Moral Appraisal for Everyone: Neurodiversity, Epistemic Limitations, and Responding to the Right Reasons

Abstract

De Re Significance accounts of moral appraisal consider an agent’s responsiveness to a particular kind of reason, normative moral reasons de re, to be of central significance for moral appraisal. Here, I argue that such accounts find it difficult to accommodate some neuroatypical agents. I offer an alternative account of how an agent’s responsiveness to normative moral reasons affects moral appraisal – the Reasonable Expectations Account. According to this account, what is significant for appraisal is not the content of the reasons an agent is responsive to (de re or de dicto), but rather whether she is responsive to the reasons it is reasonable to expect her to be responsive to, irrespective of their content. I argue that this account does a better job of dealing with neuroatypical agents, while agreeing with the De Re Significance accounts on more ordinary cases.

I make trouble for accounts of moral appraisal that understand a particular kind of moral reason, normative moral reasons de re, to be particularly significant for moral appraisal. I show how these accounts are unable to satisfactorily accommodate cases involving neuroatypical agents, and I offer an alternative account, the Reasonable Expectations Account, that avoids placing any particular weight on which normative moral reasons agents are expected to respond to. I show how this does a better job of accommodating cases involving neuroatypical agents while agreeing with De Re Significance accounts on the cases they get right.

Normative Moral Reasons De re

Normative moral reasons de re are the features of a situation that make actions morally right or wrong. For example, the fact that insulting someone is unkind is, perhaps, a normative moral reason that makes insulting people morally wrong. Normative moral reasons are to be distinguished from both motivating reasons – the reasons an agent is motivated by in acting-- and explanatory reasons – the reasons that explain why an agent acted as she did. What, exactly, the normative moral reasons are depends on what, in fact, makes actions right or wrong. For example, if the correct first-order moral theory is utilitarianism, then the normative moral reasons are whatever contributes to happiness maximisation. If Kantianism is true instead, then the normative moral reasons are whatever contribute to treating others as ends in themselves.¹ Normative moral reasons de re are to be contrasted with normative moral reasons de dicto, which concern only moral valence. For example, that action X is morally wrong is a normative moral reason de dicto not to do it. Appropriate responsiveness to normative moral reasons de re is thought by some to be particularly significant for moral appraisal.² For example, Elizabeth Harman emphasises its importance for determining blameworthiness:

An action is blameworthy just in case the action resulted from the agent’s caring inadequately about what is morally significant – where this is not a matter of de dicto caring about morality but de re caring about what is in fact morally significant (2011: 460).

And, Nomy Arpaly emphasises its importance for determining praiseworthiness:

For an agent to be morally praiseworthy for doing the right thing is for her to have done the right thing for the relevant moral reasons—that is, the reasons for which she acts are identical to the reasons for which the action is right. […] For an agent to be morally praiseworthy for her right action it is not sufficient that her action be motivated by a desire to do what is right (2002: 73-4).

These accounts – De Re Significance accounts – hold that blame is always deserved when the agent fails to respond appropriately to normative moral reasons de re, and praise is deserved only if the agent responds appropriately to normative moral reasons de re. In other words, responsiveness to normative moral reasons de re is central to determining moral appraisal.³

De Re Significance Claim: Moral appraisal is determined entirely by the agent’s responsiveness to normative moral reasons de re.

The model of appraisal that typically underpins De Re Significance accounts makes three core claims:

1. Moral appraisal is determined by the quality of the agent’s will.

2. Quality of will is deficient when the agent is insufficiently responsive to normative moral reasons.

3. Normative moral reasons are always de re and never de dicto.

In short, agents are blameworthy when and to the extent that they exhibit deficient quality of will, and agents who are unresponsive to normative moral reasons de re manifest deficient quality of will.⁴ Here, I argue that De Re Significance accounts are mistaken because not every case in which an agent fails to respond to a normative moral reason de re should be viewed as a case in which the agent exhibits deficient quality of will.

The De Re Significance Claim is often used as part of an explanation of why moral ignorance cannot excuse. Moral ignorance that leads to wrong action has been thought necessarily incompatible with appropriate responsiveness to normative moral reasons de re and so unable to mitigate an agent’s blameworthiness for that wrong action.⁵ According to De Re Significance accounts, agents who fail to be appropriately responsive to normative moral reasons de re are always blameworthy – regardless of their epistemic situation. One important result of the discussion will be that we have reason to resist this strong claim about moral ignorance.

The following section discusses some cases that are particularly difficult for De Re Significance accounts to deal with. Section 3 discusses the strategies for dealing with these cases that are available to proponents of De Re Significance accounts and argues that they are unsuccessful. Section 4 adapts these cases so as to put further pressure on the idea that failing to be appropriately responsive to normative moral reasons de re always indicates deficient quality of will. Section 5 presents an alternative account of moral appraisal – the Reasonable Expectations Account – that does a better job of dealing with these cases, while agreeing with De Re Significance accounts on the more ordinary cases they get right. On this account, what is significant for appraisal is responsiveness to normative moral reasons that it is reasonable to expect the agent to recognise. This allows moral ignorance to excuse in some cases.

Moral Limitations

This section discusses cases of agents who are unresponsive to the normative moral reasons de re of their situation because their epistemic circumstances make it very difficult for them to be appropriately responsive. These cases present a problem for De Re Significance accounts of moral appraisal. On the one hand they seem committed to saying that these agents are blameworthy because they fail to be appropriately responsive to normative moral reasons de re. On the other hand, the agent’s epistemic circumstances make it doubtful whether these cases should be interpreted as cases of deficient quality of will.

To illustrate the shape of the examples, here is a non-moral example. My capacity for culinary excellence is, sadly, limited. One explanation for this is that I am not sufficiently responsive to what is gastronomically important – normative gastronomical reasons de re. This makes me bad at cooking. Suppose, as many people think, that one should add salt when cooking pasta. If this is true, there will be reasons for this. Perhaps, “that it improves the texture of the pasta”, or “that it improves the taste of the pasta”. Whatever these reasons are, I am not sufficiently responsive to them. I have never been able to discern the difference between pasta cooked with added salt and pasta cooked without it. I also know that I am no expert – I assume there is some reason to add salt to pasta, and I know I fail to appreciate it. Nevertheless, it is possible for me to cook well, or well enough, without ever properly appreciating the reasons there are in favour of adding salt. When the gastronomical stakes are high – for example, when I am cooking for others – I know not to rely on my own culinary abilities. I use recipe books, or ask for advice. Given what I know about my own limitations, it seems that this is exactly what I should do – defer to culinary testimony, and not risk relying on my own flawed understanding of what is gastronomically important. Nevertheless, recipe books sometimes contains misprints. If my recipe contains a misprint, and I fail to add salt because of this, I do the wrong thing but – perhaps – do not deserve blame. The cases I discuss are structurally similar, but the relevant reasons are moral rather than culinary.

This section focuses on fleshing out possible examples of agents whose position with respect to moral reasons is similar to mine with respect to culinary reasons and shows how De Re Significance accounts will find these difficult to deal with. I focus on cases involving autism and psychopathy. These are the most widely discussed by philosophers, and the neuroatypicalities most obviously relevant to the capacity to appreciate moral reasons. However, the arguments I make here are applicable beyond these cases, to any robust psychological features that ground limitations in appreciating moral reasons. Section 3 pre-empts some strategies that De Re Significance accounts might use to deal with these cases, arguing that these will not be successful. Section 4 goes on to argue that these accounts face further problems when appraising agents like this who defer to moral testimony but do the wrong thing because that testimony turns out to be misleading.

Autistic Agents

Autism seems to present epistemic limitations in responding to normative moral reasons de re.

Truthfulness. Mike is autistic, and this makes it difficult for him to imagine the internal mental lives of other people, including the more complex aspects of how others will feel in response to his actions. Since he finds other people’s emotions difficult to imagine, he also finds it difficult to see them as reasons for and against actions. Mike desires the well-being of others, and he believes that others’ well-being is always served by their knowing the truth. Sometimes people’s feelings are hurt when he is too truthful, but he finds it difficult to predict such occurrences. He usually only notices that he has ‘put his foot in it’ when his friends and family explain to him why that person was upset, and tell him that he must try to be more sensitive. Moreover, explanations of why these people are upset strike him as confusing and a little far-fetched. While he tries to be charitable, he struggles to believe that it could really be so morally important to avoid hurtful assertions, particularly when this comes at the expense of saying things that are relevant and true.

Autism is characterised by impaired social interaction, deficits in empathy, and impaired understanding of other people.⁶ These all contribute to difficulties in understanding the social world, including aspects of the social world that are morally significant. Mike’s autism thus makes him prone to failing to be appropriately responsive to some normative moral reasons de re, as in Truthfulness. Assuming that, ceteris paribus, ‘that someone’s feelings would be hurt’ is a normative moral reason de re to avoid doing that which would cause hurt feelings, Mike systematically fails to be appropriately responsive to this. Not only is Mike unable to reliably recognize hurt feelings, he is sceptical of the importance of avoiding hurting feelings in some cases.

Of course, just as I can use a recipe book when cooking, Mike has strategies available to him that can help him understand other people well enough to avoid hurting them. For example, he can ask people he trusts, or spend time learning different facial cues. However, while these strategies will help to some extent, they are unlikely to be as effective as the resources for social understanding available to neurotypical people. When he inevitably slips up, this does not indicate deficient quality of will. The problem for De Re Significance accounts is that they cannot accommodate this. In so far as he fails to be appropriately responsive to the normative moral reasons de re against asserting hurtful truths, De Re Significance accounts are forced to say that Mike is blameworthy.

There are two ways De Re Significance accounts might attempt to avoid this implication. First, they might respond that the features of the situation that Mike’s autism causes him to miss are mere details, and not among what is morally important (de re). Provided he cares about more general morally important things, such as the well-being in others in general, then De Re Significance accounts need not say think that he has deficient quality of will, and can avoid saying he is blameworthy. This is what Arpaly and Schroeder (2013) suggest. They  see no conflict between their account of moral appraisal and neuroatypical psychology, saying, “there are times when the virtuous person fails to feel the right emotions, think the right thoughts, or attend to the right things. […] autism, absent-mindedness, and the like are all morally neutral. Even a perfectly virtuous person can be mentally retarded, manic, depressed, autistic, or absent-minded.” (2013, 201). Their view is that quality of will is determined entirely by the agent’s desires – whether they intrinsically desire the good, correctly conceptualized. Provided Mike has the right desires, he need not be blameworthy.

This approach is plausible in some cases. For example, compare the following case, in which the agent fails to appreciate the normative moral reasons de re because he is distracted.

Distraction. Stefano is trying to decide how to allocate office space. The available space is such that not everyone can have an ideal office. Various considerations contribute to determining the fair allocation of the available office space (how much each researcher uses the office, their accessibility needs, etc.). Working out the fairest allocation is complicated, and to help him, Stefano has taken detailed notes of these various considerations. Part way through the allocation process, Stefano is distracted by a seagull crashing into his window. This causes him to miss a line in his notes, meaning that some of the researchers’ needs are not factored into the allocation fairly, and those researchers end up  worse off than the others.

De Re Significance accounts should not want to say that Stefano is blameworthy. Assuming he has taken appropriate precautions against becoming distracted, it seems incorrect to attribute deficient quality of will to him. So, De Re Significance accounts must find some way of denying that Stefano is unresponsive to the normative reasons of the situation. To this end, they might argue that the details he misses are not, in fact, among the normative moral reasons de re. Perhaps the normative moral reasons de re include only more abstract things such as fairness in general. Stefano then need not be blameworthy so long as he is appropriately responsive to these. Alternatively, perhaps the details are among what is morally important (de re), and Stefano is appropriately responsive to them in virtue of, for example, writing them down, and aiming to take them into consideration. Perhaps it is also relevant that he would have responded appropriately to these considerations had he not become distracted.⁷

Assuming that De Re Significance accounts have the resources to accommodate cases like this, we might think that agents like Mike can be accommodated similarly. However, it is not clear that they can. Unlike Stefano, Mike does not quite have the right desires. For example, he does not desire that he avoid asserting hurtful truths – he is sceptical of the importance of this. Furthermore, some aspects of the situation that he is not sufficiently responsive to are not mere details – they are morally significant. Being appropriately responsive to the consideration ‘that it would hurt someone’s feelings’, requires direct responsiveness to the moral badness of hurting other peoples’ feelings. Proponents of De Re Significance accounts emphasise the importance of this intrinsic concern for normative moral reasons de re. However, in some situations Mike is oblivious to the moral badness of hurting other people’s feelings. Mike is not responding to the right reasons if he only avoids saying the hurtful thing in response to some other, more general consideration. If indirect responses to more general reasons could also count as appropriate responses to normative moral reasons de re, this would put pressure on the significance of normative reasons de re, and thus on the central project of De Re Significance accounts.

Another way that De Re Significance accounts might attempt to deal with agents like Mike is by denying that they are appropriate targets of moral appraisal. Some have thought that full moral agency requires a degree of emotional empathy, and the ability to enter into another’s perspective-- something that autistic people characteristicaly lack (Blair 1995; Hobson 2007; Shoemaker 2015, 168).⁸ Others have seen moral agency as requiring the ability to participate in a moral conversations involving the exchange of moral reasons⁹ – which according to De Re Significance accounts would have to be de re and not de dicto. De Re Significance accounts might on these grounds argue that autism precludes full moral agency. However, autistic people often seem very able to participate in moral life by making choices, deliberating, and adopting moral rules. Temple Grandin, well-known both for her autism and her work on humane cattle slaughterhouses, is clearly guided by a deep moral commitment to improving animal welfare,¹⁰ and so seems to possess the necessary abilities to participate in moral life. In support of this, Krahn and Fenton (2009) argue that high-functioning autistic people have the capacity for cognitive (but not emotional) empathy, and this is sufficient for moral agency, and Shoemaker (2007) argues that autistic lack of empathy need not imply a lack of caring about other people, and it is this (rather than emotional empathy) that is needed for full participation in moral practices.¹¹

Taking a slightly different approach, Kennett (2002) argues that empathy (cognitive or emotional) is not a necessary condition of moral agency, but merely one method to discover information about the moral landscape. It is primarily an epistemic tool. This implies that autistic people need not be thought to have their moral agency compromised, so long as they can find alternative methods to discover the relevant information – for example, relying on testimony, or studying moral rules and theories.¹² While autistic people may lack the emotional resources to understand other people as well or as quickly as neurotypical people, they face no particular barriers to using these other methods.

If this this is right, then we should not think of the limitations involved in autism as putting agents beyond moral evaluation. However, the alternative methods that autistic people are able to make use of will not always be as effective epistemic routes to the relevant moral information as ordinary neurotypical empathy. It is implausible that relying on these routes will never cause autistic agents to fail to appreciate moral reasons – Mike’s case is one example of this.¹³ So long as it is possible that some of the information that these alternative routes miss is more than mere details, autistic agents will sometimes face an impaired ability to be appropriately responsive to the normative moral reasons de re. Lack of responsiveness to normative moral reasons de re in these circumstances does not seem to be a manifestation of deficient quality of will, but rather a different – neuroatypical – way of approaching the world.

In other words, and as Kennett puts it, “there is more than one way to be a moral agent” (2002: 357). It would be a problem for De Re Significance accounts if they could not accommodate this, and it turned out that they were forced to treat behaviour that is merely neuroatypical as if it were blameworthy. De Re Significance accounts risk doing this because they insist that to avoid blame, the agent must be responsive to a particular set of considerations (the normative moral reasons de re). So long as there are possible neuroatypicalities that prevent proper appreciation of this set of considerations, these neuroatypicalities will be difficult for De Re Significance accounts to accommodate. They will be forced to say that agents who fail to respond appropriately to these considerations are either blameworthy or less-than-full moral agents.

Psychopaths

Psychopathy presents another kind of neuroatypicality that De Re Significance accounts find it difficult to deal with.

Distinction. Bonnie consistently fails to appreciate the distinction between moral and conventional wrongs. She has managed to learn by heart most of the actions typically considered wrong, either morally or conventionally, and has found that avoiding these actions is usually a good idea (for example, it avoids unpleasant confrontations with the law). She knows that some of these actions are considered ‘morally’ wrong, and therefore more important to avoid than the others, but she is unable to feel the force of the distinction herself. One day, she is waiting for a taxi in the rain when a pregnant woman arrives at the taxi rank, urgently needing a taxi to take her to the hospital. While most neurotypical people would feel morally compelled to offer the first taxi to the pregnant woman, this does not occur to Bonnie. The situation does not fall under the description of any of the actions that she has learned are wrong, either morally or conventionally. Bonnie takes the first taxi when it arrives, and does not allow the pregnant woman to go first.¹⁴

Psychopaths characteristically fail to appreciate the distinction between ‘moral’ and ‘conventional’ norms.¹⁵ Some have taken this to imply that psychopaths lack crucial competencies relevant to morality. They are unable to “grasp moral concepts”, and they lack “sensitivity” to moral harms and wrongings (Levy 2008: 166).¹⁶ These limitations would seem to preclude psychopaths from being appropriately responsive to normative moral reasons de re. Bonnie is unable to appreciate the moral reasons to let the pregnant woman have the taxi, because she is unable to appreciate the moral significance of considerations such as the woman’s suffering, or the value of alleviating it.

How should De Re Significance accounts deal with this case? On the one hand, since Bonnie is unresponsive to normative moral reasons de re, De Re Significance accounts seem committed to evaluating her as blameworthy. Indeed, many proponents of De Re Significance accounts have thought that psychopathic indifference towards moral considerations is paradigmatic of what it is to exhibit a negative quality of will, and is a manifestation of deficient quality of will no matter the circumstances (Scanlon (1998: 284); Arpaly (2002); Watson (2004: 266); Harman (2015); Mason (2015); Weatherson (2019)). The problem, they hold, is that Bonnie does not care enough about what is important. Bonnie manifests deficient quality of will because failing to respond appropriately to normative moral reasons de re always constitutes deficient quality of will.

However, an important feature of this case is that the moral transgression Bonnie commits is relatively subtle. Unlike other, more egregious moral transgressions, failing to offer taxis to pregnant women is not prohibited by law, and there is no one else around offering their judgments on Bonnie’s actions. She lacks information about the normative features of the case – she is unable to appreciate the salient moral reasons for herself, and she has very little other input. Compare the following case:

Murder. Bonnie is waiting in the rain, under an umbrella, for a taxi at a currently empty taxi rank. She is in no particular hurry. Soon she is joined by pregnant woman who needs to get to the hospital as soon as possible. To lessen the boredom of the wait, she decides to murder the pregnant woman and hide her body before the first taxi arrives.

Murder is a striking and obvious example of something thought to be morally wrong. Most legal systems prohibit it, and even the most basic knowledge of society’s conventions indicate that it is thought to be wrong. One does not need to appreciate the distinctions between moral and conventional prohibitions to work out that murder is thought by most to be wrong, and if they are right, then it is very wrong. Just as I can use a recipe book when cooking, Bonnie can use information about legal or conventional prohibitions to avoid actions that might be wrong.¹⁷ It would be more plausible that she was blameworthy in a case like this.

Having presented the cases under discussion, the following section pre-empts some strategies that De Re Significance accounts might use to deal with these cases.

Possible Responses to the Cases

Of course, it is open to proponents of De Re Significance accounts to accept these implausible verdicts about blameworthiness and claim that any intuitive resistance we might have to them is mistaken.

Indeed, some have done exactly this. Arpaly and Schroeder emphasise that although agents with autism (or other neuroatypical traits) face no barrier to being praiseworthy – provided they are responsive to what is in fact morally important (2013, 201). Arpaly and Schroeder do not entertain the idea that autism might make appropriate responsiveness to salient moral considerations very difficult. However, elsewhere they make clear that in so far as an agent fails to be appropriately responsive to what is morally important, she is blameworthy – even if it is not her fault that she is unresponsive (2013, 186). So, they are forced to conclude that although autistic agents can in principle be appropriately responsive, if they are not, the difficulties they face afford them no excuse. I do not have much to say in response to this, other than that it seems implausible. While my opponents will not be persuaded by this, it is helpful to highlight the implications of De Re Significance accounts for those who might want to adopt them. Section 4 goes on to argue that there are additional and hitherto unanticipated implausible implications of these accounts when agents who face moral limitations try to use moral testimony to do the right thing.

Alternatively, De Re Significance accounts might respond that Bonnie is not an appropriate target of blame, because she is not a fully responsible moral agent. As Levy puts it, “psychopaths do not possess the relevant moral knowledge for distinctively moral responsibility; lacking this knowledge, they are unable to control their actions in the light of moral reasons.” (2008, 129). Similarly, Shoemaker argues that psychopaths are ‘marginal’ agents because they lack sufficient regard for others (2015).

However, there is a sense in which neither of these seems like the right response. The psychopath’s limitations are structurally similar to Mike’s (and to some extent Stefano’s) limitations, so we might expect similar treatment. All three are cut off from some morally relevant aspects of their environment, and all three seem to possess sufficient agency to manage their actions (and so do not seem beyond moral appraisal entirely). If we were reluctant to blame Mike, we should also be reluctant to blame Bonnie. And, if we think that Bonnie’s limitations in appreciating moral reasons put her beyond moral appraisal, we should also think that Mike’s put him beyond moral appraisal. Like Mike, she fails to weigh the importance of moral considerations correctly. Bonnie does not think that moral considerations are more important than conventional wrongs, and Mike does not think that hurt feelings are more important than truth-telling. The difference between them is that Bonnie’s failures are more general. Furthermore, psychopaths are in many respects ordinarily responsible agents. They make decisions for reasons, act freely, and understand conventional and legal rules against various wrong actions. Some have thought this is enough to ground moral appraisal (Vargas and Nichols 2007; Blair 2008; Greenspan 2016). This route is not open to De Re Significance accounts – for them, it is only responsiveness to normative moral reasons de re that is relevant in moral appraisal. So, unless there are moral reasons to obey conventional norms, psychopaths cannot be blameworthy for disobeying conventional norms. So, De Re Significance accounts must either say that psychopaths are exempt from moral appraisal, or that they are blameworthy for disregarding normative moral reasons de re.

Viewing Bonnie (and perhaps other neuroatypical agents) as possessing less-than-full moral agency is problematic for a further reason – it is at odds with our usual approaches to limitations in competence. It means banishing many neuroatypical individuals from the moral community. Compare our approaches to physical limitations, for which it is usually considered appropriate to accommodate individuals within the community, rather than banish them. We typically expect wheelchair ramps to be provided to allow physically limited individuals to access public buildings and participate in the moral and social practices that take place there. Were we to, instead of doing this, take something like Strawson’s (1962) ‘objective stance’ towards wheelchair users and simply exclude them from these practices, this would not be thought acceptable.

Not only this, but exempting neuroatypical people from the demands of morality, or taking the objective stance towards them, depends on the possibility of drawing a clean line between neuroatypical agents and the rest of us. It is not obvious that this is possible. The psychological characteristics that contribute to moral competence reasoning plausibly come in degrees, and vary among the population. Even the most neurotypical of us will have found ourselves in morally unfamiliar situations in which our epistemic access to salient moral considerations is limited. For example, when travelling in an unfamiliar culture, or puzzled or mistaken about the correct first-order moral theory.¹⁸ Neuroatypical agents can be thought of as extreme examples of these more everyday situations.

The previous section showed that there are possible cases of unresponsiveness to normative moral reasons de re in which it is both appropriate to treat the agent as ordinarily responsible, and it is questionable whether the agent should be interpreted as manifesting deficient quality of will, because they face epistemic barriers to accessing the normative moral reasons de re. De Re Significance accounts cannot easily accommodate these cases. As this section has argued, they are forced to either blame neuroatypical agents who fail to respond to normative moral reasons de re, thus dismissing the moral relevance of the psychological limitations these agents, or inappropriately view them  as less-than-full moral agents. In Section 5 I will argue that responsiveness to normative moral reasons de re is not all that matters for moral appraisal, and propose an account that avoids having to endorse either of these unappealing options for dealing with neuroatypical agents.

The following section shows how these cases present a particular challenge for De Re Significance accounts when agents with these limitations try to do the right thing by relying on testimony.

Bad Advice

This section argues that there are additional respects in which De Re Significance accounts are forced to make implausible claims about neuroatypical agents – once we develop the cases from the previous section into cases of acting wrongly on the basis of misleading testimony. I highlight how the agents’ epistemic situations are such that responding to the normative moral reasons de re of the situation is neither the only nor the best way for them to act well. If they are aware of this, and deliberately act in a way they have good reason to think will enable them to do the right thing, they deserve positive moral appraisal (that is, they at least deserve an excuse, and we might additionally think they deserve praise). In contrast to De Re Significance accounts, I argue that this is true even if they are only responsive to normative moral reasons de dicto. I argue that De Re Significance accounts are forced to evaluate neuroatypical agents who are trying to do the right thing as blameworthy, and this is a problem because these agents are not manifesting a deficient quality of will. I argue that to accommodate these cases successfully, we should think that on some occasions, moral ignorance does excuse (pace De Re Significance accounts).

These cases are constructed in the following way. An agent with a limitation in their capacity to be appropriately responsive to some relevant normative moral reasons de re faces a difficult moral choice. They also know that this is the sort of case in which they are prone to making the wrong choice. Preferring to do the right thing rather than the wrong thing, but remaining neutral on what that is, they seek out and act on moral advice. Unfortunately, the advice is misleading and they do the wrong thing. These cases are much like the example of the incompetent chef cooking pasta for others and following a recipe. It is worth noting that acknowledging the moral commendability of this behaviour is consistent with the widespread claim that in general there is something morally non-ideal about deference to moral testimony. Philosophers have offered various ways of cashing out exactly what is wrong with deference to moral testimony. For example, that it undermines moral worth (Fletcher 2016), or inappropriately outsources our moral agency (Nickel 2001; Hopkins 2007).¹⁹ Agreeing that there is something non-ideal about moral testimony need not preclude thinking that there are some situations in which one should defer and would be praiseworthy for doing so. For example, when the agent knows that her judgment is likely to be compromised (see Enoch (2014)). Here, I am claiming only that in some situations deference to moral testimony is the best option an agent has, and if this leads them to do something wrong, they should not be evaluated as blameworthy, because trying to do the right thing using the best available means you have does not constitute deficient quality of will.

Indeed, it is plausible that this is exactly what they  should do. Agents who have limited capacities for moral reasoning should avoid relying on their own abilities when they find themselves in high stakes moral situations – just as I should use a recipe book when cooking. In so far as they are aware of their limitations, they should make use of strategies to compensate for them – for example, seeking out reliable moral testimony. Since they should do this, it would be inappropriate to blame them. Even the wisest moral advisors are occasionally mistaken. When this happens, agents who relied on their advice will do the wrong thing. And, it seems correct to say that they acted out of moral ignorance – not knowing what the right thing to do is, they do whatever the advisor says is the right thing to do, because they believe it was the right thing to do, while ignorant of what the right thing actually is. However, if the above is correct, then they are not blameworthy because they were doing what they ought to do, and not manifesting deficient quality of will. So, this seems to be a case in which moral ignorance excuses.

However, De Re Significance accounts cannot agree with these approving claims about employing strategies to compensate for moral limitations, because the testimony that the agents would need would be of the following form:

“Do X, because X is morally right.”²⁰

This testimony cites only the action’s moral value – that X is the right thing to do – as the reason to do X. This makes it a normative moral reason de dicto (and not de re). Furthermore, we can assume that this is the only reason the agents are responsive to – we have stipulated that these are agents who lack the capacity to be responsive to the relevant normative moral reasons de re. This means that when the advice is misleading, the agents are blameworthy, and their attempts to do the right thing cannot redeem them. However, this seems like the wrong assessment. That X is morally right is a good reason for agents with limited moral abilities to do X, something for which they should be praised rather than blamed.²¹ Indeed we might blame them for not following the advice.

The problem is that the idea that moral appraisal depends entirely on responsiveness to normative moral reasons de re is plausible, at best, only for neurotypical agents. It is perhaps plausible that in ordinary cases, when an agent cares only about what is morally right, and does not care about any of the morally right things (de re), this is sufficient to show that she is manifesting deficient quality of will. However, consideration of neuroatypical agents reveals possible cases in which motivations to do what is morally right de dicto accompanied by unresponsiveness to what is morally right de re is not an instance of deficient quality of will. Consider the following ordinary case of wrongdoing:

Scrooge. Scrooge is the CEO of a large company. He had a pleasant childhood and a good education. He does not care about the well-being of others at all – he cares only about his own interests. However, he cares very much about his profit margins. His marketing team tells him that if he is seen to be doing what is morally right this will improve profits in the next quarter. So, he hires a team of ethicists who tell him that improving conditions for his overworked employees is the right thing to do. Aiming to do the morally right thing, whatever it is, he makes some changes that improve conditions.

There is no particular barrier to Scrooge caring more about important moral considerations, for example, the welfare of his employees. Nevertheless, he does not care. His motivation to do the right thing is clearly self-serving and does not make up for this. De Re Significance accounts deliver the correct evaluation here – he is blameworthy. He is not sufficiently responsive to considerations that he should be more responsive to. However, it is plausible that the inference from Scrooge’s unresponsiveness to normative moral reasons de re, to his deficient quality of will, is cogent only because he is failing to care about important considerations that he could easily care about.

There are important differences between Scrooge and neuroatypical agents like Mike and Bonnie. Unlike Scrooge, when they seek out and act on misleading advice, they are doing the best thing that they could do, given their limited capacities. Doing one’s best is not correctly characterised as deficient quality of will. On the contrary, their desires and motivations seem to deserve praise – even if not maximal praise (see Hills (2009), Sliwa (2012)). If this is right then De Re Significance accounts cannot be right, even if we accept the model of moral appraisal that typically underpins them.

It is worth noting that De Re Significance accounts do have some resources for accommodating non-blameworthy cases of  acting on moral advice. Harman (2015) and Weatherson (2014) divide such cases into two categories. First, sometimes acting on moral advice is not blameworthy because the advice-taker is factually ignorant, and there are some non-normative facts that her moral advice-giver has that she does not. She should therefore defer to their judgment not out of the de dicto motivation to do the right thing, but as a way to take into consideration the non-normative facts that she is unaware of (and that ideally she would investigate before reaching a judgment). The second category includes all other cases of acting on moral advice. These, they argue, are blameworthy because they are objectionable cases of responsiveness (only) to moral reasons de dicto, which reveals deficient quality of will and can never excuse wrongdoing. The cases discussed here fit comfortably into neither category.

Another way that De Re Significance accounts might seek to accommodate the cases is by agreeing that morally limited agents can blamelessly act on advice but denying that this could involve unresponsiveness to normative moral reasons de re. Instead, they might argue, agents who act on moral advice blamelessly would not seek out testimony about what the morally right thing (de dicto) to do is, but about what the normative moral reasons de re are. This advice would be of the following form:

“Respond to consideration Y, because it is one of the normative moral reasons de re”.

However, it is not clear that this kind of advice would be of much use to morally limited agents. For this advice to be useful, the agent would need to be able to draw the right conclusions about what the relevant normative moral reasons de re require, but it is not clear that the agents under discussion could do this. To insist that blamelessness requires these agents to seek only advice in de re form, regardless of its usefulness, itself suggests a kind of fetishism for moral reasons de re that we should reject.

What this suggests is that the central claims of De Re Significance accounts are true only of some, largely neurotypical, agents. It is only for these agents that we can infer from unresponsiveness to normative moral reasons de re to deficient quality of will. De Re Significance accounts thus generate  implausible moral appraisals for  neuroatypical agents. In the following section I offer an alternative account of moral appraisal which overcomes this problem by making greater reference to the agent’s capacities to be responsive to particular reasons (whether de re or de dicto).

Reasonable Expectations

This section presents an alternative to De Re Significance accounts, according to which blameworthiness depends on whether the agent is responsive to the aspects of the situation that it is reasonable to expect her to be responsive to. This account, the Reasonable Expectations Account, can accommodate neuroatypical agents who act on bad advice, while agreeing with De Re Significance accounts on the cases involving neurotypical agents that they get right. Crucially, the Reasonable Expectations Account acknowledges a wider range of considerations that determine blameworthiness, not only normative moral reasons de re. In short:

Reasonable Expectations Claim: Moral appraisal is determined entirely by whether the agent is responsive to the normative moral reasons that it is reasonable to expect her to be responsive to.

The idea that blameworthiness is determined by what it is reasonable to expect of the agent is by no means new, and there are different ways of understanding this notion.²²

For example, Rosen (2002, 2004) holds that blameworthiness depends on whether, in moral deliberation, the agent has managed her beliefs as it is reasonable to expect her to. He takes this to be cashed out in terms of procedural obligations on belief formation. Namely, to take steps to “inform [one]sel[f] about matters relevant to the moral permissibility of [one’s] conduct”, and to “reflect” to the degree deemed appropriate by the situation (2002, 63–5), where more reflection is demanded in response to “serious criticism”, “known diversity of opinion”, or “perceived tension in one’s moral view” (2002, 65). In Rosen’s view, provided an agent has met these obligations, and does not believe that her actions are wrong, it is not reasonable to expect her to avoid those actions (even if they are wrong).²³ As others have pointed out, one implication of this is that agents can meet these conditions while holding moral beliefs that lead directly to wrongdoing. Ancient slaveowners,²⁴ cold-hearted capitalists,²⁵ and 1950s sexists²⁶ are standard examples. Given the social circles they move in, situations are possible in which these agents reflect as much as the situation demands and nevertheless reach the conclusion that some bad action is permissible. Rosen’s view is that these agents meet the procedural obligations on belief formation, and so are not blameworthy for the resulting bad actions.

Some have thought Rosen’s view of what it is reasonable to expect too lenient. For example, Fitzpatrick (2008) argues that it is reasonable to expect agents to revise their beliefs, even if those beliefs are commonplace within their social circles. Fitzpatrick takes this blameworthiness to be explained by its manifesting the vice of failing to subject one’s beliefs to sufficient critical scrutiny.²⁷ The Reasonable Expectations Account agrees with Fitzpatrick on this point, but has no need to appeal to vices. Rather, there are often good reasons to revise beliefs, and it is reasonable to expect agents to recognise and respond appropriately to them. For example, agents deliberating today need only consider the status of historical moral beliefs about slavery or the rights of women to realise that what one’s society thinks about moral permissibility is often not a good guide to what is true about moral permissibility.²⁸ This should tell us that merely relying on the commonplace beliefs of one’s society does not count as adequate moral reflection – it is reasonable to expect more scrutiny.²⁹ Acknowledging this means disagreeing with De Re Significance accounts, since it implies that not only moral reasons de re are significant. Information about what the right thing de dicto is likely to be, and how the agent responds to that information, is also significant.

So, in general it is reasonable to expect agents to be responsive to various morally important aspects of their situations, including information about what is morally right de dicto, and not only what is morally important de re. Additionally, what it is reasonable to expect of any particular agent is affected by various factors. For example, her capacities. Uncontroversially, if S completely lacks the capacity to do X, then it is unreasonable to expect S to do X. However, there is some leeway in how limited S’s capacity to do X must be before it becomes unreasonable to expect her to do X. Philosophers have also disagreed over which capacities are relevant to determining what it is reasonable to expect. Sher (2009) holds that physical, but not mental, capacities are relevant. Sher borrows this demarcation from the legal reasonable person test for negligence. However, while orthodoxy in legal philosophy has generally been reluctant to take an agent’s mental capacities into consideration when attributing legal culpability,³⁰ this seems a puzzling import into a theory of moral blameworthiness. In so far as we are interested in the agent’s genuine capacity to respond to reasons, mental capacities are of obvious relevance to this (as the cases involving neuroatypical agents show). The Reasonable Expectations Account departs from Sher’s, and considers the agent’s mental capacities relevant in determining what it is reasonable for the agent to recognise. With this in mind, one might whether and how more everyday neurotypical psychological features affect an agent’s mental capacity, and therefore what it is reasonable to expect. The short, though perhaps unsatisfying, answer is that it depends. Forgetfulness, for example, may make it more difficult to respond appropriately – but we need more detail to determine this precisely. Forgetful agents will in most cases have strategies available to help them remember morally important things, and it is reasonable to expect them to use these strategies. Dispositions towards laziness or unreflectivity might impact mental capacity, but again, it depends on the details. Such dispositions could not form the basis for an excuse unless they significantly impacted the agent’s capacity to respond to all the moral reasons available to her (where this includes moral reasons de dicto as well as de re).³¹

Another relevant factor is the moral or practical stakes. As Guerrero argues, “[t]he more morally significant the actions that a belief in p […] will support or license, the more stringent the epistemic demands that must be met” (2007, 69). So, it is reasonable to expect agents to be more careful in forming their beliefs if something important depends on it. The Reasonable Expectations Account agrees with this general point but given its focus on the reasons that it is reasonable to expect agents to recognise, this expectation to take greater care applies not only when the stakes are actually high, but also when the agent has reasons to think they are high.

By focussing on what it is reasonable to expect, rather than whether the agent is responsive to normative moral reasons de re, the Reasonable Expectations Account can say the right thing about the cases discussed in Sections 2 and 4. When the agent’s capacity to be responsive to the normative moral reasons de re is impaired, we need not evaluate them as blameworthy, even if they do the wrong thing. This is because there are usually other aspects of the situation that it is reasonable to expect them to be responsive to. For example, Mike lacks the capacity to respond to moral reasons involving other people’s feelings, but faces no barrier to appreciating the de dicto reason ‘that X is the morally right thing to do’.

Additionally, the Reasonable Expectations Account can agree with De Re Significance accounts that responding to normative moral reasons de re is often praiseworthy, and failure to do so is often blameworthy. For example, Huck Finn is often used by proponents of De Re Significance accounts as an example of someone who is praiseworthy in virtue of doing the right thing in response to the normative moral reasons de re, in spite of his mistaken belief that he is acting wrongly in doing so. Proponents of De Re Significance accounts have appealed to this case to argue that responsiveness to normative moral reasons de re is all that matters for moral appraisal, and the agent’s epistemic situation does not matter.³² However, we can agree that Huck is praiseworthy without taking this to imply that it is only responsiveness to normative moral reasons de re that matters. We should also note that Huck has ordinary capacities to recognize and respond to moral considerations, although given the society he lives in, it is unreasonable to expect him to know that he is doing the right thing.³³ So, the Reasonable Expectations Account can agree that he is praiseworthy for managing, against the odds, to do the right thing for the right reasons, while also acknowledging that were Huck to not free Jim, his misguided moral beliefs may constitute an excuse,³⁴.³⁵

It is less clear that the accounts can agree on how to evaluate agents like Bonnie. By stipulation, she has no capacity to appreciate the significance of moral reasons. The Reasonable Expectations Account says that Bonnie’s lack of capacity to appreciate the normative importance of moral reasons makes it unreasonable to expect her to be responsive to them, and so not blameworthy if this results in wrongdoing. De Re Significance accounts cannot agree with this result. However, it is worth pointing out that this does not imply that Bonnie will be excused every time she does wrong – only in cases where the wrongdoing is due to her lack of capacity.³⁶ Many cases will not be like this. For example, when Bonnie fails to offer the taxi to the pregnant woman, this is explained by her inability to appreciate the normative force of moral reasons, because in this case she has no other information available that would indicate that this might be a wrong thing to do. The case is subtle, there are no laws governing it, and we assume Bonnie has no prior experience to draw on.

Had Bonnie done something less subtle, like murder, the Reasonable Expectations Account would not necessarily imply she is excused. In Murder, there are reasons that it is reasonable to expect Bonnie to appreciate against murdering. We can assume that she knows that there are conventional norms against murder – most legal systems prohibit it, and most people think that one morally ought not murder. Of course, Bonnie thinks this is all mistaken. However, we should expect her confidence in this to be similar to most ordinary people’s confidence in their own moral beliefs. Few people are completely certain about their own moral beliefs, and those who are rarely have good reason to be certain.³⁷ So, we should expect Bonnie, just like anyone else, to have some credence in the possibility that she is wrong about the permissibility of murder. If she does, then she also knows that by murdering, she would be taking a risk. She knows that if she is wrong, then committing murder is  very wrong. Furthermore, if she is right, and murder is permissible, then what she stands to gain is relatively trivial – she merely alleviates the mild boredom of the wait for the taxi. Bonnie does not need the capacity to appreciate the normative force of moral considerations to realize that this is not a risk worth taking. She only needs to appreciate that it might turn out that she is mistaken about how important it is to avoid morally wrong actions. These considerations concern what is morally important de dicto, rather than de re. Unlike De Re Significance accounts, the Reasonable Expectations Account is able to consider responsiveness to this kind of information as relevant to moral appraisal. This is particularly useful in cases like Bonnie’s where agents are unable to respond to normative moral reasons de re. While De Re Significance accounts are forced to evaluate agents like Bonnie as either beyond moral evaluation, or blameworthy by default, the Reasonable Expectations Account can give a more nuanced and satisfying account of whether agents like Bonnie are blameworthy. By including normative reasons de dicto among what it is reasonable to expect agents to recognise, the Reasonable Expectations Account can say that agents like Bonnie are sometimes, but not always, blameworthy.

Summary

I have argued that De Re Significance accounts face problems when appraising neuroatypical agents, because of their focus on the moral significance of a particular kind of moral reason, normative moral reasons de re. Consideration of neuroatypical agents revealed possible cases of agents who are unresponsive to the normative moral reasons de re of their situation, but for whom this unresponsiveness does not imply deficient quality of will. After exploring the options available to De Re Significance accounts for accommodating such cases and highlighting their weaknesses, I offered the Reasonable Expectations Account as a better way to accommodate these cases while nevertheless agreeing with De Re Significance accounts about the cases they get right.