Against Empathy: The Case for Rational Compassion

by Paul Bloom

The idea I’ll explore is that the act of feeling what you think others are feeling—whatever one chooses to call this—is different from being compassionate, from being kind, and most of all, from being good. From a moral standpoint, we’re better off without it.

It might feel, at least to some of us, that our opinions about issues such as abortion and the death penalty are the product of careful deliberation and that our specific moral acts, such as deciding to give to charity or visit a friend in the hospital—or for that matter, deciding to shoplift or shout a racist insult out of a car window—are grounded in conscious decision-making. But this is said to be mistaken. As Jonathan Haidt argues, we are not judges; we are lawyers, making up explanations after the deeds have been done. Reason is impotent. “We celebrate rationality,” agrees de Waal, “but when push comes to shove we assign it little weight.”

Empathy is a spotlight focusing on certain people in the here and now. This makes us care more about them, but it leaves us insensitive to the long-term consequences of our acts and blind as well to the suffering of those we do not or cannot empathize with.

Empathy is the act of coming to experience the world as you think someone else does. Empathy in this sense was explored in detail by the philosophers of the Scottish Enlightenment, though they called it “sympathy.” As Adam Smith put it, we have the capacity to think about another person and “place ourselves in his situation . . . and become in some measure the same person with him, and thence form some idea of his sensations, and even feel something which, though weaker in degree, is not altogether unlike them.”

Our decisions about what’s right and what’s wrong, and our motivations to act, have many sources. One’s morality can be rooted in a religious worldview or a philosophical one. It can be motivated by a more diffuse concern for the fates of others—something often described as concern or compassion and which I will argue is a better moral guide than empathy.

There are all sorts of real-world acts of kindness that are not prompted by empathic concern. We sometimes miss these cases because we are too quick to credit an action to empathy when actually something else is going on.

But spotlights have a narrow focus, and this is one problem with empathy. It does poorly in a world where there are many people in need and where the effects of one’s actions are diffuse, often delayed, and difficult to compute, a world in which an act that helps one person in the here and now can lead to greater suffering in the future. Further, spotlights only illuminate what they are pointed at, so empathy reflects our biases.

Intellectually, we can value the lives of all these individuals; we can give them weight when we make decisions. But what we can’t do is empathize with all of them. Indeed, you cannot empathize with more than one or two people at the same time.

Empathy is particularly insensitive to consequences that apply statistically rather than to specific individuals.

I’ll argue that what really matters for kindness in our everyday interactions is not empathy but capacities such as self-control and intelligence and a more diffuse compassion. Indeed, those who are high in empathy can be too caught up in the suffering of other people.

But this understanding of the minds of others is an amoral tool, useful for effecting whatever goals you choose. Successful therapists and parents have a lot of cognitive empathy, but so too do successful con men, seducers, and torturers. Or take bullies.

I believe that the capacity for emotional empathy, described as “sympathy” by philosophers such as Adam Smith and David Hume, often simply known as “empathy” and defended by so many scholars, theologians, educators, and politicians, is actually morally corrosive. If you are struggling with a moral decision and find yourself trying to feel someone else’s pain or pleasure, you should stop. This empathic engagement might give you some satisfaction, but it’s not how to improve things and can lead to bad decisions and bad outcomes. Much better to use reason and cost-benefit analysis, drawing on a more distanced compassion and kindness.

Empathy is used as a tool by charitable organizations, religious groups, political parties, and governments, and to the extent that those who spark this empathy have the right moral goals, it can be a valuable force. While I think empathy is a terrible guide to moral judgment, I don’t doubt that it can be strategically used to motivate people to do good things.

Or consider a recent study by Abigail Marsh and her colleagues, of people who choose to donate their kidneys to strangers. Consistent with my argument, these exceptionally altruistic individuals do not score higher on standard empathy tests than normal people. But they are different in another way. The researchers were interested in the amygdala—a part of the brain that is involved in, among other things, emotional responses. Their previous research had discovered that psychopaths had smaller than normal amygdalae and lessened response when exposed to pictures of people who looked frightened, so they predicted that these do-gooders would have larger than normal amygdalae and greater than normal response to fear faces. This was exactly what they found.

We do best when we rely on reason. Michael Lynch defines reason as the act of justification and explanation—to provide a reason for something is to justify and explain it, presumably in a way that’s convincing to a neutral third party. More specifically, reasoning draws on observation and on principles of logic, with scientific practice being the paradigmatic case of reason at work.

It turns out, though, that this the-whole-brain-does-it conclusion arises because neuroscientists—along with psychologists and philosophers—are often sloppy in their use of the term empathy. Some investigators look at what I see as empathy proper—what happens in the brain when someone feels the same thing they believe another person is feeling. Others look at what happens when we try to understand other people, usually called “social cognition” or “theory of mind” but sometimes called “cognitive empathy.” Others look at quite specific instantiations of empathy (such as what happens when you watch someone’s face contort in disgust), and still others study what goes on in the brain when a person decides to do something nice for another person, which is sometimes called “prosocial concern” but which one normally thinks of as niceness or kindness. Once you start pulling these different phenomena apart, which I’ll do below, things get more interesting, and you see how these different capacities relate to one another.

An empathic response can be automatic and rapid. If you see someone hitting his finger with a hammer, you might flinch, and this seems to be a reflexive response. But for the most part, whether or not we are consciously aware of it, empathy is modified by our beliefs, expectations, motivations, and judgments. This is the second finding from neuroscience: Our empathic experience is influenced by what we think about the person we are empathizing with and how we judge the situation that person is in.

We mirror the feelings of others, but this mirroring is limited: Empathic suffering is different from actual suffering. Empathy is also contingent on how one feels about an individual. It’s not always the case, then, that we feel empathy and thereby treat someone well. Instead, we often think that someone is worth treating kindly (because he or she treated us nicely in the past or is simply like us) and then we feel empathy. And finally, emotional empathy—the sort of empathy that we’re obsessing about here—can be usefully disentangled from the essential capacity of understanding other people.

I favor Batson’s own analysis that empathy’s power lies in its capacity to make the experience of others observable and salient, therefore harder to ignore. If I love my baby, and she’s in anguish, empathy with her pain will make me pick her up and try to make her pain go away. This is not because doing so makes me feel better—it does, but if I just wanted my vicarious suffering to go away, I’d leave the crying baby and go for a walk. Rather, my empathy lets me know that someone I love is suffering, and since I love her, I’ll try to make her feel better.

Another issue is that the standard empathy scales are imperfect measures of empathy. The most popular measures include questions that are related to empathy in the sense of mirroring others’ feelings, but they also have questions that tap other capacities, such as kindness or compassion or interest in others.

A recent paper reviewed the findings from all available studies of the relationship between empathy and aggression. The results are summarized in the title: “The (Non)Relation between Empathy and Aggression: Surprising Results from a Meta-Analysis.” They report that only about 1 percent of the variation in aggression is accounted for by lack of empathy. This means that if you want to predict how aggressive a person is, and you have access to an enormous amount of information about that person, including psychiatric interviews, pen-and-paper tests, criminal records, and brain scans, the last thing you would bother to look at would be measures of the person’s empathy.

Being high in empathy doesn’t make one a good person, and being low in empathy doesn’t make one a bad person. What we’ll see in the chapters that follow is that goodness might be related to more distanced feelings of compassion and care, while evil might have more to do with a lack of compassion, a lack of regard for others, and an inability to control one’s appetites.

The effect was strong. Three-quarters of the subjects in the high-empathy condition wanted to move her up, as compared to one-third in the low-empathy condition. Empathy’s effects, then, weren’t in the direction of increasing an interest in justice. Rather, they increased special concern for the target of the empathy, despite the cost to others.

All of these laboratory effects can be seen as manifestations of what’s been called “the identifiable victim effect.”

This effect also illustrates something more general about our natural sentiments, which is that they are innumerate. If our concern is driven by thoughts of the suffering of specific individuals, then it sets up a perverse situation in which the suffering of one can matter more than the suffering of a thousand.

Plainly, then, the salience of these cases doesn’t reflect an assessment of the extent of suffering, of their global importance, or of the extent to which it’s possible for us to help. Rather, it reflects our natural biases in who to care about. We are fascinated by the plight of young children, particularly those who look like us and come from our community. In general, we care most about people who are similar to us—in attitude, in language, in appearance—and we will always care most of all about events that pertain to us and people we love.

While writing this book, I discovered that there is a field of study called “disaster theory.” A lot of the work in this area explores self-interested motivations. In the United States, for instance, presidents are more likely to declare national disasters during election years, and battleground states get more donations than others; money allocated to address disasters is used as an inducement and a reward. Other research in this area illustrates the arbitrariness of what we focus on, the way our interests fail to coincide with any reasonable assessment of where help is needed the most or where people can do the most good. This is the sort of thing Isaacson was complaining about.

Empathy’s narrow focus, specificity, and innumeracy mean that it’s always going to be influenced by what captures our attention, by racial preferences, and so on. It’s only when we escape from empathy and rely instead on the application of rules and principles or a calculation of costs and benefits that we can, to at least some extent, become fair and impartial.

But doing actual good, instead of doing what feels good, requires dealing with complex issues and being mindful of exploitation from competing, sometimes malicious and greedy, interests. To do so, you need to step back and not fall into empathy traps. The conclusion is not that one shouldn’t give, but rather that one should give intelligently, with an eye toward consequences.

Expanding on this point, Alexander makes a distinction between “man versus nature” problems and “man versus man” problems. Healing the sick is an example of “man versus nature,” and this is the sort of thing that effective altruists now focus on. Fighting global capitalism is “man versus man.” This has the potential for long-lasting change for the better, but the outcome is less certain. After all, many people are in favor of global capitalism, and many honestly believe that the spread of market economies is what will make the world a better place.

These observations bring us back to a complaint I’ve made before—empathy is innumerate and biased. Hearing that my child has been mildly harmed is far more moving for me than hearing about the horrific death of thousands of strangers. This might be a fine attitude for a father—we’ll return to that question at the end of the next chapter—but it’s a poor attitude for a policy maker and a poor moral guide to our treatment of strangers.

We are not psychologically constituted to feel toward a stranger as we feel toward someone we love. We are not capable of feeling a million times worse about the suffering of a million than about the suffering of one. Our gut feelings provide the wrong currency through which to evaluate our own moral actions.

Scarry suggests that we do the opposite. Don’t try to establish equality and justice by raising others up to the level of those you love. Don’t try to make them more weighty. Rather, make yourself less weighty. Bring everyone to the same level by diminishing yourself. Put yourself, and those you love, on the level of strangers.

Or consider why economics is sometimes called “the dismal science.” It’s a derogatory description thought up by Thomas Carlyle in the 1800s, coined to draw a contrast with the “gay science” of music and poetry: “Not a ‘gay science,’ I should say, like some we have heard of; no, a dreary, desolate and, indeed, quite abject and distressing one; what we might call, by way of eminence, the dismal science.” Carlyle has a specific issue in mind, a case where he wanted to ridicule economists for objecting to something that was the subject of considerable feeling and heart, something that Carlyle had defended with great emotion. What was this issue that the economists were being so negative about? Slavery. Carlyle was upset because the economists were against slavery. He argued for the reintroduction of slavery in the West Indies and was annoyed that the economists railed against it. Think about this when you’re tempted to scorn economists and the cool approach they take to human affairs, and when you hear people equating strong feelings with goodness and cold reason with nastiness. In the real world, as we’ve seen, the truth is usually the opposite.

Indeed, some believe that a political continuum from left to right might be universal. John Stuart Mill pointed out that political systems have “a party of order or stability and a party of progress or reform.” Ralph Waldo Emerson wrote that “the two parties which divide the state, the party of conservatism and that of innovation, are very old, and have disputed the possession of the world ever since it was made,” and he went on to conclude that such “irreconcilable antagonism must have a correspondent depth of seat in the human condition.” This antagonism is stronger with social issues. Our political natures seem to manifest themselves most clearly with, as one set of scholars put it, “matters of reproduction, relations with out-groups, suitable punishment for in-group miscreants, and traditional/innovative lifestyles.” Less intimate issues, such as free trade or deregulation of banks, are less predictable and aren’t as reliably related to one’s broader political orientation.

A different analysis of the liberal-conservative contrast is proposed by Jonathan Haidt, based on his theory that humans possess a set of distinct moral foundations—including those concerning care, fairness, loyalty, authority, and sanctity. These are evolved universals, but they admit of variation, and research by Haidt and his colleagues suggests that liberals emphasize care and fairness over the others, while conservatives care about all these foundations more or less equally. This is why, according to Haidt, conservatives care more than liberals do about respect for the national flag (as this is associated with loyalty), children’s obedience toward parents (authority), and chastity (sanctity). And again this perspective has conservatives drawing upon nonempathic values more so than liberals.

Still, the cost of a politics of empathy is massive. Governments’ failures to enact prudent long-term policies are often attributed to the incentive system of democratic politics (which favors short-term fixes) and to the powerful influence of money. But the politics of empathy is also to blame. It is because of empathy that citizens of a country can be transfixed by a girl stuck in a well and largely indifferent to climate change. It is because of empathy that we often enact savage laws or enter into terrible wars; our feeling for the suffering of the few leads to disastrous consequences for the many. A reasoned, even counterempathic analysis of moral obligation and likely consequences is a better guide to planning for the future than the gut wrench of empathy. This is not a partisan point; it’s a sensible one.

associated with. I’m no longer so certain. A careful look at empathy reveals a more complicated story. Here as always it’s important to distinguish empathy from understanding. It’s undeniably a good thing when the people in our lives understand us. And it’s even more important to distinguish empathy from compassion, warmth, and kindness. Nobody could deny that we want the people in our lives to care about us.

So what’s the difference between people who are high in communion (positive) and those who are high in unmitigated communion (negative)? Both sorts of people care about others. But communion corresponds to what we can call concern and compassion, while unmitigated communion ends up relating more to empathy or, more precisely, empathic distress—suffering at the suffering of others.

In his book on Buddhist moral philosophy, Charles Goodman notes that Buddhist texts distinguish between “sentimental compassion,” which corresponds to what we would call empathy, and “great compassion,” which is what we would simply call “compassion.” The first is to be avoided, as it “exhausts the bodhisattva.” It’s the second that is worth pursuing. Great compassion is more distanced and reserved, and can be sustained indefinitely. This distinction between empathy and compassion is critical for the argument I’ve been making throughout this book. And it is supported by neuroscience research. In a review article, Tania Singer and Olga Klimecki describe how they make sense of this distinction: “In contrast to empathy, compassion does not mean sharing the suffering of the other: rather, it is characterized by feelings of warmth, concern and care for the other, as well as a strong motivation to improve the other’s well-being. Compassion is feeling for and not feeling with the other.”

There is a neural difference: Empathy training led to increased activation in the insula and anterior cingulate cortex (both of which we discussed in relation to the neuroscience-of-empathy studies in an earlier chapter). Compassion training led to activation in other parts of the brain, such as the medial orbitofrontal cortex and ventral striatum. There is also a practical difference. When people were asked to empathize with those who were suffering, they found it unpleasant. Compassion training, in contrast, led to better feelings on the part of the meditator and kinder behavior toward others.

This connects nicely with the conclusions of David DeSteno and his colleagues, who find, in controlled experimental studies, that being trained in mindfulness meditation (as opposed to a control condition where people are trained in other cognitive skills) makes people kinder to others and more willing to help. DeSteno and his colleagues argue that mindfulness meditation “reduces activation of the brain networks associated with simulating the feelings of people in distress, in favor of networks associated with feelings of social affiliation.”

There is a world of difference, after all, between understanding the misery of the person who is talking to you because you have felt misery in the past, even though now you are calm, and understanding the misery of the person who is talking to you because you are mirroring them and feeling their misery right now. The first, which doesn’t involve empathy in any sense, just understanding, has all the advantages of the second and none of its costs.

This is a better model for what should go on in therapy—the trick, then, is not for the therapist to have empathy; it’s for the patient to have it.

Then there is the wish to restore balance. Pamela Hieronymi puts it like this: “A past wrong against you, standing in your history without apology, atonement, retribution, punishment, restitution, condemnation, or anything else that might recognize it as a wrong, makes a claim. It says, in effect, that you can be treated in this way, and that such treatment is acceptable.” Those practices she lists, starting with apologies, serve to repair the victim’s status—to use that lovely legal expression, they serve to make the victim whole again. From this perspective, an apology involves an acknowledgment that it is unacceptable to harm someone without just cause. For this to work, it has to be somehow costly; you need to know that the person means it, so some suffering is needed. Empathy allows for a perfect eye-for-an-eye correspondence, where the perpetrator experiences the very same suffering as the victim.

As an intelligent utilitarian, Singer appreciates that some parochial actions and attitudes might serve to maximize overall happiness. If you and I both have babies, they are most likely to survive if I take care of mine and you take care of yours. But a utilitarian like Singer—in direct opposition to someone like Asma–would insist that this bias has no intrinsic value. Like our appetite for punishment, our relatively greater concern for those close to us might be a necessary evil. Singer is in good company when he dismisses the intrinsic value of intimate relationships, and it’s not just Gandhi.

Yes, empathy is biased and parochial—but in a stupid way. Even if we decide that certain individuals are worthy of special treatment, even here empathy lets us down, because empathy is driven by immediate considerations, making us too-permissive parents and too-clingy friends. It’s not just that it fails us as a tool for fair and impartial moral judgment, then, it’s often a failure with intimate relationships. We can often do much better.

If I’m in pain because I’m feeling your pain, there is a much easier way to make my pain go away than helping you—I can turn my head and stop thinking of you; the empathic connection is broken, and I’m right as rain. Then there’s Batson’s research, which shows that people tend to help even when escape is readily available. This is a problem for the selfishness theory of the power of empathy and is more consistent with the view that empathy motivates good behavior (when it does) by exploiting positive sentiments that are already present.

The claim that we actually only care about survival and reproduction confuses the goals of natural selection (again, metaphorically speaking) with the goals of the creatures who have evolved through natural selection, including us. The difference between the two is obvious when you think about other domains.

From this perspective, violence is a glitch in the system, something gone wrong. Adrian Raine has likened violent crime to a kind of cancer, as both are products of a combination of genes and environment, and both can be seen as diseases that deserve treatment. But there is another, opposite, view, popular among economists and evolutionary theorists. This is that violence is an essential part of life, an often rational solution to certain problems. Cancer is an aberration, an illness, something that could be cleanly excised from the world: If it were eradicated tomorrow, the rest of human life would remain happily intact. But violence is part of human nature, shared with other animals, evolved for punishment, defense, and predation. And unless we are transformed into angels, violence and the threat of violence are needed to rein in our worst instincts. You can have a world without cancer, but there will never be a world without violence. Since the recipients of violence are rarely pleased with the violence directed toward them, then, in the eyes of some, at least, there will never be a world without evil.