Addiction is widely considered to be a disease ¹ . It has been consistently listed in every edition of the DSM, including the latest one (American Psychiatric Association, 2022), it is listed in the World Health Organization’s classification of diseases, the ICD-11 (World Health Organization, 2019), and it is commonly referred to as a disease by numerous other influential medical and research institutions around the globe. In psychiatric literature, support for the view is almost universal, and it is also widely accepted among treatment providers (Barnett et al., 2022). While popular conceptions of addiction have traditionally considered it to be a moral failing, the view of addiction as a disease has of late gained greater acceptance among the general public (Connelly et al., 2018; Pescosolido et al., 2010; Rise & Halkjelsvik, 2019). In addition, many people with addiction think of their condition as a disease (Snoek, 2017b, p. 189). Even non-medical institutions and support groups such as AA and NA refer to addiction as a disease (Flanagan, 2013).

Contrary to the prevailing opinion, some have remained agnostic or outright skeptical about the appropriateness of the disease label. Many have criticized the view of addiction as a disease of the brain (Field et al., 2019; Heather et al., 2022; Levy, 2013; Pickard, 2022a, 2022b), while others have questioned whether it is aptly seen as a disease at all (Heather, 2013, 2017; Heyman, 2009; Lewis, 2015, 2017; Ross, 2020; for an early defense of the view, see Fingarette, 1988). Skeptics have challenged the arguments advanced in defense for the disease view, pointing to disanalogies with paradigmatic medical conditions and claiming that addiction does not meet plausible criteria for inclusion. In addition, they have sometimes suggested that labeling addiction as a disease may not be in the best interest of those who suffer from it.

So, is addiction a disease? Any attempt to answer this question directly is complicated by its relationship to several other contentious issues, both on the side of theories of addiction and on the side of theories of disease. My primary aim in this paper is to lay out what the major points of contention are. My secondary aim is to offer a limited defense of the disease view. The crux of the debate is whether addiction is aptly pictured as the result of psychological dysfunction, and the main hurdle for the psychological dysfunction claim is that we currently lack a relatively unified account of the behavioral trait that most strongly suggests dysfunctional processes: the loss of expectable ability to control behavior. I argue that there is a sense in which the dysfunction claim may be warranted even in the absence of a unified causal explanation. This requires, however, assumptions that are to some extent controversial—most notably, the impaired control view of addiction and the acceptability of dysfunction-talk based on personal-level behavioral traits.

I begin by looking at the disease question itself. I suggest that it is not exhausted by either institutional or purely pragmatic considerations, and I introduce the hybrid account of disease as harmful dysfunction in terms of which the ensuing discussion is framed (Section 2). Subsequently, I turn to the question of whether addiction satisfies harm and dysfunction criteria. Addiction is obviously harmful, but one might ask whether it involves the right sort of harm to qualify as disease—namely, harm that is intrinsic to the condition. Although addiction-related harms are significantly exacerbated by modifiable socioenvironmental factors, I argue that it is highly unlikely that addiction is a condition whose only relationship to harm is mediated by extrinsic factors (Section 3). I then turn to what is arguably the central issue in the disease debate: whether addiction is aptly pictured as the result of dysfunction in the sense relevant for disease. I first discuss brain dysfunction. Critics of the disease view have objected on the grounds that we currently have no neural-level explanation of what is dysfunctional in the neural underpinnings of addiction. Although I accept this observation as true, I argue that on a plausible view of brain dysfunction, a claim of brain dysfunction need not presuppose that we have a way of identifying dysfunction that is independent of psychological or behavioral criteria (Section 4.1.). The final battleground, then, is whether we have reason to believe that addiction involves psychological dysfunction. If the impaired control view of addiction is correct, I argue, then the loss of expectable ability to control behavior strongly suggests psychological dysfunction. One might worry that dysfunction-talk cannot be warranted unless we have a precise etiological explanation for impaired control. I concede that we do not currently have one, and it is an open possibility that a precise etiological explanation will not be forthcoming—the causes of impaired control in addiction may be too complex, varied and multifactorial to satisfy the requirement that we identify a precise etiological mechanism. I suggest that impaired behavioral control may nevertheless be sufficient for a claim of dysfunction if one drops the requirement of a relatively unified causal account (Section 4.2.).

At risk of letting the reader down at the outset, I’ll begin by stating the inevitable: whether addiction is a disease depends in large part on what one takes a disease to be. This is a question that has long been debated by philosophers of medicine. The main contenders in the debate are naturalism, normativism, and several variations thereof. Naturalist accounts take disease to be definable in purely medical or scientific terms. In a popular articulation of the view, a key criterion of disease is dysfunction, which naturalists believe can be glossed over in purely medical or scientific language (e.g., Boorse, 2014). Normativists see disease as a fundamentally value-laden concept. On most normativist accounts, calling a condition a disease implies that the condition is undesirable, or that the person suffering from it is unfortunate for having the condition, with judgments about these matters involving an inescapable reference to evaluative or otherwise normative standards (e.g., Cooper, 2020).

One point on which these venerable foes agree is that the question ‘what is a disease?’ should be approached as one whose answer depends on philosophical or otherwise conceptual considerations. This implies that the question is distinct from, and in some sense prior to, the question of whether a given condition is treated by competent clinicians and recognized as a disease by relevant medical institutions. It also implies that the question of whether a condition is a disease is not exhausted by pragmatic considerations, such as whether it is beneficial to regard the condition as a disease. I will assume as much in what follows. Before moving on, it is worth defending this assumption briefly.

Consider a coarse institutionalism about disease, the view that there is nothing to being a disease other than being accorded the status of disease by the world of medical institutions and/or competent medical practitioners. If being a disease is identified with its being treated as such, this view seems to commit us to the unpalatable implication that conditions become diseases when the relevant institutional arrangements are made and cease to be diseases when they are no longer regarded as such, rather than these decisions tracking prior facts relevant to disease status.

Institutionalism allows, however, for more interesting variations that leave enough room for the possibility of questioning actual institutional arrangements and practices. One stems from the observation that when clinicians and medical institutions debate whether a condition should be included in nosological classifications, they tend to focus on whether it would be beneficial to adopt the relevant practices and institutional arrangements. One might think then that all there is to the question of whether a condition is a disease is whether it would be beneficial for medical institutions—or for society at large— to regard it as such.

Pragmatic considerations of this sort play a major role in medical thinking. Consider, for instance, the distinction between clinical and subclinical problems in regulating drug use. The distinction between a disordered and a normal-range difficulty in regulating drug use is largely driven by considerations of whether the person would benefit from consultation or treatment by a competent physician. Even though treatability is an important consideration, some reflection suggests that it cannot be all there is to being a disease. For some diseases, we do not currently know of any beneficial medical intervention. And people can sometimes benefit from consultation with health professionals even though they do not suffer from a recognizable pathology. Many people seek therapy to cope better with a range of ‘problems in living.’ They benefit from therapeutic intervention without having, or being seen as having, a psychiatric disorder (Bortolotti, 2020).

Similar considerations apply to variations of the argument that focus on other potential benefits of seeing a condition as a disease. For instance, claims about the impact of the disease label on social attitudes toward addiction have often been presented as central to the disease debate. For the disease view, it has been argued that framing addiction as a disease leads to less stigmatizing attitudes among the public, as people come to see addiction as the purview of doctors rather than law enforcement (Leshner, 1997; McLellan et al., 2000; Volkow & Koob, 2015) ² . Those who argue against the disease view have countered these claims. The alleged responsibility-undermining implications of the disease view have been cited as a reason against it (Holden, 2012). And whether adopting the disease label reduces stigma has also been disputed, with empirical studies yielding mixed results, at best (Corrigan & Watson, 2004; Kvaale et al., 2013; Meurk et al., 2014; Pescosolido et al., 2010). What is the impact of the disease label on social attitudes remains an open empirical question. Suppose, however, that the real effect of adopting the disease label is to increase the perception of ‘otherness’ of people with addiction in the eyes of the public. Conceptually, it would make perfect sense to continue to maintain that addiction is a disease, acknowledging at the same time some undesirable effects of the disease label. We would then be very concerned about combating the stigma of addiction (as we should be), but this concern does not seem to be a proper basis for arguing that addiction is not a disease.

It seems then we cannot ask whether addiction is a disease without taking a stance on what it is for a condition to be a disease in the first place. One might hope to do so, however, without taking a substantive stance. Consider the view of disease as a family resemblance concept (Lilienfeld & Marino, 1995). In this view, asking whether addiction is a disease amounts to asking about the extent to which it shares the features that conditions we think of as diseases typically have. Addiction arguably presents both analogies and disanalogies with paradigmatic medical conditions. For instance, the fact that pharmacological treatments are in use and that hereditary factors account for a portion of susceptibility to addiction seem to bring us closer to core components of the disease concept (Leshner, 1997; McLellan et al., 2000; Volkow & Koob, 2015). On the other hand, the fact that recovery from addiction, when successful, is a highly agential achievement seems to distinguish addiction from many paradigmatic diseases (J. Davies, 2018; Holden, 2012; Satel & Lilienfeld, 2017). Moreover, many people seem to benefit from 12-step programs, which approach addiction in a non-medical way—even though they often refer to it as a disease. In sum, it seems that addiction is in some ways similar to paradigmatic medical conditions, but it is also in some ways similar to life problems that we tend to approach in non-medical ways. If we could do no better than place addiction in this similarity space, we would probably find its place somewhere on the fringes of the domain of disease. How well addiction fits into the disease category would then be a matter of degree.

The above reasoning highlights an important point: if addiction is a disease, it is at best a somewhat unusual member of the category ³ . I resist, however, the family resemblance approach for much the same reasons given above: there seems to be content to the concept of disease that is not exhausted by placing particular conditions in varying degrees of approximation to paradigmatic medical conditions in resemblance space.

So, we are back at square one, facing a choice between substantive philosophical accounts of disease. In what follows, I will adopt a hybrid theory of disease (Wakefield, 1992). The hybrid account renders the core of the disease concept as the presence of a harmful dysfunction, thus integrating both naturalistic and normativist elements. My choice of a hybrid account is based on considerations of intrinsic plausibility, but it also has a strategic payoff. The two main criteria for disease on the hybrid account —harm and dysfunction— coincide with the main candidate criteria favored by other accounts (which nonetheless find reason to object whether one or the other is necessary). Thus, discussing whether addiction meets harm and dysfunction criteria will make the ensuing discussion still relevant to people committed to most other accounts of disease.

One more terminological note: I will not be making a principled distinction between disease and disorder. Although there is not universal agreement on the use of these terms, it is sometimes assumed that calling a condition a disease implies that there is a known explanation for its defining symptoms, whereas calling it a disorder implies no such commitment. I do not follow this usage here. One reason is that, I think, we are often content to call certain conditions diseases even when we lack a consensus explanation of their defining symptoms. Moreover, whether it is appropriate to call a condition a disease in the absence of a precise causal explanation of its symptoms is one of the key points under debate. It is therefore preferable not to settle the issue by terminological fiat. ⁴

In the standard formulation of the hybrid account, for a condition to be a disease it must be harmful to the individual who has it in the light of appropriate sociocultural standards (Wakefield, 1992, 2021). The harm criterion stands for the normative component in the theory of disease: part of what we say when we call a condition a disease is that it is bad for you. Consider situs inversus totalis, a rare condition in which the lateralized position of organs within the thorax is inverted (Eitler et al., 2022). Situs inversus is plausibly seen as a case of a physiological (developmental) process gone wrong. However, people who have it typically live their entire lives without experiencing any harm as a result of the condition, and therefore it is arguably inaccurate to say that they suffer from a disease.

Whether addiction meets the harm criterion might seem like a point of no debate. No one familiar with severe cases could argue that having an addiction is not harmful. Some see harm as a definitional feature of the condition, arguing that someone cannot properly be said to have an addiction if the way they use drugs does not cause them significant harm (e.g., Sinnott-Armstrong & Pickard, 2013). Indeed, harmful consequences are part of the definition of the Substance Use Disorder label in the DSM-5 and are reflected in the diagnostic criteria. Even if it were better thought of as not defining of addiction, it would be hard to deny that harm is a central element in typical addiction case-histories. In severe cases, people with an addiction often enter a destructive spiral that leads to death or other dramatic consequences, including loss of health, loss of housing and livelihood, damage to important relationships, among others. Even theorists who reject the view of addiction as a disease on other grounds agree that it is “unquestionably destructive” (Lewis, 2015, p. x).

A more difficult question arises when we ask whether addiction involves the right sort of harm to be a disease. Intuitively, it must result from the condition in a way that does not depend on contingent features of the social, environmental, or cultural setting. Consider the neurodiversity view of autism. Proponents typically acknowledge that people with autism experience difficulties in navigating a social world constructed to meet the needs of neurotypical individuals. But—the argument goes—it is not the condition that is intrinsically harmful, but rather the social attitudes and environmental arrangements that fail to adequately accommodate the needs of neurodiverse individuals (Dominus, 2019). These contrasting ways of thinking about the relevant harms have importantly different implications. The former suggests that it is the person who should be treated to better adapt to environmental conditions, whereas the latter suggests that it is the environment, not the person, that should be modified.

A further complication arises when we consider what goes into an environment being modifiable. Upon closer examination, the question points to more than the mere feasibility of envisioned changes in environmental conditions. Any such changes are bound to involve costs of various kinds—economic, social, moral. Thus, when we ask, ‘Is the harm experienced by people with this condition an intrinsic feature of the condition?’ the question we pose is partly normative in nature. Consider homosexuality, another locus classicus in this debate. In the context of a homophobic society, being gay can certainly lead to harm. But it is the homophobic attitudes that need to be amended, not gay people. In saying this, one is not simply assuming that changes in homophobic attitudes are possible. Indeed, it is clear that the struggle against homophobic mindsets and attitudes remains a challenging task. To say that it is the homophobic environment rather than the person that needs to be changed is to take a moral stance. One way to articulate this stance is to claim that the cost of getting people to leave homophobic attitudes behind is morally justified.

Neil Levy (2013) has suggested that something resembling the above argument may apply to addiction. What gets the argument off the ground is that there is considerable evidence that environmental circumstances have a significant impact on addiction. For instance, the proximity of opportunities to use are known to modulate the intensity of craving (Dar et al., 2010; Juliano & Brandon, 1998) and craving episodes are highly context-sensitive in further ways, as they are easily triggered by a variety of environmental cues that the person associates with drug use (Skinner & Aubin, 2010). Moreover, social and economic variables also play a role. Substance use problems disproportionately affect those suffering from housing and employment instability (Saloner & Cook, 2013), and recovery is much more likely for those who can benefit from social support networks and who have credible alternatives to a drug-focused lifestyle (Hart, 2013). In addition, many believe that stigmatizing attitudes toward people with addiction play a role in explaining why it is so difficult to quit (Hadland et al., 2018; Yang et al., 2017).

Addiction problems would look very different in a more just society, one in which opportunities were available to all and in which people with addiction were not rejected and stigmatized. Levy’s argument does an important service by undermining the comfortable assumption that society as a whole has little to do with harm in addiction—there is indeed much that could be done to alleviate these harms. Nevertheless, it is a stretch to conclude that addiction-related harms would disappear under ideal social conditions. It is difficult to imagine a scenario in which addiction might turn out to be like homosexuality as a trait whose only relationship to harm is mediated by modifiable factors.

First, some of the harms associated with addiction do not seem to depend on conditions external to the agent in any robust sense. Many people experience various types of health problems as a result of a history of drug abuse. There is evidence of neurotoxic effects associated with sustained long-term use of some drugs, particularly alcohol and stimulants (Ersche et al., 2013; Xiao et al., 2015), and long-term drug use can have adverse effects on the organism beyond the brain. The adverse health effects associated with long-term alcohol and nicotine use are well known.

Other types of harm result from social and environmental factors that do not appear to be truly modifiable. It is probably true that drug availability can make a difference in people’s ability to control their use, but it is hard to imagine a scenario in which drugs could be made truly unavailable (as opposed to outlawed). It is also hard to imagine changing the environment so that it is completely free of cues that trigger craving, since craving can be triggered by all sorts of things that the person has come to associate with drug use, including the time of day (Palij et al., 1996). For other types of cues, the magnitude of the envisioned environmental change may be so great as to be almost unfeasible.

To recapitulate, on a plausible construal of the harm criterion for disease, whether addiction is a disease depends in part on the extent to which we think of addiction-related harms as the result of modifiable social and environmental conditions. The more we are inclined to think that there are accessible possible worlds—as Levy would put it—in which relevant changes in environmental conditions would lead to the elimination of these harms, the less addiction will seem to involve the kind of harm relevant to disease. But it is difficult to imagine a scenario in which suitable changes to environmental conditions would lead to the elimination, rather than reduction, of addiction harms. Some of these do not seem to depend on environmental conditions in a robust sense, and some of the environmental conditions that do play a role appear not to be truly modifiable. Some of the harms experienced by people with addiction are best thought of as intrinsic to the condition, even though stigma and social distancing undoubtedly add unnecessary pain to the harms experienced by people with addiction.

This is not the same as saying that addiction is a disease, since not all inherently harmful conditions are aptly called diseases. It is time now to move on to the other main component of the hybrid account: the dysfunction criterion.

The question of whether addictive behavior is the result of dysfunctional processes naturally leads to the question of how we determine what the relevant norms of proper functioning are. This is, alas, another notoriously contentious issue. Naturalistic views of function and dysfunction have tended to cluster around two main families of views. Selectionist or etiological views identify the function of a trait or mechanism with the production of the effect for which it was selected over evolutionary history (Millikan, 1984; Neander, 1991; Wakefield, 1992). Systemic or causal views see function as depending not on historical facts but on the actual causal role of the trait or mechanism in the context of wider organismic operation (Boorse, 1975; Cummins, 1975). As noted above, Wakefield himself favors a selectionist account, and thus the standard formulation of the hybrid account of disease incorporates a selectionist understanding of dysfunction. In principle, however, one could side with Wakefield in thinking that disease requires both harm and dysfunction, and still endorse a different account of the latter. Another way in which I will depart from the standard hybrid approach is that I will not assume that an acceptable account of function must necessarily be value-free. ⁵

(My own view is that the causal account comes closer to capturing the sort of contrast between well-functioning and dysfunctional traits that is at play in judgments about health and disease. A case discussed by Wakefield (2020) serves as an illustration. Imagine a gosling that, upon emerging from its shell, first encounters a fox and subsequently develops the stereotypical mother-oriented behaviors of geese toward the fox. In a sense, the imprinting mechanism is failing to do what it was evolutionarily meant to do, e.g., it has led the gosling to pair with the wrong creature in a way that will negatively affect its fitness. On the other hand, there is a sense in which the mechanism is performing as expected: it has successfully registered the visual impression of the first creature it encountered and has subsequently been successful in tracking and following around that creature. My intuitive reading of the case is that, while there is a clear sense in which something has gone awry with the gosling’s development, the sense in which her imprinting mechanism has failed her is not of the sort we think of as disease or disorder ⁶ . Regretfully, I cannot properly defend this claim here.)

The discussion of brain dysfunction and psychological dysfunction raises different issues, so I will discuss them in turn. To foreshadow, my view is that whether brain dysfunction is involved in addiction depends, among other things, on whether we have reason to believe there is dysfunction at the psychological level. With respect to psychological dysfunction, the main difficulty is not how we define norms of proper functioning, but the prior question of what mechanism is responsible for the dysfunctional outcomes. I argue that a plausible case for psychological dysfunction can be made given two critical assumptions: that addiction involves a loss of expectable ability to control behavior, and that dysfunction-talk is acceptable in the absence of a relatively unified etiological explanation.

I have articulated and, in a limited sense, defended a version of the view of addiction as a disease. The main limitation of my argument is that it hangs on several assumptions concerning contentious issues in theories of disease, theories of biological or mental function, and theories of addiction. Nevertheless, I hope to have made some progress in identifying some of the major points of contention and the assumptions that go into articulating a plausible way of defending the disease view.

I began by assuming a hybrid theory of disease and then looked into the grounds for claiming that addiction meets its two criteria: harm and dysfunction. On a plausible interpretation of the harm criterion, this requires that the condition reliably leads to harm in a way that does not depend on modifiable features of social and environmental context. While the harmfulness of addiction is beyond doubt, one might wonder about its dependence on socioenvironmental factors. I argued that some of the harms associated with addiction do not seem to depend much on environmental factors, and that some of the environmental factors that do seem to play a critical role in addiction—such as drug availability and the presence of craving-triggering cues—do not seem to be modifiable in the sense at issue. This, I suggested, is sufficient to claim that addiction meets the harm criterion for disease.

I then turned to dysfunction. I sided with critics of the BDMA in their observation that we currently lack a neural-level explanation of what is dysfunctional about the types of brain processes that correlate with addictive behavior. On a plausible view of brain dysfunction, however, claims of brain dysfunction need not require that dysfunction be identifiable by neural-level means. This transfers the question to the psychological level. I focused on what is arguably the most important candidate for psychological dysfunction: the impairment of behavioral control. The loss of expectable ability to control behavior certainly suggests that something is wrong with the mechanisms that lead to behavior, but it may not be enough if one requires that dysfunction-talk is backed by a precise etiological explanation. Indeed, the main hurdle for the psychological dysfunction claim is that we do not currently have a sufficiently unified explanation for impaired control. One might continue to bet that a relatively unified explanation will be eventually be uncovered, and thus defend the dysfunction claim on the basis of that promise. I have suggested that there is a way in which we might endorse the dysfunction claim even if a relatively unified explanation of impaired control were not forthcoming. This requires accepting dysfunction-talk in the absence of a precise etiological story. For those willing to make such a move, the loss of expectable ability to control behavior that we see in addiction may be sufficient for psychological dysfunction. This way of articulating the dysfunction claim avoids some important criticisms raised by those who resist the disease view—reductionism and excessive internalism—, thus faring better in this regard as well.