Blameworthiness for Non-Culpable Attitudes

ABSTRACT

Many of our attitudes are non-culpable: there was nothing that we should have done to avoid holding them. I argue that we can still be blameworthy for non-culpable attitudes: they can impair our relationships in ways that make our full practice of apology and forgiveness intelligible. My argument poses a new challenge to indirect voluntarists, who attempt to reduce all responsibility for attitudes to responsibility for prior actions and omissions. Rationalists, who instead explain attitudinal responsibility by appeal to reasons-responsiveness, can make sense of blameworthiness for non-culpable attitudes. In response, voluntarists could propose a revision of our actual practices. This would lead us into a quite different debate.

KEYWORDS:

• blameworthiness
• responsibility
• attitudes
• belief
• voluntarism
• rationalism
• reasons
• forgiveness
• apology

1. Introduction

We show reactive sentiments in response to our own and other people’s attitudes. We sometimes feel guilty about not intending what we believe we ought to do. We might resent someone for wishing us harm. And we can feel hurt by what others think and feel about us. According to Strawson [1962], such reactive sentiments reveal that we hold each other responsible. Since we show them in response to attitudes, it seems that we don’t merely regard each other as responsible for actions and omissions, but also for attitudes.

However, philosophers have argued that we don’t control our attitudes as we control our actions: we cannot choose what we believe, feel, desire, or intend.¹ Indirect voluntarists therefore argue that responsibility for attitudes can at most be indirect. We can control beliefs by inquiry, emotions by going for a walk, and desires and intentions by deliberating about what is good and right. For the indirect voluntarist, the observation that we show reactive sentiments towards attitudes reveals merely that we regard each other as blameworthy for not properly managing our mental life. It allows us to conclude only that we can be indirectly responsible for attitudes in virtue of being directly responsible for actions and omissions that caused the attitude.²

There has recently been opposition to voluntarist accounts, coming from epistemology and moral psychology. The common theme is that voluntarists misconceive of attitudes as if they were nothing but brute states that can be managed indirectly—like headaches that we can manage by taking a painkiller. Yet our attitudes, like our actions, are often direct responses to reasons, and we are evaluated as rational, or as irrational, in light of them. It therefore seems that responsibility for attitudes is as direct as is responsibility for actions. Rationalist accounts, which explain responsibility for attitudes by appealing to their reasons-responsiveness, develop this idea.³

However, voluntarists insist that genuine responsibility presupposes voluntary control, and that anything that rationalists talk about is therefore not direct responsibility for an attitude: either it is merely rational evaluation of an attitude, and thus doesn’t even amount to responsibility, or it is derived from responsibility for prior attitudinal self-management, and thus doesn’t amount to direct responsibility for an attitude.⁴

This paper proposes a way out of this stalemate by considering our practice of apology and forgiveness. I argue that this practice is sometimes fully intelligible when a person holds a ‘non-culpable’ attitude (NCA)—that is, an attitude that the person had no duty to avoid by prior actions or omissions. This reveals that we sometimes regard each other as morally blameworthy for NCAs. Voluntarists cannot reduce this blameworthiness to mere negative evaluation or to indirect blameworthiness: the latter strategy fails because the person had no duty to avoid the attitude; and the former strategy fails because our practice of apology and forgiveness makes sense only under the presumption of genuine moral blameworthiness. It follows that voluntarism is a false theory of how we regard each other as responsible. Voluntarists therefore shouldn’t understand themselves as proposing an analysis of our practices of holding responsible, but a revision of these practices. This places the argumentative burden on voluntarists, and it changes the nature of the debate.

I first characterize NCAs and frame my discussion (section 2). I then present my argument against indirect voluntarism (section 3). Next, I argue that a rationalist account can make sense of blameworthiness for NCAs (section 4).

2. Non-Culpable Attitudes and Reactive Sentiments

I introduce NCAs (section 2.1) and discuss what blaming responses can be appropriate towards NCAs (section 2.2). I frame my discussion by arguing that, although non-culpability should affect the intensity of reactive sentiments, it remains open whether NCAs can impair relationships in such a way as to warrant blaming responses. Section 3 argues that NCAs are relationship impairing in the relevant way.

2.1 Non-Culpable Attitudes

I will work with the following definition.

NCA. An attitude A of a person S is non-culpable iff S did not possess decisive practical (that is, prudential or moral) reasons to engage in practices of attitudinal self-management (that is, actions and omissions with foreseeable effects on attitudes) that probably would have led S to not holding A.

There are also non-culpable absences of attitudes. For instance, S’s not believing that p is non-culpable when S lacks a true belief about p because S didn’t inquire into p, but had sufficient reasons not to inquire into p, because S had more important things to do. Furthermore, aspects of an attitude can be non-culpable even if the attitude is culpable. For instance, S might culpably fail to manage their anger but then experience an uncontrollable increase in the anger’s intensity due to further provocations. For simplicity, I focus on cases where people hold an attitude that is fully non-culpable, and that still has problematic consequences for themselves or others. In these cases, the attitude could probably not have been avoided by engaging in reasonable practices of attitudinal self-management. That is, there was no course of action that S ought to have performed that would probably have led S to avoid the attitude. That makes the attitude non-culpable.

I employ the notions of ‘ought’, ‘reason’, and ‘duty’ (also ‘justification’, ‘permission’, and ‘allowed’) in their subjective or perspectivist sense. For I am interested in possessed reasons: violating objective ‘oughts’, such as when you fail to leave a burning house because you have no clue about the fire, doesn’t give rise to blameworthiness. By contrast, it is at least closer to a sufficient condition on blameworthiness that the reasons that explain why it was true that you ought to have done something that you failed to do were, in a sense, possessed by you when you violated the duty. This is the use of ‘ought’ that is central to discussions about responsibility and blame. Using the term in this perspectivist sense doesn’t commit me to any view about which use of ‘ought’ is the ‘central deliberative ought’.⁵

Note that most of your attitudes are non-culpable. It is seldom true that you had decisive reasons to avoid an attitude by engaging in self-management practices. Such practices are normally not worth the effort. Most obviously, you had no duty to avoid most rational attitudes and attitudes with good consequences. Furthermore, beliefs that you formed reflectively by means of careful inquiry into an issue, but also beliefs that you just acquired spontaneously by moving around in the world, are mostly non-culpable: you had decisive reasons to acquire reflective beliefs, and spontaneous beliefs were mostly unforeseeable, because you normally don’t know what you will encounter.⁶ By contrast, suppose that you decide to call your ex-partner, who recently broke up with you, because you want to know whether they have a new romantic partner. It is reasonable for you to avoid the relevant belief and to avoid feeling the associated emotions (like sadness or anger) by not talking to your ex-partner. In such cases, acquiring the attitudes is culpable, and you will be (at least prudentially) blameworthy. Similar considerations hold for emotions, desires, and intentions: while many are reflectively and responsibly acquired, or arise unforeseeably and spontaneously in our everyday life, and were thus acquired non-culpably, other attitudes should have been avoided by various strategies of self-management and are thus culpable.

Among NCAs are deeply ingrained implicit biases and problematic emotional patterns that we display due to our education and socialization. For instance, if an adolescent is amused by a sexist joke, this amusement might stem from an implicit bias, and the emotion and the laughing might well have come about so spontaneously that both were non-culpable, while the implicit bias was so ingrained that the young man had not yet had opportunities to reasonably get rid of it by exercising control over his own mind. The argument that I develop here will allow us to say that the adolescent could be blameworthy for his bias and amusement (given certain conditions) even though he had no duty to get rid of the bias before now.

More generally, we can harm each other with behavioural responses that stem from our attitudinal dispositions that are deeply ingrained, and that are thus often non-culpable. Becoming clear about which (if any) blaming-responses to adopt when we harm each other in these non-culpable ways will help us to see how we should relate to one another.

2.2 Blameworthiness and Reactive Sentiments

To see what it could mean to be blameworthy for a NCA, consider a case from Smith [2005: 267–8]. While Abigail adopted a racist ideology through growing up in a racist environment, Bert grew up in a tolerant family but then later reflectively endorsed racist attitudes. According to Smith [ibid.: 268],

understanding of the circumstances in which a person’s evaluative tendencies were formed may … have a very important influence on the kind or degree of moral criticism we think it appropriate to make. We can appreciate how difficult it might be for Abigail to come to recognize the viciousness of her own evaluative judgments, given their early entrenchment in her psyche, and also how difficult it might be for her to modify these judgments once their viciousness is recognized. For this reason, we are likely to be less critical of Abigail than we are of Bert, who adopted his racist-intolerant commitments in a fully reflective way (after being exposed to the morally appropriate values of tolerance and inclusiveness).

On Smith’s rationalist account, Abigail is blameworthy for holding racist attitudes because they reflect vicious evaluative judgments. Voluntarists baulk at the idea that a person who lacked reasonable opportunities to get rid of an attitude can be blameworthy for it. Even though Abigail’s attitudes are morally bad, voluntarists regard it as unfair or inappropriate to blame her for attitudes that she acquired merely by being raised in a certain way. Smith’s case can elicit the intuition that we cannot be responsible, and thus that we cannot be blameworthy, for NCAs. Note that to ‘appreciate how difficult it might be for Abigail’ to recognize that her attitudes are morally problematic is to acknowledge that Abigail, due to her distorted perspective, had no duty to engage in attitudinal self-management that would have led her to non-racist attitudes. Her racism is, up to a point in her life, non-culpable. Voluntarists will argue that, although Abigail might become blameworthy for these attitudes when she gains access to information that casts doubt on her racist outlook, which will provide her with reasons that will then create an obligation for her to actively reconsider her view (by inquiry and deliberation), she is blameless right now.

Smith’s case suggests that the degree to which a person had opportunities to engage in reasonable attitudinal self-management (that is, the degree of culpability of an attitude) is proportional to the degree to which it is appropriate to blame them. Even Smith acknowledges that we respond differently to Abigail and to Bert. But if the emotional intensity of, say, our indignation diminishes with fewer reasonable opportunities to change, then it’s natural to suppose that our indignation won’t be appropriate when there were no reasonable opportunities for the person to change. Intuitively, therefore, it seems that we are blameless for NCAs.

However, this becomes less convincing if we see that blame need not involve strong feelings of resentment, indignation, or guilt. As Smith [2013: 32] points out, some reactions deserve the label ‘blame’ without involving passionate components. In particular, blaming loved ones for moral failures commonly happens without indignation. Furthermore, we can modify a relationship by ‘dispassionately “unfriending” someone on one’s Facebook page, for example, or by simply refusing to trust anymore, and these too should qualify as blame’ [ibid.]. Proponents of such an account of blame argue that blame’s primary function is to mark impaired relationships [Scanlon 2008]. Regarding one’s relationship to someone as impaired because the person failed to show proper regard just is the blame [Hieronymi 2004].

Employing this notion of blame, we could reply to the voluntarist that it is appropriate to treat one’s relationship with Abigail as impaired due to her racist attitudes, and that she is therefore blameworthy, even if indignation might not be appropriate. Alternatively, one could agree that indignation appropriately diminishes when we realize that an attitude was less culpable than we thought. However, this observation doesn’t imply that indignation becomes fully inappropriate in response to NCAs. Milder forms of resentment or indignation might still be appropriate.⁷

Could voluntarists explain the verdict that Abigail is blameworthy? It doesn’t seem so. For if someone’s reasons against an action are not decisive—if the action isn’t forbidden—then the action is allowed. But it is incoherent to blame someone for doing something permissible (see Kiesewetter [2017: 29]). Since Abigail lacked decisive reasons to avoid her racist attitudes, she was permitted to let them develop, and is thus blameless for not managing her mental life better. Furthermore, even if we accept that one can be blameworthy for doing something that was allowed, note that Abigail was unaware that she was becoming racist, and avoiding becoming racist in her social environment is difficult. She therefore didn’t even have very strong reasons to avoid her NCA. I will argue that she can still be blameworthy in such cases. Voluntarists cannot explain this, even if there were some blameworthy permissible acts.

Voluntarists must then argue that Abigail isn’t blameworthy: the relationship modification that is appropriate, due to Abigail’s racist attitudes, is not a kind of blame. I agree that we can modify relationships without blame, as, for instance, when people drift apart when they live in different places. So, the fact that we can appropriately modify our relationship towards Abigail does nothing, by itself, to show that she is blameworthy.

In reply, I argue that NCAs sometimes make a specific kind of relationship modification appropriate that counts as genuine blame. The literature on blame contains proposals about what turns a relationship modification into blame.⁸ I will employ a sufficient condition that is compatible with most proposals: if a relationship modification makes the full practice of apology and forgiveness intelligible, then it’s blame. I’ll argue that, since a person’s NCAs can make this practice intelligible, we can be blameworthy for NCAs. This poses a new problem for voluntarists.

3. A New Argument against Indirect Voluntarism

I will now argue that NCAs give rise to our full practice of apology and forgiveness, and that they therefore can impair our relationships in the way required for moral blameworthiness. I first introduce the cases that will be counterexamples to voluntarism (section 3.1). I then defend my argument in three steps (sections 3.2–3.4): first, it is appropriate for the protagonists to apologize for a harm that they have caused; second, the specific kind of apology makes the full practice of forgiveness intelligible, thereby revealing that we regard the protagonists as blameworthy; third, this blameworthiness is best understood as blameworthiness for their NCA.

3.1 Blameworthy Non-Culpable Attitudes

This section presents cases in which a NCA manifests in non-culpable behaviour that causes harm to others. I will make the following assumptions in each case.

1. The person (S) holds an attitude (A) that they acquired through a process in which there was nothing that they should have done to avoid holding it: that is, A is non-culpable.

2. S displays a behaviour (B) that was (partially) caused by A and there was nothing that S should have done to avoid B: that is, B is also non-culpable.

3. A is not pathological, but instead is responsive to reasons to some degree.

4. A is unjustified: that is, it is insufficiently supported by S’s possessed reasons for A.

As I will argue, each of the following cases is a counterexample to the voluntarist’s view that we are blameless for NCAs if we assume (a)–(d):

1. John (S) was raised with the sexist belief that men are supposed to lead (A). As a result, he often treats women unfairly in job interviews (B).

2. Sonja (S) is under stress this morning. She couldn’t reasonably avoid the stressful situation. As a result, she has no time to think clearly about the hairs that she discovers in the sink. She forms the unjustified belief that they are from her roommate Sarah (A). She feels angry at Sarah, but she manages her behaviour towards Sarah to the best of her abilities. Yet Sonja cannot manage to avoid behaving in an unfriendly way to Sarah and leaves the house with a suspicious look at Sarah that displays an element of contempt (B).

3. Tim (S) intends to become a successful entrepreneur in the oil and gas industry (A). He idolizes his father, who leads such a company. In his twenties, Tim becomes head of the company, and leads it with success (B). However, he gains evidence that his company is destroying our planet. Since his intention is constitutive of his identity, it is very difficult for him to appreciate this evidence, and to revise his intention.

4. Carla and Jack (S) are visiting their daughter. Upon noticing the state of her garden, Carla and Jack wish to give her a lesson about proper gardening (A). While they can reflectively avoid giving such a lesson, this desire poisons the social interaction (B). Their daughter notices the desire.⁹

5. Ramona (S) is a successful leader in higher management. Due to her stressful job, she has become irascible. On one particularly stressful day, she becomes angry at a subordinate (A) without any offense, and she behaves in a very unfriendly way (B).

The cases cover non-culpable beliefs, intentions, desires, and emotions. Notably, it is impossible to describe cases that involve merely one kind of attitude. For if an attitude is responsive to reasons (condition (c)), it enters into rational relations with other attitudes that the person holds. Tim’s intention will come with a belief that he ought to become a successful entrepreneur, and Carla’s and Jack’s desires will lead to certain emotions that influence their behaviour. There is room in each case to argue that blameworthiness for one kind of attitude is more basic than blameworthiness for another. For instance, maybe blameworthiness for a particular emotion traces back to blameworthiness for a belief—which is plausible in (2). However, the position that I sketch in section 4 will imply that we can be directly responsible for any attitude that is sufficiently responsive to reasons.¹⁰ This is compatible with there being hierarchies of responsibility for attitudes.

Direct responsibility and blameworthiness presuppose some degree of reasons-responsiveness. A pathological phobia of spiders doesn’t indicate that the phobic is a coward, and therefore doesn’t, by itself, make the phobic criticizable in any substantial sense (although they might be criticizable for not having therapy). The protagonists (S) in (1)–(5) would be perceived differently if we learned that their attitudes aren’t responsive to reasons at all: they would become more like obstacles to deal with than persons to whom to relate. But can NCAs be reasons-responsive?

Remember that evaluating an attitude as non-culpable is to look at the practical reasons that the person had to manage this attitude, and then to judge that the person responded correctly to their reasons: their actions and omissions were permissible. That is, someone with a NCA complied with their duties of attitudinal self-management, at least with respect to that attitude. By contrast, when evaluating an attitude as reasons-responsive (that is, as non-pathological), we consider whether the attitude could now change in response to reasons. If rational discourse can have an influence on John’s non-culpable sexist views in an appropriate subset of close possible worlds, then his non-culpably acquired attitudes are sufficiently responsive to reasons to be rationally evaluable. As I will argue, his attitudes can then give rise to our full practice of apology and forgiveness, and thus to genuine blameworthiness.

I turn to condition (d) in section 4, which I think follows as a precondition on direct blameworthiness from a broadly rationalist account of responsibility. For now, I elaborate on (a) and (b).

By assuming (a), we conceive of our cases in such a way that S lacked decisive reasons to engage in attitudinal self-management that could have avoided A. For instance, in (1) and (3), we assume that the sexist belief and the capitalistic intention were so deeply ingrained that the young protagonists could not reasonably be expected to get rid of them by long and careful reflection. In cases (2), (4), and (5), we assume that the attitudes that were the immediate cause of unfriendly behaviour were either not foreseeable for S (say, Carla and Jack didn’t know that their daughter has a garden) or were tough to avoid for S (say, Sonja is sensitive about cleanliness, and Ramona needs things to be done in her way). Here we additionally suppose that S didn’t yet have reasonable opportunities to get rid of the underlying disposition that caused them to form A. That is, we assume that A is truly non-culpable.

Concerning (b)—the non-culpability of behaviour B—we need to assume either that B would be justified if A was justified, or else that B was too difficult to control by S, given A.

To illustrate the first assumption, we might suppose that if John’s sexist belief in (1) was justified, then it would also be justified to treat women in the way that he does in job interviews. Here the original moral fault seems to lie in his holding a sexist belief rather than in his behaviour. This also holds for Tim’s capitalistic intention in (3): leading his company to success would be justified if his intention was justified. Therefore, his behaviour itself is non-culpable: any blameworthiness for it must originate in some attitude or deed before the behaviour at issue occurred.

To illustrate the second assumption, we might suppose that, in (4), Carla and Jack cannot reasonably avoid being distracted from the conversation by the neglected flowers outside the window, thereby revealing their judgmental desire; or we might conceive of (2) and (5) in such a way that Sonja or, respectively, Ramona cannot reasonably control the involuntary and automatic expressions of her anger, given the other duties that she has right now, like getting ready for work, or, respectively, distributing tasks among her subordinates. Generally, the fact that it is difficult for a person to control a behaviour can make it the case that they lack decisive reasons to avoid the behaviour, thereby rendering it non-culpable.

3.2 Apologies and Excuses

My argument starts by noting that S owes an apology to the harmed party for B. Assume that Carla and Jack realize that their desire poisoned the social interaction, or that Ramona realizes how her irascibility together with the stressful situation led her to be unfriendly to her subordinates. They should then apologize. Providing merely an excuse would be inappropriate: explaining to the harmed party why they couldn’t do otherwise would not repair the impaired relationship (see Hieronymi [2019]). Only an apology can do this work. That is, an apology, rather than an excuse, is appropriate even though A and B were non-culpable. This is so, even if we assume that the harmed party knows about the non-culpability: they know how hard it is for S to change their views on a matter (in (1) and (3)), or to control their behaviour (in the other cases).

So far, this is merely an observation about how we expect each other to interact: we want apologies for harms that people caused, not just excuses. This holds even in cases where the person is blameless for the harm caused (see Sussman [2018: 788]). The second, more controversial, step of my argument is that the specific kind of apology that S owes to the harmed party implies that S is blameworthy. I now turn to this claim.

3.3 Apologies, Blameworthiness, and Forgiveness

We often apologize conventionally for behaviour for which we aren’t blameworthy. Conventional apologies range from harmless cases to severe ones. Among the harmless cases are apologizing for accidentally touching a stranger while sitting down on public transport and apologizing for hitting someone with a door after opening it with the normally expected amount of attentiveness and care. Here we are causally responsible for harm. But we didn’t violate any norms with which we were (normatively) expected to comply. There are severe cases with the same structure. You might hit someone by opening the door with normal care, and yet the person might get hurt so badly that they end up in hospital. An honest apology is appropriate. Providing an excuse wouldn’t satisfy our moral expectations: we demand that you apologize, rather than explain yourself, but we do not demand that you acknowledge any blameworthiness. Of course, you might become blameworthy for failing to apologize conventionally.

Thus, not all appropriate apologies imply blameworthiness. These conventional apologies differ from what I call, somewhat technically, authentic apologies: by apologizing authentically, one acknowledges blameworthiness for violation of a norm with which one was rightly expected to comply. In (1)–(5), S should apologize authentically for the harm they have caused, thereby acknowledging their blameworthiness for a norm violation. For now, I leave aside the nature of this norm violation; I return to it in section 4. Instead, I argue that S indeed owes an authentic apology to the harmed party.

How can we determine whether the apology that S owes is authentic? The initial problem is that one can apologize authentically in the same manner as one apologizes conventionally: both can be done very seriously, honestly, etc. It is therefore often impossible to distinguish between the apologies by looking at how they are physically conducted (which is why my label ‘authentic’ is somewhat technical). Instead, we need to look at the wider context of the practice in which authentic apologies are embedded. For only authentic apologies—those that imply blameworthiness—open the space for the full practice of forgiveness.

According to a prominent understanding, to forgive is to let go of resentment while still acknowledging that the offender committed a genuine wrong for which they are blameworthy (see Hieronymi [2001]). Forgiving is possible, normally, only if the offender realizes the wrong that they have committed as a wrong, apologizes, and feels remorse about what they have done. The offender changes the reflective stance on their behaviour. If the wrong is then indeed no longer a threat to the wronged person's standing within the moral community, a continuation of resentment wouldn’t be appropriate. Since this account explains paradigm cases of forgiveness, it will serve for my purposes.¹¹

The claim is that the full practice of forgiveness is pointless when the subject is blameless. If all involved parties know that your touching was unintentional, or that you opened the door with normal care, then there is nothing to forgive, because you are not to blame. It then doesn’t make sense for the harmed party to say ‘I forgive you’ and to really mean it. In the harmless cases, this sentence could be at best a joke. And, in the more serious cases, it would indicate that the situation was misunderstood. There is something to forgive only if there is someone to blame. Furthermore, it doesn’t make sense to not forgive you, or to postpone forgiveness. If no one was wronged, then there is no point in denying forgiveness, because there is nothing to forgive in the first place. By saying that the full practice of forgiveness is intelligible I mean that each of these moves—forgiving, denying, or postponing forgiveness—makes sense. Importantly, each move isn’t always justified: sometimes you ought to forgive, but then denying forgiveness is still intelligible.

I proceed as follows. First, I show that the full practice of forgiveness is intelligible in (1)–(5). Second, I argue that this implies blameworthiness, by replying to counterexamples. This will also clarify the distinction between authentic apologies and conventional apologies.

An important feature of my cases is that behaviour B reveals a problematic aspect of the person’s character or self. In (1) and (5), a person (the job applicant, the subordinate) was directly disrespected by S’s participation in sexist or elitist structures. In so far as S’s attitudes aren’t just cases of pathology, they reveal a feature of S’s self on which the disrespected must take a stance. They could, for pragmatic reasons, view S as a victim of their social environment, thereby avoiding the ‘strains of involvement’ [Strawson 1962]. But this would only be a strategic move in order to deal psychologically with the wrong. Alternatively, they could hold S accountable for S’s behaviour, deny S forgiveness, and thereby continue to view the relationship as impaired. Doing so would amount to moral protest (see Smith [2013]), and to emphasizing one’s dignity and worth (see Hieronymi [2001]), and therefore to blame. Similar responses can be appropriate to Tim in (3) if we imagine that he apologizes to the public after realizing how he deceived himself into thinking that business as usual wouldn’t contribute to planetary destruction. The public, and especially those most affected by climate change, can intelligibly deny forgiveness.

In (2) and (4), we can imagine that the protagonists decide to reduce their involvement with S even though S apologized. The daughter in (4) might not invite her parents over to her place for a while. By doing so, she could intelligibly hold that her parents aren’t just blameless victims of the conservatism of their generation. Instead, she might judge that the parents’ desire to lecture her about the garden reflects a character trait that negatively affects the relationship: the parents do not treat their daughter as a peer who is in charge of her own life. Plausibly, the parents display the vice of judgmentalism. Gary Watson [2013] points out that this entails non-acceptance, rejection, and the distance of superiority.¹² Plausibly, Sonja’s quick judgment in (2) manifests similar flaws in character. Sonja isn’t merely a victim of her stress, as the voluntarist makes the case appear. Rather, her behaviour reveals an aspect of her self. She thus needs to apologize, thereby acknowledging her blameworthiness. She then makes herself vulnerable by opening the room for Sarah to decide whether she will forgive her. Sarah’s continuation of blame might be manifested in her decision to move out (suppose that Sonja is often non-culpably unfriendly). This decision could be a response to Sonja’s self, rather than a mere strategic way of dealing with her.

In all of these cases, the full practice of forgiveness is intelligible. The second step of my argument is that the intelligibility of this practice implies that the person is blameworthy. The recent literature suggests two counterexamples to this second step in my argument. First, one might appeal to cases in which a person is merely causally responsible for a harm, but in which making amends still makes sense. Take Williams’s [1981] famous case of the lorry driver who non-culpably kills a child. The driver should feel ‘agent-regret’ and it seems intelligible that he apologizes to the child’s parents, even in full knowledge of his blamelessness. This apology doesn’t seem to be ‘merely conventional’, and the practice of forgiveness seems intelligible.

However, such cases don’t make the full practice of forgiveness intelligible. As Sussman [2018: 806–7] argues, only ‘quasi-apology’ and ‘quasi-forgiveness’ are appropriate. Such quasi-apologies ‘need not express any sort of change of heart, any resolution to act differently in the future’, and ‘although the driver should ask for forgiveness … , the parents should respond by telling him that none is needed’ [ibid.: 806]. The parents might feel overwhelmed after the accident, making it difficult to respond appropriately to the driver; doubts about whether the driver is blameless might plague the minds of all parties involved; and the driver should take up the parents’ blame although he is blameless: it isn’t the right moment to insist on excuses [ibid.: 800–1]. However, if the situation was transparent, and if the parents are no longer overwhelmed, then they should (quasi-)forgive the driver: his apology does not make him vulnerable to the parents’ judgment, because he is blameless. It therefore wouldn’t even be intelligible to deny him forgiveness. This is a crucial difference to the authentic apology that is appropriate in (1)–(5).¹³

Second, Driver [2017] argues that we sometimes morally ought to violate relationship-based duties. In her example, Agamemnon has a moral duty (we assume) to sacrifice his daughter Iphigenia for the greater good—the survival of his whole army. He ought to violate a pro tanto relationship-based obligation towards his daughter. For Driver, this is a resolvable tragic dilemma of blameless wronging. Yet Agamemnon should ask Iphigenia for forgiveness after attempting to sacrifice her. According to Driver, his obligation to his daughter is not eliminated by the moral duty to save the lives of his army; it is merely outweighed. She thinks that Agamemnon is blameless because he (a) does what he morally ought to do, and (b) is not of bad character, but acts on a reflective decision (or so we assume).

However, Driver’s analysis of this case is problematic. Note first that, if Iphigenia agrees that Agamemnon is blameless, how could she intelligibly deny him forgiveness? Their relationship might then be impaired just because it is difficult to get along with someone who attempted to kill you, even though his reasons were decisive. Again, only quasi-forgiveness would be intelligible. Driver would object that Agamemnon’s violation of his pro tanto relationship-based obligation not to sacrifice Iphigenia makes the full practice of forgiveness intelligible. But it is puzzling why his pro tanto obligation has any such significance when it is outweighed.

The account that I develop here offers a better explanation of why genuine forgiveness could still be intelligible even though Agamemnon does what he ought to do: a good father doesn’t even think about sacrificing his daughter for the greater good. The fact that he attempts to sacrifice her reveals that his attitudes fall short of the ideal of the parent–child relationship. While Agamemnon indeed performs all of the actions that he ought to perform, he violates norms that govern his attitudes: as a father, he should think and feel in a way that prevents him from sacrificing his daughter. That his attitudes fall short of this ideal makes him blameworthy, and at least his daughter is justified to blame him. That is, he holds blameworthy non-culpable attitudes that impair his relationship.

One might object that if Agamemnon ought to sacrifice Iphigenia, then he must also be permitted to have attitudes that allow him to go through with it. However, note that this would be the wrong kind of reason for attitudes (see Hieronymi [2006]): the fact that he can only sacrifice his daughter if he doesn’t love her is not a reason against loving her. It is at best a reason to get rid of his love. If he gets rid of it, he still fails to live up to the ideals of the parent–child relationship. Tragically, Agamemnon violates either a duty to act or a duty to think and feel. Both duties are all-things-considered duties, in the sense that neither is outweighed by the other.

One might generally doubt that parents have an attitudinal obligation to love their children so much that they cannot sacrifice them for the greater good. If so, then Agamemnon is indeed blameless. But then only quasi-forgiveness is intelligible: Iphigenia couldn’t intelligibly deny him forgiveness. The intuition that she can intelligibly deny him forgiveness, and that she furthermore has a right to do so, just indicates that parents do have such an attitudinal obligation: parents should love their children to such a degree that sacrificing them is not an option.

Thus, Driver’s counterexample ends in the following dilemma. Either Agamemnon wrongs Iphigenia or he doesn’t. If he doesn’t, then he is indeed blameless; but then only quasi-forgiveness is intelligible. If he does, then the full practice of forgiveness is intelligible; but then he is blameworthy because his wronging Iphigenia consists in violating an all-things-considered attitudinal duty, rather than a mere pro tanto duty not to sacrifice her. More generally, whenever forgiveness seems to be intelligible even though the person seems to be blameless, it turns out either that only quasi-forgiveness is intelligible, or else that an attitudinal relationship-based duty was violated, and therefore the person is in fact blameworthy.

I have argued that, in (1)–(5), S owes an authentic apology to the wronged party—that is, an apology that makes the full practice of forgiveness intelligible and thus implies blameworthiness. I have also defended the claim that the intelligibility of this practice implies blameworthiness: in any potential counterexample, either mere quasi-forgiveness is intelligible, or the person is, after all, blameworthy for violating an attitudinal norm. I thus conclude that S is blameworthy.

3.4 From Behaviour to Attitudes

I now argue that if S is blameworthy, then S cannot merely be blameworthy for their non-culpable behaviour B but must also be blameworthy for their attitude A that caused B. The best explanation for S’s blameworthiness for B is that S is originally or directly blameworthy for A. Remember that we have stipulated that there was nothing that S should have done to avoid A, and that S already managed B as well as we can reasonably expect of S. I have argued that this does nothing to annul S’s blameworthiness: B still reveals an impaired relationship that calls for the full practice of apology and forgiveness. But does that show that S is blameworthy for A?

First, a clarifying remark: when S apologizes, they don’t normally apologize for the attitude. Rather, they apologize for their behaviour and the resulting harm.¹⁴ This doesn’t imply that S is blameworthy only for B but not for A. If S indeed managed B to the best of their abilities, then S is praiseworthy in one respect: S cares about not letting their mental state go on a rampage in the external world. But if S is praiseworthy for managing B, then S’s blameworthiness cannot be explained merely by appealing to B. Furthermore, much of B concerns involuntary expressions that can, by themselves, no more provide a basis for blame than automatic reflexes would do (especially in (2), (4), and (5)). Finally, the proposal that S is blameworthy only for B but not for A won’t help the voluntarist. For they are committed to the view that S is blameless for anything non-culpable—whether attitude or behaviour. So, it isn’t promising, and especially not for voluntarists, to locate the original or direct blameworthiness in the non-culpable behaviour.

We are left with the task of looking for another original locus of S’s blameworthiness. There are two reasons why S’s attitudes are a plausible candidate. First, attitudes are, like actions, governed by norms and reasons. If blameworthiness presupposes the violation of a norm or a failure to respond correctly to possessed reasons, then attitudes are, like voluntary actions, promising for locating direct blameworthiness. Second, vices are manifested not only in actions and overt behaviour, but also in attitudes. The irascible person tends to form unfavourable beliefs about others on the basis of which they become disproportionately angry. Such vices impair relationships, and we can therefore be blameworthy for displaying them. If attitudes can manifest vices, then our attitudes are plausible loci of original blameworthiness.

I have no conclusive argument that there is no other good explanation for S’s blameworthiness. However, I have argued that S cannot be blameworthy only for B. At the very least, voluntarists cannot retreat to this claim. Attitudes, which are part of the space of reasons and which can be manifestations of vicious character, seem the most plausible original locus of S’s blameworthiness. I conclude that we can be blameworthy for NCAs, and that therefore voluntarism is false.

4. A Rationalist Alternative

When considering blameworthiness for NCAs, we experience an ambiguity. On the one hand, the person holding a NCA had no duty to avoid it, and thus they seem blameless. On the other hand, the NCA might still be responsive to reasons, and it might cause harm to others. As I have argued, this can give rise to our full practice of apology and forgiveness, which in turn implies that we regard each other as blameworthy for NCAs. This section proposes that we can capture this ambiguity by adopting a rationalist account that acknowledges that both reasons for actions as well as reasons for attitudes shape our responses towards people who hold unjustified NCAs. Given my previous argument, this allows us to see that answerability, or, more precisely, failures of reasons-responsiveness, can appropriately give rise to moral blame. Voluntarists therefore cannot reply that rational evaluation doesn’t give rise to ‘genuine’ responsibility and blame (see section 1).

Return to Smith’s case (section 2.2). Abigail’s racist attitudes are, earlier in her life, non-culpable, and later, when she has more access to relevant information that casts doubt on her racist outlook, still less culpable than those held by Bert, who adopted them reflectively after growing up in a tolerant family. As granted earlier to the voluntarist, the non-culpability, and thus the lack of voluntary control that Abigail had over her mind, affects the emotional intensity of our blaming responses. I suggested that this is because the difficulty of exercising control over her mind (due to lack of information, but also because her attitudes are recalcitrant) makes it the case that she lacks decisive reasons to engage in attempts at attitudinal self-management.

We can see now how a rationalist account can make sense of the intuitive ambiguity that we experience when it comes to NCAs. First, a rationalist can argue that our reactive sentiments are rightly affected by Abigail’s lack of opportunities to exercise voluntary control: they are affected by whether she responded correctly to her practical reasons for attitudinal self-management. Since we assume that Abigail’s attitudes are non-culpable, we assume that she didn’t commit any such fault in self-management when she responded to her practical reasons with her actions and omissions. We are less critical of her than of Bert because, in so far as it comes to how Abigail conducted her actions and omissions in shaping her mental life, she is faultless.

Furthermore, a rationalist will argue that Abigail is still answerable for her racist attitudes: it is intelligible to request her reasons for them, rather than just her reasons for self-management. When asking ‘why do you believe that p?’ or ‘why are you angry about p?’, we often ask not merely for explanation but for justification: we ask for the reasons that the person took to support believing p, or the reasons that the person took to support being angry about p. A good reason for belief is evidence for p, and a good reason for being angry is an offense. If the person is unable to provide a good answer to our request for justification, this can give rise to blame. Therefore, if we suppose that Abigail’s reasons do not support a racist attitude, a rationalist can locate a genuine shortcoming in Abigail’s reasons-responsiveness: she fails to acknowledge reasons against holding racist attitudes. This failure, however, isn’t a failure in exercising voluntary control.¹⁵

The rationalist assumes that, if Abigail is indeed blameworthy, then her reasons favour not holding racist attitudes: her attitudes are unjustified. This seems plausible. For suppose that Abigail’s perspective was so distorted by her community that all of her possessed reasons favour racist beliefs. Abigail was in this situation until a certain age. It was only when she gained more information, in light of which we would rightly expect her to revise her beliefs, that she becomes blameworthy. For, up to this point, Abigail might have responded correctly to all of her reasons and yet, due to her unfortunate circumstances, ended up with racist attitudes. However, if a person responds correctly to all of their reasons for attitudes and actions, then they are blameless.

Remember that the voluntarist made a similar claim—namely, that Abigail becomes blameworthy once she has relevant information accessible that would then make it the case that she ought to engage with this information actively and to consider it until her beliefs are revised in light of it (see section 2.2). The difference to the rationalist’s proposal is that only the voluntarist requires that Abigail has reasonable opportunities to revise her attitude. To see this difference, suppose that, after Abigail reads information that should rationally dislodge her racist stance, she irrationally fails to revise her beliefs. She would have to invest more time in engaging with the information. However, suppose that she has no such duty to engage further with the information—say, because the library is on fire and she should leave. The voluntarist would claim that she is blameless, for she fulfilled all of her duties to manage her attitudes. The rationalist, by contrast, can argue that she is blameworthy because her racist attitudes are now unjustified in light of her new information.¹⁶

5. Conclusion

My argument against indirect voluntarism from section 3 concluded that Abigail’s NCAs might still warrant moral blame. When she gains awareness of her failure to acknowledge her reasons against her racist attitudes, she owes the people whom she harmed an authentic apology, thereby making herself vulnerable to their decisions about whether they forgive her. Again, this practice wouldn’t make any sense while knowing that Abigail’s attitudes were a result of her responding correctly to all of her reasons—for both her actions and her attitudes. However, if the NCAs themselves are unjustified by her possessed reasons, then, by denying forgiveness and continuing resentment, victims of her racism might appropriately blame her, thereby expressing moral protest and emphasizing their own dignity. Abigail’s racist attitudes can then reflect various vices, thereby revealing features of her character or self, thereby warranting relationship modifications that count as moral blame.¹⁷

If this is correct, then indirect voluntarists can no longer insist that rationally evaluating an attitude by noting that it is insufficiently supported by reasons cannot give rise to ‘genuine’ blame. Rather, my argument reveals that failing to respond correctly to one’s reasons for attitudes can give rise to reactive responses that are very similar to the responses that are appropriate when people fail to respond correctly to their reasons for actions. In order for these reactive responses to be appropriate, it is often sufficient that an attitude is non-pathological, is unjustified by the subject’s reasons, and causes harm to another person. It isn’t necessary that one violated any duty of attitudinal self-management. The voluntarist therefore cannot account for our blameworthiness for NCAs.

In reply, the voluntarist might propose that our practice of apology and forgiveness needs a fundamental revision (see Nussbaum [2016]). Nothing that I have said excludes this possibility. However, if I am right, then our actual practice commits us to the view that we are sometimes blameworthy for NCAs, and thus to the falsity of indirect voluntarism. The argumentative burden is on the voluntarist to argue that we should revise our actual practice.¹⁸

Disclosure Statement

No potential conflict of interest was reported by the author.

Notes

1 See Hieronymi [2006, 2009a]. See Kavka [1983] and Owens [2000: 81–2] on why we cannot choose intentions. I put this issue aside. If we can choose some attitudes, then some reactive sentiments are appropriate for this reason.

2 See Meylan [2013, 2017] and Peels [2017] on doxastic responsibility, Oakley [1992] on emotional responsibility, and Jacobs [2001] on responsibility for character. See also Rosen’s [2004] view that blameworthiness always originates in akratic action, and Fischer and Tognazzini’s [2009] view on tracing back all responsibility to voluntary action.

3 See Owens [2000, 2017: intro.], Smith [2005], Hieronymi [2006, 2008, 2014], McHugh [2013, 2017], McCormick [2015], Roberts [2015], Portmore [2019], White [2019], Schmidt [2020a], and Osborne [2021]. Adams [1985] is the locus classicus of the opposition to indirect voluntarism. He did not appeal to the idea of reasons-responsiveness at the time.

4 For both responses, see Peels [2017: 46–8, 159–60]. For the latter response, see Fischer and Tognazzini [2009].

5 See Kiesewetter [2017: ch. 8] for a perspectivist account.

6 Some aspects of spontaneous beliefs are reasonably foreseeable: it is foreseeable that you will acquire beliefs about the environment through which you are planning to walk, but you don’t know which concrete beliefs you will acquire.

7 Wallace argues that ‘it would indeed be strange to suppose that one might blame another person without feeling an attitude of indignation or resentment toward the person, or that one might blame oneself without feeling guilt’ [1994: 75]. The second reply to the voluntarist in this paragraph does justice to this point. That is, I am not committed to the view that we can blame without feeling reactive sentiments, although I sympathize with this idea due to the arguments brought forward by Smith, Scanlon, and others. Hieronymi [2014, 2019] claims that there is a reactive attitude, ‘resentment+’, that is only appropriate towards someone for holding an attitude M if this person had a reasonable opportunity to avoid M, but she denies that ordinary resentment presupposes reasonable opportunity.

8 For instance, that it is a response to the fact that one’s attitudes fall short of the normative ideal of the relationship [Scanlon 2008], or that it is an expression of moral protest [Smith 2013].

9 This case is motivated by a case in Tognazzini’s [2020] discussion of blameworthiness for judgmental thoughts.

10 Plausibly, one might argue that, in (4), there would be nothing problematic if the parents merely believed that the garden isn’t properly tended—that is, without desiring to lecture. This indicates that the desire is the problem, rather than the belief. Furthermore, Ramona’s anger in (5) need not be based on any fully fledged belief.

11 Fricker’s [2019] notion of gifted forgiveness isn’t captured by the account.

12 See Tognazzini [2020] for a more detailed discussion of Watson’s take on judgmentalism.

13 One might take issue with calling the driver’s apology ‘conventional’. I sympathize with this. Calling all non-authentic apologies ‘conventional’ is technical. There are important differences within the camp of non-authentic apologies.

14 However, maybe we sometimes do apologize for ‘believing badly’ about someone (see Basu [2019]), and plausibly we apologize for having been angry, where we need not be precise about the object of the apology.

15 Hieronymi [2006, 2014: 22–4] and Roberts [2015], especially, have worked out these two dimensions of holding each other responsible for our attitudes, by distinguishing between reasons for attitudinal self-management and genuine reasons for the attitudes themselves, and by connecting them to different exercises of agency (see Hieronymi [2009b]).

16 I say ‘can argue’ because the rationalist might formulate further necessary conditions on blameworthiness. Elsewhere [2020a, 2020b, forthcoming], I spell out some aspects of the connection between blameworthiness and reasons.

17 Am I denying that blaming racists who, due to their distorted perspective, responded correctly to all of their reasons is legitimate? Not necessarily, for the sense of ‘appropriateness’ here is narrower than ‘all things considered’ legitimacy (see Calhoun [1989] Coates [2020]). Furthermore, one can protest systemic oppression without blaming these individuals.

18 I am grateful to the audience at the conference on The Ethics of Emotions at the University of Bern in July 2021, as well as the audience at the Karlsruhe-Saarbrücken Colloquium in February 2020, for their very helpful feedback. Special thanks to Pamela Hieronymi who first made me aware of the conflict between indirect voluntarism and our practice of apology and forgiveness. For detailed written comments on earlier versions, I thank Jonas Blatter, Marie van Loon, Veli Mitova, Tom Roberts, two anonymous referees, and an associate editor. Thanks also to Gerhard Ernst, Anne Meylan, Katia Sapori, Peter Schaber, and Peter Schulte for helpful suggestions. Thanks to Kathleen McCully for her competent proof reading. Finally, I thank the Zurich Epistemology Group on Rationality (ZEGRa) and the African Center for Epistemology and Philosophy of Science (ACEPS) for their ongoing support.