong desires or appetites for pleasure, then being in love, enjoying friends, commitments to and delights in one’s children, passionate pursuit of hobbies, even the desire to remain conscious, and acquire nutrition and hydration are addictions as one cares strongly for such things. Intuitively that shows that intensity of desire is insufficient for addiction. If addiction is just strong desire than we can even addicted to the avoidance of pain without the alternative being pleasure. If you thought addiction to exercise was going too far, surely you should think addiction to avoiding exercise is going even further astray. But one can have a strong desire to avoid the pain of exercise and it is not a desire for resulting pleasure, just the absence of discomfort. The disease model, properly construed a la Wakefield or Boorse, not as Leshner or Volker are represented, will allow there to be strong attractions without addiction if they involve mental systems doing what they were designed to do (evolved, created, historically have done for millennia etc.) I suspect this is a problem not just for the liberal view but also willpower and perhaps Rob’s disposition to control models. Can they distinguish powerful additive drives from intense healthy drives like love or must treat them all as addictions?


Robert Kelly November 4, 2020 at 10:48 PM


The short answer is, I don’t care about distinguishing so-called “healthy drives” if those healthy drives involve a systematic disposition to fail to control them. But again, I don’t know if you take the systematic lack of control over desires itself to be dysfunctional. I guess I think you can have healthy addictions, depending on what you mean by that. What I care about is whether the individual has a systematic disposition, and that disposition is towards failures to control certain types of their desires. If those desires are “healthy”, whatever that means, then there are healthy addictions. Your explanations above seem to belie the fact that, like other disease proponents, you are almost exclusively focused on harm (and also perhaps naïve examples of addicts as sore-infested heroin junkies living in alleys because their lives are destroyed). As explained above, I think (and I thought you did as well) that addiction is not necessarily harmful. If it isn’t, then the appeals to addiction jeopardizing survival and reproduction are unfounded. Maybe this is common or typical (but this ignores the problem with the data, such as selections effects and the problematic DSM being the inclusion criteria for most studies on addiction), but it is not necessary.

David H November 4, 2020 at 10:50 PM


I defend a naturalist account of disease that doesn’t require harm. See my article in the current issue of Theoretical Medicine and Bioethics entitled something unimaginative and unmemorable like “a naturalist response to Kingma’s critique of naturalism.” I certainly understand the attraction of the normativist’s account of disease as harmful since conditions that make it more likely you die young (and for many people are infertile) will be harms. But in my article, I argue that the epistemic possibility of Epicureanism and its competitor Universalism means that fatal diseases can be harmless. I certainly would still want to argue that people died of diseases even though it wasn’t a harm because they don’t exist or are in Heaven.

Robert Kelly November 4, 2020 at 10:51 PM


F&S are not so naïve to say that addiction is merely a strong appetite in the sense that it simply amounts to strong desires. They are sneaky, but not that naïve. By saying “strong” appetite, they are sneaking in a degree condition that will separate “mere” akratics from addicted akratics. They contend there is no difference in kind, but their view allows that there are differences of degree, even big ones. This is what Neil, Phil, and I were discussing above concerning stability and dispositions. The desire needs to be stable in some way—like being significantly stronger, more frequent, or both—and I think this is just them getting at the fact that addicts are strongly disposed to akratic acts, whereas akratics (one who commits an akratic act) is not necessarily *strongly disposed* towards those acts, or at least not with respect to some specific type of behaviors. If they were, then they would be promoted from mere akratic to addicted akratic. There “no difference” claim is really about individual actions, I think. And so this leaves room for systematic, dispositional differences.

My view distinguishes normal urges (if they can be called that) to breath and drink when thirsty because these are not desires in Arpaly and Schroeder’s sense. Concerning desires to exercise and loving your kids, I have no prior motivations to exclude these things from the category of what we can possibly be addicted to. Just because it is normally healthy to love your kids or get some exercise, this doesn’t mean you can’t develop desires for those things that turn into a systematic disposition to fail to control those desires. That would be an addiction. If that becomes unhealthy on your view, then it seems like your view entails that a systematic lack of control over desires is unhealthy. Maybe that is true. If so, then addiction is unhealthy. But I would want to know why this is true, and this would also seem to make it unhealthy in a different way that you have been describing (vis-à-vis survival and reproduction).

David H November 4, 2020 at 10:55 PM


I believe healthy people would have certain levels of control. Those who are way below that level are pathological. I just don’t think lack of control is essential to addiction. I think it is possible for an addict to have a normal level of control and just an extraordinary excessive drive. So he gives in. It need not be that he has less control than non-addict. It could be. But it could also be he has normal or super control but the urge is not normal. overpowering. I wonder if my account is in trouble with the case of someone who had normal level of desire or appetite but suboptimal control. Do I have to say that person is not an addict but suffers from a different pathology? Hmmm.


Robert Kelly November 4, 2020 at 10:56 PM


Thank you for the counterexample! In all seriousness, on your first point, fair enough. If you think there is reason to believe that impaired control may itself be dysfunctional, then my account may render addiction a disease. The question would just be whether *systematic failure to control desires* matches up with the level of impaired control that counts as *dysfunctional control loss*. Seems plausible. Once you convince me of the reasons for thinking impaired control is dysfunctional, I may just change my view and count addiction as a disease. However, and most importantly, my framework for what addiction is would be the same. This is why I have tried to emphasize that, although I think we should refrain from *defining* addiction as a disease, my account is consistent with every instance of addiction being an instance of disease.

Another thing to point out is that your “mere” pathological need view seems like it would have a lot of trouble distinguishing itself from other disorders that seemed to be defined in terms of intrusive thoughts or obsessions (that can include desires and urges). You seem to characterize addiction as a sort of constant thought/feeling intrusion, where it doesn’t matter whether or how one is able to respond to those thoughts or feelings. Why aren’t people with certain forms of OCD, PTSD, dementia and so on addicts on your view, since these conditions can be manifested in the form of recurring, intrusive needs? Maybe it is because I still am unsure what a “need” is on your view. My account avoids this because these are not desires in the Arpaly & Schroeder sense (and intentionally so, as they are interested in the reward system and have a chapter on addiction). Robbins and Everett have demonstrated some of the neurological and behavioral differences between compulsions and desires (or “wants” in the Robinson and Berridge sense). These differences map on to Arpaly & Schroeder’s view of desires, and so I can rule out compulsive urges from things like OCD and the urge to breath.


David H October 30, 2020 at 2:50 PM

Misunderstanding of hijacking our motivational system:

The authors reject talk of hijacking as addiction is accounted for as other pursuits of desire that affect our brain, training pathways and producing proclivities through similar reward structures. But hijacking need not involve different mechanism but just different object. Instead of flying to Miami, hijacker flies to Cuba –, so instead of desiring healthy X,Y,Z which he evolved to desire, he desires intoxicating drug. The latter is not what those learning faculties were designed to pursue as they make it less likely the addict survives and reproduces, unlike other enjoyable learned behaviors


Robert Kelly November 4, 2020 at 11:01 PM


What is the reason for thinking that we can’t be addicted to exercise, or sugar, or a beloved (think obsessed stalkers)? This seems possible, but these are things we probably evolved to desire.

David H November 4, 2020 at 11:02 PM


I agree. We could be. But my worry was that Foddy and Savulescu will treat normal love given its strength and the way it blinds us and outweighs other considerations to be an addiction. Isn’t their main point that there is a rationality to many addictions that moralizing about pleasure being aberrant keeps us from seeing?

Also, I don’t think this is how anyone usually talks about hijacking in the literature. Volkow is probably the most famous for this, and she doesn’t mean it in this way. It is about the chemicals in drugs having direct causal effects on the reward system (specifically the dopaminergic system) that are typically produced indirectly. Also, they mean that the reward system fails to respond to certain negative effects of engaging in the behavior the way it normally does—in other words, the feedback loop is not effective. Usually, if bad things start happening when we do X, we will learn to stop doing X. One reason is that we get punishment signals when we do X (Brain: “Hey, that was worse than we predicted!”). With addictive drugs, they directly stimulate the dopaminergic system so that you get a reward signal no matter what, which strengthens the disposition to do that behavior again, and it does so independently of other considerations or effects that behavior may have, and so the feedback loop gets messed up. That is what they mean by hijacking.

I am unfamiliar of the literature on hijacking. I was just responding to Foddy and Savulescu treating addiction like other cases of rationally strongly valuing certain things as to be neurologically similar to such learned behaviors. I was then trying to save hijacking by appealing Wakefield style to the object of the desire rather than a neurological difference


David H October 30, 2020 at 2:50 PM

Foddy and Savulescu’s mistaken notion of mental disease:

1) They sound like Szasz in the demand for a lesion for there to be a brain disease or like Marc Lewis in their appeal to other forms of learning being similar in rewards and training and brain changes as addictions. Nor does it matter that addicts are no more restricted in their choices than we are in our everyday limitations of choices. I think this all misses the lesson of Wakefield’s gosling imprinted on a fox that looks like the gosling’s mother counterexample. Diseases involve dysfunctions and these can be anatomically indistinguishable from healthy brains just as two artifacts can be physical duplicates but one is a functional x (given its design history) and the other is a dysfunctional y (given its design history). We can’t tell from the brain scans that the gosling imprinted not on its mother but on a fox that looked like a goose. Addiction is a disease when it produces beliefs and desires that are dysfunctional due to the substitution of the wrong object of one’s attitudes. Disease produced cognitive states (beliefs, reasoning, inferences, deductions, inductions, associations etc.) and affective states (desires, needs, inclinations, appetites etc.) won’t be making the contributions to survival (and reproduction) that they were designed to do. It doesn’t matter that the addiction is similar to learning and rewards. So it is false to say “whether or not we think addictive desires are truly irresistible, we must agree that they are biologically the same as strong desires which are not oriented towards some other rewarding outcome.” (P. 4) The creature is desiring something that is a threat or keeps the creature from desires what will keep it alive. 1) Likewise, learning to be aroused by only shoes or to eat exclusively paint may be similar learning but the objects render the diseases dysfunctional undermines respectively reproduction and survival. It doesn’t matter that it produced endorphins and brain patterns like sugar and chicken. 3) Or if one can’t function at school or work without drugs or drink, that is a disease as healthy cognitive and affective systems are designed to depend upon food and sleep, not pills and booze.


David H October 30, 2020 at 2:51 PM

Weakness of Will, Autonomy, and Disease:

weakness of will need not be autonomy or responsibility undermining as we are accountable and blameworthy for giving into temptation and taking a second and the last piece of cake when someone else has not had any. If autonomy and responsibility diverge, maybe we are responsible for instances of weakness of the will but aren’t autonomous as the latter is an ideal. But I construe autonomy primarily as healthy mental functioning and think most weaknesses of will are not unhealthy as they are a product of our design like being fatigued after mental or physical exercise. Our inability to think well or exert ourselves when fatigued is not malfunctioning but a byproduct of our design. I see most weakness of will as likewise just a byproduct of our design. However, it could be if one is pathologically weak willed. I imagine nature didn’t select our cognitive equipment never to be weak-willed but to have a certain amount of will power. So, some weak-willed folks are diseased and not autonomous but by no means is that essential to addiction

David H October 30, 2020 at 2:51 PM

Addiction as pathological strong need (appetite?):

I choose to define disease in terms of need (I am open to a better term) rather than desire as that latter sounds too conceptual as one needs propositional content to desire x. – one can’t desire what one can’t conceptualize. I want a newborn to be addicted I even want to say the unconscious in withdrawal are addicts so they need the drug just as trees need water. Am I committed to the seemingly absurd possibility of addiction in the permanently mindless? Addicted plants? I might be but I will wait on the details about the pathologically strong needs of such plants. I can always abandon the non-psychological reading of need that I favor because I want the mindless in withdrawal with no possibility of mindedness in the future to be considered addicted. See below


Robert Kelly November 4, 2020 at 11:03 PM


The difference between physical dependence (tolerance and withdrawal) and addiction takes care of this. Distinguishing them is now standard. The change from DSM-IV to DSM-V incorporated this distinction concerning addiction (the DSM’s “substance use disorders”). We know people can become dependent and not addicted, as when they develop tolerance and withdrawal for benzos, pain meds, anti-depressants yet have no problem stopping once the treatment is done. We also know some drugs (cocaine, ecstasy) are much weaker in terms of producing dependence, but people can become addicted to them nonetheless because they can lose control over their desires for them.

Your view on dependence being sufficient for addiction is not shared in the literature.

David H November 4, 2020 at 11:05 PM

Response to Rob – Reflective Equilibrium and Nosology:

I was trying to say what addiction is not be loyal to the literature. I had in mind a small sample of the literature that couldn’t capture what we lay people consider to be addictions – newborns of opiate addicts who passed their addiction onto their offspring. So given the distinction between dependence and addiction, then I am using the latter in a folk rather than scientific way. But philosophy of medicine doesn’t have to leave all usage of medical practitioners intact. We, or at least the Boorses amongst us, are better at conceptual analysis than non-philosophers in the sciences. So the question comes down to nosology – disease individuation. Is the lack of control a symptom of someone with a pathological (dependent causing) need for opiates, alcohol etc.? Or is it a separate disease – assume for the sake of argument addiction is a disease. My earlier posts were appeals to how counterintuitive it would be if neonates became addicts later when their control systems developed enough to fail. I also thought it counterintuitive if their dependency was so strong that they couldn’t become addicted or if someone was addicted, he could cease to be by removal of control system

Robert Kelly November 4, 2020 at 11:06 PM


I see. I agree with the Boorsean approach, but I don’t see why this would make newborns addicted. When we are doing our philosophical analysis of this stuff, using our intuitions and so on, we should still come to the idea that physical dependence is distinct from addiction. I am not just relying on what the medical literature says for what addiction is. I would be a brain disease theorist if that were the case. I am just pointing to the facts that we have about physical dependence and using my intuitions and philosophical theorizing to help me utilize that evidence. If dependence were sufficient for addiction (helping you to capture what you call the “folk” idea that newborns are addicted), then I think you are going to get a way too permissive account of addiction. Do you just mean tolerance and disposition to withdrawal (dependence) by pathological need? Plants will be able to be addicted on this account. It doesn’t seem like any sort of psychology is required for tolerance (a substance having less and less of its typical effects after repeated exposure) or withdrawal (disposition to unhealthy or otherwise negative processes when exposure ceases). Withdrawal is often described as involving psychological symptoms in addition to physical, but this is because people are typically the referent. Physical withdrawal seems like it could occur in a plant. The point is that I am not just trying to say what the medical community thinks. Non-disease proponents from non-medical disciplines who think addiction largely involves choice do not equate addiction with dependence. Philosophers like myself and others who come at this as Boorse tried to come at the phil med literature do not equate addiction with dependence. I think this should force you to provide a clearer picture of what you actually mean by “pathological need” (the “need” part, in particular). This seems to be where your view is running addiction and dependence together


David H October 30, 2020 at 2:52 PM

Addictions without Strong Desire:

The authors write “The liberal view is not so minimal that it cannot say what addictions are they are strong appetites to pleasure” (p. 17) and they earlier wrote “…to produce addictions. All that is required is a high dose of any rewarding behavior, be it eating, sex or anything else the subject finds attractive.” (pp. 6-7) Elsewhere they write “addictive desires are just strong, regular appetitive desires” (p. 15) and they entitle one section Addictive Desires are just Strong Desires towards Pleasure” (p. 16). I am not even sure that a strong desire is a necessary condition as I think there can be addictions without desires in the very young newborn who can’t conceptualize and thus desire the drug or cognitively impaired and maybe even in the comatose in withdrawal.

David H October 30, 2020 at 2:52 PM

Addiction without pleasure seeking:

This is related, I think, to Jack Freer’s comment about relief rather than pleasure motivating some long-term addicts. I think we read an article in a pantc or blameless buffalo meeting where addicts ceased to enjoy the drug but still craved it (perhaps “crave” not the best word, but I don’t think they used ‘relief’ as it didn’t provide any or much or for long term. Maybe ‘urge’ is better). (Maybe Rob is aware of the reference even if we didn’t read it together in Pantc or Blameless Buffalo but it was an article in one of the addiction anthologies I own. ) The drug use wasn’t pleasant but they still needed it and it wasn’t just to avoid withdrawal. The authors had some explanation that assumed it wasn’t pleasure seeking. If that is phenomenologically correct about some long-term addicts who use without euphoria or pleasure, then that doesn’t fit into Savulescu and Foddy’s account of addiction described in their p. 16 section title Addictive Desires are Just Strong Desires Towards Pleasure.


Robert Kelly November 5, 2020 at 10:13 AM


Yes, it is Robinson and Berridge’s “incentive sensitization theory” that describes how liking (pleasure-based) and wanting (motivation-based) come apart. My view—that is, Arpaly and Schroeder’s view of desire—can accommodate this. If you don’t like the word ‘desire’, you are free to substitute your favorite word. But when I say desires are necessary, I don’t mean they have to be intellectualized or that they even involve pleasure. Desires on A&S’s view are not the same as appetites or habits even. They *typically* cause pleasure and promote habits, but they need not. They are representations of states of affairs as rewards (which has to do with their causal role in the reward and motivation system producing learning signals). I agree that this is a place where F&S (who I ultimately disagree with on most things in their account) have a problem.


David H October 30, 2020 at 2:52 PM

Differences of Degrees and Kinds – Misunderstanding the Medical Spectrum:

The authors write “terms like addiction and dependence can be reasonably employed when a person’s like become particularly strong, but it should be understood that these terms denote a difference indegree, not a difference in kind.” We don’t say someone with 84 degrees Fahrenheit temperature is a little feverish or anyone with some amount of iron is somewhat anemic, everyone with some memory loss is a somewhat dementia, every blood pressure is a degree of hypertension etc. So why say it with addiction?


David H October 30, 2020 at 5:04 PM

The author say everyone who eats is somewhat addicted to food. to quote the great Scranton philosopher JB: "Come on, Man!"

Robert Kelly November 5, 2020 at 10:19 AM


I totally agree. To be fair, I think they do want to have a more rigorous distinction than that (again, see the comments from Phil, Neil, and I above). I think they want “strong” to be doing a lot of work, much of which they don’t flesh out. But my account helps with this. The difference comes to systematicity. I actually don’t even care if addiction is a “natural kind” or not. Maybe the line gets drawn somewhat pragmatically. Maybe not. But we are circling a particular type of disposition that akratics (and other non-addicts) lack. I’m not that interested in whether that disposition is a natural kind or not.


David H October 30, 2020 at 2:53 PM

Addictions without loss of control:

I agree with the authors that addiction need not involve a loss of control but for very different reasons as I am not requiring different levels of control than those found in non-addicts. I think the mistake of defining addiction in terms of control is conflate a common symptom with the disease, making a common disposition of addicts into the nature of addiction. Most addicts will be disposed to lose control but there are young neonatal addicts with no control to lose but their addiction will prevent them from acquiring self-control. Addictions shouldn’t be limited to those capable of control or autonomy. I would think the concept addiction should also apply to non-human animals who don’t ever exercise something like self-conscious control which can be overwhelmed by an addiction


Robert Kelly November 5, 2020 at 10:22 AM


Again, you keep appealing to neonates. This just is not a problem, and no one in the literature thinks it is (as is evidenced by the fact that no one talks about this when discussing control and so on in addiction). Neonates are dependent, and no one thinks this is sufficient for addiction. Problem solved.

Second, control need not be self-conscious. We have more control than apes, and they have more control than rats, and they have more control than beetles. Dan Dennett and Bruce Waller each have (different) accounts of control (Waller calls his free will) that are based in evolution and are compatible with determinism. Each can be used to explain how certain mammals can have control, and how that control can evolve. Waller thinks another degree of control is required for responsibility, and so he thinks we have free will but nor responsibility (a reverse semi-compatibilist). Anyhow, just as you don’t want to over-intellectualize desire, we should not over-intellectualize control. Rats can typically resist urges to do things they like if you give them certain incentives. “Addicted” rats (if they are) seem to lose this ability. Maybe it is not addiction. But it sure does look like they lose certain abilities that they previously had (and that rats typically have), and those abilities look an awful lot like varying degrees of control. I even think Fischer’s reason-responsiveness framework can be adapted to such cases to explain the control loss there.

David H November 5, 2020 at 10:23 AM

Symptoms and Disease Individuation:


You state that addiction could be a disease and I assume you accept that there are diseases are not the same as their symptoms. Would you agree that someone with an obsessive compulsion handwashing disorder is still disordered with the obsessive handwashing disorder if we give her a pill that leaves the desire to wash her hands just as intensive and disruptive and frazzled, but the will power to resist the urges? (Assume the compulsion doesn’t migrate but remains oriented to hand washing.) Would you fall back on the disposition to failure of control still being there if she ceases to take the drug? That sounds to be a too promiscuous use of disposition. Anyway, let’s say the pill’s effects don’t wear offs so she is no more disposed to handwashing than I am. Is what remains a different disease? I think not. Wouldn’t it be bad nosology – confusing symptoms with diseases. Why isn’t this analogous to addict struggling successfully with maintaining control due to will power or a will power drug? (Savulescu elsewhere in an article with Ingvar Persson talks about fatigue in resistance to temptation setting in but glucose pills offsetting this moral fatigue). So my lay person’s suspicion is that essential to obsessive compulsive disorder is not a loss of control but the pathological urge. Why treat addiction differently? Why not make it the need or appetite (which is perhaps not captured by “dependence” as one can be dependent on something without an urge or drive for it). Imagine the opiate “dependent” neonate is nursed for years by an addicted mother and so remains opiate “dependent.” (I have a new age relative who nursed her daughter for six or seven years or something like that.) The child starts to develop control mechanisms but they frequently fail due to the need for opiates. Did the child just acquire a new disease or a new symptom of an old disease? If the latter, then one shouldn’t appeal to usage to declare “Problem solved.” You wouldn’t appeal to usage to claim Boorse is wrong about universal diseases or whether aging is a disease. Perhaps those of more peripheral versions of countering medical usage than my accepting lay language of addicted infants.

David H October 30, 2020 at 2:53 PM

Addicts with Control:

There are also addicts who maintain control but they have to be super vigilant and are barraged by thoughts and urges for the drug and constantly struggling. Imagine virtually everything is a distracting cue motivating them to take a drink or drug. They can’t concentrate on work or family, they can’t sleep well, and are stressed and distraught. But they never give in? They have superior will power or control. (Or is control so broadly construed that having unwanted thoughts is a lack of control?) Such constant drive for a drink and the disruption of one’s daily life because of the relentless impulse sounds like an addiction to me. Such drives or appetites or needs are surely a disorder and the disorder should be addiction not a different kind of disorder, addiction being reserved for control failures. I see no benefit to save a favored theory of addiction as disposition to lose control and create a new pathology for such disordered drives. The disordered excessive drives or needs explain why there will be a loss of control just as other diseases explain their symptoms


Robert Kelly November 5, 2020 at 10:25 AM


If they had full control over their desires, they could (indirectly) start modifying them if they wanted. Also, it is really unclear if you are describing addiction. I don’t think the ICD or DSM would make this a case of addiction (or substance use disorder). At least not necessarily. There is certainly lots of overlap between criteria, but it sounds like they have the obsessive component of OCD. There are non-addict disorders that have to do with intruding thoughts and desires (and no subsequent behavior)—PTSD can even be this way.

Also, in short, if you just stipulate that the person has full control over acting on their desires, I just fail to have the intuition that they are addicts (especially given the above).

David H October 30, 2020 at 2:53 PM

Addicts in recovery:

Do not some people who never relapse still not describe themselves as lifetime addicts? They don’t undergo a failure of control but they have a pathologically excessive need so they think they are always addicts even if they are sober addicts or addicts in recovery. I would want to restrict “recovered addict” to someone who doesn’t have the abnormal drives and dispositions to go on benders or breakdowns or blackouts or other excesses characterizing their days as drunk or addicts. If they no longer had those dispositions but would just be casual drinkers than I would say they were no longer addicts. But other people may be addicts for their lifetime even though they never give into the need.

David H October 30, 2020 at 2:54 PM

Late acquisition of addiction and pseudo-Cures of Addiction:

If fetuses and newborns can’t be addicts because they don’t have control mechanisms that can be overwhelmed, that might mean that they become addicts only when they start to develop control mechanism which can then fail. That is quite odd and a rather late onset of their addiction. Moreover, if it is losing control defined an addict then one cease to be addict by becoming mindless or minimally minded. But surely one doesn’t cease to be an addict by an injure that leaves one comatose or with an infant-like mind and no failures of control. Likewise, for animals who don’t have higher order impulse control. So, removing control mechanism and making someone like a baby shouldn’t remove addiction.


Robert Kelly November 5, 2020 at 10:29 AM


They would lose their addiction and (presumably) maintain physical dependence.

David H October 30, 2020 at 2:55 PM

Withdrawal and Permanently Unconscious Addicts:

Do advocates of addiction defined in terms of control want to say someone ceases to be addict because they go into a coma and never leave it as there is no control to exercise now or ever again.? Imagine their body is wracked with withdrawal, shaking or doing whatever is characteristic behavior. Is the absence of a conscious desire that can’t be controlled a reason to say the person is no longer an addict? I think understanding addiction as a pathologically strong need of the body would allow us to say the unconscious in withdrawal was an addict. It is a too a promiscuous use of disposition to lose control that applies to such individuals if they were to become conscious again, though not physically possible, they would not have control. Better to make disposition to control likely consequence or symptom of addiction than its definitional core. Loss of control is a symptom of the disease not its constitutive feature


Robert Kelly November 5, 2020 at 10:31 AM


Again, I think you have dependence in mind most of the time. This can do a lot of work for you in all of these cases. Everyone makes this distinction, so it would benefit your view to take it on.

Who knows what would have to be true in the coma case to know that the disposition is gone. But if there are sufficient changes such that it is now false that there is some disposition to fail to control their desires—and maybe false that they have desires, given the coma—then they are not an addict. Maybe they could be in a coma but their disposition to control their desires or not is merely dormant and unaffected by what is causing the coma. The case seems underdescribed. But if there just is no capacity for control at all, it seems like they are going through a purely physical withdrawal, and this does not require addiction.


David H October 30, 2020 at 2:55 PM

Lack of Control Thought experiment and Conceptual Reflective equilibrium:

Imagine someone with just impulses of normal control but without any typical human control. Imagine normal strength impulses but something happens to the individual’s control mechanism. So he acts on any impulse whatsoever, eating your food, drinking your water, taking your desert off your plate etc. But he has no more frequent desires or stronger desires for food and water and sugar than anyone else. Is he an addict? I think not. If you agree, that suggests addiction involves pathologically excessive needs. I suppose one can respond I have only shown such pathology to be necessary, not sufficient. It isn’t sufficient for control failure is required. See my other arguments against control. Or imagine a world where it is healthy to indulge in sugar, booze and drugs at levels of our addicts at the expense of other activities. Would that person be an addict. He had no more control than our addicts and gave up many other activities. But if it is a healthy craving, then it is like our strong desire for sleep, air, water and nutrients.


David H October 30, 2020 at 2:58 PM


Foddy and Savulescu on lay person’s view and drug treatments:

Although authors don’t endorse layperson’s view, they’re response is curious, writing that the laypersons’ views is “that addicts are morally corrupt hedonists…But if drug addicts are just wanton hedonists, then there is no solution except perhaps punishment. Thus, it should come as no surprise that the lay view is not discussed in the addiction literature.” Why not some form of moral education, moral exhortation, or moral habituation? Why just punishment?

Misinterpreting NA, AA?

They take these and other 12 step programs that have participants admit they are powerless over the drugs or alcohol to have a flawed conception of the involuntary addict. But maybe they are just to be construed as saying that one can’t beat the addiction on one’s own but needs help of others, not necessary the miraculous grace of God. It is probably an effective strategy even if not literally accurate. Does Rob or anyone else know the full story here?

Should Addicts Abstain or Engage in Moderate Use?

Maybe because I am not well read in the addiction treatment literature, but I was surprised by the pp 18-19 claim “…controlled drinking has been shown to be no less effective than abstinence as a treatment strategy for alcoholics but the liberal view entails that it is often a more desirable outcome” Does Rob or anyone else know whether this is true?

Too liberal a Liberal Account:

"We call this the liberal view of addiction because it permits people to nominate their own desires and values.” (p. 15) Does this mean that whether someone is an addict is up to that individual? Or is it that the accuracy of their self-report that are addicts just depends upon them sincerely report the strengths of their desires?

Aging out of Addiction?

The authors repeat the well-known claim that many addicts age out of their addiction. They construe this as problem for disease model as the brain damage or altered brain would seem not to predict this. They just explain it as changing desires with age, which, incidentally, is compatible with the disease model that understands addiction as pathologically strong needs. But what I was wondering is whether this aging out of addiction is also characteristic of pain killer addictions? Perhaps the greater numbers of middle-aged prescription drug addicts is due to their addiction starting later and they too will age out. Does Rob or anyone know the relevant data with pain killers? Do the patterns supporting the aging out of addiction thesis?


Jack Freer November 1, 2020 at 2:57 PM


Miscellaneous Response

I believe the loss of control is accepted as an essential feature of addiction (the other commonly cited is continued use despite harm). That said, there is still a common usage equating addiction with physical dependence (as when a newborn child of an addicted mother is described as "born addicted" -- the child is really born drug dependent).

The claim that controlled drinking is no less successful than abstinence probably reflects the overall dismal success rate (including abstinence) rather than highly successful results with controlled drinking. Anecdotally, I have heard more than a few instances of an abstinent addict (usually to alcohol or tobacco) who "has just one drink" or "one cigarette" and is off and running--returning to a pack a day smoker or a pint a day drinking within days. I don't see how such a person could maintain controlled use.

Re: 12 step and acknowledging powerlessness: In the 1980s, CFI was the home base for James Christopher and Secular Organizations for Sobriety ( Christopher would say that the difference between his system and AA was not just the acknowledgement of a supreme being, but also the locus of power. The AA reliance on a higher power, Christopher argues, relieves the individual of responsibility). Tim Madigan tells me Christopher died in CA a few months ago.



David H November 5, 2020 at 12:37 PM


Yes. I may be misusing language and calling "addiction" what is properly described as "dependency." I may be mistakenly thinking I am in a substantial debate over the nature of addiction and really a dispute over language – I am an outsider and probably too influenced by laypeople who talk about addicted babies and recovering addicts and life-long addicts and dry alcoholics who don’t give in to one’s addiction. I also assume folks were correct to speak of and animal addicts who I take lack mechanisms of control but just have competing drives that I perhaps construe on a hydraulic model in which stronger current wins. So I want a theory that accounts for that and couldn’t understand why professionals are defining addiction in ways that can’t capture such usage. You and Rob make distinctions the folk do not and thus distinguish between dependence and addiction, as do experts in the field.

I also don’t have any knowledge about the control literature. I suspect some losses of control are pathological and some are symptoms of pathology. If addiction really involves the latter, I don’t want to expand the number of pathologies and so want to say the pathology of the dependence causes the symptomatic loss of control. So this leads me to stick to my model even though you and Rob point out I am using “addiction” the way experts use “dependence.” I don’t think the loss of control is always its own pathology, unlike cases such as when heart disease or injury to the windpipe cause some downstream pathologies because other organs fail as a result of the earlier.

If control losses were pathologies in their own right, then there is much less reason for me to insist addiction is essentially the dependence pathology rather than the control pathology. I would just be speaking with the folk while you and Rob speak with the experts. But my thought experiments or faint memories/anecdotal cases of people with different levels of control but overwhelming needs for substances makes me want to say there are addicts without control pathologies. So if someone had normal control powers (perhaps this is a mistaken to speak in such quantities) but overwhelmingly strong desires that neither they or anyone else could resist, I would want to say they were addicts without control deficits. Likewise, I want to say someone with exceptional will power who resists cravings for drug which the rest of us couldn’t is an addict. And I don’t want to say someone who loses their capacities for control with say a stroke that reduces them to infant-like mind ceases to be addicts. I want ONE disease to do a lot of explanatory work and have control issues as symptoms.


Stephen Kershnar November 2, 2020 at 9:17 AM



I gather you give the following argument against control.

1. Animals, infants, and comatose can be addicted but not lose control (they don't have the capacity for control).

2. An addict can maintain control.

3. If 1 and 2, then a loss of control is neither necessary nor sufficient for addiction.


If I understand you correctly, and perhaps I do not, here is your theory of addiction.

(1) A person is addicted to something if and only if he has a pathologically strong need.

(2) A need is pathologically strong if and only if it is unhealthy.

(3) A need need not be a conscious desire (and, likely, not a desire at all).

Here are my concerns.

First, without desire, it would seem that plants could have addictions (assuming they have needs). Could a plant be addicted to heroin?

Second, imagine a person who will die if he doesn't get an insulin injection. He does not like to get it, does not go through withdrawal (imagine that he just feels very sleepy). It is an odd notion that he is addicted to insulin shots when he doesn't desire them, has control about whether to get them, and does not go through discomfort without it (that is, paradigmatic withdrawal).

Third, I do not see what the word "need" does in your analysis. It would seem that on your account an addiction is a pathology. You can't fill it out via behavior (see, for example, the comatose), control, or desire (see, again, comatose and possibly plants if we want to be promiscuous about needs). How does an addiction differ from a pathology?


Jack Freer November 3, 2020 at 10:15 PM

Steve, there are a number of medications that produce sickness if stopped abruptly (including insulin, certain blood pressure medicines, and corticosteroids). These are physical dependencies at best and nothing like addiction.


David H November 4, 2020 at 5:26 PM

I mentioned in my post that if I am committed to unconscious addicts then I could be committed to addicted plants as I had in mind a tree with pathological need for water. It is great if there is no such thing or anything else analogous to bodily needs in a mindless human being for a drug. If I am committed to addicted plants and this is so embarrassing and can’t be saved by limiting addiction to beings that were or could be conscious (which seems ad hoc) then I would drop it and say the unconscious guy in withdrawal was no longer an addict. I just wanted to capture the addiction in his body with a notion of appetite or need that wasn’t conscious and wasn’t intellectualized like desire (desires have contents, thus involve concepts, future directed as the kid would have a desire for the drug or for the pain to stop, neither of which I think a simple minded newborn has). I still can give the control people a hard time with newborns and adult addicts whose injuries made them newborn-like and thus no longer addicts on the control view. Also, the recovering addict who doesn’t lose control but is tormented by the need and can’t function well. Is this really just dependency but not addiction? When does an infant become an addict – only after it develops self-conscious control mechanisms that then fail?

I think addiction is a disease (for reasons similar to you to distinguish it from other strong appetites) and loss of control is a symptom. Otherwise we have two diseases – pathological drive and then loss of control. Rob and Jack tell me that addiction experts distinguish drug dependency (my need) from addiction (loss of control) so I may be misled by lay person’s talk of kids born addicted. I am worried that I am out of touch with expert opinion since I want to engage in reflective equilibrium. But if addiction is a disease, then pathological need seems to the disease which explains loss of control as a symptom. There might be more explanatory power to my outlier outsider approach. But again, I may just be misled by lay people talk of addicted animals and addicted newborns

Why is someone addicted to insulin shots? Is it pathological need to require insulin? That just seems like the need for water or a water substitute or a need for blood or a blood substitute. I don’t think those are diseases

I think there is a sense of need (maybe there is a better word) that captures the body’s drives that are conscious and not conscious which is different from other pathologies like a broken arm. Think of desires for bodily needs arising when one becomes conscious or self-conscious of those needs. There is something a need for hydration before and after consciousness has in common. Pathology is too general but merely qualified or restricts something else like need. One wouldn’t say the break was a drive. pathology is too large a notion so I need a type of pathology to capture what is there with self-conscious addicts, newborns needing but not desiring heroin, and bodies in withdrawal.


Pat D November 2, 2020 at 3:10 PM

Sorry to chime in late. I've learned a lot catching up with the discussion.

One question for Rob: Do you take different levels of (dys)function into account in your understanding of addiction - how they differ and how they interact?

For instance, as people have indicated, infants with neonatal withdrawal syndrome are not considered to be addicted. Although they manifest biological and behavioral disorders, we wouldn't consider their "affective and cognitive systems" (as DH puts it) to be disordered as such, even at this early stage of development of those systems. It seems that addiction pertains to dysfunction of these (? higher-level) systems. In addition, it's clear that social and political factors play a role not only in the treatment, but also in the etiology of addiction.


Robert Kelly November 5, 2020 at 11:27 AM


This is a good question. In one sense, I don't take dysfunction into account at all since I don't think we need to in order to say what addiction is. In other words, I think we can say what addiction is (systematic loss of control over certain desires), and then philosophers of medicine can tell us whether that is a dysfunction (in our world, given our history and laws of nature). Maybe it is. Wakefield and Hershenov seem to think so. I am not opposed to that turning out to be true since I am not trying to keep people's addictions from counting as diseases. I think it is quite plausible actually that some addictions are diseases/dysfunctions, just as some instances of grief are dysfunctions while others are not.

But more to your point, in another sense, yes, I am happy to consider the fact that there are different "levels" of dysfunction, as you call them. I guess I would want to make sure I understood what you mean, though. Using your example of infants that are dependent (a dysfunction) but not addicted, I think I understand. I tried to use just this kind of example in exchanges with David H. and so I think you and I are on the same page in that respect. But then you go on to say that addiction is a dysfunction of higher-level systems. Two things. First, again, I think an account of addiction can be given that does not appeal to dysfunction at all (even if what addiction is ends up, due to our history and laws of nature, being dysfunctional in our world, or in some instances in our world). Second, I'm not sure I would say it is necessarily about higher-level systems, if by this you mean something like more sophisticated thinking, reasoning, planning, etc. Of course, such higher-level cognitive systems can certainly end up being affected, as Volkow and others have shown. But I am not adverse to saying that other mammals like rats can be addicted. But maybe the systems involved there are still "higher-order" on your view, as compared to those of infants.

On the last point, yes of course I agree that politics and social factors play a role in treating and causing addictions. Hanna Pickard has written a lot about this. But just as clearly, these are about how addiction develops and how it can be affected in treatment once it comes into existence (or prevented before it does), and not about what addiction itself is.


David H November 7, 2020 at 8:42 AM


Imagine an addictive drug becomes "extinct" leaving the addicts with no chance to obtain it and so there will be no lapses of control. Has the addict ceased to be an addict and is just dependent? Is there some sort of disposition in normal environments that will save the notion of addiction as loss of control?


Robert Kelly November 8, 2020 at 11:59 PM


See my above replies about different types of control and desert islands. The fact that the drug has gone “extinct” (or, as above, that we drop an addict on a desert island) does not mean the person ceases to become an addict. I think you agree, and it seems that this is meant to be a counterexample to the control condition I require. However, nor would either of these cases entail that the individual immediately gains control. Control is understood dispositionally on my view of addiction—it is about a capacity which is systematic. So, because control is understood dispositionally and because there are other ways to fail to control one’s addictive desires that are independent of the drug existing (e.g. concerning choices to seek it, choices to go into recovery or not, etc.), your counterexample is unsuccessful. My dispositional account can handle this case rather easily.


David H November 7, 2020 at 8:52 AM


Could not someone with a normal amount of control be overwhelmed by an extremely powerful drug? It is a drug so powerful none of would resist it after one dosage. There need not be a weakening of our control faculties in the sense that comes from chronic use and brain damage or a loss of control for any other drug.

then there is the sense of loss of control in which one has less of a reserve of will power and discipline and delay gratification etc. than normal person.

the first loss of control is that of a normal person. It is like a crime of passion that any normal person might be susceptible to if they were put in a certain scenario. I think it is a mistake to say the addict in the first sense has a diminished capacity of control. His capacity is the same as everyone else's. But intuitively he is an addict. That suggests to me the key to addiction is the (pathologically) powerful need (appetite, desire, want) and not a loss of control. by hypothesis, the addict is no different from the rest of us. None of us can resist the drug if say we use it once or just get a whiff of it. It is beyond normal control to resist. Normal control means no loss of control in the sense of reserves, only loss in the second sense of giving in to it


Robert Kelly November 8, 2020 at 11:59 PM


I again think the dispositional account can save us here. I have not delved much into the notion of systematicity in these posts, and I will save this for the talk on Friday, but roughly my thought is that there is an equivocation or something similar happening regarding control. You have described the individual such that they are overwhelmed by the desire (for behavior type X, let’s say), yet that they have normal control. It seems like what you need to mean by the latter (in order to say they are overwhelmed by the desire for X) is that, with respect to *other* desires, deliberations, behaviors, and so on, they have normal control. This is perfectly consistent with my account. The person can systematically lack control over their desires for X (and behaviors relevantly related to X), and have totally normal control over their desires for Y. An alcoholic is not necessarily a gambling addict, or water addict, or golf addict, or sex addict or whatever addict. You start with the most general description of control (they have “normal capacity”) and then move to the conclusion that their control over the drug (or behavior X) must be normal, too, or indeed that they haven’t even lost control over X. Well, once we focus on desires for X, we no longer have the most general sense of capacity for control in mind. In other words, when we test whether the person has control by plopping them into possible scenarios, we are either interested in every possible scenario there is, or every possible scenario related to engaging in behavior type X. The latter is the relevant scope because that is the addiction we are interested in testing. Even if in the grand scheme of things, the person could still be appropriately described as having normal control (maybe their desires for X and the triggering conditions themselves are few and far between, and otherwise they have perfect control), they would still be an X-addict if they never controlled their desires regarding X. Hence, the compatibility here suggests that the use (or scope) of “control” has shifted through your argument.