What the Science of Morality Doesn’t Say about Morality

What the Science of Morality Doesn’t Say about Morality


Gabriel Abend

date de sortie





In recent years scientists have devoted increasing efforts to the study of morality. As neuroscientist Moll and his colleagues (2003, p. 299) say, “morality has been at the center of informal talks and metaphysical discussions since the beginning of history. Recently, converging lines of evidence from evolutionary biology, neuroscience and experimental psychology have shown that morality is grounded in the brain.” In this article I ask what exactly this new science of morality can and can’t claim to have discovered about morality; what it can and can’t tell us about morality on the basis of the work it has done. I argue that the object of study of much recent work is not morality, but a particular kind of individual moral judgment. Most data and analyses are about something very specific: an individual’s judgment about the rightness, appropriateness, or permissibility of an action, made in response to a stimulus at a particular point in time. But this is a small and peculiar sample of morality, whose incidence in people’s actual moral lives is uncertain. There are many things that are moral, yet not moral judgments. There are also many things that are moral judgments, yet not of that particular kind. If moral things are various and diverse, then empirical research about one kind of individual moral judgment doesn’t warrant theoretical conclusions about morality in general—i.e., morality’s nature, functioning, origins, causes, or effects. If that kind of individual moral judgment is a peculiar and rare thing, then it is not obvious what it tells us about other moral things. What is more, it is not obvious what its theoretical importance is to begin with—i.e., why we should care about it at all.


Thus, my arguments raise questions about the theoretical meaning and value of research about individual moral judgment. My claim is not that the numerous new experimental findings about this object will necessarily turn out to be inconsequential or useless—that we can’t know at this point. But at present it’s not very clear what larger conclusions follow from them, nor what their implications for a scientific theory of morality are, much less what their practical or policy implications might be (if any). In this respect, the literature is rife with questionable claims and non sequiturs. Indeed, several recent papers seem unaware of the crucial distinction between individual moral judgment and morality, or contain problematic argumentative transitions from moral judgment to the ambiguous “moral decision-making.”


In what follows, I begin by identifying what I call “moral judgment-centric approaches” (MJA), i.e., research approaches that have individual moral judgment (MJ) at their methodological center. Then, I spell out my claim that MJ is a peculiar moral thing and there’s much moral life beyond it. Last, I argue for a pluralism of methods and objects of inquiry in the scientific investigation of morality, so that it transcends its problematic overemphasis on a particular kind of individual moral judgment.

Scientists of morality have done a great deal of research that revolves around individuals making moral judgments. Subjects may be in the lab, inside the scanner, or at home on their computer. In most studies they are healthy (or “normal”) persons, but in some they have brain damage or a psychiatric condition. In a few studies their brain’s activity or chemistry has been manipulated, but in most it hasn’t. The subjects’ task is to make a moral judgment about statements or situations they are presented with. Oftentimes these studies elicit judgments specifically about moral dilemmas. That is, subjects are presented with a situation where two or more courses of action are possible, or, more often, a situation where the two alternatives are doing a certain thing or abstaining from doing it. Then subjects are asked questions such as: Would it be permissible for person A to do action X in situation S1?; Would it be okay for you to do Y in S2?; or something along these lines. These answers are their moral judgments. For example, “It is not permissible for A to do X in S1,” is a subject’s moral judgment.


The most famous of these moral dilemmas is the “trolley problem,” originally crafted by philosopher Philippa Foot (2002). But psychologists and neuroscientists have conducted experiments using many other dilemmas, some of which they drew from the ethics literature, and some of which they expressly designed to manipulate variables of interest. For instance, Judith Thomson’s “loop” and “fat man” variants on the trolley problem; more recent reformulations by Marc Hauser and John Mikhail; Joshua Greene’s “crying baby” and “infanticide” dilemmas; in earlier moral psychology Kohlberg’s “Heinz dilemma” and “the captain’s dilemma”; and Jonathan Haidt’s ingenious cases (although not dilemmas) about sex among siblings, eating one’s dog, or wiping the toilet with a national flag.

The research questions that these studies have sought to address are diverse, but some of the main ones are as follows. What brain areas are “activated”, “recruited,” “implicated,” “responsible for,” or “associated with” making moral judgments? What brain areas or circuits “subserve” particular kinds of moral judgments (e.g., deontological and consequentialist ones)? What are the neural “correlates,” “basis,” “foundations,” “underpinnings,” or “substrates” of moral judgment, decision-making, and emotions? What are the specific functional roles of specific brain areas? What causes individuals’ moral judgments: hot intuition, affect, and emotion, or rather cold reason and reasoning? Is there a moral faculty, organ, or universal grammar, comparable to the language faculty, organ, or universal grammar?

What is common to most of these research projects, though, is that individual moral judgment (MJ) is at their methodological center. What subjects do is to make moral judgments. What researchers account for, predict, and find neural correlates of are moral judgments. Indeed, the focus is not just on moral judgment, but on a specific kind of moral judgment. Its prototypical features as found in this literature are:
a moral judgment is made by and is an attribute of one individual
it’s made in response to a specific stimulus
the stimulus is an imaginary situation and a question about it
the judgment is about an action (rather than, say, a person or state of affairs)
a moral judgment is a statement (indicative mood)
it is in essence an utterance or speech act (even if not in fact uttered)
it makes use of “thin” ethical concepts only (okay, appropriate, permissible, acceptable, wrong, etc.)
it’s fixed, settled, verdict-like
it’s clear (not conceptually or semantically muddled, incoherent, etc.)
it’s made at a specific, precise, discrete point in time


I argue that MJ is a peculiar kind of moral thing, hence not a good sample of moral things. It’s one among the many moral things that are part of people’s moral lives. I further argue that there’s no reason to suppose that all of the members of the moral class work the same way. It follows that investigations about MJ—what I call “moral judgment-centric approaches” (MJA)—don’t have the resources to make claims about the nature and functioning of morality as a whole. Evidently, what is and isn’t part of people’s moral lives is an empirical question, not an armchair one. Further, it can’t be answered unless you specify just what persons, because this seems to vary from society to society, as well as historical period, social class, age, gender, and many other variables. It’s also an empirical question whether, given a group of people, MJ is a relatively small part of their lives, or rather is a large or the largest one. My aim here is simply to make some suggestions as to what other moral things there might be besides MJ. Future work—both empirical investigations and retrospective reviews—should put my suggestions to test, and specify where, to whom, and to what extent they apply (if at all). At this stage, plausibility is their main test.


So, what else might there be? Two kinds of things: (i) things that are moral, or reasonably called “moral,” yet not moral judgments; and (ii) things that are moral judgments, yet not of the particular MJ kind—i.e., they don’t meet one or more of the above conditions. In the next section I suggest some distinctions that compare and contrast the MJA picture with what it leaves out of sight. Even before the empirical evidence comes in, I believe that leaving these things out of morality by fiat, without a convincing argument, is unacceptable. But this is precisely what MJA have done.


In the recent empirical work, MJ is conceived of as a declarative sentence. I argue that moral questions and exhortations are different from judgments in theoretically important respects. Consider these three types of sentences: (i) Declarative: It is morally permissible/right for A to Φ; (ii) Interrogative: Is it morally permissible/right for A to Φ? / What ought A to do, morally speaking?; (iii) Imperative: Φ! / You must/ought to Φ! Not only is MJ conceived of as a declarative sentence in the indicative mood, such as sentences of type (i). I think it’s also assumed that moral judgments are in essence utterances or speech acts. Even if in practice subjects check boxes, push buttons, or left-click, they would be prepared to utter something like (i) under the appropriate speech-act conditions.

I argue that, besides judgments, one component of people’s moral lives may be questions, including but not limited to questions such as (ii). These questions people pose to and about other people in ordinary conversations and personal ruminations, and, perhaps more importantly, to and about themselves. Moral questions may or may not lead to answers, such as the MJ that the experimenter’s or survey researcher’s question leads to. Sometimes they may remain unanswered, yet still function as tentative guides to thought and action, because they suggest what’s important, worrisome, worth thinking about. Or they may remain unanswered without really guiding anything—just obstinately live on in one’s mind. Phenomenologically, the experience of having or posing or struggling with a moral question is arguably a distinct one, and surely much unlike uttering a moral judgment. Thus, moral questions are a distinct kind of moral phenomenon, worth attending to in and of themselves. Consider a few more concrete examples. Should I regularly give money to charity? Why would I not eat meat if I like it? Why would someone who likes eating meat not eat it? Is it wrong for me to have an abortion? Is she a good spouse? Why am I doing what I’m doing with my life? What’s really important to me? Is life fair to me? Is he such-and-such kind of person?


Moral exhortations—sentences (iii) above—seem to be yet another kind of empirical phenomenon and work in yet another way. Your saying to your friend, “Don’t lie to your spouse!,” is presumably based on a judgment, belief, conviction, or feeling to the effect that it’d be morally wrong or bad for her to lie to him (for the sake of the argument, I set aside non-moral reasons she might have). Besides exhortations, there’s also what one might call self-exhortations—e.g., “Let me improve this aspect of my life!,” or “I really shouldn’t do that thing anymore!” They, too, are presumably based on some sort of judgment to the effect that that is a morally bad thing to have or do. However, neither exhortations nor self-exhortations seem reducible to or understandable in terms of the judgments that apparently underlie them. For one, as empirical phenomena, exhortations and judgments don’t look alike at all. Nor does the first-person experience, phenomenologically speaking. Conceptually, they have a different aim, job, or point; assuming speech, a different speech act. Although exhortations implicitly or explicitly contain theoretical evaluations, their point is practical in a straightforward and unmediated manner: they try to bring about a specific thing. Morality—however defined or understood—seems to have to do primarily with practice and action, and only secondarily or derivatively with theory about practice and action (more on this below). In this respect, exhortations fare better than judgments, whose practical aim is less direct. Why are moral judgments but not exhortations empirically studied?


Finally, there are narratives. As extensive literatures in sociology, psychology, anthropology, history, and communication have shown, human beings are narrative creatures. People tell stories about themselves and about others, about their lives and identities, about their community, its past and its origins. For our purposes, it’s important that narratives can’t be decomposed or analyzed into a set of judgments, rules, points, arguments, questions, or thoughts. A narrative’s meaning is tied to its unity. The meaning of one part depends on its relationship to the meaning of other parts—much like indexicals, networks, and relational properties in general. Some particular types of narratives, such as fables, conclude with a moral or a rule of conduct. Yet, again, it’s unclear what the moral’s meaning would be, were its narrative context to disappear, and thus the moral remained as a self-standing sentence. Along these lines, you could argue that all real-life moral judgments are embedded in a narrative context—isolated moral judgments never occur, or even aren’t possible at all. Similarly, you could raise these empirical questions about people’s moral lives: Do they go about making independent judgments about things, one after the other? Or, rather, do they go about concocting and telling stories and bits of stories, to make sense of things and weave them together, and in the context of which they morally evaluate things?

I’d like now to consider what MJA might be neglecting in light of three traditional distinctions in moral philosophy: (i) the right v. the good v. the virtuous; (ii) permissible v. obligatory v. supererogatory; and (iii) ethics of doing v. ethics of being. As mentioned above, MJ are about what it’d be “okay,” “appropriate,” “wrong,” “morally wrong,” “permissible,” “morally permissible,” or “morally acceptable” for someone to do. These are the concepts that the experimental tasks employ. All of these questions try to get at some undoubtedly relevant moral things: rightness, acceptability, permissibility, “okay-ness,” etc. Yet, might they all be missing some other relevant moral things? Perhaps in some societies questions about wrongness and permissibility are prominent in some people’s moral lives. The practical significance of the concepts of rightness, wrongness, permissibility, and impermissibility in contemporary Western societies is evident, for example, in their ubiquitous institutional and cultural incarnations. Besides, as a matter of fact, these concepts have had an elective affinity with a particular form: rules or law-like principles.


However, there is a different sort of moral relationship that people can have to things—viz., finding them not morally right but morally good, either in themselves, or as means to further moral ends. You can say that people are after these moral goods, even when they lack a plan as to how to get them and this is a matter of practical sense. These are things that someone may hold dear; they aren’t her mere desires and preferences, but things she finds worthy of being had, desired, preferred, sought, cherished, or chosen. People seem to have moral projects, hopes, aspirations, ideals, and commitments, where goods and “the good” play the key role. Some examples of goods of diverse types are: liberty, truth, knowledge, community, solidarity, faith, health, wealth, honor, pleasure, excellence, love, family, friendship, security, ataraxia, work, self-expression. Further, there is a plurality of goods and of conceptions of the good, even within a single moral community, which at times are incommensurable. Moreover, besides doing what’s right and avoiding what’s wrong, people may wish to live a good, fulfilling, life; live it well, not misspend it. Some people might even have one project or commitment that is the most important of all. For example: have a family, write a novel, become rich, help the poor, find God, fight against evil, reach a state of contentment and serenity, or bring about the revolution. None of these things are encompassed by the concepts of rightness and permissibility; MJA are blind to them.


Nor do these concepts encompass the moral relationship of admiration, and the class of acts known as supererogatory. When presented with morally extraordinary people, acts, or states of affairs, people may feel or express admiration. Consider the lives of the saints and the heroes. Their acts are obviously permissible and obviously not obligatory. But that doesn’t help us describe and account for how people seem to react and relate to them. The same can be said about abhorrent, abominable, and despicable acts and people. The experience and social consequences of the morally admirable/heroic and despicable/monstrous don’t seem understandable in the same terms as the good/bad, much less the right/wrong. They are not quantitatively more intense, but qualitatively different.


I’ve suggested that MJA miss the good and the supererogatory completely or almost completely. I believe it doesn’t cover, either, people’s judgments about character, virtue, and, more generally, what some ethicists call “ethics of being.” It seems that sometimes people’s moral judgments, questions, and experiences are not about what to do, but about what or how to be. Someone may be referred to as a bad, good, self-absorbed, vain, generous, fair, open-minded, or irresponsible person. Or as someone who has guts, or is depraved, or sly, or manipulative, which in turn may become reasons for action (e.g., “I decided not to engage in business with Jones, because he’s an irresponsible fellow”; or, “Try to help him if you can—he’s a good man”). Interestingly, these judgments about being are presumably based, inductively, on individual instances of doing. But they are nonetheless not reducible to them. Moreover, there is a sort of stickiness to judgments about being or character, precisely because they are seen as referring to relatively stable traits or dispositions. Indeed, a judgment about how someone is might shape future instances of judgment about what they do. In addition, moral judgments and questions about yourself, and moral exhortations to yourself, may come in the language of being and character, not in the language of doing. You may hope and strive to be an honest, courageous, pious, or respectful person, without it being possible to translate this into a set of concrete and exhaustive judgments, principles, or action-maxims. Your moral self-conception or identity—what kind of person you see yourself as, what kind of person you hope to become, who are your moral heroes and antiheroes—is missed by talk about right, wrong, permissible, and forbidden, which is what MJA investigate.


Furthermore, you may be unable (or unwilling) to bracket the fact that moral life and moral action are (1) the moral life and moral action of particular people, which (2) necessarily take place within a context. As it’s been often observed, agent-neutral ethical theories such as utilitarianism fail to take this into account. So does a methodological approach such as MJA. To return to the example of the trolley-problem experiments, without any (social, cultural, religious, legal) context, it might be hard for subjects to decide what the right thing to do is, or even to find the question meaningful and see its point at all. “Well, it really, really depends,” they may think to themselves. In addition, subjects may wonder whether the fat man to be sacrificed for the greatest good of the greatest number is an HIV/AIDS researcher who may discover a cure for it, or a ruthless Uruguayan dictator. Not being analytic philosophers, subjects may ask themselves: what in the world does it mean that he is nobody in particular—neither a medical researcher nor a dictator, neither kind nor unkind, neither old nor young?


I’ve been making some empirical conjectures about ordinary people’s moral lives, and specifically whether MJA may be missing some of its components. I have not been talking about the academic field of moral philosophy and its normative and metaethical debates. Yet, many of my points have mutatis mutandis a counterpart in moral philosophy. For instance, if the good ought to have priority over the right, if the “law conception of ethics” is misguided, if Kantians can accommodate the concept of the supererogatory, if utilitarians can accommodate the concepts of identity and integrity. More precisely, many of my points have a counterpart in several strands of criticisms leveled at mainstream analytic ethics. This is not a mere coincidence. For MJA’s conception of morality heavily draws on the particular conception of morality that consequentialism and deontology share, yet which is not shared by other traditions. For instance, a pragmatist, existentialist, communitarian, particularist, Buddhist, or virtue ethicist would probably see the disputes between consequentialism and deontology as pointless. Today’s scientists of morality have framed their question and modeled their object of inquiry after these two schools, thereby unwittingly taking sides on a substantive issue in ethics.

In this paper I’ve asked what moral judgment-centric approaches (MJA) in psychology and neuroscience tell us about morality. I’ve argued that their object of study, MJ, is not a good sample of moral things. There are many things that are moral, yet not moral judgments. There are also many things that are moral judgments, yet not of the particular MJ kind (as defined above). If these arguments are correct, it follows that MJA research doesn’t license theoretical conclusions about morality in general. Rather, it only licenses conclusions about the specific moral object it has actually investigated. To be sure, many suggestive experimental findings about this object have been reported. But I don’t think enough thought has been given to what to make of them theoretically, what social phenomena they illuminate, what neural correlates are correlates of, or what to do with these findings to build a satisfactory understanding of morality. To be sure, the new science of morality is still in its infancy. But that doesn’t make unwarranted or unclear claims any more acceptable.


Here a pluralist approach would argue that if morality is made up of many different kinds of things, then a compelling science of morality should consider all of them as objects of inquiry, and take them all into account when making claims about what morality is or how morality works. Because they are different kinds of things, they call for different questions and methods. Further, they may yield theoretical claims that aren’t subsumable under a single, comprehensive theory of morality, or parsimonious principle about the nature of morality. At least, you shouldn’t start with the assumptions that: (a) there is one such theory or principle; and (b) there is a basic micro-unit, common to all moral things.


I think such pluralist view is on the right track. Whatever the theoretical meaning and value of MJA findings turn out to be, they are about and shed light on one particular kind of moral thing. In order to make claims about morality in general, many other objects must be included. And many other methods must be used: psychological and neuroscientific, as well as anthropological, historical, and sociological. For example, for certain research questions there’s no way around the ethnographic observation of action and interaction in their natural setting. For certain research questions there’s no way around statistically representative surveys of a population. If you intend to use scientific knowledge about morality to make practical recommendations to policymakers, then an organizational analysis is unavoidable—individual-level differences may be inconsequential or even irrelevant. In some social situations, it’s not individuals’ automatic intuitions or reactions to stimuli, but deliberation and debate that carry the day regarding what ends up being done and even what ends up being believed and felt by those very individuals. Similarly, if you’re interested in moral narratives, you’ll have to patiently listen to them or read them, give people some time and freedom to express themselves, and figure out their meanings.

In sum, I believe that there is neither a single best method to study morality, nor a single object of inquiry on the basis of which to make claims about morality. Nor is there a single path toward an understanding or theory of morality. Therefore, a plurality of theoretical and methodological approaches should coexist, each aware of its strengths, but also of its partial perspective and limited scope. In practice, there should be more variation in objects of inquiry and methods. Individual judgment about rightness and wrongness is one among many objects of inquiry scientists of morality need to do research on; so are moral questions, questions about good rather than right, questions about being rather than doing, thick judgments, narratives, institutions, and behavior in natural settings. Neuroimaging and experimental methods are two among many methods scientists of morality need to use; so are surveys, ethnographic observation, and the history of moral concepts and practices.


In these respects, moral psychologists and neuroscientists can benefit from more engagement with historians, anthropologists, and sociologists of morality. Certainly, in one sense it’s entirely sensible that there be a division of labor between, say, neuroscience, sociology, and history. However, this division of labor entails two perils. First, it’s a consistent historical pattern that scientific disciplines tend to forget the incompleteness of their standpoint and disregard findings that don’t fit with their approach. Second, some of the phenomena and processes of interest to a science of morality can’t be broken into separate parts—e.g., the social and the neural—because these parts mutually influence or even constitute each other. Hence, studying them separately may lead to misleading results. Unfortunately, a pluralist science of morality entails costs, too. First, its accounts and theories are unlikely to be simple and parsimonious, much like other versions of methodological, explanatory, and ontological pluralism in science. Second, as in politics, this sort of pluralism is time-consuming and hard to realize. For it requires interactions among people who differ not just in the content of their answers, but in the form of the questions they ask, and in the questions they find worth asking to begin with. Yet, just as in politics, this might still be the best way to prevent oversights and build a robust scientific understanding of morality, not just of a particular kind of individual moral judgment.



Gabriel Abend | résident à l’IEA de Paris

Gabriel Abend is an Assistant Professor of Sociology at New York University. His book, The Moral Background: An Inquiry into the History of Business Ethics, was published in April 2014 by Princeton University Press. This text is an abridged version of an article originally published in 2013 in Philosophy of the Social Sciences, volume 43, number 2, pp. 157-200.


01/10/2013 - 30/06/2014