Psychopathology and the Ability to Do Otherwise


  • The copyright line for this article was changed on 14 May 2015 after original online publication.


When philosophers want an example of a person who lacks the ability to do otherwise, they turn to psychopathology. Addicts, agoraphobics, kleptomaniacs, neurotics, obsessives, and even psychopathic serial murderers, are all purportedly subject to irresistible desires that compel the person to act: no alternative possibility is supposed to exist. I argue that this conception of psychopathology is false and offer an empirically and clinically informed understanding of disorders of agency which preserves the ability to do otherwise. First, I appeal to standard clinical treatment for disorders of agency and argue that it undermines this conception of psychopathology. Second, I offer a detailed discussion of addiction, where our knowledge of the neurobiological mechanisms underpinning the disorder is relatively advanced. I argue that neurobiology notwithstanding, addiction is not a form of compulsion and I explain how addiction can impair behavioural control without extinguishing it. Third, I step back from addiction, and briefly sketch what the philosophical landscape more generally looks like without psychopathological compulsion: we lose our standard purported real-world example of psychologically determined action. I conclude by reflecting on the centrality of choice and free will to our concept of action, and their potency within clinical treatment for disorders of agency.


When philosophers want an example of a person who lacks the ability to do otherwise, they turn to psychopathology. Addicts, agoraphobics, kleptomaniacs, neurotics, obsessives—indeed, on occasion, even psychopathic serial murderers—are all purportedly subject to irresistible desires: desires so strong that the person is compelled to perform the action in question. No possible alternative is supposed to exist. Harry Frankfurt, for instance, suggests that certain circumstances, such as hypnosis, coercion, and ‘inner compulsion’ make it impossible for a person to avoid doing something (2003 [1969], 167). Addiction is such a compulsion according to Frankfurt: it is a ‘physiological condition’ that means a person ‘inevitably succumbs’ to the desire to use, which is ‘too powerful […] to withstand’ and results in that person potentially being ‘helplessly violated by [their] own desires’ (2003 [1971], 328). As Carl Elliott describes it, the addict ‘must go where her addiction leads her, because the addiction holds the leash’ (2002, 48, quoted in Levy 2011a). Al Mele and Neil Levy concur that agoraphobics ordinarily ‘cannot resist’ their desire to remain in the house ‘no matter how hard [they try]’ (Mele 1990, 456–7 and Levy 2011a). Keith Lehrer imagines a neurotic who has a pathological aversion to candy and so is ‘utterly unable’ to touch it because they ‘could not possibly bring themselves to choose’ to (1968). Justin Capes invokes a man whose OCD, in the circumstances, renders him ‘incapable of acquiring motivation to refrain from washing [his hands] or to do anything incompatible with his washing them’ (2012, 10). Michael Fara invokes a neurotic who ‘could not try to lift a spider’ because of the strength of their phobia (2008, 851), an addict who ‘because [they are] an addict […] lacks [the] ability [to resist taking the drug] (ibid., 858), and, finally, a psychopathic serial murderer ‘addicted to […] horrific activities in just the same way, and with just the same force, as the drug addict is addicted to taking the drug. [They are] driven inexorably towards killing [… which is …] an action [they] are powerless to prevent’ (ibid., 860). In all such cases, it is supposed that either it is impossible for a person to try or choose or be motivated to do otherwise, or, even if they can try or choose or be motivated to do otherwise, it is impossible for them to succeed. One way or another, psychopathology is thought to strip people of the ability to do otherwise: no possible alternative course exists. In this respect, this view understands psychopathology as stripping people of free will, understood as a distinctive ability of rational agents to ‘choose a course of action from among various alternatives’ (O'Connor 2011).

Philosophy is not unique in its conception of psychopathology as rendering a person powerless in the face of its demands. No doubt, popular culture takes a similar view of many psychiatric disorders. The primary aim of this paper is to argue that this conception of psychopathology is false, and, in so doing, to offer an empirically and clinically informed understanding of disorders of agency, as I shall call these conditions, where core symptoms or maintaining factors of the disorder include actions and omissions. Addicts, agoraphobics, kleptomaniacs, neurotics, obsessives, psychopathic serial murderers, and, further, patients diagnosed with disorders whose symptoms include impulsive behaviour, such as personality disorders, eating disorders, and paraphilias, have the ability to do otherwise: it is possible for them to refrain from performing the actions constitutive of the disorder. If this is right, then psychopathology does not strip people of free will.1

There is, of course, no question that people suffering from disorders of agency have severe behavioural problems, and, in many cases, impaired control relative to the norm. That is (part of) why they are disordered. It may be very difficult for them to resist performing the action in question, and the cost of doing so may be very high. For these (and perhaps other) reasons, even if they could have done otherwise, they may yet be partly excused for their behaviour. Powerlessness is not the only excuse or mitigating circumstance (see §2 and §3 below). Equally, people suffering from disorders of agency may be rightly deserving of care and compassion: even if their power to do otherwise renders it appropriate for us to hold them responsible for their behaviour, it may not yet be appropriate to blame them, for these concepts are distinct (Pickard 2011a, forthcoming; Lacey and Pickard 2012). But, crucially for the purposes of this paper, impairment of control relative to the norm is not extinction. Psychopathology does not render a person powerless: subject to irresistible desires that leave no possible alternative choice or action. The secondary aim of this paper is thus to sketch what the philosophical landscape looks like once this misconception is corrected.

The paper is structured as follows. In the first section, I return again to Fara's examples above, to try to bring to the fore how pre-theoretically jarring they are, and to suggest a diagnosis of what motivates this conception of psychopathology. I begin the task of correcting this misconception by appeal to standard clinical treatment for disorders of agency. In the second section, I focus on one particular kind of psychopathology, namely, addiction, which I discuss at some length.2 I argue that we have no good reason to think that addiction is a form of compulsion. And I briefly suggest five factors that can together explain how addiction impairs control without extinguishing it. There are, beyond doubt, important empirical differences between disorders, which we must bear in mind. Where relevant, I shall point out some of these differences. But it will help us to have a single, clear example before us, and addiction provides a particularly good one, both because it is arguably the prototypical example of irresistible desire in the philosophical literature, and because we have better neurobiological knowledge of the mechanisms underpinning addiction than we do, say, of kleptomania. It is not absurd to think that, if even addicts can do otherwise, then, surely, so too can kleptomaniacs. In the third section, I step back from addiction, and briefly sketch what the philosophical landscape more generally looks like without psychopathological compulsion. To anticipate, I suggest we lose our standard purported real-world example of psychologically determined action: we have no examples of real actions over which people are powerless. I conclude by reflecting on the centrality of choice and free will to our concept of action, and their potency within clinical treatment for disorders of agency.

§1. Psychopathology and Philosophy

Fara's suggestion that a psychopathic serial murderer is powerless to prevent themselves from killing, just like an addict is powerless to prevent themselves from using drugs, is, to put it bluntly, shocking. We should pause, and ask ourselves some flat-footed questions. How is it possible to conceive of murder in this way? How could a person not be able to stop themselves intentionally killing someone else? How could murder be an action a person is powerless to prevent?

These questions gain urgency when we reflect on aspects of our ordinary concept of action. As both Maria Alvarez (2009) and Helen Steward (2009, 2012) have recently emphasised, this concept arguably applies only to behaviour over which we can exercise a degree of control. Pre-theoretically, we commonly hold that what makes a piece of behaviour an action, as opposed to a mere bodily movement, like an automatic reflex, is that it is voluntary. This means that there is the capacity for genuine choice between possible courses of action. Minimally, there must be at least two choices: to act in a particular way at a particular time, or not to, that is, to refrain from performing that particular action.3 Perhaps the agent has no idea what else to do. Perhaps the agent cannot do anything different. But, if the behaviour is an action, they can at least not perform that very action.4 Of course, we may sometimes struggle to tell whether a piece of behaviour is voluntary or involuntary and so whether it counts as action or automatic reflex. Nonetheless, the conceptual point stands. If we are genuinely powerless over our behaviour, with no capacity whatsoever for control, it cannot just be taken for granted that the behaviour counts as intentional action: at the very least, the idea needs elucidation or argument. And murder is, by definition, intentional action. It is intentional killing.

Fara's comparison between the addict and the serial murderer is, nonetheless, apt. Despite the important moral differences between the behaviours and the corresponding differences in our responses, there are yet similarities. Both addictive drug-taking and serial murder are associated with kinds of psychiatric disorder. Both involve complicated, diachronic planning and execution. Both typically involve ambivalence and regret, if not, at least in the psychopathic case, consistent and genuine remorse. Both typically lead to negative consequences for the agent, which we might normally expect to act as a deterrent, but which in these cases apparently do not.

So, if an addict can be powerless over the kinds of actions that comprise drug-taking, then it is indeed difficult to see why a psychopathic serial murderer could not be equally powerless over the kinds of actions that comprise murder. We are familiar with this view of addiction, less comfortable with this view of murder. But, rather than extend a misconception of addiction to psychopathic serial murder, we should use the comparison between them to query our conception of addiction. For, we might ask the same sorts of flat-footed questions of addiction. How is it possible to hold that a person could be powerless to prevent the complicated, diachronic, planned and executed instrumental actions that comprise drug-seeking and drug-taking behaviour? Is it really credible that at no point is there the possibility of an alternative course of action for an addict, that these desires are irresistible?

I want to suggest one diagnosis for what has gone wrong in philosophical and indeed popular conceptions of psychopathology, which, for moral reasons, never mind philosophical clarity, it is important to state explicitly. The diagnosis is that, in thinking about psychopathology, people are often rightly motivated by moral concerns for those who suffer from it. Most psychiatric disorders are associated with impoverished backgrounds and early psychosocial adversity, sometimes to a truly harrowing extent.5 And predisposing environmental factors aside, people who have psychiatric disorders typically suffer tremendously, experiencing extreme degrees of distress and dysfunction. That is the nature of psychiatric disorder. When psychiatric patients do wrong or perpetrate harm, whether to themselves or others, we may therefore rightly wish to temper our natural tendency to judge and to blame, and instead try to maintain an attitude of sympathy and compassion, in light of these considerations. It is unusual to find people ready to extend this sort of attitude to murderers, but it is perhaps natural to find people who feel it, or who believe they ought to feel it, when confronted with addicts, agoraphobics, neurotics, obsessives, and possibly even kleptomaniacs, where wrong or harm to others is less obvious and less severe. And, clearly, one way to try to maintain this attitude is to deny patient agency: to claim that addicts and other psychiatric patients are compelled to behave as they do and literally cannot do otherwise. For, if this is right, then they cannot help it, as we naturally say, and cannot be held responsible never mind to blame for their actions and any wrongdoing or consequent harm. But, again, there are reasons apart from compulsion for holding that responsibility is diminished or blame inappropriate: powerlessness is not the only possible excuse or mitigating condition, let alone ground for an attitude of compassion and concern (again, see §2 and §3 below). Indeed, clinical practice with patients with disorders of agency typically distinguishes sharply between the appropriateness of holding a patient responsible for behaviour, and blaming them (Pickard 2011a, forthcoming; Lacey and Pickard 2012). Responsibility is central to effective treatment: agency cannot be denied. But blame is detrimental: sympathy and compassion must nonetheless be maintained.

For many disorders of agency, standard psychological and group treatment programmes employ a three-pronged approach, alongside the prescription of medication if appropriate. To take a concrete example, consider two behavioural patterns diagnostic of forms of personality disorder: deliberate self-harm and violence towards others. Such behavioural patterns have typically developed as a habitual, unhealthy way of acting on and managing anger. To effectively treat such a behavioural pattern, first and foremost, the patient is encouraged or, as a condition of treatment, straightforwardly required, to agree to reduce or simply outright stop self-harming or violent behaviour.6 To use the language of addiction, they need to go cold turkey. If they breach this agreement, negative consequences, to which they have typically also antecedently agreed, may be imposed. Second, the patient is helped to identify triggers of anger so they can stop it from escalating in the first place, and to develop alternative, healthy coping strategies to express and manage it instead. Third, the patient is throughout offered empathy and understanding despite the nature of the behaviour, and a non-judgmental environment within which to understand how this behavioural pattern has developed, to reflect on any acts of self-harm or violence the patient subsequently commits,7 and to gain a narrative sense of the role and effect of self-harm or violence in their lives (for further discussion of the nature of such therapy, see Pearce and Pickard 2010, 2012; Pickard 2009, 2011a, forthcoming; Pickard and Pearce forthcoming).

All three prongs are important to effective treatment. But the point to emphasise here is that such treatment programmes begin by asking patients to agree to stop problematic behaviours: to refrain from self-harming or acting violently towards others.8 The clinical presumption inherent in such treatment programmes is that patients have the ability to do otherwise at the start of treatment and can therefore appropriately be asked to refrain from the problematic behaviour. Hence from a clinical perspective, the sort of denial of agency typical in philosophical and popular conceptions of psychopathology is inconsistent with the presumptions underlying effective treatment. If a patient really cannot do otherwise, it is not acceptable for a clinician to encourage let alone demand that they change their behaviour, never mind the fact that it would be miraculous for the patient actually to manage to do so: the first prong of treatment should literally be impossible.

Indeed, if we pause and reflect on the nature of disorders of agency, the need to directly and immediately target patient agency in effective treatment should be unsurprising. If the core symptoms or maintaining factors for a disorder consist in actions and omissions, then improvement or recovery depends on doing things differently. Patients who abuse drugs and alcohol, self-harm, or act violently towards others, need to stop. Patients who don't eat need to start eating. Given the diagnostic criteria by which these disorders are defined, this change in behaviour is necessary for improvement let alone recovery. No doubt, sometimes behavioural change can be effected subliminally, randomly, or through purely pharmacological means. For instance, if a patient is kept heavily sedated, they are unlikely to be violent, because they are unlikely to do much of anything. But the normal way to clinically effect behavioural change for the benefit of the patient, and so promote genuine improvement or recovery from disorders of agency, is through mobilising personal decision, will, and resolve. Patients must decide to change how they behave and work to see that decision through, despite inclinations to revert to old patterns and any actual relapses. They need to ‘take responsibility’ for their behaviour as it is often put in the clinic. Psychiatry is not a recherché intervention in this regard. Many of its techniques are, quite simply, common sense methods of bolstering agency, delivered together with a good dose of care, sympathy, and compassion.

§2. Addiction9

It is not only philosophers who treat addicts as lacking the ability to do other than consume. Addiction is widely viewed as a chronic, relapsing neurobiological disease (a ‘physiological condition’ to use Frankfurt's term above) characterised by compulsive use of drugs or alcohol (National Institute on Drug Abuse 2009, World Health Organization 2004). In a common metaphor, the addict's brain is ‘hijacked’ by the drug, destroying the capacity for voluntary choice or control of substance use (Charland 2002; Hyman 2005; Leshner 1997). Addiction is considered to be a form of compulsion. Although there is no clearly agreed definition of what this term means, it is standardly understood to consist in an urge, impulse or desire that is irresistible: so strong that there is no possible alternative choice or action, as in the examples above.

As suggested (§1), from a pre-theoretical perspective, we should immediately be cautious of this claim on conceptual grounds. Drug-seeking and drug-taking behaviour appears to be deliberate, flexible, and involve complicated diachronic planning and execution. It bears all the hallmarks of intentional action. But, again pre-theoretically, this seems to suggest that alternatives must be available: minimally, it must be possible for addicts to refrain. Hence the claim that addiction is a form of compulsion needs both argument and elucidation. We need reasons to believe that it is true. And, if we are convinced it is true, we need to revise our pre-theoretical understanding of action accordingly, elucidating a concept of action that does not demand choice between alternatives.

There are three reasons commonly offered in support of the claim that addiction is a form of compulsion: (i) the nature of withdrawal; (ii) the neurobiological effects of drugs; (iii) the testimony of addicts themselves. In this section, I argue that they do not succeed in supporting this claim, and offer an alternative explanation of the impairment in control that characterizes addiction, before returning in the following section to the question of what the philosophical landscape more generally looks like without psychopathological compulsion.

The Nature of Withdrawal

Cultural portrayals of withdrawal, especially from alcohol and heroin, are dramatic: shakes, fever, retching, delirium. Heroin withdrawal is sometimes described as agony or torture: something no one should be asked to undergo, and which could break the will of the strongest of us (cf. Arpaly 2006, 20). Equally, withdrawal from severe alcohol dependence, without medical monitoring, can be dangerous: people sometimes die. But withdrawal, even from alcohol and heroin, is rarely so extreme. Heroin withdrawal is typically similar to a bad cold or, at worse, flu. Indeed, some addicts choose to abstain and suffer withdrawal simply in order to lower their tolerance (Ainslie 2000). Furthermore, the physical symptoms of all withdrawal can also now be minimized pharmacologically through the use of various drugs: benzodiazepines for alcohol withdrawal; and either a long-acting opioid, such as buprenorphine, or symptomatic treatment, such as anti-nausea drugs, for opioid withdrawal. If an addict wants to stop using, medication is available to make withdrawal a safe and physically manageable option.

It is thus simply not true that addicts are compelled to use drugs because of the physical need to avoid withdrawal. Of course, during withdrawal and indeed afterwards, addicts may yet have a strong, unfulfilled desire for the drug, which may be psychologically difficult to endure. But that is a different point: the physical effects of withdrawal are usually moderate and, in more severe cases, can nonetheless be medically well-managed to minimise physical pain and risk, even if the psychological effects of abstaining remain.


The conception of addictive desires as compulsive is often linked to neurobiological research on the long-term effects of immoderate drug and alcohol consumption on the brain. For example, Louis Charland suggests that ‘the compulsive drug-taking that defines addiction is a direct physiological consequence of dramatic neuroadaptations produced in the reward pathways of the brain’ (Charland 2002, 40–1; cf. Leshner 1997 and Hyman 2005). These act to ‘nullify any semblance of voluntary choice’ (Charland 2002, 41). So according to Charland, the addict is powerless over the causal force exerted by drugs on their brain: there is no possibility of doing otherwise.

It is important to be clear that immoderate long-term drug use can certainly affect neural mechanisms. Many drugs directly increase levels of synaptic dopamine, which may affect normal processes of associationist learning related to survival and the pursuit of rewards (for a review see Hyman 2005, but note that certain foods, like sugar, have a similar effect, cf. Foddy and Savulescu 2006, 2010). Once drug-related pathways are thus established, cues associated with drug use cause addicts to be motivated to pursue the reward of drugs to an unusually strong extent. Moreover, there is increasing evidence that as drug use escalates, control devolves from the prefrontal cortex to the striatum, in line with a shift from action-outcome to stimulus-response learning (for a review see Everitt and Robbins 2005). In rats, drug use that is initially goal-directed and sensitive to devaluation of outcome becomes increasingly habitual: triggered automatically and insensitive to (mild) devaluation.

However, this neurobiological evidence does not establish that addictive desires are compulsive. First, although neurobiology may explain how cues associated with any substance that directly increases levels of synaptic dopamine strongly motivate behaviour, it remains unclear why these mechanisms would be sufficient to render desires for drugs different in kind, and not simply in strength, from more ordinary appetitive or reward-driven desires which we do not regard as irresistible. Neurobiology may explain why addictive desires are very strong and thus hard to resist, but it does not thereby explain why they should be impossible to resist. Second, although increasing striatal control and insensitivity to (mild) devaluation of outcome does show that the behaviour has become more automatic and habitual, it does not show that control is fully lost. Automatic, learned habits can not only be deliberately altered over time, but can be resisted in the moment when attention is focused and motivation exists. Moreover, human motivation is typically complicated and sensitive to more than devaluation of immediate outcome: the lessons from experiments with rats do not clearly apply. Third, it is usually open to addicts, unlike experimental rats, to avoid drug-associated cues and stimuli. This is a standard intervention in all effective treatments for disorders of agency, and it is well-known by addicts: identify triggers and avoid them (Petersen and McBride 2002). For example, alcoholics know that if they genuinely want to abstain, it is much better not to go to the pub in the first place: don't court temptation.

Fourth, and in ways most importantly, this view of addiction is challenged by large-scale national survey data (for a comprehensive review of these findings, see Heyman 2009; cf. Foddy and Savulescu 2006 and Peele 1985). Data from The Epidemiologic Catchment Area Study 1980–1984, The National Co-morbidity Survey 1990–1992 and 2001–2002, and the National Institute on Alcohol Abuse and Alcoholism demonstrate that addiction, as defined by the DSM-IV criteria for substance dependence (APA 2000) peaks in adolescence and early adulthood, but, in the majority of cases, has resolved permanently, without clinical intervention, by the late twenties or early thirties (Anthony and Helzer 1991; Compton et al. 2007; Kessler et al. 2005a, 2005b; Stinson et al. 2005; Warner et al. 1995). Addicts tend to ‘mature out’ as the responsibilities and opportunities that characterise adult life increase. Moreover, research equally suggests that the majority of addicts will abstain from using over prolonged periods of time when offered immediate but modest monetary incentives (Higgins et al. 1991, 1994, 1995). This finding has led to the development of various forms of Contingency Management Treatment for addiction. Such treatment is very simple: vouchers, money, or small prizes are given to patients who produce clean urine samples. Typically, patients submit urine thrice weekly, with increasing value for each clean sample. The samples are tested and the reward offered immediately. Contingency Management Treatment reduces risk of disengagement from treatment and increases periods of abstinence compared to other standard treatments (for a review, see Petry et al. 2011). If addictive desires are irresistible, and drug-taking and drug-seeking behaviour is a direct consequence of a neurobiological disease, then spontaneous recovery and motivated abstinence should be surprising and rare. Yet both are not only possible but common. The natural explanation is that such addicts choose to abstain when they are sufficiently motivated to do so: they are not compelled to use.

Philosophers often suggest that spontaneous recovery and motivated abstinence fail to establish that addicts are not compelled to use. The reason offered is that the capacity for control must be relativized to a motivational and epistemic context (cf. Mele 1990; see too Capes 2012). Otherwise, as Neil Levy puts it, ‘we get the absurdity that, say, agoraphobics are not compelled to remain indoors, since, given the appropriate incentives [e.g. the house is on fire], they would leave’ (2011a, 271). Applying this lesson to addiction, the claim is that the fact that addicts refrain from use in particular circumstances (e.g., when offered immediate but modest monetary incentives, or when they secure a good job, or become parents) does not show that they have control over their use outside of these circumstances: all it shows is that they have control in these circumstances (cf. Levy 2011a). Control must always be relativized to circumstance.

For the record, note that, contra Levy, from a clinical perspective there is no absurdity whatsoever in the claim that, at least typically, agoraphobics can leave the house. Effective treatment for agoraphobia (or indeed any phobia) is likely to include a form of exposure therapy which, in this case, involves nothing other than the patient leaving the house, with increasing duration and regularity, and decreasing support from the therapist (Gros and Antony, 2006; Ito et al. 2001). Repeated exposure to anxiety-provoking stimuli reduces anxiety. The more one does it, the easier it gets; but one has to do it for exposure therapy to work. Like the clinical presumption in treatment for self-harm and violence as discussed above, the clinical presumption in exposure therapy is that agoraphobics can leave the house, however much they desire not to. In this, agoraphobia is like other disorders of agency: the first prong of treatment simply demands the patient change the problematic behaviour, despite the intensity of the emotion driving it. This of course is perfectly compatible with clinical recognition of the degree of the agoraphobic's anxiety, and the consequent difficulty for them in facing it.

Returning to the philosophical claim, we should agree that extreme circumstances affect people's abilities. In order to save a child from death, a parent may have the ability to move a crushing weight even though in standard conditions they lack the requisite physical strength. After withstanding harrowing physical torture, a prisoner may lose the ability to further resist the demand for information. Extreme circumstances no doubt affect what people can and cannot do. But this point should not bar us from holding that, in less extreme circumstances, behavioural change following motivational change provides strong evidence of a general ability for behavioural control.

Consider, for instance, a man who has no diagnosable disorder but who ‘sees red’ and routinely resorts to physical violence in drunken disputes—except when in view of a policeman. On such occasions, he is highly motivated not to hit, which he would otherwise do, out of fear of being detained and charged with common assault. Does his restraint in this context show only that he can control his aggression when in view of a policeman, but not necessarily otherwise? This is not our natural understanding of this man's behaviour. The more natural understanding is that it shows that the man has a general ability to control his aggression, but that he only exercises it when he wants to. As Michael Smith puts it, ‘capacities [abilities] are essentially general or multi-track in nature, and […] therefore manifest themselves not in single possibilities, but rather in whole rafts of possibilities' (2003, 26). We expect that the ability to do something in one situation generalizes, even if there are some circumstances where a person genuinely lacks an ability that they otherwise have, or possesses an ability that they otherwise lack. There is a basic, common-sense distinction, between what a person can do but won't (because they don't want do) as opposed to what they want to do but can't (because they lack the ability). We must recognize extremes, but relativizing control too strongly to motivational and epistemic circumstances threatens the cogency of this distinction.

With respect to addiction, modest monetary incentives and the ordinary aspects of adult life that motivate ‘maturing out’ (such as employment opportunities and parenthood) are not extreme or unusual circumstances. They are standard, commonplace reasons for abstaining that abound in ordinary life. They thus provide strong evidence that addicts have the general ability to control their use, in a broad range of ordinary conditions, despite the neurobiological effects of drug-use.10 Of course, addicts will only refrain from use if they want to. Having an ability does not require that one choose to exercise it. Below I suggest why there may be compelling reasons why addicts often don't choose to abstain, despite the terrible consequences they may suffer because of their consumption. Just so, there may be compelling reasons why we should not judge them or blame them for using. But the link between motivation and abstinence should not cause us to hold that, unless motivated, addicts cannot abstain from drug use—any more than we should hold that aggressive men cannot refrain from hitting, and agoraphobics cannot leave the house.


The final reason offered for the view that addiction is a form of compulsion is the testimony of addicts. Louis Charland famously reports a conversation with a heroin addict named Cynthia, who treats the idea that heroin addicts have the capacity to consent to heroin prescription with utter disbelief: ‘if you're addicted to heroin, then by definition you can't say “No” to the stuff’ (Charland 2002, 37). Cynthia is not exceptional: especially when initially engaging with psychiatric services, it is not unusual for patients to say they ‘can't’ control their drug-taking and other impulsive behaviour. But there is good reason to treat self-reports of compulsive drug-taking with scepticism. First, not all addicts agree: for every story of compulsive use, there is a story of deliberate abstinence and hard-won recovery (cf. the first-person narratives in Heyman 2009). Second, not only does our cultural conception of addiction invite this self-image, but, as suggested above, adopting it can serve to excuse addicts from the responsibility for drug-related behaviours: addicts have reason to claim to be compelled (cf. Ainslie 1999; Davies 1992; Foddy and Savulescu 2006, 2010). Third, clinical practice lends support to such scepticism. Part of the clinical aim with patients who struggle to control problematic behaviour is precisely to help them to see that it is not that they can't control their behaviour, but that they don't, and to help them understand why they don't: what they gain from the problematic behaviour and its role in their lives. This is part of the third prong of treatment. It is, in ways, the most painful, for it may require patients to confront aspects of themselves about which they feel guilt and shame. Finally, ‘can't’ may have multiple meanings in this context (Sinnott-Armstrong and Pickard, forthcoming). When we say that we cannot do something, we often mean that the costs of doing it are high, or that the reasons for not doing it are compelling. This may be what addicts mean by ‘can't’. They may not be saying that it is impossible to refrain from use: that the desire to consume is irresistible. Rather, they may be expressing how hard it is to choose to refrain because of the costs of abstinence, and the many good reasons they have to continue to use (see below).11

In sum, the evidence is strong that addicts possess the general ability to control their use in a broad range of circumstances. On the one hand, both the epidemiological data and clinical practice clearly point towards this claim. On the other, the standard considerations appealing to withdrawal, neurobiology, and testimony that are typically offered in favour of the opposing view of addiction as a form of compulsion fail to establish its truth. We should thus proceed, until further evidence or argument to the contrary is mustered, on the assumption that addiction does not provide an example of irresistible desire and hence a challenge to our pre-theoretical conception of action. Addicts are not compelled to use by irresistible desires: they have the ability to do otherwise. Of course, as with any ability, its possession is consistent with the possibility of particular circumstances in the lives of particular individuals when it cannot be exercised. And, once again, there may be compelling reasons why we should not blame addicts for using, even if they could indeed have done otherwise (see below and §3). The point is rather that an empirically and clinically informed understanding of addiction does not suggest it provides a real-world example of a case where a person acts—but lacks the ability to do otherwise.

Five Folk Psychological Factors Explain Chronic Addiction

Treating addiction as neurobiological disease characterised by compulsive drug use bars understanding of the psychological reasons why addicts use drugs and alcohol, and, consequently, why their control is impaired, and their choice not to abstain comprehensible. I suggest that five rough-and-ready folk psychological factors explain chronic addiction, in conjunction with neurobiological data and environmental and genetic factors.

Factor One: Strong and Habitual Desire

There is no question that, for all addicts, the desire to use their drug of choice is strong and habitual. As suggested above, we are starting to understand some of the neurobiological mechanisms underpinning the formation of desires and the establishment of strong stimulus-response associations between cues and behaviour. But even without this understanding, common sense tells us that strong habits are hard to break. When desire is strong and one is in the habit of satisfying it, it is not easy to resist.

Factor Two: Willpower

Resisting a strong desire requires a conscious effort at control: it requires will. There is increasing empirical evidence for what we might metaphorically construe as a faculty of willpower that acts much as a muscle does. It is effortful to exercise, and its exercise depletes its strength in the short-term, but can increase it in the long-term (for a review see Muraven and Baumeister 2000). Self-control, especially in relation to strong habits, requires this faculty, which is typically not well-exercised in addicts: conscious and sustained effort to resist the pull of the drug. Furthermore, it may also be that the conscious and sustained effort required to resist the pull of the drug can create “judgement shifts” whereby addicts, tired of resisting, reassess the value of abstinence and abandon prior resolutions in face of the present value of use (Levy 2011a, 2011b). The willpower and strength of resolve needed to break the habit is great.

Factor Three: Functional Role

As detailed above, addiction peaks in adolescence and early adulthood, but, in the majority of cases, has resolved permanently, without clinical intervention, by the late twenties or early thirties, when addicts tend to ‘mature out’. The exceptions to this finding are addicts who suffer from additional psychiatric disorders. They do not ‘mature out’ (Regier et al. 1990).

Non-addictive drug consumption serves multiple functions, such as improvement in social interaction, sexual behaviour, and cognitive performance (for a review, see Muller and Schumman 2011). In non-addicted users, one salient function is the management of psychological distress. This is common knowledge in our culture: we ‘reach for the bottle’ or ‘drown our sorrows’ when in need. Within the general population, research demonstrates that alcohol, barbiturates, benzodiazepines (and other sedative anxiolytics) and especially cannabis are self-administered to cope with stress (Bonn-Miller et al. 2007; Boyd et al. 2009; Boys et al. 1999, 2000; Cooper et al. 1988; Kuntsche et al. 2005; Perkins 1999). Within the psychiatric population, there is evidence of increased consumption of alcohol, cannabis, benzodiazepines (and other sedative anxiolytics), nicotine and opioids (Jacobsen at al. 2001; Hughes et al. 1986; Khantzian 1985, 1997; Markou et al. 1998). The ‘self-medication’ hypothesis has long been a staple of clinical understanding of psychiatric patients' use of drugs and alcohol (see especially Khantzian 1985, 1997).12 Psychiatric patients use drugs and alcohol to gain relief from intense negative emotions and other symptoms.

Hence one reason why psychiatric patients do not ‘mature out’ of addiction at the same rate as the normal population may lie in the purpose served by drugs and alcohol for these patients, together with the nature of their lives (Pickard 2012, Pickard and Pearce forthcoming). Drugs and alcohol provide a habitual and, in the short-term, effective way of managing the severe psychological distress typically experienced by patients with co-morbid psychiatric disorders and associated economic, social, and relationship problems. Put crudely, drugs and alcohol offer a way of coping with intense negative emotions and other psychiatric symptoms and problems. Hence, unless recovery from co-morbid disorders is achieved, better life opportunities are available, and alternative ways of coping are learned, patients are unlikely to forgo the use of drugs and alcohol as a way of managing their intense negative emotions and other symptoms. The cost of abstinence for such patients is likely to be very high.

Factor Four: Motivation and Incentive

Addiction is not only associated with co-morbid psychiatric conditions. It is also associated with lower socio-economic status (Compton 2007; for a review of the data see Heyman 2009), and, of course, the problems attendant upon the acquisition and use of the drug itself, such as poor employment opportunities, and loss of good regard of family and friends. The life choices and alternatives available to addicts are typically meagre: even if they succeed in abstaining, they will still need to pick up the pieces and squarely face some of the worst of life's various miseries. Bruce Alexander's infamous experiment ‘Rat Park’ is instructive in this light (Alexander et al. 1978; Alexander et al. 1985). Caged, isolated rats addicted to cocaine, morphine, heroin and other drugs will self-administer in very high doses, foregoing food and water, sometimes to the point of death (Woods 1978). Alexander placed morphine-addicted rats in an enclosure called ‘Rat Park’ which was a spacious, comfortable, naturalistic setting, where rats of both sexes were able to co-habit, nest and reproduce. Rats were offered a choice between morphine-laced water and plain water. On the whole, they chose to forego the morphine and drink plain water, even when they experienced withdrawal symptoms, and even when the morphine-laced water was sweetened to significantly appeal to the rat palate. Recent studies complement Alexander's findings. Environmental enrichments protect against relapse in rats (Solinas et al. 2008) who even when addicted will typically choose not to self-administer drugs if provided with alternative goods (Ahmed 2010, 2012)

Addicts who abstain from drug use are not typically offered the immediate option of a human version of ‘Rat Park’. The good life does not spring forth ready-made; help with housing, employment, psychiatric problems, and social community, does not tend to be promptly available. The opportunities and choices available to many addicts may reasonably impede their motivation to control their use, for the alternative goods on offer are poor.

Factor Five: Decision and Resolve

Controlling use typically requires not just willpower, but perseverance and resolve. Addicts must overcome any natural ambivalence they might feel about whether or not to stop using. They must decide to change, and they must form a resolution to stick with that decision in the face of future temptation. This is a substantial undertaking for many addicts. But, importantly, addicts cannot even make such a decision if they genuinely believe that they are powerless over their desire to use: that their behaviour is the effect of a neurobiological disease. For one cannot rationally form an intention or make a decision to do something if one believes that one cannot succeed: that it is simply not in one's power to do so.13 In this respect, the characterization of addiction as a neurobiological disease impedes recovery, for it is an obstacle to the rational formation of intentions or decisions to abstain. A belief in one's own self-efficacy may be crucial (cf. Bandura 1997).14 This belief is undermined by adhering to a disease model of addiction and thereby assigning addicts to the sick role (Pearce and Pickard 2010).

Explaining Impaired Control in Addiction

All addicts, by definition, have a strong desire to use their drug of choice, which they are in the habit of fulfilling. Restraint is thus hard: it requires willpower. Most addicts ‘mature out’ during their late twenties and early thirties. Once the view that addiction is a form of compulsion has been rejected, the natural explanation is that they are motivated by the responsibilities and opportunities that characterise adult life to exercise the willpower necessary to abstain. Those addicts who do not ‘mature out’ typically suffer from additional psychiatric disorders. For these patients, substance use is likely to serve a particular purpose: it provides a habitual and, in the short-term, effective way of managing the severe psychological distress typically experienced by patients with co-morbid psychiatric disorders and associated economic, social, and relationship problems. Put crudely, drugs and alcohol offer these patients a way of coping with intense negative emotions and other symptoms: it is a chosen if habitual means to desired ends. Hence, unless recovery from co-morbid disorders is achieved, better life opportunities are available, and alternative ways of coping with psychological distress have been learned, patients are not likely to forgo the use of drugs and alcohol. The cost is too great, the alternative goods on offer too few. This is a compelling reason to continue to use. Chronic substance use is, to some degree, a rational choice for such patients, unless they can be given hope for a better life.

Hence, although addictive desires may be strong and habitual, they are not irresistible. Addicts are agents who use drugs and alcohols as means to understandable ends. They can, and often do, choose to abstain, when sufficiently motivated and supported to do so. Note, again, that this leaves open the possibility that they can be excused when they don't. For example, if drugs and alcohol are indeed used to manage severe psychological distress, then, in absence of alternative coping mechanisms, addicts may be justified in choosing to take drugs, with the crucial caveat that such justification depends on the nature and degree of any harm caused to others by their doing so. In other words, addicts may be excused not by compulsion, but by duress, as the cost of abstinence may be too high to be reasonable to ask addicts to bear (Pickard 2011b, 2012; Yaffe 2011).

What is true of addiction is true of compulsion in psychopathology more broadly. Many factors, such as strength of desire and habit, willpower, functional role, motivation and incentive, decision and resolve, contribute to why control is impaired relative to the norm. But impairment is not extinction. Patients are not powerless over their desires: they have the ability to do otherwise, which they exercise, often successfully, when they so choose. Effective treatment for disorders of agency starts with patients choosing to do things differently. Only then is recovery possible.

§3. The Philosophical Landscape

Suppose, then, that addicts, agoraphobics, kleptomaniacs, neurotics, obsessives, and of course psychopathic serial murderers, are not subject to irresistible desires: it is possible for them to do otherwise. Doing so may be very hard. The cost to them may be very high. For these (and perhaps other) reasons, they may be excused if they don't. But they are not powerless in face of their desires: it is not impossible for them to do things differently. What is the philosophical import of this fact?

At the start of this paper, I quoted Harry Frankfurt as claiming that hypnosis, coercion, and ‘inner compulsion’ can make it impossible for a person to avoid doing something. How does compulsion relate to these other circumstances? Frankfurt presumably means to appeal to a popular conception of hypnosis, whereby people enter a zombie-or sleep-like state so that any action is like an automatic reflex, lying wholly outside of their conscious control. Although less is known about hypnosis than addiction, it is again important to note that this is unlikely to be an empirically accurate picture of the phenomenon. Under hypnosis, subjects are suggestible, but they cannot be forced to do anything they do not want to do, and, although peripheral consciousness is decreased, they are fully awake and experience heightened conscious attention or focus (Spiegel and Spiegel 2004 [1978]). But, if the popular conception of hypnosis (or the Hollywood fiction of zombies) was real, then the behaviour of hypnotized or zombified subjects would approximate an involuntary, automatic reflex in response to suggestion or stimulus, over which the subject lacked all control. Such subjects could not do otherwise. But nor would their behaviour count as intentional action.

The effect of coercion on the power to do otherwise is more complicated. We need to distinguish two kinds of case. Consider the standard example in the literature: the hold-up.

Version One: A hardened and experienced bank teller is confronted (not for the first time) with an armed thug: their till or their life. They are not overly emotionally or psychologically affected by the thief. Instead, they are capable of making a reflective choice in the moment and acting upon it. They calmly hand over the cash in the till. They could have done otherwise: they could have given their life instead of the money. Obviously, they are justified in not doing so, because the obligations of a bank teller do not include self-sacrifice in the line of duty—unlike, say, the obligations of a soldier. (Compare again the way an addict might be justified in taking drugs: we do not expect them to bear the cost of the other alternative currently on offer, namely, abstinence.)

Version Two: An anxious and jejune bank teller is confronted with an armed thug: their till or their life. Petrified, overwhelmed and on the verge of a panic attack, they frantically throw the bank's money at the thug, and then promptly faint. They would have done the same even if there had been plate glass protecting them. They were beside themselves, as we say, with fear. The emotional and psychological impact of the thief's threat deprives them of the power to do otherwise: they momentarily lack the capacity for behavioural control. (Compare an atypical agoraphobic who actually has a panic attack and faints when leaving the house.)

In the first version of the hold-up, the bank teller retains the capacity for behavioural control: they could have done otherwise. Their behaviour is an action, pure and simple, the chosen option between alternatives, however unfortunate and unfair those alternatives are (cf. Mackie 1977).

In the second version of the hold-up, the bank teller's behaviour approximates an automatic reflex. They are frantic, out of control, beside themselves. Some psychological and physical states, such as extreme panic, fear, rage, stress, and exhaustion, can affect our brains and bodies, altering the neurobiological and physical bases that realise the capacity for executive function and behavioural control. In this, they are akin to states of sleep, popular hypnosis, or zombie-hood. When one is sleeping, hypnotized (as it is popularly conceived), transformed into a zombie, or in a state of terror or rage or utter panic, one may lack the capacity for behavioural control even though, if one was awake, or fully and normally conscious, or not suffering from extremes of emotion, one would possess it. Typically, the behaviour resulting from these psychological and physical states is either a form of relatively short-lived frenzy, or paralysis. When the state abates, the control returns.

Note that it is possible that addicts and other psychiatric patients—just like the rest of us—can suffer from such extreme emotional or physical states (as with the atypical agoraphobic mentioned above). Indeed, if drug and alcohol consumption is a coping mechanism, then abstinence may itself produce panic, fear, and exhaustion, contributing to the distress patients may already be experiencing and which is driving the desire to use. If so, it is at least possible that addicts might momentarily lose their capacity for behavioural control, and, in a sort of frenzy, use whatever drugs or alcohol they find in their immediate proximity.15 But the explanation of this powerlessness is not the irresistibility of their addictive desires per se, but rather, their emotional or physical state, which temporarily removes executive function and with it, behavioural control. It is also the case that such frenzied, out of control behaviour is by no means typical of the behavioural patterns constituting addictive consumption.

Hypnosis (as it is popularly conceived) and cases of coercion which lead to extreme emotional or physical states can cause behaviour which comprises or which approximates involuntary, automatic reflexes. Other cases of coercion leave the capacity for choice intact while yet affording an excuse, through the unjust restriction of alternatives. The importance of ‘inner compulsion’ in the philosophical landscape is that it promises something in-between behaviour that comprises or approximates automatic reflexes, and voluntary behaviour that is characterised by choice between alternatives. Taking drugs and alcohol, refusing to leave the house, refusing to eat, stealing, cleaning, hand-washing, checking multiple times to make sure a door is locked, self-harming, killing—from the outside, these look just like voluntary actions and omissions if ever there are any. But if, from the inside, the desires driving these actions are irresistible—if there is genuinely no possibility to choose or act otherwise—then they are also like automatic reflexes. They look like cases of action that are psychologically determined by the pathological desire. A key function of psychopathology within the philosophical landscape is thus to provide us with a real, commonplace example of what action is like in absence of free will. No doubt, philosophers have constructed hypothetical examples of non-pathological agents, such as imagined victims of evil brain-manipulating neuroscientists, who purportedly lack the power to do otherwise.16 But psychopathology is the central example of irresistible desire in the philosophical literature that purports to be real. It thus appears to offer us a model of what our world would genuinely be like if there is no free will. If that is indeed our world, then we are just like addicts and others who suffer from disorders of agency and have no possibility of doing otherwise when acting: we act, but not out of choice or of our own free will.

Psychopathology does not provide us with such a model. People who suffer from disorders of agency can do otherwise: they are not subject to irresistible desires. Above, I suggested that our pre-theoretical concept of action connects it to choice and control: action involves the possibility of doing otherwise, at least in so far as one could refrain from performing that very action.17 We labour to make sense of the claim that there could be genuine action in absence of such choice and control. Psychopathology is no help with this task: it does not offer us a real case of action without choice between alternatives, from which to develop an alternative conception of what action is. Sometimes people do indeed lack the capacity for behavioural control due to the effect on executive function of their emotional or physical state, in which case their behaviour approximates an automatic reflex rather than being self-evidently an action. This can be true of all of us, including those who suffer from disorders of agency. But in psychopathological cases, there is no compulsion or impossibility of choosing or doing otherwise based on irresistibility of desire. Rather, there is impaired control relative to the norm due to a range of interacting psychological factors and hard choices in difficult life circumstances.18 Psychopathology does not show that there are actions over which patients are powerless. It thus fails to provide a model of a world without free will, in which desires can by-pass or ‘helplessly violate’ the agent, destroying all power to do otherwise while yet issuing in something recognizable as action, not just automatic reflex.

One source of potential pressure on the centrality of choice and free will to our concept of action is thus relieved. But the lesson for philosophy from psychopathology that I suggest we should draw is stronger than this. Clinical interventions for disorders of agency are various and complicated, and include pharmacological alongside social and psychological means. It is difficult enough to assess outcomes, let alone to gain knowledge of the mechanisms by which outcomes are achieved. Nonetheless, as described above, many of the more successful treatment strategies typically depend on mobilizing the person's own agency for change through supporting them explicitly to see that it is in their power to do things differently and to choose to do so. Of course, that is not all there is to it. People need to overcome their ambivalence and stay the course despite temptation to revert to past habits, self-doubt, and fluctuating desires. Motivation and resolve is no doubt crucial. Medication and common sense techniques for structuring the environment and improving willpower to resist temptation can help too. So also can the increased self-esteem, hope, and sense of belonging, that comes from receiving compassion from others and being part of a community, which is an essential component of many forms of group-based treatments (Pearce and Pickard 2012). But empowerment and choice is central to this process of improvement and often marks the moment when things start to change. Quite often, the best explanation there is for why an entrenched pattern of behaviour suddenly shifts in the clinic appeals to the patient's decision, here and now, to shift it. Psychopathology does not provide an example of what the world would be like without free will. Rather, the clinical world provides an example of where the potency of free will, whether or not this is ultimately illusory, is on clear display.19

  1. 1

    Note for clarity that I do not here address the question of whether or not free will ultimately exists and is or is not compatible with determinism. The claim that psychopathology in particular does not strip people of free will is consistent with the claim that, appearances notwithstanding, no one has free will, as much as with the claim that, appearances withstanding, we all have free will, including those of us who suffer from disorders of agency, and those of us who do not.

  2. 2

    Some readers may be impatient with the empirical and clinical detail of this discussion: I ask that they bear with me. Again, the primary aim of this paper is to establish that the way philosophers have conceived of psychopathology is false: this is why it cannot play the role assigned to it within philosophical debates. Establishing this requires real and sustained engagement with the facts of psychopathology.

  3. 3

    Although both Alvarez and Steward emphasise the necessity of the possibility of restraint to the concept of action, there are important differences between them. Alvarez holds that in order for a piece of behaviour to count as an action, the agent must be then able to refrain from performing that type of action. Steward holds that in order for a piece of behaviour to count as an action, the agent must be then able to refrain from performing that token action. But this distinction aside, the spirit of the accounts is very similar. Indeed, Steward suggests that this spirit is found throughout the history of philosophical writing on action and free will and can be discerned in Aristotle (1984), Hobbes (1999), Hume (1975), Reid (1994), and Kant (1960). See too Williams (1995).

  4. 4

    Arguably this is why certain expressions and gestures, such as those typically associated with emotions, are not considered to be involuntary, automatic reflexes. Although they are not actions because the subject does not choose to initiate them or intend to perform them, neither are they wholly involuntary, because, should the subject become aware of their expressive behaviour, they may be able then to choose to suppress it or stop (cf. Goldie 2000).

  5. 5

    For a general survey of this data across psychiatric disorders see Aneshenshel and Phelan 2006. For data and discussion of this association in relation to addiction in particular, see §2 below.

  6. 6

    Although there are differences between community, in-patient, and forensic management strategies for self-harm and violence, and of course between different practicing clinicians, the use of contracts for managing these behaviours is common within treatment courses with a strong evidence-base, such as Dialetical-Behavioural Therapy (DBT) (Lineham and Dimeff 2001), Systems Training for Emotional Predictability and Problem Solving (STEPPS) (Blum et al. 2008), and Therapeutic Communities (TCs) (Lees et al. 1999). Contracts can be verbal or written, undertaken between patient and therapist or patient and group, more or less detailed, and more or less tailored to the individual patient. For an example of a standard DBT therapy contract see: For an excellent collection on understanding and managing self-harm see Motz 2009.

  7. 7

    Note that toleration of breaches of the agreement is an important part of the therapeutic relationship. Patients may commit to ending self-harm or violence but yet on particular occasions lapse. Managing such lapses is crucial, as hope and resolve need to be maintained, and criticism avoided. Crucially, clinicians do not typically achieve this by holding that patients were unable to do otherwise on the occasion of the lapse. Rather, they are much more likely to hold that the patient chose to behave as they did and break the agreement, and that part of the aim of the therapy is to help them understand why, on that occasion, they made that choice, and to help them choose differently in future when they are again tempted towards self-harm or violence.

  8. 8

    Cf. Alvarez 2009 p. 74 who suggests that psychological therapies could not be deployed if the problematic behaviour was genuinely unavoidable.

  9. 9

    For a more detailed and clinically-focussed discussion of the view of addiction put forward in this section see Pickard 2012, Pickard and Pearce forthcoming.

  10. 10

    Indeed, this point is found even in Edwards and Gross' seminal discussion of the Disease Model of Alcohol Dependence. They write that ‘it is unclear, however, whether the experience [of alcoholism] is truly one of losing control rather than one of deciding not to exercise control’ (Edwards and Gross 1976, 1060).

  11. 11

    For a reflective first-personal account of addiction that is broadly in keeping with the view presented here, see Flanagan 2011 and forthcoming.

  12. 12

    For a review of how the neurobiological effects of drugs and alcohol may act to alleviate psychiatric distress and symptoms, see Muller and Schumann 2011.

  13. 13

    Note that this posits only a modest connection between intention and belief: the claim is only that it is not rational to form an intention if one believes that one cannot succeed, not that it is impossible. For discussion see Holton 2009.

  14. 14

    Self-image may be correlated with self-efficacy. Robert West reports a study finding that within one week of quitting, half of all participating smokers thought of themselves as ex-smokers. This self-image is optimistic: on average 75% will be smoking again within the year. However, 50% of those who thought of themselves as ex-smokers were still abstinent at 6 months, as compared with 0% of those who did not immediately embrace the label (West 2006, 163).

  15. 15

    See Kennett 2001 for a discussion of whether a similar explanation of lack of capacity for behavioural control can apply to certain instances of murder.

  16. 16

    For arguments against the general use of these sorts of hypothetical examples within the philosophy of action and the cogency of some particular well-known cases, see Alvarez 2009 and Steward 2009 and 2012.

  17. 17

    Indeed, even rival neo-Humean accounts of action which do not explicitly employ notions of choice and control, such as the account put forward by Donald Davidson (1980 [1963]), may implicitly depend on these notions for their plausibility. Davidson holds that what makes a piece of behaviour an intentional action is that it is caused—in the right way—by the agent's reasons for acting, that is, their (instrumental) beliefs and desires. On the one hand, it is natural to spell out the demand that the causation proceed ‘in the right way’ as opposed to allowing for deviant causal chains precisely by appeal to choice over action. Choice guarantees the right kind of causal chain: it is only when the behaviour is caused by a desire that the subject has chosen to act on that it counts as action. On the other hand, as Davidson himself emphasises, the beliefs and desires employed by this account of intentional action are supposed to belong to a self-conscious, rational subject. Cognitive psychology depends on the idea that lower animals have non-conscious representations of the world and non-conscious representations of outcomes at which their behaviour aims. No doubt, so too do higher animals, like us. But the resulting behaviour does not count as intentional action according to Davidson. Whatever the beliefs and desires are that together cause intentional action, they are not non-conscious representations of this sort, but rather conscious, personal-level states that are evaluated and acted on by a self-conscious, rational subject. Again, this distinction seems to appeal to the possibility of choice over action.

  18. 18

    Following Bernard Williams, we might express this point by saying that, although free will does not come in degrees, freedom does: ‘Why does freewill, unlike freedom, not come in degrees? Presumably it is because its assertion consists only of an existence claim. How exactly that claim should be expressed is notoriously disputed, but it is something to the effect that agents sometimes act voluntarily, and that when they do so they have a real choice between more than one course of action; or more than one course of action is open to them; or it is up to them which of several actions they perform… [this] merely requires that there be, in the appropriate sense, alternatives for the agent, and that it is indifference to their number, their cost, and so forth. That is why the freewill that it introduces is different from the freedom that comes in degrees and is opposed to constraint’ (1995, 5).

  19. 19

    This paper was presented at a Workshop on Abilities, Agency, and Freedom at the Humboldt-Universitat zu Berlin, and I am grateful to the audience and, in addition, Christian List, John Maier, Michael Smith, and Barbara Vetter, for very helpful comments. It also inherits debts owed to all those who have contributed to my research on responsibility without blame, and addiction, whom I thank again here. Finally, special acknowledgement is due to Ian Phillips for his contribution to articulating the two ways coercion can function as an excuse in §3, and special thanks for countless, constructive discussions over the years. This research is funded by a Wellcome Trust Biomedical Ethics Clinical Research Fellowship.