I. INTRODUCTION
A growing body of empirical research suggests that preferences for utilitarian solutions to certain moral dilemmas arise from a mental process that is more cognitive and less affective than the mental process leading to deontological preferences.Footnote 1 In light of these findings, some researchers interested in moral psychology have argued that utilitarian judgements are optimal. A well-publicizedFootnote 2 study by Daniel Bartels and David Pizarro entitled ‘The Mismeasure of Morals’ calls such arguments into question.Footnote 3 The study purportedly provides evidence that utilitarian solutions to the relevant moral dilemmas are primarily endorsed by individuals who possess traits typically associated with psychopathy such as callousness and a proclivity for manipulation. According to the authors, these findings give researchers reason to refrain from classifying preferences for utilitarian solutions to these dilemmas as instances of optimal moral judgement. As they put it: ‘these approaches lead to the counterintuitive conclusion that those individuals who are least prone to moral errors also possess a set of psychological characteristics that many would consider prototypically immoral’.Footnote 4
Bartels and Pizarro are careful not to overstate the implications of their findings. They caution that they have not provided evidence that endorsing utilitarianism is pathological. They also note that their results say nothing about whether utilitarianism is the correct normative ethical theory. However, the purported implications of the study do pose a direct threat to utilitarianism. If the results provide reason to doubt the correctness of utilitarian solutions to the target moral dilemmas, they thereby provide reason to doubt the correctness of utilitarianism itself.
In what follows I provide a two-part response to ‘The Mismeasure of Morals’. The first part comprises concerns about the methodology used and the adequacy of the data provided in support of the authors’ conclusions. A major source of concern is that while the data show that individuals with higher scores on psychopathy measures had a greater tendency to endorse utilitarian solutions, there is no indication of how high the scores were. Without this information it is unclear whether any of the subjects were anti-social in any meaningful sense. Moreover, even if some subjects with utilitarian preferences did have robust psychopathic traits, this would not be enough to justify the conclusion that utilitarian solutions were endorsed primarily by individuals with these traits. Thus, the study does not provide evidence that could undermine the neuropsychological arguments for utilitarianism.
The second part of my response focuses on the relationship between character and moral judgement. Although there is no evidence that most individuals who endorse utilitarian solutions have psychopathic traits, it is plausible that most people with robust psychopathic traits would endorse utilitarian solutions. Some of the claims found in ‘The Mismeasure of Morals’ suggest that the fact that anti-social individuals are highly likely to endorse utilitarian solutions (if it is indeed a fact) is a strong reason for scepticism about the correctness of those solutions. In order to refute this suggestion I consider various examples which illustrate the point that character traits of individuals most likely to make a given judgement are an unreliable indicator of the quality of that judgement.
The discussion proceeds as follows. I begin in section II with an overview of the recent influx of empirical research and corresponding arguments in favour of utilitarian judgements. In section III, I briefly recount the methods and results of the Bartels and Pizarro study. In section IV, I raise some methodological concerns about the study, and I explain why it fails to undermine neuropsychological arguments for utilitarianism. In section V, I argue that even if non-clinical psychopaths are substantially more likely to endorse utilitarian solutions, this does not cast doubt on the correctness of those solutions. I conclude with a brief discussion of a separate but related concern some may have about neuropsychological arguments for utilitarianism.
II. THE NEUROPSYCHOLOGICAL CASE FOR UTILITARIANISM
The rise of the practice of classifying utilitarian judgements as optimal is attributed in large part to the work of Joshua Greene et al.Footnote 5 Green et al. hypothesized that characteristically deontological judgements are driven by emotion, while characteristically utilitarian judgements are driven by rational deliberation.Footnote 6 They tested their hypotheses by observing participants’ neural activity while making judgements about moral dilemmas in which utilitarianism and deontology call for starkly contrasting responses. What the dilemmas have in common is that in order to maximize utility the subject would have to sacrifice an innocent person. The following is a representative example:
Crying Baby: Enemy soldiers have taken over your village. They have orders to kill all remaining civilians. You and some of your townspeople have sought refuge in the cellar of a large house. Outside you hear the voices of soldiers who have come to search the house for valuables. Your baby begins to cry loudly. You cover his mouth to block the sound. If you remove your hand from his mouth his crying will summon the attention of the soldiers who will kill you, your child, and the others hiding out in the cellar. To save yourself and the others you must smother your child to death. Is it appropriate for you to smother your child in order to save yourself and the other townspeople?Footnote 7
Greene et al. found that when subjects indicated a preference for non-utilitarian solutions to sacrificial dilemmas (in this case refraining from smothering the baby) they had increased neural activity in the areas of the brain associated with emotional response. In contrast, those who indicated a preference for utilitarian solutions had greater activity in regions of the brain associated with ‘higher cognition’.Footnote 8
Greene cites numerous studies as corroborating evidence for his hypothesis.Footnote 9 Motivated by the thought that rational deliberation is a more reliable method for making moral judgements than immediate emotional responses, some researchers have concluded that utilitarian moral reasoning is optimal, while non-utilitarian reasoning is suboptimal. Utilitarian moral judgements are considered optimal because they are more cognitive and less affective than deontological judgements. The advantage of ‘cognitive’ representations is their behavioural neutrality. As Greene puts it, cognitive representations ‘can be mixed and matched in a situation specific way without pulling the agent in multiple behavioral directions at once, thus enabling highly flexible behavior’.Footnote 10 In contrast, emotion-laden representations bring with them behavioural dispositions that are difficult to overcome.
In sum, utilitarian judgements are considered optimal by some researchers because they appear to be driven by a mental process that is highly flexible – in terms of both the capacity to consider more information and the behavioural responses that result. Deontological judgements are considered suboptimal because they are apparently driven by a mental process that brings with it a strong behavioural disposition and a limited ability to consider a wide range of factors.
III. ‘THE MISMEASURE OF MORALS’
Bartels and Pizarro hypothesized that utilitarian preferences in sacrificial dilemmas may not be the product of rational deliberation guided by an equal concern for the welfare of everyone. Rather, such preferences may be driven by selfishness and aided by a muted aversion to causing another person's death. One might have expected the authors to validate their concern by pointing to empirical evidence indicating that individuals with clinical-level emotional deficits tend to endorse utilitarian judgements.Footnote 11 However, Bartels and Pizarro claim that we should avoid making generalizations about the quality of moral judgements based on whether psychopaths endorse them because the deficits of clinical populations may cause them to make their moral judgements ‘through qualitatively different psychological mechanisms than those at work in non-clinical populations’.Footnote 12 Yet the authors do think we can evaluate the quality of moral judgements based on the character traits of non-clinical individuals who tend to make them. If most non-clinical individuals who prefer utilitarian solutions are selfish, manipulative and emotionally detached, this should make us think twice about the moral optimality of such preferences.
In order to assess the association between anti-social character traits of non-clinical individuals and utilitarian preferences, Bartels and Pizarro gave 208 undergraduates a battery of fourteen sacrificial dilemmas. Subjects indicated their preference for a particular solution by clicking on one of four boxes as in the following example:
Subjects were given a set of three personality measures: psychopathic personality, Machiavellianism and perceived life meaninglessness. Psychopathic personality is indicated by low empathy, callousness and thrill-seeking. Machiavellianism is characterized by cynicism, emotional detachment and manipulativeness. The No Meaning scale measures the degree to which subjects find life meaningful and worthwhile.Footnote 13
The authors’ data showed a correlation between higher preference for utilitarian solutions to sacrificial dilemmas and higher scores on measures of psychopathy, Machiavellianism and No Meaning. A preference for utilitarian solutions showed a correlation of .38 with psychopathy, .35 for Machiavellianism and .21 for No Meaning. Those who scored high on the psychopathy scale, relative to their sample, had a mean preference for utilitarian solutions of roughly 0.3 on a scale ranging from −2.0 (strong non-utilitarian preference) to +2.0 (strong utilitarian preference). Those who scored high on the Machiavellianism and No Meaning scales had mean utilitarian preferences of roughly 0.2 and 0.1 respectively.
Bartels and Pizarro believe these correlations confirm their hypothesis that there is more than one route to arriving at a preference for utilitarian solutions to sacrificial dilemmas. While some may arrive at these preferences via deliberation aimed at maximization of welfare, others form them as a result of emotional deficits including a muted aversion to causing death. The authors maintain that this conclusion has important implications for researchers interested in moral psychology. They claim that in light of their findings the practice of classifying utilitarian preferences as instances of optimal moral judgement is highly dubious. They put the point as follows: ‘We should be wary of favoring a method that equates the quality of a moral judgment with responses that are endorsed primarily by individuals who are likely perceived as less moral (because they possess traits like callousness and manipulativeness).’Footnote 14
IV. METHODOLOGICAL CONCERNS
Bartels and Pizarro suggest that their study provides evidence that most individuals who prefer utilitarian solutions to sacrificial dilemmas possess anti-social traits like callousness and manipulativeness. But do their data really support this conclusion? A close examination of their methods and findings suggests that they do not.
A major shortcoming of the study is that the published data do not provide any indication of what it means to be ‘high’ on a measure of psychopathy. The authors split their sample into three equal groups, identifying those with scores falling in the highest third, relative to their sample, as ‘high’. Without presenting norms for the measure of psychopathy it is difficult to know how to interpret the authors’ classification of those ‘high’ in psychopathic traits. To see why this is problematic we can consider a possible set of data that is consistent with the reported findings. Suppose the measure for psychopathy ranged from zero to 100. Suppose further that the mean score of subjects with non-utilitarian preferences was five, and the mean score of subjects with utilitarian preferences was twelve. While this could be a statistically significant difference in some cases, it may not be a conceptually meaningful difference. Suppose that the average score of all participants was eight, and the threshold for clinical psychopathy is a score of eighty. If so, the difference between a score of twelve and a score of five is less meaningful. At the very least, it would be a stretch to conclude that members of the group of subjects whose mean score was twelve actually possess psychopathic character traits. There is a difference between being slightly more cold and callous than average, and being properly described as a cold and callous individual. As the data are presented, the extent to which the groups reflect different character traits is difficult to determine.
It is possible that the divergence on the personality measures between the two groups was more substantial than this, and that the average score on the psychopathy measures among the group with utilitarian preferences approached the threshold for clinical psychopathy. But even if this were the case, it would not be sufficient evidence for the conclusion that most individuals who prefer utilitarian solutions possess anti-social personality traits. It would be unsurprising to discover that people who are genuinely cold, selfish and manipulative would prefer utilitarian solutions to sacrificial dilemmas. Not only are such individuals less prone to an emotion-based aversion to killing innocent people, in nine of the fourteen target dilemmas the utilitarian option allows the agent to avoid her own impending death. Of course we should expect people who are selfish and emotionally detached to endorse these options – it would be puzzling if they did not.Footnote 15 Indeed, the preference for these solutions among virtually all robustly anti-social individuals would give rise to a statistical correlation between preferring utilitarian solutions and having these traits. But it would be a mistake to infer from this correlation that the majority of individuals who prefer utilitarian solutions to sacrificial dilemmas can properly be described as anti-social.
A further source of scepticism about the authors’ conclusions arises from the presentation of the target dilemmas. In the Greene et al. studies subjects were asked whether it is ‘okay’ (i.e. permissible) or ‘appropriate’ to make the relevant sacrifice in the given circumstances. In contrast, subjects in the Bartels and Pizarro study were asked to predict how they would act if they found themselves facing the dilemma (e.g. ‘In this situation, would you smother the baby?). A prediction of one's behaviour if faced with a set of circumstances is not necessarily an endorsement of that behaviour. It is possible that subjects who expressed non-utilitarian preferences actually judged the utilitarian solutions to be morally correct. Their selection of non-utilitarian solutions could have been merely a prediction of their inability to execute what they judged to be the morally optimal yet emotionally difficult utilitarian solution. There is nothing in the published data to rule out the possibility that the majority of individuals classified as ‘low’ on the anti-social measures considered the utilitarian solutions morally optimal. Conversely, subjects who chose the utilitarian solutions might have been merely predicting that they would be unable to perform the non-utilitarian solution despite believing that it would be morally optimal to do so. Thus, it is possible that the percentage of subjects classified as ‘high’ on the anti-social measures who ultimately endorse the non-utilitarian solutions is much greater than the authors indicate.
If we had good reason to think that the majority of subjects with utilitarian preferences arrive at their judgements solely via self-interested calculation (perhaps assisted by a lack of aversion to causing harm) this would undermine the suggestion that utilitarian preferences are the product of optimal moral reasoning. But the considerations I have raised reveal that ‘The Mismeasure of Morals’ does not provide strong evidence that most non-clinical individuals who endorse utilitarian solutions to sacrificial dilemmas possess psychopathic traits. Thus, there is little reason to believe that most subjects who expressed utilitarian preferences engaged in purely selfish reasoning aided by a callous disposition.
V. CHARACTER AND MORAL JUDGEMENT
Although we lack evidence that (in non-clinical populations) most subjects who endorse utilitarian solutions have anti-social characteristics, it would be unsurprising to discover that those who do have anti-social traits (at subclinical levels) are more likely to favour utilitarian solutions than people without these traits. Most of the utilitarian options are prudentially optimal, and the more cold and selfish one is, the more likely it is that one's practical decision-making will be dominated by considerations of self-interest. Suppose empirical data supported this conjecture. Would this constitute compelling evidence against the moral optimality of the utilitarian solutions? Some of the claims found in the Bartels and Pizarro study seem to imply that it would. For instance, the authors argue that classifying utilitarian solutions as optimal leads to the ‘counterintuitive inference that correct moral judgements are most likely to be made by the individuals least likely to possess the character traits generally perceived as moral’.Footnote 16 This raises an important theoretical issue about the relationship between character and moral judgement. In the rest of this section I shall argue that the character of individuals who are most likely to make a given judgement is an unreliable indicator of the correctness of that judgement. Thus, even if it is true that people with psychopathic traits are highly likely to endorse utilitarian solutions, this does not gives us compelling reason to doubt that these solutions are correct.
At first blush, the thought that a judgement which cold and selfish individuals have the highest likelihood of endorsing could be morally optimal seems counterintuitive. However, when the morally optimal solution is also prudentially optimal, having anti-social traits is not a hindrance to correct moral judgement at all. In fact, a proclivity to selfishness often increases one's likelihood of making the correct judgement in such situations, while strong pro-social traits make correct judgement less likely. We can call such cases counter-character scenarios. It will be helpful to consider a few examples. To avoid begging any questions I have selected examples for which utilitarians and deontologists are likely to agree about which solution is morally optimal. Consider first an example from Derek Parfit:
Earthquake – You are trapped with a stranger in slowly collapsing wreckage. The stranger's life is not in danger, but she is in danger of losing her leg. Your life is in danger. You can prevent the stranger from losing her leg, but doing so will cost you your life. Alternatively, you can save your own life, but doing so will preclude you from saving the stranger's leg (though she will survive either way). What should you do?Footnote 17
The utilitarian solution to this dilemma is to allow the stranger to lose her leg so that you can save your own life. Although the stranger losing her leg will likely decrease the total hedonic value of the rest of her life, there will be greater net utility if both of you survive than if you give your life to save her leg. Saving yourself is also the recommended course of action from a deontological perspective. The value of a human life cannot be traded for the instrumental value of a person's leg. Of course, if the loss of the leg were the result of using the stranger without her consent in order to save oneself, the action would be deemed impermissible on the Kantian view. But as the case is described saving oneself would not involve using the stranger in any way.
I suspect that most individuals who are presented with this dilemma would express a preference for self-preservation, which is the correct choice according to both utilitarianism and deontological views. But certainly there are some individuals who would endorse the incorrect self-sacrificial option. What sort of characteristics might such individuals have? Although this is an empirical question, it is plausible that most people who would endorse sacrificing one's life to save a stranger's leg have strong pro-social character traits. Such people are likely to be highly altruistic, generous, compassionate individuals. The greater the value one places on being selfless, the more likely it is that one would endorse sacrificing oneself to save a stranger's leg.
What about individuals with the anti-social traits that Bartels and Pizarro discuss? Which solution would they endorse? This is another empirical question, but the smart money says that people who are genuinely cold, callous and selfish would endorse saving one's own life rather than sacrificing oneself to save a stranger's leg. Thus, Earthquake is an example of a moral dilemma in which individuals with traits typically perceived as immoral will almost certainly make the correct judgement (from both deontological and utilitarian perspectives), while having certain pro-social character traits presumably decreases one's likelihood of making the correct judgement.
Consider next an example from Justin D'Arms and Daniel Jacobson:
Nursing Home – Mother has grown older, and grown mentally ill. She makes increasingly exigent demands on the family. Her illness is degenerative. She always feared being ‘put away’; you know she wants to stay at home, but you have real doubts about your ability to care for her. And you also see the effects of the tension, pity, and finally resentment on your family. Should you put mother in a nursing home?Footnote 18
As D'Arms and Jacobson rightly suggest, any plausible normative view will hold that putting Mother in the nursing home is the right thing to do. It is clear that under these circumstances the net utility of putting Mother in a home is greater than the net utility of letting her stay. Thus, the utilitarian judgement is that one ought to place her in a home. One might worry that placing Mother in a home is a violation of her autonomy, and therefore impermissible from a Kantian perspective. However, given her degenerative mental illness, Mother's autonomy is greatly diminished. Although cases of diminished autonomy are controversial, this particular case is one in which Kantians are likely to agree that doing the right thing necessitates violating the expressed wishes of the less than fully autonomous individual.
For present purposes the relevant question is what sort of character traits we would expect to find in people who make the incorrect judgement about Nursing Home. As was the case in the previous example, it is eminently plausible that possession of certain pro-social character traits increases one's chances of making the incorrect judgement. The emotional complexities of the case raise special difficulties for those individuals whom we admire precisely because of their sensitivity. A person who is extremely loyal, grateful and sympathetic will likely have a more difficult time reaching the conclusion that the right thing to do is to violate the expressed wishes of one's mother. We can imagine such a person thinking: ‘I owe everything to my mother. I do not care what the situation is; she can stay with me for as long as she wants.’ What makes Nursing Home an interesting case is that it is an example in which doing the right thing is likely to generate feelings of guilt. Even if we are convinced that we would be acting on the side of reason, it is easy to imagine the pangs of guilt we would feel as we drive away from the nursing home leaving Mother behind. This might seem irrelevant given that we do not have to act out the option we endorse when presented with a hypothetical moral dilemma. But emotions often do play a role in our assessment of hypotheticals. It is not a stretch to suggest that the more sympathetic and loyal one is, the greater the role emotions like guilt will play in making a moral judgement about this case. Imagining how guilty one would feel after putting Mother in the home could easily lead to the judgement that doing so would be wrong.
None of this is to say that everyone who is sympathetic and loyal will make the incorrect judgement about Nursing Home. The claim is merely that the more robustly sympathetic and loyal one is, the greater the chances that emotional responses will have a detrimental influence on one's judgement about the case. Consider once again individuals with the anti-social traits of coldness, callousness and manipulativeness. Individuals with these traits will have no trouble in coming to the correct moral judgement. Emotionally detached individuals are not at risk of being led astray by the guilty feelings that might lead to the mistaken judgement that putting Mother in a nursing home would be wrong. If the moral judgements of individuals with anti-social character traits are going to be unduly influenced by anything it is going to be self-interested considerations. And if self-interested considerations play a strong role in the anti-social individual's judgement about Nursing Home, this will manifest itself in the individual making the correct judgement that one ought to put Mother in the home.
Perhaps some will be unmoved by my discussion of Earthquake and Nursing Home in the absence of empirical data to support my conjectures about the traits of those who endorse the different alternatives. For those who are sceptical that highly selfless individuals are more likely to endorse self-sacrifice, and that extremely loyal individuals will experience greater difficulty in deciding that one ought to put one's mother in a home, we can note that the general point I have been arguing for holds independently of the verification of these conjectures. The important claim is that drawing conclusions about the correctness of a moral judgement by examining the character traits of those who are most likely to make that judgement is highly precarious. We need not conduct any empirical investigations in order to see that this is true. We can simply suppose that certain results are obtained and ask what conclusions would follow.
Suppose empirical research confirmed my hypothesis that individuals who are exceedingly altruistic and compassionate are significantly more likely than average to judge that one ought to sacrifice one's life in order to save a stranger's leg. Or suppose that subjects who are cold and callous are the ones most likely to endorse saving oneself. Would these results give us reason to doubt that the right thing to do in Earthquake is to save oneself so that both can live? Of course not. Likewise with Nursing Home. If research showed that exceptionally loyal and sympathetic individuals are the ones most likely to judge that Mother ought to be able to stay at home, or that cold and callous individuals are the ones most likely to endorse putting her in a home, this should not undermine our confidence that the right thing to do is to put Mother in the nursing home.
What these examples reveal is that the character traits of individuals most likely to make a judgement can be a highly misleading indicator of the correctness of that judgement. If we have independent reason for believing a given judgement is correct, and the judgement in question is one we would antecedently expect anti-social individuals to make, the fact that they are the ones most likely to make the judgement should not undermine our confidence. In the case of sacrificial dilemmas, researchers have argued that neuropsychological evidence tells in favour of the optimality of utilitarian judgements because the mental process involved is more cognitive and less affective. Yet we would not be surprised to discover that most anti-social individuals would indicate utilitarian preferences. If this turned out to be the case, this would not be strong reason to doubt that the utilitarian solutions are correct. Sacrificial dilemmas may simply be another instance of counter-character scenarios.
Although deontologists and utilitarians are deeply divided over the fundamental principle of morality, proponents of these theories generally agree that when making moral judgements we can easily be led astray by the influence of emotions, inclinations and prejudices. Thus, to the extent that we can draw conclusions about the quality of moral judgements by examining the individuals who make them, it is the process of reasoning used that is most illuminating. Sometimes people with highly amiable dispositions have greater difficulty engaging in unadulterated reasoning. Conversely, individuals with unsavoury traits are sometimes better able to arrive at the judgements that would result from pure rational reflection. This is why we should be wary of drawing conclusions about the quality of a moral judgement by examining the character traits of those who tend to make that judgement.Footnote 19
Bartels and Pizarro are right to suggest that it would be a mistake to classify an individual as an optimal moral judge simply in virtue of the fact that she endorses utilitarian solutions to sacrificial dilemmas. But proponents of neuropsychological arguments for utilitarianism claim only that those who opt for utilitarian solutions make optimal moral judgements about the relevant cases. They do not make the implausible claim that anyone who makes an optimal moral judgement about those cases is thereby an optimal moral judge. If utilitarianism is the correct normative ethical theory, then psychopathic individuals are likely to choose the right option when presented with sacrificial dilemmas. But they would still be far from being optimal moral judges because they are highly unlikely to choose the utilitarian option when presented with a wide range of other moral decision problems. For instance, we should not expect cold and selfish individuals to endorse donating one's disposable income to famine relief rather than purchasing a luxury item for oneself. Nor should we expect subjects with robust psychopathic traits to endorse abstaining from meat consumption for the sake of animal welfare.
VI. CONCLUSION
I have argued that the empirical findings put forth in ‘The Mismeasure of Morals’ do not undermine the neuropsychological arguments for utilitarianism, and that if anti-social individuals are indeed most likely to prefer utilitarian solutions, this tells us nothing about the correctness of those solutions. However, the question of what theoretical implications can be drawn from the empirical findings of Greene et al. remains controversial.Footnote 20 While a full discussion of these issues is beyond the scope of this article, I would like to conclude by considering the role of emotional responses in ethical decision-making and moral agency.
One of the underlying assumptions of the arguments of Greene et al. is that if there are truths about what we have most reason to do when faced with moral-decision problems, we are most likely to arrive at these truths when our reasoning is unencumbered by affective responses. Some may find this denigration of the role of emotions in moral judgement to be unpalatable. There is much plausibility to the thought that a capacity to feel emotions such as sympathy and compassion is a precondition for successful moral agency. This may seem incompatible with the suggestion that emotional responses are a hindrance to optimal moral reasoning. But we must be careful not to conflate optimal moral agency with optimal moral judgement. The claim that emotions are a hindrance when making moral judgements is entirely compatible with the belief that emotions are essential for successful moral action.
There are many situations in which emotions can help motivate us to act in morally optimal ways. Sympathy and compassion can prompt us to sacrifice our own interests for those who need our help, and these feelings can also be instrumental in making our efforts at helping more effective (e.g. when emotional support is needed). Individuals who are devoid of sympathy and compassion are unlikely to behave in morally optimal ways. Thus, even if Greene et al. are right in thinking that moral judgements resulting from affective responses are less reliable, this should not undermine our conviction that warmth and compassion are among the traits possessed by ideal moral agents.Footnote 21
