Why severe moral transgressions are often difficult to understand

ABSTRACT

When we learn about a severe moral transgression that has been committed, we are often not only horrified but also puzzled. We are inclined to raise questions such as ‘Why did they do this?’ or exclaim: ‘I cannot understand why anyone would do such a thing!’. This suggests that there is something difficult to understand about severe moral wrongs. In this paper, I offer an explanation of the phenomenon that severe moral transgressions are often difficult to understand. I begin by arguing that the relevant sense in which we try but often fail to understand morally bad action is that we try but fail to empathize with the agent at the time of the action. Then, I introduce a distinction between two broad kinds of morally bad action. Finally, I show that each of these kinds of morally bad action gives rise to obstacles to empathizing with the agent at the time of the action. Based on this explanation, I consider whether we should try to overcome this limitation to our understanding.

KEYWORDS:

• morally bad action
• evil
• empathy
• simulation
• imaginative resistance
• empathic resistance

1. Introduction

When we are confronted with a severe moral transgression that has been committed, we are often not only appalled and horrified but also puzzled. We are inclined to raise questions such as ‘Why did they do this?’, ‘How could they do this?’ or exclaim: ‘I cannot understand why anyone would do such a thing!’. While not everyone will experience this reaction equally often and with respect to the same cases, I take it that this phenomenon is familiar enough to merit investigation. It seems to suggest that there is something difficult to understand about severe moral wrongs. Because what we understand is not in general restricted to what we agree with or approve of, it is not immediately obvious why this should be the case.

In this paper, I offer an explanation of the phenomenon that severe moral transgressions often appear resistant to our understanding. My discussion will be restricted to severe moral wrongs. This is because trivial moral wrongs or morally wrong actions for which we have an excuse do not normally give rise to problems of understanding.¹ I begin by arguing that the sense in which we frequently cannot understand severe moral transgressions is that we try but fail to empathize with the agent at the time of the action (section 2). Then, I introduce a distinction between two broad kinds of morally bad action (section 3). Finally, I show that each of these kinds of actions gives rise to obstacles for empathy (section 4). Based on this explanation, I then consider whether we should try to overcome this particular limitation to our understanding (section 5).

2. The sense in which severe moral transgressions are difficult to understand

If we grant that severe moral transgressions are often difficult to understand, this still leaves open in which sense such actions can be difficult to understand. I want to suggest that the sense in which we often struggle to understand morally bad actions is that we try but fail to understand the agent at the time of the action.

Minimally, to understand another person, I must form a sufficiently accurate representation of them, which includes their perspective on the world and how it relates to their environment. To represent another person’s perspective, I must ascribe mental states, such as beliefs, desires, and emotions to them. This raises the question of how we ascribe mental states to others, a task that has been described as ‘mindreading’.

In recent decades, there have been two main approaches to this question. According to ‘theory theory’ we ascribe mental states to others by relying on a tacit folk psychological theory. According to ‘simulation theory’, by contrast, we understand others by imagining being in their situation and recreating their cognitive processes in our own minds. While according to theory theory, we understand others in the same way we understand objects of scientific study, according to simulation theory, the way in which we understand others is distinctive. Because we are agents ourselves, rather than having to rely on a theory, we can rely on our own capacities for theoretical and practical reasoning to understand others.

Consider the case in which A wants to eat a sandwich, believes that Joe’s Café is the only place in the vicinity that sells sandwiches and decides to go to Joe’s Café (Goldie 2011). According to simulation theory, B can predict or explain A’s behavior by forming the ‘pretend desire’ to eat a sandwich, the ‘pretend belief’ that Joe’s Café is the only place in the vicinity that sells sandwiches and relying on their own capacities for practical reasoning to generate the ‘pretend decision’ to go to Joe’s Café.

In cases where B already has information about A’s initial mental states, simulation allows B to predict A’s behavior. In cases where B observes A’s behavior, they can engage in an abductive process to come up with potential initial mental states and reasoning processes that would have produced the behavior (Goldman 2006; Gordon 1986).

The process described can be divided into three distinct and consecutive phases. In the ‘matching phase’ B generates ‘pretend’ mental states; in the ‘simulation phase’ B generates new mental states based on their own reasoning capacities; in the ‘attribution phase’ B attributes these new mental states to A. The matching phase allows B to adjust for differences between A and B by adopting ‘pretend’ mental states that might conflict with B’s own perspective. In this case, to generate the right ‘pretend decision’, B must ‘quarantine’ their conflicting mental states in the simulation process.

Simulationists agree that high-level simulation amounts to imaginative perspective-taking. The ‘pretend’ mental states involved in simulation are thus the mental states involved in imagination. This identification of high-level simulation with a specific kind of instructive use of the imagination (Kind 2016b; Kind and Kung 2016) leads to a straightforward account of simulation in the case of belief. We recognize a distinction between believing that p and imagining that p in everyday language, and these attitudes resemble yet differ from each other in exactly the right ways for propositional imagining to constitute a counterpart to belief for the purposes of simulation.

While beliefs and imaginings play analogous roles in theoretical and practical reasoning, in contrast to believing, imagining is unconstrained by evidence regarding p and thus subject to a higher degree of voluntary control. In addition, what we imagine can contradict what we believe without this creating cognitive conflict. This allows us to match the beliefs of others in simulation even though we do not share them. To adopt the ‘pretend belief’ that Joe’s Café is the only place in the vicinity that sells sandwiches is thus to imagine that Joe’s Café is the only place in the vicinity that sells sandwiches.²

While simulation theory and theory theory set out as mutually exclusive rivals, more recently, many theorists have advocated hybrid views, according to which we sometimes rely on theory and sometimes on simulation and simulation and theory can also interact in single instances of mindreading (Goldman 2006; Spaulding 2020). In addition, there has been a tendency towards pluralistic approaches to folk psychology, which recognize a range of strategies. While these recent developments are plausible, they suggest that simulation is one of the mechanisms we use for understanding others and that we use it when understanding is important and difficult to achieve (Wolf, Coninx, and Newen 2021). Since these conditions are met in the case of severe moral transgressions, it is plausible that we rely on simulation in these cases.

I will refer to high-level simulation as ‘empathy’. This use of ‘empathy’ differs from two paradigmatic uses of the term. On the one hand, ‘cognitive empathy’ is often used to refer to any strategy of mindreading (Spaulding 2017). My use of ‘empathy’ differs from this in that it specifies simulation as the relevant strategy. On the other hand, ‘affective empathy’ is often used to refer to a range of processes that involve affect sharing (Maibom 2017). My use of empathy is broader in that it refers to simulation of cognitive processes in general, where this can include simulation of affective mental states. At the same time, it is narrower in specifying that these simulated mental states occur in the context of imaginative perspective-taking. While identifying simulation with empathy is thus controversial, I take it that this use tracks at least one of the meanings ordinarily associated with ‘empathy’.

My suggestion is thus that the sense in which severe moral transgressions are often difficult to understand is that we try but fail to empathize with the agent at the time of the action.³ To work towards an explanation for why this is the case, in the next section, I will introduce a distinction between two broad kinds of morally bad action.

3. Two kinds of morally bad action

In philosophical theory, we frequently encounter two different conceptions of morally bad or evil action. According to ‘privative’ conceptions, morally bad action can only be understood as derivative of morally good action; it is essentially a failed version of the good. According to ‘perversive’ conceptions, morally bad action can be understood independently of morally good action; it can be seen as affirming an alternative and equally real order (Korsgaard 2009; Noller 2017).

These different theoretical conceptions seem to respond to different kinds of agents who commit moral wrongs, which can be discerned in real life cases, but find their clearest expression in fiction. The first kind of character is the ‘wanton’. They are characterized by inability and weakness. They are unable to think clearly, deliberate, and act for reasons. They are unable to control their impulses, and, consequently, they cannot pursue long term plans. The second kind of paradigmatic ‘villain’ is the ‘evil schemer’. They are characterized by skill and strength. They are able to think clearly, deliberate, and act for reasons. They can control their impulses and are able to follow through on long-term projects. However, they use all of these abilities in order to consciously and deliberately do things that are morally horrible.

The distinction between these two kinds of paradigmatic wrongdoers can be brought out by appealing to Karen Jones’s notion of our ‘normative conception of agency’. According to Jones, at least from the first-person point of view, we must conceive of ourselves as guiding our actions in accordance with our best reasons understood as reasons (Jones 2003, 188f). While the evil schemer can be understood as meeting the requirements that stem from this conception of agency, that is, as guiding their actions by what they take to be their best reasons understood as reasons, the wanton fails to meet these requirements. Their actions are akratic, impulsive, compulsive, committed on a whim or out of thoughtlessness.⁴

Closely related to the evil schemer are the ‘fanatic’ and the ‘egoist’. While an evil schemer might choose morally wrong actions while being under no illusion as to their moral status, the fanatic subscribes to a warped ideology, which allows them to see their morally wrong actions as morally laudable or even required. The egoist, by contrast, is simply indifferent to moral considerations. They choose actions solely based on what would serve their own interests. In spite of these differences, for my purposes, the fanatic and the egoist fall in the camp of the evil schemer. These types of agents meet the requirements that stem from our normative conception of agency, that is, they choose their actions based on what they take to be their best reasons understood as reasons.⁵

When considered as an account of all cases of morally bad action, both the privative and the perversive conception of morally bad action face problems. This is because, in spite of their caricatured features, both types of paradigmatic wrongdoers appear equally real. By providing an account of morally bad action that adequately captures only the actions of one of these types of agents, both conceptions are thus subject to counterexamples.

This general dialectic comes out, for instance, in discussions of Christine Korsgaard’s account of morally bad action. According to Korsgaard’s version of constitutivism, the norms that are constitutive of agency ultimately commit us to act in accordance with the moral law (Korsgaard 2009). As Herlinde Pauer-Studer points out, this type of view makes it difficult to think of morally bad action as action (Pauer-Studer 2018). While Korsgaard recognizes this challenge and attempts to meet it by distinguishing different kinds of failures of agency, the central problem remains. On a constitutivist view, agents who commit morally bad actions cannot be seen as fully exercising their agency. However, some wrongdoers seem to be doing just that.⁶

Attempts to account for all instances of morally bad action along the lines of a perversive conception of morally bad action risks running into the same type of problem. In his interpretation of the history of philosophical thought on evil, Jörg Noller suggests that we find a development from privative to perversive conceptions, which he takes to be preferable (Noller 2017). However, it seems that even severe moral wrongs can be committed akratically, impulsively, compulsively, on a whim or out of thoughtlessness. When understood as an account of all morally bad action, the perversive conception is thus subject to counterexamples as well.⁷

In light of this dialectic situation, I suggest that, rather than trying to subsume all cases of morally bad action under one conception, we should distinguish between two broad kinds of morally bad actions: those that are better described by a privative and those that are better described by a perversive conception of morally bad action.⁸ In the next section, I will show that each of these broad kinds of morally bad action gives rise to obstacles to empathy.

4. Two kinds of obstacles to empathy

Consider, first, the case of morally bad action that is best characterized by a privative conception. This kind of action is committed by agents who, in one way or another, fail to meet the requirements that stem from our normative conception of agency, that is, they do not guide their action by what they take to be their best reasons understood as reasons. However, to the extent that an action is not guided by a sequence of practical reasoning that involves deliberation and leads to a decision, it becomes increasingly difficult to understand via the kind of empathy I have described.

To see why this is so, assume that in trying to understand an agent who commits a morally bad action of this kind, an interpreter comes up with a sequence of practical reasoning that would explain the action. Even if the interpreter were able to do this, in ascribing the corresponding mental states to the agent, they would not arrive at an accurate understanding of what led to their action, but rather generate a specific kind of misunderstanding through projection. This is because the sequence of practical deliberation that leads to a decision that the interpreter came up with does not in fact reflect what was going on in the mind of the agent at the time of the action.

In general, simulation enables us to understand others’ actions by allowing us to access the considerations that recommended the action to the agent first-personally. However, in cases where actions are not preceded by practical deliberation leading to a decision, it is at least possible that the agent’s own understanding of why they acted is inhibited. These are the kinds of cases in which the agent is likely to say in retrospect, ‘I don’t know why I did this’. There thus remains something puzzling about the action, even from the perspective of the agent committing it.

This reflects a known limitation of simulation that proponents of the view have noted. Jane Heal, for instance, has stressed that the domain of simulation is the domain of contentful mental states that are rationally (or at least intelligibly) linked (Heal 2003, 76). Since not all of our action is preceded by explicit deliberation, this might limit the domain in which simulation can help us understand others significantly. However, intelligibility is a matter of degree. In many cases, even if we cannot understand an agent by recreating their reasoning process in imagination because their actions were not guided by deliberation, we might still be able to partially empathize with them, for example, by being able to imaginatively recreate a desire or an emotion that influenced them.

Christine Tappolet has suggested that akratic actions that are committed under the influence of an emotion become intelligible through the presence of an occurrent emotion (Tappolet 2003). Empathizing with the relevant emotion might thus make akratic action at least somewhat intelligible. However, note that the claim is comparative: the presence of an emotion might make the akratic action more intelligible than it would have otherwise been. In addition, not all akratic actions occur because of the influence of an occurrent emotion and not all actions that fall under the general category of morally bad actions best captured by a privative conception are akratic.

Relatedly, Adam Morton has argued that we can understand evil action by relying on analogies with ordinary behaviors in which we overcome boundaries (Morton 2011). While this might contribute to some sort of understanding, what seems to do the work here is not empathy but rather a kind of analogical reasoning. This is presumably also why many ordinary cases of akratic action – such as smoking in spite of having resolved to quit – are not puzzling to us. The degree of intelligibility afforded by analogical reasoning based on our own experience with akrasia is deemed sufficient in these contexts. However, Morton’s strategy is prone to fail precisely because we do not take cases of severe moral transgressions to be similar enough to mundane cases.

While we might thus resort to other strategies that help us attain at least partial understanding, the point remains that there is a degree of understanding through empathy which we cannot attain to the extent that an action is best described by the privative conception of morally bad action. These types of action are not fully intelligible even to the agent who commits them. Therefore, it seems plausible that empathy will at most lead to a partial understanding as well.

Now, consider the case of action that is best described by a perversive conception of morally bad action. It will quickly become obvious that this is the more difficult case to explain. This is because, in this case, there is a sequence of practical reasoning to recreate in imagination, and yet we often seem unable to do so. I want to suggest that this is because, in attempting to understand agents who commit severe moral transgressions, we often encounter a phenomenon that can be described as ‘empathic resistance’.

The phenomenon of empathic resistance is closely related to the phenomenon of ‘imaginative resistance’, which occurs in reading fiction. It has been observed that there seems to be a puzzling asymmetry between our ability to imagine fictional worlds that differ with respect to descriptive facts and our ability to imagine fictional worlds that differ with respect to moral facts. While we happily go along with any stipulations an author may make concerning descriptive facts – including things we take to be physically or even metaphysically impossible, such as time travel – we face imaginative resistance when we are supposed to imagine that what we take to be deeply morally wrong is morally unobjectionable or even praiseworthy.

Different explanations have been offered for this phenomenon. Tamar Szabó Gendler has suggested that imaginative resistance is ultimately due to an unwillingness to take on certain perspectives even in the context of engaging with fiction (Gendler 2000). Against this, others have argued that we cannot imagine deviant moral propositions being true because moral concepts are higher-order concepts. Whether they apply is determined by the lower-level facts (Walton 1994; Weatherson 2004). More recently, Gendler has argued that, even if this is right, we can ignore this kind of inability to imagine in detail and continue the imaginative project by attending to local bits of coherence. Thus, ultimately the right explanation of resistance has to do with our unwillingness, which is based on an unwillingness to be corrupted into taking on certain perspectives on the actual world (Gendler 2006).

While there are important differences between our engagement with fictional worlds and our engagement with the perspectives of real others on the actual world, the phenomenon of imaginative resistance suggests that something similar might play a role in our difficulty to empathize with agents who commit severe moral transgressions.⁹

In the context of imaginative perspective-taking or simulation, resistance might occur at different points and in different ways. First, empathic resistance might occur in the ‘matching phase’. While, in general, imagining allows us to match the beliefs of others we do not share, the phenomenon of imaginative resistance challenges whether this also applies in the case of moral beliefs. If we cannot imagine moral facts we do not take to obtain when reading fiction, perhaps we also cannot do so when trying to recreate someone else’s reasoning in imagination. Because taking on the ‘pretend belief’ that p is just imagining p being the case, this means that we cannot simulate someone else’s moral belief if it conflicts with our deeply held convictions. Like in the case of imaginative resistance, this can be explained by appealing to a combination of an inability to imagine higher-order concepts applying in situations with conflicting lower-level facts in detail, and an unwillingness to be corrupted into taking on certain perspectives.

Similar restrictions might apply to other types of mental states, such as desires and emotions. As noted above, there is no consensus on whether in simulation, we rely on desires and emotions in imaginative contexts or on imaginative counterpart states to desire and emotion. However, regardless of which option one favors, we can account for resistance we encounter in attempting to recreate certain desires and emotions. Even those who have opted for a counterpart state to desire have argued that which ‘desire-like imaginings’ we are willing to adopt is constrained by our own attitudes (Currie 2002).

Alternatively, empathic resistance might occur in the simulation phase. This is because we might be unable to quarantine some of the attitudes associated with our own deeply held moral convictions. According to Karsten Stueber, when we hold moral commitments, we not only form the corresponding belief, but also become ‘emotionally attuned’ to them, in the sense that we tend to react emotionally in accordance with these commitments. Stueber suggests that we might be unable to quarantine these attitudes in simulation and that they thus interfere with the simulation process so that we are unable to recreate the thought processes that lead to the decision to commit a morally bad action (Stueber 2006, 213).¹⁰

These different mechanisms underlying empathic resistance might play a role in different cases. Perhaps we are unable to match the distorted moral beliefs of the fanatic, cannot recreate the desires of a sadistic evil schemer, and fall short of weighing considerations as the egoist does in light of their imperviousness to moral concerns. In any case, we encounter empathic resistance because our imagination is constrained by our own moral commitments. Therefore, each of the two kinds of morally bad action I distinguish give rise to obstacles to empathizing with the agent. While morally bad actions that are best captured by a privative conception are difficult to understand because there is no coherent practical reasoning to recreate in imagination, morally bad actions that are best captured by a perversive conception are difficult to understand because we encounter empathic resistance.

This allows me to offer the following disjunctive explanation of the phenomenon that we often encounter difficulties in trying to understand severe moral wrongs: Severe moral transgressions are often difficult to understand because we seek to understand the agent by empathizing and fail either because there is no coherent practical deliberation to recreate in imagination or because we encounter empathic resistance. In the next section, I will consider whether we should try to overcome this limitation to our understanding.

5. Should we empathize with agents who commit severe moral transgressions?

My discussion thus far can be understood as identifying a distinctive kind of limit of our capacity to empathize. Extending our empathy to a wide range of perspectives is often considered something we should strive for. This raises the question whether we should try to overcome this specific kind of limitation. The explanation offered is relevant to this question because it indicates whether overcoming this limitation to our understanding is possible and what would be required to do so. In the case of morally bad actions best characterized by a privative conception, it suggests that we encounter a limitation to our ability to empathize that is impossible to overcome. Our only option is to resort to other strategies of understanding. In the case of morally bad actions best characterized on a perversive conception, by contrast, it indicates that we encounter a limitation to our ability to empathize that we could in principle overcome. But should we try to overcome this particular boundary?

One way in which we could try to overcome this boundary is by giving up on our moral commitments. However, assuming that our moral judgments are reliable at least concerning severe moral transgressions, this would not be advisable.¹¹ But the explanation offered above suggests that there are other ways we could attempt to expand our ability to empathize: we could instead work on our unwillingness to entertain certain perspectives even in imagination or better quarantine our own moral commitments. But would it be right for us to practice our imagination in this way?

This question can be approached from different directions, which indicate different and partially conflicting considerations. When we consider our relation to the agent, we might see reason to extend empathy. Olivia Bailey has argued that there is at least some reason for the virtuous to empathize with mildly vicious perspectives because being empathized with is a basic human need (Bailey 2021). While this argument provides some consideration in favor of empathizing, it has less force in the context of severe moral transgressions.

When we consider our relation to the victim, we might see a reason to resist extending empathy. As Morton has pointed out, the point of view of the victim is usually that of incomprehension (Morton 2004, 14). Empathizing with the victim might thus require at least being able to see the action as incomprehensible. In addition, as my discussion of empathic resistance indicates, empathizing with the perpetrator might require us to take on a perspective on the victim that we might not want to have available.

When we consider our relation to ourselves, another set of considerations becomes salient. Morton and Stephen Grimm have argued that understanding wrongdoers allows us to see ourselves as similar to them (Grimm 2018; Morton 2004). In a similar vein, Anja Berninger has suggested that empathizing with fictional wrongdoers undermines our confidence in our own moral character (Berninger 2021). These authors have suggested that this is something we should seek out because it enables a self-critical stance. However, Berninger also cites evidence that we are more likely to forgive wrongs if we see ourselves as capable of doing similar actions and marks it as an open question whether this would always be the most appropriate attitude (Berninger 2021, 249).

Finally, we might consider wider effects on society. When we consider this, it seems that at least some (for example, therapists and detectives) should be able to understand wrongdoers if this contributes to prevention or moral repair. However, it is not clear whether the relevant kind of understanding in these contexts is empathic understanding. Even if the relevant understanding turns out to be empathic understanding, this might still allow for a division of labor, in which only some members of society empathize with wrongdoers and those who do receive special kinds of training and support.

On balance, it seems that there are at least some reasons, having to do with our relation to the victims of severe moral transgressions, to avoid seeking empathy with agents who commit them. While there might be countervailing reasons stemming from our relation to the perpetrator, ourselves, or wider effects on society, these considerations are inconclusive. There is thus a sense in which severe moral transgressions are and must remain incomprehensible: A certain degree or kind of empathic understanding is beyond our reach either because the action does not allow for this kind of understanding or because there are ethical reasons to refrain from the exercise of imagination understanding would require. However, because empathy is not the only way to achieve some sort of understanding, we can give up our aspiration to understand the action ‘from the inside’ and resort to different, more theoretical ways of understanding.

6. Conclusion

My starting point was the observation that, sometimes, when we learn about a severe moral transgression that has been committed, we are not only appalled but also puzzled. In these situations, we are likely to express an inability to understand. This reaction suggests that there is something difficult to understand about severe moral wrongs. My aim was to offer an explanation of this phenomenon. I have started by arguing that the relevant sense in which we try but often fail to understand morally bad action is that we try but fail to empathize with the agent at the time of the action. Then, I introduced a distinction between two broad kinds of morally bad action. Finally, I showed that each of these kinds of morally bad action gives rise to obstacles to empathizing with the agent at the time of the action. According to the explanation offered, severe moral transgressions are often difficult to understand because we seek to understand the agent by empathizing and fail either because there is no coherent practical deliberation to recreate in imagination or because we encounter empathic resistance. Finally, I have considered whether we should try to overcome this limitation and suggested that we have some reason to avoid empathy in these contexts.

Acknowledgments

For helpful comments on earlier versions of this paper, I thank Daniel Sharp, members of Neil Roughley’s Research Group Colloquium, audiences at the Virtual Metaethics Colloquium, the Conference Interpersonal Understanding: Transcending Boundaries? at the University of Duisburg-Essen, and the FS Philosophy Forum at the Frankfurt School of Finance & Management as well as two anonymous referees.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

Work on this paper funded by the DFG/AHRC-Project ‘How Does it Feel? Interpersonal Understanding and Affective Empathy’.

Notes on contributors

Katharina Anna Sodoma

Katharina Anna Sodoma is a Postdoctoral Research Assistant at LMU Munich. Her research focuses on topics at the intersection of ethics and political philosophy, epistemology, and philosophy of mind.

Notes

1 That there is a close connection between severe moral wrongs and failures of intelligibility has been pointed out by philosophers interested in evil (Russell 2012; 2020; Morton 2004; Neiman 2015). However, the notion of evil raises many important questions, which remain matters of ongoing controversy, such as what distinguishes evil from ‘merely’ morally wrong actions and whether we ought to use the category of evil at all. While, in what follows, I will draw on the work of philosophers interested in evil, I will not take a stance on these issues.

2 Matters are admittedly less straightforward when it comes to the simulation of other types of mental states, such as desires and emotions. Authors disagree on whether the best account of simulation (as well as other uses of the imagination) must stipulate imagination counterpart states to desires and emotions or just appeal to desires and emotions themselves (Currie and Ravenscroft 2002; Currie 2002; Kind 2011; 2016a; Roelofs 2023). However, these questions are beyond the scope of this paper and need not be settled for the purposes of the argument to follow.

3 This suggestion is closely related to Luke Russell’s claim that the most plausible sense in which evil actions might be incomprehensible is that we cannot imagine choosing them (Russell 2012, 68f; see also 2020, 41f). However, while in Russell’s scenario we try to imagine ourselves in another person’s situation, choosing the action they chose, and fail, in empathy we try to recreate another person’s perspective.

4 This characterization of the wanton naturally raises questions of responsibility. Addressing these questions is beyond the scope of this paper. However, it seems that, in general, an agent’s responsibility will only be mitigated in case they cannot exercise their capacities for normative guidance but not if they simply fail to do so.

5 Of course, this is not to say that they act in accordance with what are in fact their best reasons (Worsnip 2019).

6 Even though it cannot be understood as an account of all morally bad action, but rather must be understood as an account of the actions of a particular historical individual against the background of collective wrongdoing, these different conceptions of morally bad action also play a role in Hannah Arendt’s influential reflections on evil in connection with the Eichmann trial (Arendt 2006; 2003) as well Bettina Stangneth’s more recent interpretation (Stangneth 2016).

7 Strictly speaking, Noller is discussing different theories of ‘evil’ rather than of all morally bad action. However, my aim in this section is not to refute any specific theory of morally bad or evil action, but rather to map out two possible positions and argue in favor of a view that combines them.

8 In suggesting this, I do not mean to preclude that there are other possible distinctions (or further distinctions within each of the broad kinds) that it might be useful to draw. In addition, I take it to be plausible that many real-life cases of morally bad action will have both privative and perversive elements.

9 A full treatment of the extensive literature on imaginative resistance and its implications for empathy is beyond the scope of this paper. However, in what follows I hope to make a start on theorizing ‘empathic resistance’. For a different exploration of the phenomenon see (Szanto 2020).

10 In a different context, Stueber has argued that it is in light of the humanity of the victims that certain perspectives become impossible to recreate (Stueber 2011, 178). One way to interpret this interesting remark is that empathic resistance might occur because we face difficulty in stabilizing the result of empathy while at the same time relating the perspective of the agent to the environment – in particular, to the agents affected by their action.

11 I take it that the idea that only becoming disposed to do morally bad actions could enable us to understand them implicitly underlies statements such as the following made by Russell: ‘Perhaps you can imagine yourself choosing to shoplift and commit other ordinary wrongs, at least in some circumstances. In contrast (I hope!) you cannot imagine ever choosing to strap on a suicide bomber’s vest, walk into a church in Sri Lanka on Easter Sunday, and kill hundreds of innocent worshippers.’ (Russell 2020, 41f, my emphasis) Although note that Russell is here concerned with imagining choosing oneself, which is slightly different from, albeit closely related to, simulating another person’s decision.