A Case for the Guilty: A Defence of Blameless Guilt

Imagine you receive a call from a friend; while driving carefully and responsibly, she struck and killed a child who darted out in front of her vehicle. The police and witnesses agree there was nothing she could have done to prevent the accident, that this was not a case of even minor negligence, and the accident was completely out of her direct control.¹ When you arrive at your friend’s home you find her wracked with guilt. To comfort her, you remind your friend the accident was not her fault as there was nothing she could have done to prevent it, and, since she is not to blame, she should not feel guilty. Having not considered blame as necessary for guilt, your friend perks up is relieved of all her guilty feelings. She no longer feels a sense of personal responsibility or special relationship to the child or the accident. Or, imagine your friend has already come to this conclusion on her own by the time you arrive, free of guilt she proclaims her mere causal role is unfit for blame.

To think one’s friend would feel no guilt, despite having no direct control, about her causal role in the child’s death is unsettling, if not disturbing, but this reaction does follow from Kantian blame and blame as necessary for guilt’s fittingness. The Kantian holds an agent may only be blameworthy for the orientation of her will; therefore, as long as she has acted in good will the agent cannot be blameworthy, regardless of the consequences of the act (therefore, morality exists within actions themselves, as opposed to evaluations of the outcome of actions):

A good will is not good because of what it performs or effects, not by its aptness for the attainment of some proposed end, but simply by virtue of the volition, that is, it is good in itself, and considered by itself is to be esteemed much higher than all that can be brought about by it in favour of any inclination, nay, even of the sum total of inclinations. Even if it should happen that, owing to a special disfavour of fortune…this will should wholly lack power to accomplish its purpose, of with its great efforts it should yet achieve nothing, and there should remain only the good will (not, to be sure, a mere wish, but the summoning of all means in our power), then, like a jewel, it would still shine by its own light, as a thing which has its whole value in itself.²

That the friend did everything she could to prevent the accident (was a vigilant driver, had her car serviced regularly, etc.), not merely that she did not intend to cause harm, leads the Kantian to conclude the friend is not blameworthy. It follows, if blameworthiness is the proper object of guilt, she ought not feel guilty.³ But this is highly counterintuitive; there is something unsavoury about the friend’s guiltlessness towards the child’s death she is causally connected to. Further, this intuition seems grounded in more than mere social norms; the friend appears to fail to recognize something morally important about her role in the child’s death, despite her lack of blameworthiness. The discomfort brought by the absence of guilt cannot be explained away; though she is not blameworthy, intuition suggests the friend has some sort of moral responsibility to the child’s death brought about by her causal role, and this appears to warrant guilt.

This sticky situation leaves the Kantian with three options: 1) intuitions about guilt and moral luck are mistake; the agent is right not to feel guilty, 2) there must be revisions to the Kantian account of blame to accommodate feelings of guilt when one has merely a causal role in harms (she must accept it is possible for an agent to be blameworthy for what is beyond her control), or 3) the Kantian should uncouple blameworthiness and guilt’s fittingness as to allow for situations in which guilt is fitting for an agent who is not blameworthy, and understand blameworthiness as a sufficient, but not necessary condition for guilt to be fitting. Thomas Nagel argues Kant was wrong to deny the existence of moral luck (that an agent may be judged as praiseworthy or blameworthy for events out of her control);⁴ and that judgments of praiseworthiness and blameworthiness must, at least somewhat, relate to outcomes of an agent’s actions, as it is consequences which are the typical object of moral evaluation:

Prior to reflection it is intuitively plausible that people cannot be morally assessed for what is not their fault, or for what is due to factors beyond their control…However jewel-like the good will may be in its own right, there is a morally significant difference between rescuing someone from a burning building and dropping him from the twelfth story window while trying to rescue him.⁵

This paper will suggest the Kantian does not need to revise her understandings of blameworthiness in order to account for responsibility to harms one has played a causal role, and maintains the traditional Kantian account of blame. I argue she should accept the third option above, and revise her understanding of blameworthiness as necessary for guilt’s fittingness in order to preserve the Kantian account of blame without compromising the fittingness of guilt in causal roles played in harms. This paper argues the judgement elicited by feelings of guilt are that genuine threats to the agent’s interpersonal relationships are present and sincere sympathy and regret for harms caused to others, and functions to motivate reparation of damage to social bonds done by the harms to which one is causally connected. Therefore, the proper object of guilt is not one’s own blameworthiness (as most fittingness account of guilt suggest), but it is a special obligation to amend the harms which one has caused, regardless of the agent’s blameworthiness for the event. The Kantian should then recognize an agent ought to feel guilty for her role in the harms she is causally connected to, but not blameworthy for, as duties of social obligation create a moral responsibility to repair and reconcile these harms and express genuine care.

Though there is some dispute as to the exact definition and criteria for fittingness, for simplicity I will use a more generic understanding.⁶ To judge an emotion as a fitting response is to hold the proper object of the emotion is present, and the judgement the emotion functions to elicit is accurate.⁷ Fittingness evaluations are primarily concerned with the rationality of emotions; fittingness examines how one ought to feel, and why. The Kantian understands morality and reason as inseparable, for what it is to be moral is to act with reason,⁸ and so understanding the rationality of moral emotions is an important issue for the Kantian.⁹ To be fitting, the emotion must fulfil its purpose; they are reasons for why one ought to feel in certain ways, and offer an explanation of the relationship between reasons and values.¹⁰ For example, losing a favourite pair of mittens, or a close friend moving away are scenarios when sadness is fitting. If the proper object of sadness is the loss of something which is valued by the agent (the mittens, the friend’s proximity) then sadness’ proper object is present in both scenarios. If the judgement elicited by sadness is a valued object is lost, this also occurs in both scenarios. The presence of the proper object and the appropriateness of the judgement makes sadness fitting in both cases. The proper object and judgement are distinct but clearly connected; the emotion must appropriately correspond to the reality of the situation (the presence of the proper object) and the agent must herself react properly (the judgement elicited). Proportionality is also a factor; that which is of more value to the agent should cause stronger and longer lasting sadness; if her friend’s proximity is valued more by the agent than the mittens, it would be fitting for the agent to be more sad regarding the move than the mittens. Fittingness also evaluates when one ought not feel an emotion; sadness is unfitting when the garbage man takes away the trash because, though there an object is lost, it is not a loss of a something valued by the agent nor would the judgement an object of value is lost be appropriate. I will regard an emotion as fitting if and only if the proper object is present and the judgement elicited by the emotion accurately corresponds to the situation.

The origin of emotions may lie in their adaptive value, but this does not compromise the importance of their contemporary social and moral role as eliciting value judgements. D’Arms and Jacobson suggest rational sentimentalism to explain the interaction between the evolutionary history of emotions and their contemporary functions. Rational sentimentalism refers to the idea that emotions correspond to “human values”, which are held by every morally regular human and serve as the basis for morality (these values are, however, still subject to reason). Human emotional tendencies do not need to be assumed fitting, fittingness evaluations do need to correspond to actual human concerns.¹¹ D’Arms and Jacobson argue “almost all humans have the emotional capacities necessary for being sensitive to … human values”.¹² Fittingness theories function to account for what judgement, corresponding to a deeply held “human value”, is being elicited when one experiences an emotion. These human values play important roles in human contemporary social and moral life; they correspond to what is important to agent, and serve as the basis for moral action and care. Fittingness evaluations of emotions should be ground in these universal human values. The intuitions regarding guilt for harms one has merely a causal connection to should not be dismissed because they alert us to a abnormal reaction which does not express the human value connected to the situation. Concerns regarding relationships and values, such as needs of belonging and social acceptance, and of genuine care for the wellbeing of others are examples of human values. Guilt should correspond to a near-universal human survival problem, but this alone does not determine fittingness; reason and contemporary function are also major factors. The rough-and-ready definition of guilt is the admittance to wrongdoings and the acceptance of moral blame, and the uncomfortable sensations associated. Cases such as the one discussed earlier reveal the social and moral function of guilt to be much more complex than this common conception. Most fittingness theorists hold it is fitting to feel guilt if and only if one is blameworthy. Allan Gibbard’s account relies on this relationship between guilt and blame so much so that he understands blameworthiness in terms of guilt’s fittingness.¹³ Gibbard himself could hold it be fitting for an agent to feel guilt for events out of her control (as he does not subscribe to the Kantian account of blame), but this would necessarily entail she is blameworthy for the event. To preserve Kantian blameworthiness in Gibbard’s account, one would have to counterintuitively concede guilt is only fitting for harms one is in direct control of; a stark deviation from the traditional Kantian account.

Contemporary evolutionary psychology argues guilt evolved as a response to fear of social rejection when an agent has been unfair to and/or harmed others,¹⁴ but intuitions about cases of moral luck reveal guilt has a much more complex function and evolutionary history than these more shallow accounts. If emotions are understood to be the manifestation of judgements regarding human values, rather than simply evolved behavioural mechanisms to help ensure survival, emotions must be understood both through their evolutionary history and to what “human value” they correspond to, and that these are deeply intertwined. D’Arms and Jacobson’s account meshes well with common attitudes and intuitions about emotion: “people do not care simply about whether emotions do them good, or are useful to nicety at large; we also care about having fitting feelings and the values they decry”.¹⁵ It is not solely about what emotions do for us (as argued by evolutionary psychology accounts), but also what they mean to us. We feel sad when a friend moves away because it shows we recognize that something of value, the friend’s proximity, is lost; even though sadness is unpleasant and temporarily impacts functioning, it would be odd to desire one’s friend or oneself not to feel sad about the move, or that they not be consoled by the fact they may call or visit. In our original case, we do not want the friend to feel guilty because of what guilt does (it appears to do nothing but make her feel rotten), but we want her to feel guilty because that means she holds the shared human values and properly recognizes and responds to them (as it represents that she values the child’s life and acknowledges her role in the accident). Her lack of guilt suggests she does not hold or cannot recognize and respond to these human values, and indicates a problem in moral functioning. Intuitions about the judgement guilt elicits suggest one should feel guilt for any harms she is attached to, despite blameworthiness. Therefore, the judgement corresponding to guilt is not that one is blameworthy, but must correspond to harms one has played a role in more generally.

I purpose to combine rational sentimentalism with the structure of adaptive fittingness proposed by Allan Gibbard. Gibbard suggests a fittingness account of guilt which explains guilty feelings as originally arising as a socially adaptive reaction towards other’s anger when one has acted in a blameworthy fashion. Gibbard understands emotions as having evolved to coordinate social interactions; they are meant to both convey a judgement one has made (when combined with rational sentimentalism, as corresponding to a particular human value) and to generate a response (when combined with rational sentimentalism, a reaction corresponding to a judgement regarding a specific human value) in another. Gibbard argues emotions must be directed in order to be fitting; in the mittens case one’s sadness is directed towards the (absence of) mittens. One is not sad globally (though experience of sadness may be overwhelming), but sad locally about the lost mittens; the feeling is eliciting a particular judgement towards a particular object. Further, since their evolved function is coordination of actions and expectations through physical expressions (for our purposes, of shared human values), emotional expressions should create responses in others¹⁶ (for our purposes, also corresponding to shared human values). In the case of the friend’s move, sadness is direct at the loss of his proximity, and therefore our sadness is meant to cause, likely, both sadness in him but also comfort; he witnesses the expression of how we value his proximity (the human value of the intrinsic worth of interpersonal relationships). Gibbard explains there are both first- and third-person fittingness, as emotions are meant to mesh with one another. In the mittens case, the first-person response is sadness, and the third-person counterpart is sympathy. Therefore sadness and sympathy mesh; sympathy works to simultaneously try to relive the sadness of another while expressing a judgement regarding the relevant human value. Emotions were not meant to exist or function in isolation; emotions necessarily fit into a complex coordination of social communication based in human values and group cohesion. Therefore, the evolutionary and moral aspects of the emotion are necessary to fittingness theories; each plays a role in the appropriateness and function of emotions.

Feelings of guilt, therefore, must serve the purpose of simultaneously expressing a judgement regarding a human value, respond to its proper object, and work to invoke a response in another. Gibbard argues “guilt meshes with anger in a special way”;¹⁷ impartial anger, responding to a blameworthy action and eliciting a relevant judgement brings about the reaction of guilt in another, which works to counter the anger of the other and hopefully reestablish social cohesion through expressing a human value relevant to care for the unjustly harmed other. Gibbard holds the proper object of guilt is one’s own blameworthiness, as it will counter impartial anger as a third-person response to the blameworthy harmful action. Therefore, guilt and anger both respond to blameworthy actions, guilt is the fitting first-person reaction, while anger is the fitting third-person reaction, and they evolved to coordinate together in a specific way to bring about reactions in the other and solve social problems. Gibbard explains:

Guilt [is the] first-person counterpart to anger … Not that guilt is self-directed anger; feeling guilty is different from feeling you could kick yourself. Rather, guilt is coordinated with anger in a special way: it aims to placate anger, and it is governed by the same norms as govern anger.¹⁸

Gibbard holds anger need not evoke guilt in others (they could react with anger themselves, or fear) in order to be fitting, but these other reactions, will not work to counter guilt and reestablish social cohesion in the same way guilt does (for our purposes, as they convey other values), which Gibbard understands to placate anger directly through motivating attempts to make amends and obtain repellence from the harmed part: expressions of guilt work as “smoothing things out and correcting for partiality”.¹⁹ Therefore, guilt works reestablish one’s belonging in the group, and regain the trust and affection of one’s fellows; its purpose is to counteract impartial anger.

However, if guilt conveys the human value relevant to genuine sympathy and regret at causing harms and functions to repair threaten relationships, guilt’s proper object cannot be one’s blameworthiness. Not every situation in which genuine threats to relationships are present is impartial anger fitting on behalf of the third-party (such as the case with the friend’s accident), or the agent blameworthy (in the Kantian sense). Though her actions do not warrant anger, as she is not blameworthy, they certainly will affect her social standing; she has caused harm to another, and therefore other’s trust and affection for her is genuinely threatened. Other’s feeling hurt, scared, or unable to trust because of harms one has experienced to which another is causally connected to are all reason which may be devoid of anger but still threaten one’s social relationships. The ability of emotion to aid in moral judgements based on intuitions, in addition to the importance placed on emotion in circumstances where one has played a causal role in harms to others is what makes fittingness evaluations particularly relevant to problems of moral luck and blame. I hold that it is possible to maintain that agent’s are not blameworthy for harms they play mere causal roles in, but playing any role (including simply a causal role) in harms are circumstance in which it is fitting to feel guilt. For this to be plausible, the blameworthiness must be a sufficient, but not necessary condition for guilt’s being fitting. One may worry about how blameworthiness could be understood besides being that which that which it is fitting to feel guilt for. In “Moral Feelings and Moral Concepts” Gibbard argues that, in societies which do not have guilt in their emotional repertoires, shame is fitting in both circumstances where one has acted in a morally blameworthy fashion, and in cases of personal failure; the proper object of shame is present and its judgement is appropriate under both of these circumstances, though morally they are very different. Therefore, shame plays multiple social roles in these guilt-devoid societies, while maintaining the same proper object and judgement.²⁰ I propose we ought to understand guilt in a way very similar to this; therefore, Gibbard is right to propose guilt has a special relationship with anger regarding one’s own blameworthiness, but this is not the only occasion where it is fitting for one to feel guilt. Therefore, guilt may mesh with various emotions which respond to harms, including one’s brought about by harms one is causally connected to, when personal relationships are threatened by events one is connected to. Of course, other’s do not experience these emotions for guilt to be fitting, but this is the proper role and function of guilt. We should understand impartial anger and blameworthiness then as sufficient conditions for guilt’s blameworthiness, rather then necessary conditions as Gibbard does.

Though Gibbard’s account may be able to account for the fittingness of guilt in cases such as our original accident case, it cannot do so without admitting the agent is blameworthy for the harmful event which was out of their causal control.²¹ This is incompatible with the Kantian account of blame, and in order to preserve both this account of blameworthiness and intuitions about guilt in cases of moral blame I suggest an account very close to that of Gibbard’s, but which uncouples guilt and blameworthiness and slightly revises both the judgement guilt elicits and its proper object, in a similar move that he makes in regards to the uncoupling of shame and impartial anger in society’s without the presence of guilt. In this revised account, guilt functions to alert agents to events which they are causally attached, regardless of the agent’s blameworthiness. Causing harm to another will threaten and compromise social relationships and group cohesion whether or not one is blameworthy for the harms done, so guilt should function in the same ways whether or not one is at fault for harms. Guilt therefore meshes with feelings of pain one is causally attached to, rather than impartial anger as Gibbard suggests. Guilt, therefore, motivates not actions which merely counter aggression but those which seek to repair the actual harm caused; guilty feelings respond to the actual harms the agent has a causal connection to, rather than the emotional reactions to the harms, as in Gibbard’s account. This account coordinates emotions with one another as well as the event itself; as opposed to merely countering other’s reactions, it serves to also express a judgement about the harm. The judgement elicited by guilt then corresponds to the genuine value of other’s wellbeing. This is much closer to how guilt’s contemporary function; we should think the friend in the original case should feel guilty because she values the child’s wellbeing, and is acknowledging her role in the harms to the child in the appropriate way. The proper object of guilt is therefore not when one is blameworthy, but when there are special obligations to make amends for harms one has caused (either deliberately or accidentally), regardless of any particular relationship. This moves the focus away from preserving social cohesion through countering anger, and onto holding guilt as eliciting the judgement that the wellbeing of others is valuable; this of course functions to show that we do genuinely care about others, and fosters trust, which is integral for social functioning and group cohesion (therefore, guilt is fitting in cases where one causes harms even when do not have a relationship with the harmed; guilt would be fitting if one were to shoot strangers on a desert island from a plane, for example). There is, therefore, a moral responsibility connected to harms one has caused that is distinct from blameworthiness for the harm.

One may wonder how an agent can have a moral responsibility to reconcile harms they are attached to, without admitting the agent is blameworthy. Jeff MacMahan offers distinctions between kinds of “attackers” to account for different kinds of causal roles an agent may play in harmful events they are connected to. MacMahan defines “culpable attacker” (or “culpable threat”) as an individual who is blameworthy for an unjustified threat they pose to the wellbeing of others, and “innocent attacker” (or “innocent threat”) as one who is not blameworthy for an unjustified threat they pose.²² McMahan provides a further sub-group of the “innocent attacker”, which he refers to as the “inadvertent attacker” who poses a threat to the wellbeing of others, which is both unforeseen, accidental, and not due to the agent’s recklessness or negligence; the “inadvertent attacker” is the victim of pure moral luck.²³ McMahan uses the term “attacker” to indicate the agent is morally responsible; therefore blameworthiness for harms is not a necessary condition for having responsibilities connected to harms they are causally connected to in McMahan’s account. Even if the harm one has caused is completely out of their control, their role in the harm creates a special responsibility to the situation: “[E]ven if the reason that the Attacker or Threat poses a threat is simply that he has bad luck, it is still true he is morally responsible”.²⁴ The fittingness account I suggest has this moral responsibility to harms one is causally connected to, independent from one’s blameworthiness, as the proper object of guilt, rather than blameworthiness as in Gibbard’s account. If this moral responsibility, which I suggest (in contrast to McMahan) is a special obligation to make amends for the harms one has caused, is present then it is fitting for the agent to feel guilty.

One may be concerned this is too weak an account of fittingness; that is, if the proper object of guilt is a special obligation to make amends for harms one is causally connected to, and the judgement feeling guilt elicits is genuine sympathy for harms one is causally connected to, one may worry this account allows guilt to be fitting in situations where it should not be. Imagine a professor finishing up final grades for the semester. One of the professor’s students, who was an attentive and polite student, but despite her hard work, constant attendance and visits to office hours, the student has performed poorly in the course. This student goes to the professor and asks her to boost her grade. The student explains that if she does not get at least a C in the course she will be put on academic probation, loose her scholarships, and will not be able to finish her degree. For her own moral integrity, and out of fairness to her other students, the professor refuses to increase the grade. Months later, the professor learns the student has lost her funding and dropped out. The fittingness account I have suggested holds it would be fitting for the professor to feel guilty in regards to the student’s failure to continue her studies. The proper object of guilt, a causal connection to harms to another, is present, and the judgement elicited by guilt, genuine sympathy for the harms to the other, are both present. For it to be fitting for the professor to feel guilt may appear counter-intuitive; the professor is not at fault for the student’s failure, and the professor acted in a morally praiseworthy way (to deny the student an unfair GPA boost). However unfair it may appear, I maintain it is fitting for the professor to feel guilt for the student’s drop out. The “fairness” of one feeling the emotion has no bearing on its fittingness; fittingness is determined solely by the presence of emotion’s proper object and whether the judged which the emotion represents corresponds to the situation. The professor has played the role of an “inadvertent attacker”, she has caused the student harm (the student did try her best) but her role in the harm was due to only bad luck. Here is D’Arms and Jacobson:

[T]here is a crucial distinction between the question of whether some emotion is the right way to feel, and whether the emotion gets it right … [A]n emotion can be fitting despite being wrong (or inexpedient) to feel. In fact, the wrongness of feeling an emotion never, in itself, constitutes a reason for that emotion to be fitting.²⁵

Therefore, though we may agree it is “unfair” for the professor to feel guilty when she has done nothing wrong (she has, in fact, done the right thing), guilt is still fitting. Neither does it matter that the professor’s actions were morally praiseworthy, as moral evaluation should not affect the fittingness of guilt. One may hold that, since it is the student’s fault she has failed the course the professor should not feel guilty; the student caused the harm to herself, and therefore the professor’s lesser causal role is unfitting for guilt. This too though does not hold, the presence of a blameworthy agent does not change that the proper object is present for the professor to feel guilt, and that her guilt would elicit an appropriate judgement.

This revised fittingness account of guilt should be accepted by the Kantian so she may perverse the traditional Kantian account of blameworthiness and reconcile it with intuitions about the fittingness of feelings of guilt to harms in which one was merely causally connected to. This account does not, of course, solve all the problems for the Kantian arising from situations of moral luck (particularly those in which one has attempted to commit a harmful act but is intercepted by mere luck from obtaining their goal) but does allow her a more complex relationship between emotions and moral evaluation. Guilt therefore serves to recognize harms other experiences, rather than an acknowledgement of the agent’s own blameworthiness.