Responsibility Without Blame: Empathy and the Effective Treatment of Personality Disorder

Abstract

Effective treatment of personality disorder (PD) presents a clinical conundrum. Many of the behaviors constitutive of PD cause harm to self and others. Encouraging service users to take responsibility for this behavior is central to treatment. Blame, in contrast, is detrimental. How is it possible to hold service users responsible for harm to self and others without blaming them? A solution to this problem is part conceptual, part practical. I offer a conceptual framework that clearly distinguishes between ideas of responsibility, blameworthiness, and blame. Within this framework, I distinguish two sorts of blame, which I call ‘detached’ and ‘affective.’ Affective, not detached, blame is detrimental to effective treatment. I suggest that the practical demand to avoid affective blame is largely achieved through attention to PD service users’ past history. Past history does not eliminate responsibility and blameworthiness. Instead, it directly evokes compassion and empathy, which compete with affective blame.

The Nature of Personality Disorder

Stigma and prejudice have long affected the science and treatment of personality disorder (PD). It is widely recognized that PD is associated with social exclusion, unemployment, homelessness, and crime, together with addiction, eating disorders, anxiety, depression, and psychoses (National Institute of Mental Health in England 2003). The financial burden placed on psychiatric, medical, social, legal, and forensic services is high. The practical and emotional burden placed on family and friends of those with PD, as well as the clinicians who seek to treat them, is equally real. PD is also associated with deliberate self-harm, suicide, mortality, violence, chaotic lifestyles, emotional instability, and severe and lasting difficulties managing interpersonal relationships. These and other aspects of PD have a profound impact on those who come into contact with service users.

The Cluster B or ‘bad’ PDs are explicitly defined and diagnosed in part via traits that count as failures of morality or virtue (Charland 2004, 2006; Pearce and Pickard 2009; Pickard 2009). For instance, the DSM-IV-TR (American Psychiatric Association 1994) defines Narcissistic PD as involving lack of empathy, grandiosity, need for admiration, and a willingness to exploit others. Histrionic PD involves an excessive demand for attention and ‘inappropriate’ sexual behavior. Borderline PD involves extreme and inappropriate anger toward self and others, instability in self-image and interpersonal relationships, and marked recklessness, impulsivity, and paranoia. Antisocial PD involves disregard for others and violation of their rights, criminal behavior, and lack of remorse. In contrast, Cluster A or ‘mad’ and Cluster C or ‘sad’ PDs may not be defined and diagnosed via traits that are so clearly connected to failures of morality or virtue. Nonetheless, they greatly affect service users’ capacity to develop relationships with others that express mutual regard, care, trust, and respect. Consider just one example from each cluster. Paranoid PD involves unjustified suspicion and distrust, and a tendency to hold grudges against others. Obsessive–Compulsive PD involves forsaking friendship for productivity, obedience to rules and authority at the expense of the good of self and others, miserliness, stubbornness, and a desire for interpersonal control. There is no question that people with PD suffer tremendously, experiencing an extreme degree of distress and dysfunction. But there should equally be no question that their behavior causes others to suffer, whether by design or by accident, and whether or not they have control over, and conscious knowledge of, the interpersonal effects of their behavior.

No doubt in part because of this financial, practical, and emotional burden, PD is heavily stigmatized within psychiatry. The stereotype of a PD service user as manipulative, demanding, attention seeking, and violent is pervasive. They are the service users ‘no one likes.’ Perhaps relatedly, PD has long been considered impossible to treat. Specialist psychiatric services have been scarce, treatment options and evidence-based research starkly limited.

This has started to change. In 2003, the UK Department of Health launched a national initiative Personality Disorder: No Longer a Diagnosis of Exclusion to combat stigma, promote awareness, and improve quality and access to services. A wave of new treatments and evidence-based research began, both in the United Kingdom and in the United States. With the exception of the appropriate prescription of medication, these treatments are psychological. Although various, they require clinicians to directly engage service users about behavior that is at once constitutive of their PD, and may be harmful not only to themselves, but also to others. Whether implicitly or explicitly, treatments require clinicians to encourage service users to take responsibility for their behavior, to choose and learn to act otherwise, if they are to improve, let alone recover (Pearce and Pickard 2010).

This requirement of effective treatment creates a clinical conundrum. How is it possible to hold service users responsible for behavior that causes harm and suffering, to the self and, especially, to others, without blaming them for it? Encouraging responsibility is central to effective treatment. Blame, in contrast, is highly detrimental. The aim of this paper is to resolve this conundrum. The paper has four parts. First, I describe the conundrum in more detail. I suggest that clinicians can often find themselves trapped between a desire to rescue and a desire to blame, despite neither response being effective. Effective treatment demands a fine balance: responsibility without blame. Clinicians do not always succeed in striking this balance, but it is an authentic stance that they genuinely adopt. Second, I offer a conceptual framework that clearly distinguishes ideas of responsibility, blameworthiness, and blame. Third, within this framework, I distinguish two sorts of blame, which I call ‘detached’ and ‘affective.’ Affective, not detached, blame is detrimental to effective treatment. I sketch an account of what affective blame is. This overall framework is central to understanding how the stance of holding a person responsible for harm but not blaming them is conceptually possible. Finally, I turn to the question of how clinicians can effectively keep affective blame at bay. It is one thing for the appropriate clinical stance to be conceptually possible, but quite another for it to be achieved in practice. I suggest that the key to striking this balance, and avoiding the trap, is understanding of an individual service user’s past history, and its power to directly evoke compassion and empathy.

The Clinical Conundrum: Striking the Balance Between Rescue and Blame

In his landmark study of staff attitudes to service users with PD in three High Security Hospitals in the UK, Bowers suggests:

The generally hopeless, pessimistic attitudes of carers can be seen to originate in the difficult behaviours of PD patients. They bully, con, capitalize, divide, condition, and corrupt those around them. They make complaints over inconsequential or non-existent issues in order to manipulate staff. They can be seriously violent over unpredictable and objectively trivial events, or may harm and disfigure themselves in ways that have an intense emotional impact on staff. If this were not enough, they also behave in the same way towards each other, provoking serious problems that the staff have to manage and contain. (Bowers 2002, 65)

Negative staff attitudes toward service users may stem from the belief that they are ‘to blame’ for this behavior. Bowers found that staff typically hold service users with PD responsible for their behavior, at least in part, because they act for reasons and “know what they are doing” (Bowers 2002, 85). Unless PD service users are psychotic or cognitively impaired, staff believe they have control over and conscious knowledge of what they are doing.¹ They are responsible, indeed they are ‘to blame,’ because they have no excuse.

Comparable findings were reached by Gallop, Lancee, and Garfinkel (1989) in a study comparing staff attitudes toward schizophrenia as opposed to Borderline PD. Otherwise identical hypothetical vignettes evoked empathy if the service user was said to be diagnosed with schizophrenia, derision if the service user was said to be diagnosed with Borderline PD. This seems to be because the behavior of service users with Borderline PD is typically seen as ‘bad’ and ‘deliberate’ in contrast with service users with schizophrenia who are seen as ‘sick’ and ‘lacking control’ (Gallop, Lancee, and Garfinkel 1989, 819; cf. Potter 2009).

In sum, staff are prone to blame service users with PD for their behavior because they believe that:

1. Service users with PD have control and conscious knowledge of their behavior.

2. Therefore, service users are responsible for their behavior.

3. The behavior causes harm.

4. In contrast with a psychotic illness like schizophrenia, a diagnosis of PD does not by itself constitute an excuse.

Blame has an adverse effect on care. The UK National Institute for Clinical Excellence guidelines on the management of self-harm, for instance, note that: “the experience of care for people who self-harm is often unacceptable” (National Institute for Clinical Excellence 2004). Service users report being left unattended, roughly handled, stitched without anesthetic, and verbally criticized and judged. Self-harm can evoke strong, negative emotions in staff. It is natural to speculate that staff do not manage these emotions well, in part because they view self-harm as deliberate, and service users as responsible: it is as if staff believe that service users are at fault, and so not deserving of care.

There is good evidence that compassion and empathy are central to good therapeutic care (Gilbert 2010). The reasons why blame is detrimental are less studied, but reflection on clinical practice provides some guide. Most simply, if clinicians blame service users, they are unlikely to offer them care, help, and a safe and trusting environment. This is not conducive to fostering a therapeutic alliance with the service user. Disengagement with services is an obvious risk. More seriously, service users with PD typically have low self-esteem and are prone to self-hatred and self-blame. This is no doubt part of the explanation for the high levels of self-harm and suicide among the PD population. Clinical blame may trigger these feelings, increasing risk of serious harm to self.

Of course, most staff recognize, at least to some extent, the adverse effect blame can have on care, and struggle not to respond in this way. To this end, they may swing to the opposite pole, and come to view PD service users as not responsible for their behavior, and in need of rescue instead. This alternative response is bolstered by the obvious fact that service users with PD suffer extreme degrees of distress and dysfunction, which they may have no clear sense how to alleviate or manage, together with the fact that it is a clinician’s duty to help and care. In this mindset, staff may hold that service users with PD ‘cannot help’ behaving as they do and so are in need of rescue because they believe:

1. Service users with PD do not have control or conscious knowledge of their behavior.

2. Therefore, service users with PD are not responsible for their behavior.

3. Further, PD is an illness, and so affords an excuse.

This alternative, rescue response may be further bolstered by the fact that service users themselves sometimes express the feeling of being out of control, compulsively driven to behave as they do (cf. accounts of addiction that treat it as compulsive such as Charland 2002; for dissenting accounts, see Foddy and Savulescu [2006] and Pickard and Pearce forthcoming). However, rescue is no more viable a clinical response than blame. It is extremely difficult, if not outright impossible, to genuinely sustain the belief that service users with PD do not have control or conscious knowledge of their behavior, no matter what they sometimes say. The reason is that this belief flouts our common sense conception of agency and action and its evident applicability to PD service users.

Our common sense conception of agency draws a basic distinction between actions and mere bodily movements, such as automatic reflexes. What makes a piece of behavior an action, as opposed to a mere bodily movement, is that it is voluntary, where this means that the agent can exercise choice and at least a degree of control over the behavior. This conception of agency and action is traditionally linked, within philosophy, to the idea of free will, and can be found in philosophers as diverse as Aristotle (1984), Hobbes (Chappell, ed. 1999), Hume (1975), Reid (1994), and Kant (1960) (for historical discussion see Bobzien [1998]; for a contemporary defense of this view, see Steward [2009]). On this view, agency and action require two capacities. First, the capacity to choose from a range of possible actions, at least in the minimal sense that, on any particular occasion, one can choose either to act, or to refrain from so acting. Second, the capacity to execute this choice: to do as once chooses, given normal circumstances (cf. Holton [2010]; for an important analysis of the nature of such capacities, see Smith [2003]). This common sense conception of agency naturally grounds judgments of responsibility: one is responsible for actions, as opposed to automatic reflexes, because it is up to one whether and how one acts. So long as one knows what one is doing, one is responsible for one’s behavior to the degree that one can exercise choice and control over it.

The behaviors that are constitutive of PD are not mere bodily movements. They are kinds of action: the kinds of behavior over which we have choice and control. On the whole, PD service users possess both relevant capacities with respect to these behaviors. On at least most, if not indeed all, occasions, they could, for example, choose not to behave so as to exploit others, demand attention, lash out in anger, drink or take drugs, self-harm, or commit a crime, and they could execute this choice: they could not do so. The evidence for this claim is relatively straightforward. Service users routinely do choose to behave otherwise and alter entrenched patterns of behavior, when they have incentive, motivation, and genuinely want to do so. Indeed, this is part of what psychological treatment both presupposes and fosters. Hence, so long as they know what they are doing, PD service users are responsible for their behavior to the degree that can exercise choice and control over it.²

However, it is important to note two caveats. First, service users with PD may not always have full conscious knowledge of why they are behaving as they do, or what the full effects of their behavior on others may be. Of course, in this, they are not unique: this is a predicament we all face to some extent. But it is possible that some kinds of PD, most obviously Borderline, will be associated with reduced capacity for such conscious knowledge. The possibility of mentalization deficits (Fonagy et al. 2004) and high levels of emotional arousal associated with BPD may have this effect. Second, it is important to recognize that, on the common sense conception of agency presented above, control is a graded notion, and the degree of control possessed by PD service users may sometimes be diminished compared with the norm. Patterns of behavior associated with PD may be habitual and strongly desired. Insofar as these patterns are ways of coping with psychological distress, service users may lack alternative coping mechanisms. Without these alternatives, alongside the hope of a better life, they may also lack the will or motivation to change their behavior, to kick a habitual pattern, and find another way of behaving that is less harmful to self and others. For these reasons, control may be diminished, and with it, responsibility. But reduction is not extinction. The difficult behavior of PD service users nonetheless counts as voluntary action. Indeed, we can understand effective treatment as, in part, augmenting service users’ existing capacity for agency (Pearce and Pickard 2010; Pickard and Pearce forthcoming).

Clinicians can acknowledge that conscious knowledge and control of action may on occasion be reduced in PD service users relative to the larger population. But it is not practical for clinicians to try to believe that PD service users have no conscious knowledge or control, as a way of avoiding blame, for it flouts our common sense conception of agency, and the evidence. Furthermore, it is also not clinically effective. There is increasing evidence for a variety of effective psychological treatments for PD (for a review of the evidence base see http://www.dh.gov.uk/en/Publicationsandstatistics/Publications/PublicationsPolicyAndGuidance/DH_4009546; Scott and Attwood, under review). These include:

1. Varieties of cognitive–behavioral therapy, such as dialectical–behavior therapy (Linehan and Dimeff 2001), STEPPS (Blum et al. 2008), and ‘stop and think’ training, to help manage self-harm and other counter-productive behaviors.

2. Motivational interviewing techniques, to engage service users and foster the desire to change (Rollnick and Millner 1995).

3. Emotional intelligence to develop the knowledge and ability to identify triggers, understand emotions, and manage behavior (Goleman 1998).

4. Mentalization-based therapy to develop self- and other- understanding and empathy (Fonagy and Bateman 2006).

5. Therapeutic Communities, often considered the treatment of choice for PD, which may employ all varieties of psychological therapy, alongside a commitment to four guiding principles of democracy, permissiveness, community, and reality confrontation that govern the community (Lees, Manning, and Rawlings 1999). Within Therapeutic Communities, service users are responsible for much of the daily running of the service, as well as directly challenged and supported to change entrenched behavior, with agreed consequences if they lapse. Therapeutic Communities also importantly provide a social community to which service users can belong, thereby reducing their isolation, and increasing their potential for well-being and support.³

These treatments are united in treating service users as responsible agents, capable of controlling their behavior and deciding to change. Indeed, post-therapeutic well-being seems to be correlated with post-therapeutic narratives of therapy that emphasize service users own agency as a force of change (Adler et al. 2008). Some interventions, like motivational interviewing, do this implicitly. The clinician adopts a submissive, nonchallenging, stance, expressing empathy and encouraging the service user to see the unwanted consequences of their behavior as motivation to change. Other interventions, like those offered by Therapeutic Communities, are highly explicit. The language of agency and responsibility permeates the culture of the group: members are not only encouraged but expected to see themselves in this light. Still other interventions, like emotional intelligence, fall in between. Service users are encouraged to distinguish emotions and behavior, to allow them to take responsibility for how they act when in the grip of strong emotions, even if they maintain, correctly or not, that they cannot take responsibility for how they feel.

If clinicians give up the belief that service users have choice and at least a degree of control over their behavior, they cannot rationally decide to work with service users to augment it. Indeed, if service users themselves come to believe that they genuinely have no choice or control over their behavior, they cannot rationally decide to try to change. For one cannot rationally resolve to change that which one believes one is powerless to change (Pearce and Pickard 2010; Pickard and Pearce forthcoming).⁴ Effective treatment for PD depends on clinician and service user believing that the service user has choice and at least a degree of control over their behavior: they are to that degree responsible agents. The cost of avoiding blame by absolving service users from responsibility is thus high. It precludes both clinician and service user alike from rationally pursuing psychological treatment, leaving only medication as an option.

The trap facing clinicians treating PD service users is real. Either PD service users are treated as having conscious knowledge, choice, and a degree of control over their behavior, or they are not. If they are, effective treatment that engages service users as responsible agents is possible, but there is high risk of blame for harm caused. If they are not, effective treatment is not possible, but there is less risk of blame. The clinical need to find a balance between these two responses is clear: there must be responsibility without blame. Clinicians regularly do succeed in striking this balance. The conundrum is how this is possible. The solution to this conundrum is part conceptual, part practical. I turn to the conceptual component first.

A Conceptual Framework for Responsibility Without Blame

Within philosophy, there is a tendency to link the idea of responsibility to morality (cf. Pickard and Phillips, under review). This link can be weak or strong. Weakly, philosophers often use ‘moral responsibility’ and ‘responsibility’ as if they were interchangeable, suggesting, if sometimes unintentionally, that all responsibility is moral. More strongly, philosophers sometimes argue that the idea of responsibility should be understood by appeal to our practice of holding others responsible via what are called our ‘reactive attitudes’ or ‘moral emotions.’ These consist in various responses we can have to the good or ill will that others display toward us, such as forgiveness and gratitude, indignation and resentment, and praise and blame (Strawson 1962). At its most radical, this link between responsibility and the reactive attitudes is thought to be constitutive. As Watson puts this view: “to regard oneself or another as responsible just is the proneness to react to them in these kinds of ways” (2004, 220). Slightly more modestly, Wallace (1994) has argued that to hold another responsible is to believe that reactive attitudes are appropriate or fitting responses to their behavior, even if one does not actually feel anything oneself.

Both weak and strong versions of this link between responsibility and morality obscure the possibility of responsibility without blame. With respect to the weaker, linguistic link, it is extremely important that clinicians are able to speak plainly to service users of their responsibility for problematic behaviors without implying that the behaviors even might be morally wrong or the person bad. Compare:

1. If you decide to self-harm/abuse substances/clean obsessively, you are responsible for that.

2. If you decide to self-harm/abuse substances/clean obsessively, you are morally responsible for that.

Note that (2) carries an implication that (1) does not. It suggests moral fault. But behaviors like self-harm, substance abuse, and obsessive rituals can be damaging to the person without necessarily damaging others. (If this is not obvious, imagine that the behavior is entirely private, all effects kept hidden from view.) They are not sins, or unequivocally and inherently morally wrong. Whatever responsibility service users have for such behavior, it is neither clinically helpful nor obviously correct to view it as moral. We are responsible for behavior that is morally neutral as well as morally good or bad, and, in clinical and other contexts that support change and reflection, there is point in emphasizing this.

Relatedly, the weaker, linguistic link obscures that fact that service users can be responsible for harm, but not blameworthy, because they have an excuse. Compare:

1. Service users may be responsible for verbal aggression toward clinicians but not blameworthy, because they are acting to relieve high levels of psychological distress, and lack alternative coping mechanisms.

2. Service users may be morally responsible for verbal aggression toward clinicians but not blameworthy, because they are acting to relieve high levels of psychological distress, and lack alternative coping mechanisms.

Note that (2) does not ring true to native ears. And for good reason. How can it make sense to be morally responsible for behavior but not blameworthy for it? Both moral responsibility and blameworthiness imply moral fault. This is what the explanation appealing to psychological distress and lack of coping mechanisms excuses, despite the fact that responsibility for the aggression yet remains.

Turn now to the stronger link. On this view, the idea of responsibility is constitutively connected, via our practices of holding others responsible, either to the reactive attitudes themselves, or to a belief about their aptness. This link makes the possibility of responsibility without blame not simply obscured, but nearly incoherent. If holding someone responsible for harm just is responding with a reactive attitude like blame, then it is not possible to hold service users responsible for harm without blaming them. Similarly, if holding someone responsible for harm just is believing that blame would be an appropriate or fitting response, then, although one may not oneself be blaming them, one is hardly adopting the blame-free, nonjudgmental stance necessary for effective clinical treatment. In practice, one might as well be blaming them, for one believes that one should. In essence, a view of responsibility that links it so closely to the reactive attitudes is not adequate to account for the clinical practice of holding service users responsible for behavior that causes harm, without blaming them for it. For according to such a view, blaming is too much a part of what it means to hold another responsible for there to be sufficient room to maneuver between them.

The moral of this discussion is that a conceptual framework that is adequate to account for clinical practice must clearly distinguish between ideas of responsibility, blameworthiness, and blame.⁵ Let us begin with responsibility.

Effective clinical treatment presupposes that service users are responsible for their behavior insofar as they have conscious knowledge of what they are doing, and can exercise choice and at least a degree of control over the behavior. As we saw, this is a traditional and common sense idea about what it means to be responsible, applicable not only to service users, but to us all.⁶ This idea of responsibility is essentially linked, not to morality and the reactive attitudes, but to agency. Crucially, on this view, we are responsible for all our actions, whether or not they are right, wrong, or neutral from a moral point of view. We are responsible for our actions because we are their agents: insofar as we know what we are doing, and can exercise choice and control our behavior, what we do it up to us.

With this idea of responsibility in mind, it is then possible to understand what, minimally, is involved in holding a person responsible. Most stringently, holding a person responsible may consist simply in judging that they are responsible, that is, that they have conscious knowledge, choice, and a degree of control of their behavior. Usually, however, the idea of ‘holding responsible’ means more than judging others to be responsible, but actually treating them thus: treating them as accountable or answerable for their behavior. What accountability or answerability consists in varies widely, depending on the context. Within clinical practice, holding a PD service user responsible for their behavior may involve asking them to explain why they made the choices they did, and encouraging them to behave differently in the future. Alternatively, it may involve the agreed imposition of negative consequences, to increase motivation, and show that the behavior, and the harm it causes, is taken seriously.

But, as the discussion of reactive attitudes makes clear, the idea of holding another responsible can involve more. It can involve judging a person not only to be responsible and therefore accountable for the behavior, but to be blameworthy, and indeed blaming them. So, let us turn now to blameworthiness.

We judge a person to be blameworthy when they are responsible for harm, and have no excuse. Excuses come in various kinds, such as bad luck, limited choices, and the intention or quality of will behind the action. As suggested above, service users who are responsible, at least to a degree, for harm to self or others may not be judged blameworthy, because they have an excuse, such as limited choices, or levels of psychological distress that we do not expect people to tolerate without taking action to alleviate it. However, sometimes they do not have an excuse.⁷ Clinicians may turn a blind eye to this, but equally, they may not: they may recognize that a service user is not only responsible, but blameworthy. However, they may still manage to avoid blame and maintain an effective clinical stance.

Distinguishing responsibility and blameworthiness is important to solving the conundrum; it allows us to see both how it is possible to be responsible, and treated thus, for actions that are not morally wrong; and how it is possible to be responsible, and treated thus, for actions that are morally wrong but for which one is not blameworthy, because one has an excuse. But we are yet left with the problem of how it is possible for clinicians to hold PD service users responsible for harm for which they are recognized to be blameworthy, and yet not to blame them. To resolve this, we need to understand what blame is.

Blame⁸

Philosophical accounts of blame are surprisingly few and surprisingly diverse, but they tend to agree on one thing. Blame carries a characteristic ‘sting.’ Being the object of another’s blame hurts. Capturing the ‘sting’ of blame is thus a constraint on any adequate account of what blame is. The ‘sting’ is also the reason why blame is so clinically counterproductive. Effective treatment is not possible if the service user feels judged, shamed, berated, attacked, or hurt.

But talk of blame is often ambiguous. When we say that another is ‘to blame’ we may mean one of three things:

1. They are blameworthy.

2. We should blame them.

3. We actually do blame them.

These three propositions are distinct. The first is a judgment about another. Whatever the conditions of blameworthiness ultimately are, they meet them. It is possible to make such a judgment about another, without also judging that we should blame them, let alone judging that we actually do. For instance, we might judge a historical figure from the distant past blameworthy for harm perpetrated, but we neither blame them, nor judge that we should—the harm is too far removed.

The second is about us and what we should do. In this kind of context, ‘should’ can have three different meanings. First, we may be saying nothing more than that blame is warranted or justified: we should blame another, because they are blameworthy. If so, (2) collapses into (1). Second, we may be saying that blame is appropriate, relative to various norms governed by the nature of our relationship with the other and the circumstances. For instance, it may be appropriate for victims to blame perpetrators for harm, when it is not appropriate for legal advocates to do so. Third, we may be saying that blame is desirable relative to a given end: whether or not blame is warranted or appropriate to our relationship and the circumstances, perhaps it would do us psychological good to vent, or perhaps it would serve an instrumental purpose, such as deterrence. In all three senses, it may be true that we should blame another, and yet we find that we do not. Perhaps we are simply too weary of battling or teaching this person, or fighting for social good: we are beyond caring at this stage to muster the energy to blame.

Finally, the third proposition is about us and what we actually do. Often enough, we feel blame toward others when we both judge them blameworthy and judge blame appropriate and desirable. But not always. Blame, like nearly all emotions, can be irrational. A moment’s reflection on the vicissitudes of personal and family relation-ships should be sufficient to establish this. When things go wrong for us, especially within long-standing personal relationships, but elsewhere, too, we often look for someone to blame, whether as a way of avoiding responsibility ourselves, or simply as a way of venting our frustrations. We sometimes blame others even when we know that the person we are blaming is not at fault, and that we should not: ‘I know it’s unfair, they don’t deserve it, but I can’t help blaming them. I’m just so angry!’

In this respect, it may be helpful to compare blame and fear. It is one thing to judge a situation dangerous. It is another to judge that fear is warranted, appropriate, or desirable. And it is another again actually to feel it. The brave soldier judges a situation dangerous, so that they can respond rationally and effectively in battle. But they do not feel fear. Nor do they judge it appropriate or desirable that they should: given their role and aim, better they should not. In contrast, the well-informed British arachnophobe feels fear even when they judge there is no objective danger or reason that they should: it is neither warranted, appropriate, or desirable to be pathologically afraid of UK spiders. Blame is like fear. It can fly in the face of considered judgments about what is true of the blamed object, and what should be true of the blaming subject.

Past philosophical accounts of blame have tended to draw on one of two ideas. Either blame is a form of punishment (Smart 1961). Or it is a sort of mental ledger or record of a person’s behavior, of use in assessing character or predicting behavior (Feinberg 1965; Glover 1970). Neither idea suffices to account for blame. As has often been remarked, blame is a mental state, and punishment is an action (Boyd 2007). It is perfectly possible to blame someone but not show it at all, let alone act so as to punish them. Similarly, it is unclear what a mental ledger or record of a person’s behavior is supposed to be, other than a memory that they did it. Moreover, we can blame people for actions we consider one-off, and would not use to assess their character or predict future behavior.

Partially in response to these deficiencies, more recent philosophical accounts of blame have focused instead on the idea that blame is essentially if not exclusively cognitive: a form of consciously accessible, personal-level judgment or belief. For instance, Hieronymi (2004) suggests that blame is the judgment that a person has shown disregard or ill will toward another. Or again, Sher (2006) suggests that blame is the belief that a person has acted badly or has a bad character, in conjunction with a desire or wish that this were not the case. Finally, Scanlon (2008) suggests that blame is the judgment that a person is blameworthy, and so has shown impaired interpersonal attitudes, which renders appropriate the revision of one’s own attitudes toward them, especially one’s intentions.

Such cognitive accounts struggle to capture both the irrationality and the ‘sting’ of blame. Consider first its irrationality. As we saw above, blame, like most reactive attitudes and emotions, can fly in the face of judgments or beliefs that a person is blameworthy (for ease of exposition, I shall ignore the differences in precise content of the various judgments and beliefs suggested, and refer to them all as judgments of blameworthiness). If irrational blame is possible, these cannot be necessary, let alone sufficient, conditions of blame. This is not to deny that blame, like fear and other occasionally irrational emotions, can clearly involve subpersonal representations, potentially of threat, harm, slight, or ill will, at some level of information-processing. The cognitive psychology of emotional information-processing is not yet unified and advanced, but theories are developing that aim to explain the varieties of rationality and irrationality, consciousness and unconsciousness, that characterize emotions (for a review of the relevant science see Dalgleish and Power [1999] and Lane and Nadel [2000]; for discussion connecting the science to more standard philosophical concerns, see Prinz [2004]). But, crucially, the representations posited to accommodate these aspects of emotions are not consciously accessible, personal-level judgments and beliefs.

Consider next blame’s characteristic ‘sting.’ One complicating factor is that individual differences in temperament and values mean that there can be no universal claims about what does and does not ‘sting.’ Some people are more sensitive than others, and some people care more about interpersonal relationships, rights, and wrongs, than others. This point is especially important with respect to PD. For instance, a service user with low self-esteem and a critical superego may be easily ‘stung’ by blame, whereas a narcissistic psychopath may be more immune. Assessing the extent to which an account of blame captures its ‘sting’ is thus the task of assessing the extent to which an account of blame captures what commonly or prototypically ‘stings.’ Disagreements are clearly possible. Nonetheless, there is good reason to hold that personal-level cognitive accounts will not adequately capture this. For judgment and belief are commonly, indeed arguably prototypically, ‘detached.’

Note that, as discussed above, we can judge or believe that a person, such as a historical figure, is blameworthy, even if we neither do nor judge that we should feel anything. Furthermore, the addition of a desire or wish that this not be so need not make the attitude any less detached. But, even when the judgment or belief is about a person with whom one is presently in relation, they may not ‘sting.’ They may be formed and expressed in a way that does not hurt or harm. For instance, good parenting routinely involves pointing out when a child has shown disregard or ill will toward a sibling, and indeed imposing negative consequences for it. That is part of bringing up children to treat others, including rivals, with regard and respect. Sometimes, no doubt, parents do this in such a way that the child feels bad and blamed. But a loving parent can often help a child to understand that their behavior toward a sibling is neither decent nor permitted, without the child feeling ‘stung.’ Furthermore, this ‘detached’ mode of forming and expressing judgments of blameworthiness can be maintained even in face of revision of interpersonal attitudes and intentions. For instance, one can rationally and politely decide to stop socializing with an acquaintance who routinely offends because one judges them blameworthy and no longer wishes to see them, without either party minding very much. This is importantly different from a situation where one party acts out of anger, writing the other off, whether justifiably or not, without due thought or consideration. There is no doubt that judgments and beliefs of blameworthiness, and the changes in relationship they license, can ‘sting.’ The point is that they can also be ‘detached.’ Presence or absence of ‘sting’ depends not only on the temperament and values of the blamed, but also on the exact nature of the change in attitude and intention of the blamer, and, moreover, how this change is experienced and expressed.

Irrationality and ‘sting’ are both secured by the same thing. The reason why blame can be irrational, and the reason why it hurts, is that it is a reactive attitude, a kind of emotion. Call blame which has these features ‘affective blame’ (cf. Smith [2007] on ‘active blame’). We need not deny that we sometimes speak of blame in a more ‘detached’ mode. As we saw, judgments that another is ‘to blame’ are ambiguous. Call this non-stinging sort of attitude ‘detached blame.’ Detached blame can consist in a judgment or belief of blameworthiness. It can be accompanied by a revision of attitudes or intentions, or a further belief that such revision would be appropriate. It can also be accompanied by the imposition of negative consequences for the action, or a just demand for accountability or answerability. The point is that it need not have any of blame’s characteristic ‘sting.’ ‘Sting’ is commonly and prototypically secured by negative affect and the potential it has to be expressed and acted on. It is affective blame that really hurts.

But there is a challenge facing this suggestion. Grant that the ‘sting’ of blame is affective. We now face the question: what kind of affect? For, it seems that affective blame can consist in a range of different emotions. Most obviously, these include hate, anger, and resentment. But the range can plausibly be extended to include certain other states that have an affective dimension without being uncontroversially identifiable as types of emotion, for instance, disapproval, dislike, disappointment, indignation, and contempt. Moreover, as expected given this range, blame’s expression can be equally various, for instance, alongside punishing, blame can also be manifest in berating, attacking, humiliating, writing off, rejecting, shunning, abandoning, and criticizing, to name but a few behaviors. The challenge is thus to unite these various emotions and manifestations thereof into a single account of blame. Each kind of reaction can occur without counting as an instance of blame. So we must explain what makes these various reactions count, when they do, as instances of blame.

It is natural to be tempted by the idea that they are united in virtue of being caused by the judgment or belief that a person is blameworthy. But this cannot be right. For, as we saw, blame can be irrational: one can blame someone in absence of such a judgment or belief. Instead, I want to suggest that the phenomenology of affective blame provides a cue. Part of what is distinctive about blame is that, when in its grip, one feels entitled to one’s blaming response, because of what the other has done: it feels as if they deserve it, even if one does not judge or believe that they do. This feeling of entitlement—of being in the right, in relation to another’s wrong—is the key to unifying affective blame. What makes a negative emotion in reaction to another count as blame is the second-order response the blamer has to their first-order emotion: their feeling of entitlement. This feeling of entitlement places the responsibility for the blaming response on the blamed. The blamer feels entitled to their first-order emotion because of what the blamed has done. It thereby gives the blamer a (defeasible and resistible but nonetheless genuine) feeling of freedom to express blame, vent, and act out of whatever negative emotion they are experiencing. The blamer acts as if, because of what the other has done, the first-order emotional reaction is deserved. In this way, although blame is not an action and so not a form of punishment, it is a punishing mental state: in reacting negatively, one feels oneself to be in the right in relation to another’s wrong.

It is important to recognize that this feeling of entitlement is not a judgment. We must eschew a consciously accessible, personal-level cognitive account of emotion at the second-order as well as the first. Rather, whatever the mature, agreed theory of the information processing underlying first-order emotions turns out to be, we need to import this understanding to the account of blame offered here. This is important, if we are to account not only for the ‘sting’ of blame, but also for its potential irrationality. For, just as one can judge that spiders are not dangerous and yet feel fear, so too one can judge that another is not blameworthy and yet not only feel anger, but also feel entitled to this anger. One can feel this, even though one knows one should not.

Of course, if the blamer views their blame as irrational and exercises their capacity for rational reflection, they may try to suppress the first-order emotion and control their behavioral tendencies. Alternatively, in the grip of the feeling, they may not. But what makes an instance, say, of anger toward another into blame, is that the blamer cannot lose the feeling that they are entitled to be angry, even if they judge that this anger is not ultimately deserved.

With the distinction between detached and affective blame in hand, we can now complete the conceptual framework, and solve the conceptual part of the conundrum. Clinicians are able to hold PD service users responsible, indeed blameworthy, for harm, without blaming them, because blame comes in two forms: detached and affective. Detached blame consists in judgments of blameworthiness, and may further involve correspondingly appropriate revisions of intentions, the imposition of negative consequences, and accountability and answerability. These can have a place within effective clinical treatment, and, insofar as they encourage responsible agency, may be essential to it. Affective blame consists in negative reactions and emotions, whether rational or not, that the blamer feels entitled to have. Effective treatment requires clinicians to avoid affective blame. Responsibility without blame is responsibility without affective blame: without a sense of entitlement to any negative reactive attitudes and emotions one might experience, no matter what the service user has done.

Impoverishment and Empathy

Part of the solution to the clinical conundrum is conceptual: we need a framework that clearly distinguishes responsibility, blameworthiness, and blame, to understand how it is conceptually possible to hold PD service users responsible for harm without blaming them. But part of the solution is practical: it is not sufficient that it is possible to avoid affective blame. Clinicians must actually manage to do so.

Clinical training and experience provide some skills that help with this task. Clinicians learn a way of speaking, which involves both a repertoire of phrases and an attitude of calm respect, that helps them to both think and speak with service users about their responsibility for harmful behavior, without blaming them. Clinicians also develop their own capacity to take responsibility for their own emotions: to reflect deeply on whether their response to a service user is warranted or necessary or even natural, and to ‘own’ their part in interpersonal engagements. Bearing in mind the nature of their relationship with service users, and the inherent power imbalance between them, no doubt further aids this task: compare again, in this respect, the nonjudgmental attitude loving parents show children. Finally, when all else fails, clinicians need a good poker face—a commitment and capacity to mask their emotions and refrain from acting out of any blame they may feel.

But, alongside these various skills, clinicians must cultivate compassion and empathy for service users (cf. Potter 2009). Quite generally, compassion and empathy are central to good therapeutic care (Gilbert 2010). They are essential when working with service users with PD. The reason is simple: a compassionate, empathetic stance is at odds with a blaming stance. Compassion and empathy push the negative emotions constitutive of affective blame aside. They simply cannot comfortably coexist.

One central way that clinicians can achieve compassion and empathy toward service users is simple: proper attention to service users’ past history. As is well known, PD is associated with dysfunctional families, where there is breakdown, death, institutional care, and parental psychopathology; traumatic childhood experiences, with high levels of sexual, emotional, and physical abuse or neglect; and social stressors, such as war, poverty, and migration (Paris 2001). PD service users often come from harrowing backgrounds, impoverished of all goods, to an extent that can be unimaginable to people who have not experienced these kinds of conditions. Effective treatment can involve helping PD service users to explore their past and recognize its effects on their personality and their present experiences and behaviors, both as a way of coming to terms with the past, and as a way of developing skills needed to better manage the present. But, in attending to service users’ past history, clinicians and service users together gain understanding of why the service users are as they are. A fuller life story or narrative comes into view, in which the service user is seen not only as one who harms, but as one who has been harmed. As Watson has put this point in relation to the psychopath Robert Harris: “The sympathy towards the boy he was is at odds with outrage towards the man he is” (2004, 244). Attention to service users’ past history is not only part of effective treatment. It also has the power to help clinicians to strike a balance between rescue and blame. It requires clinicians to keep in mind the whole of the person and the whole of their story, which undercuts a single, reactive stance, forcing affective blame to exist alongside compassion and empathy, and thereby at least reducing, if not outright extinguishing, its force.

It is important to recognize that this appeal to past history does not eliminate responsibility or blameworthiness (cf. Watson 2004). It may reduce responsibility, insofar as certain kinds of background impede the development of skills that, for instance, facilitate emotional regulation and, correspondingly, behavioral control. Equally, extreme impoverishment can limit choices, which can sometimes excuse bad decisions and the harm they cause. But such reduction is not global, and depends on the particular kind of background, skills, choices, and harm in question. Rather, the compassion and empathy that consciousness of past harm arouses directly quells and tempers affective blame. It acts as an antidote.

Effective treatment for PD is possible, but it presents a heavy burden for service users and clinicians alike. Service users must face aspects of their personalities and behavior that are harmful to self and to others, and take responsibility for them. That process typically involves painful self-reflection, and the potential for self-blame, shame, and guilt. Clinicians must themselves strike a difficult balance, encouraging responsibility, offering help and support, and tolerating harmful behaviors without succumbing to affective blame. A conceptual framework that clearly distinguishes responsibility, blameworthiness, and blame, can help both service users and clinicians in this work. But so too can clinical attention to service users’ past history, for the understanding of the person it offers, and the antidote to affective blame it provides.