Last year a striking video made its way around the Internet. In it, male sports fans sat, one at a time, opposite a female sports reporter who had been the target of abusive, misogynist tweets. Each man had to read the messages aloud to the woman who received them. One of the few printable examples was, “I hope your boyfriend beats you.” The goal of the project, created by a Web site called Just Not Sports, was to force the men to experience “the shocking online harassment happening to women in sports day in, day out.” By ripping away the protective anonymity of social media, the exercise drove home the message that if something is too offensive to say face-to-face, it is too offensive to type. The men were visibly pained as they read. They squirmed in their chairs. One guy looked like he had been punched in the gut. Every man involved appeared to come away with a better sense of how awful it was to be on the receiving end of such nastiness.

At its core, this exercise illustrated what empathy looks like: the capacity to share what someone else is feeling—a tingling fear as you watch a tightrope walker attempt to cross Niagara Falls or butterflies in your stomach because your nervous child is about to perform in a recital. In the 18th century, economist Adam Smith was among the first to name this emotion, calling it “fellow-feeling”—the sensation that something you see happening to another person is happening to you as well. The Germans call it Einfühlung, meaning “feeling into.” Yet there is more to empathy than shared feelings.

Fifteen years of neuroscientific investigation has led most scientists to see empathy as an umbrella term covering three main components. Emotional empathy—sharing another's feelings and matching that person's behavioral states (feeling afraid, for instance, when someone else is on a tightrope)—is a biological response found in many different species that evolved in the context of parental care and group living. Cognitive empathy, also called perspective taking or theory of mind, is the capacity to think about and understand another's feelings. And empathetic concern, or compassion, adds the motivation to do something about another's suffering. Taken together, these components are fundamental elements of our social lives.

“People empathize because it's absolutely critical for forming close relationships or relating to people at all,” says psychologist Jamil Zaki of Stanford University. Disentangling these components—even deciding if they should be disentangled—has been a thorny undertaking. In a 2008 paper, primatologist Frans B. M. de Waal of Emory University, a pioneer in the field, described empathy as a “Russian doll,” with “simple mechanisms at its core and more complex mechanisms and perspective-taking abilities as its outer layers.” Others take another view, focusing on the differences and preferring narrower interpretations.

Such varied definitions occupy the center of recent public debates about empathy, spurred mostly by the publication last year of Yale University psychologist Paul Bloom's book Against Empathy. Bloom devoted a lot of space to specifying which empathy he does not like: cognitive empathy is fine, but he views emotional empathy as a poor basis for moral behavior, arguing that “we’re better off without it.” We can't get rid of emotional empathy—but Bloom has a point. Empathy is not always good. Even de Waal has acknowledged, in a 2009 book, that there is “no obligatory connection between empathy and kindness.” In some situations, empathy causes emotional distress, and it is naturally biased toward those closest to us and away from others.

As a society, we do not generally see it that way. In 2006 Barack Obama was not talking about the negative side of empathy when he famously decried society's “empathy deficit.” And employers turning to “empathy training,” which is especially popular in schools, hospitals, corporations and police departments, are looking to solve problems, not create them. Yet empathy's built-in bias lies at the heart of the bitter divisions in American society after the election of Donald Trump. “It's very difficult, and painful, and uncomfortable to try to take the perspective, really understand the experience, of someone who you’ve hurt or someone whose opinion is distasteful to you,” Zaki says.

As research into empathy matures, what is emerging is a more sophisticated view of a nuanced and complex emotion that often depends on the particular context in which it is manifested. Psychologists and neuroscientists want to understand better how empathy works: when it works for us and when it works against us. The good news is that empathy, in the broadest sense, is not a pop psychological artifact. It can, in fact, be learned through training as a means to help resolve disputes. But teaching it needs to be done with care. Cognitive neuroscientist Emile Bruneau, director of the Peace and Conflict Neuroscience Lab at the University of Pennsylvania, is studying empathy's role in conflict resolution but cautions, “We need to know the pitfalls and test to make sure they’re not having an ironic effect. And then we can use that information to build interventions that are more effective.”

A multilayered phenomenon

Psychologists have been interested in empathy for decades, but the approach of bringing in neuroscience to study the emotion is only in its adolescence. The first decade or so of work focused on establishing the independent yet interacting neural networks that underlie emotional and cognitive empathy. In 2004 neuroscientist Tania Singer, now at the Max Planck Institute for Human Cognitive and Brain Sciences in Leipzig, Germany, and her colleagues published a groundbreaking paper in Science that compared brain activity in a person experiencing pain with the same person's brain activity when observing a loved one experiencing pain. Sixteen women underwent functional magnetic resonance imaging while their male partner sat nearby. Varied levels of painful stimulation were administered by an electrode to one or the other partner. A signal alerted the women when their partner was feeling pain. Some areas of the women's brains were activated only on receiving pain themselves, but others—most notably parts of the anterior insula and the anterior cingulate cortex—lit up no matter who was hurting. Empathy activated the affective, or emotional, parts of the pain network but not the physical sensation of pain. That study and the many imaging studies that followed indicate that our core ability to empathize begins with the way the brain represents our own internal states and evolved to include our perception of what others are feeling.

Cognitive empathy, in contrast, represents the more taxing effort of understanding and reasoning about the state of another individual, a capacity also called mentalizing, or theory of mind. It emerges in children around their first birthday and continues to develop into adulthood. The brain's mentalizing network has consistently been shown to include the superior temporal sulcus, temporal poles and temporoparietal junction. Another area is the medial prefrontal cortex behind the forehead, which is associated with thinking about oneself. Empathetic concern activates yet another network.

The role of each facet of empathy is most obvious when one goes missing. A person with autism spectrum disorder has little ability to assume the perspective of someone else. Psychopaths, on the other hand, understand what others are feeling but have a profound lack of empathetic concern. “They know right from wrong but don't care,” wrote neuroscientist Jean Decety and his colleague Keith J. Yoder, both at the University of Chicago, in a 2016 study. Multiple studies led by Decety have found that people with high levels of psychopathy show abnormal connections among neurons and neural activity in areas of the brain associated with empathy.

Most recently, in their 2016 study, Decety and Yoder assessed 265 people on scales of empathetic concern, psychopathy and sensitivity to moral questions related to a sense of justice. Participants then considered eight scenarios and were asked how permissible it would be to behave in a particular way. For example, when running for an infrequent bus, would it be acceptable not to stop to help a woman with a small child who has spilled the contents of her purse? Cognitive empathy, but not emotional empathy, was found to predict a sense of justice for others. Those high in “coldheartedness”—a measure of psychopathy—were the least motivated by a sense of justice, an individual's perception of injustice or the intensity of response to perceived unfairness. The researchers concluded that “it may be more effective to encourage perspective taking and reasoning [cognitive empathy] to induce concern for others than emphasizing emotional sharing with the misfortune of others.”

Credit: Tami Tolpa; Source: “Empathy for Pain Involves the Affective but not Sensory Components of Pain,” by Tania Singer et al., in Science, Vol. 303; February 20, 2004

A study published last year in Science was the first to suggest not just where but how the processing of empathy might work. James Burkett, a neuroscientist at Emory, found consoling behavior in prairie voles, a species known for its strong social nature. Pairs of male and female animals were caged together for a few weeks, then the female was removed briefly. She either was simply kept separate for a few minutes or was given a mild foot shock, a form of fear conditioning that generates stress. When the animals were reunited in the cage, Burkett's team observed their social interactions. If the female had not been stressed, neither animal seemed particularly anxious. But when she had been shocked, the male quickly began grooming her intensely—behavior interpreted as consoling because the pair did not engage in it otherwise and because it had a calming effect on the shocked animal.

The animal left behind showed a physiological response that mimicked that of the animal who had been taken away. Furthermore, the intensity of the consoling response varied from animal to animal. When Burkett looked at oxytocin, which promotes social bonding, in the brains of the voles, he discovered something interesting. The variation in behavior was predicted by the density of docking sites, or receptors, for oxytocin in the same part of the brain—the anterior cingulate cortex—that Singer had identified in humans who felt pain empathy for others. As the density of receptors increased, the amount of time animals spent consoling decreased. Burkett hypothesizes that oxytocin signaling in that area of the brain might encode for personal distress in response to the distress of others. “Some level of concern for the distress of others is necessary to motivate consoling, but too much personal distress causes individuals to avoid rather than engage,” he says.

The downside of empathy

Burkett's study provides a possible explanation for one of the negative aspects of empathy. When the emotions experienced are stressful or painful, empathy is painful—an explanation for why we sometimes avoid such feelings. “If I empathize with everyone who is in a worse state than I am, I might be motivated to donate 95 percent of my income to charity,” Zaki says. “Rather than being put in a moral double bind between guilt and poverty, I might just choose not to think about people who are less fortunate than myself.” In certain professions, such as medicine and law enforcement, where exposure to human suffering can be constant, too much personal distress gets in the way of doing the job. Physicians, for example, suffer from excessive burnout and are at higher risk than others for death by suicide.

A more universal problem is that empathy is biased. “It evolved so that we have more empathy for our family and friends than anybody else,” says de Waal, who has extensively studied the evolution of empathy. That makes sense: group living is designed to protect against predation, and individuals with strong social bonds live longer and have more reproductive success than others. Thus, we are drawn to kith and kin and naturally avoid outsiders. As Mina Cikara, director of the Intergroup Neuroscience Lab at Harvard University, puts it: “A fundamental component of human nature is the tendency to draw bright boundaries between us and them.”

A series of studies over the past decade has explored this problem and shown that the issue of in-groups and out-groups applies not only to differences of race and ethnicity but also to long-standing sports and college rivalries. Cikara and her colleagues have found that avid Boston Red Sox fans (her husband is one) are more likely to feel pleasure not only when their team plays well but also when their archrival, the New York Yankees, plays badly. And those who feel that schadenfreude most powerfully are more likely to get into a fight with a Yankees fan and inflict harm. This holds true even when the competing groups are made up. In several studies, Cikara and her colleagues randomly assigned study participants to teams they dubbed the Rattlers and the Eagles and then measured responses to positive or negative incidents that happened to members of each team. Not only did identifying with the Rattlers dampen empathy for the Eagles, it also increased the counterempathetic response—also known as not being very nice.

In a recently published study, Bruneau, Cikara and Rebecca Saxe of the Massachusetts Institute of Technology tried to determine which aspect of empathy was the best predictor of helping behavior across group boundaries. They recruited three sets of participants: Americans asked to think about Arabs, Hungarians asked to think about Muslim refugees and Greek citizens asked to think about Germans in the aftermath of the Greek debt crisis. Participants were assessed for general empathetic concern and for “parochial empathy,” the degree to which people feel empathy toward their own versus another group. Each study was slightly different, but in the American-Arab study, participants read about positive and negative events happening to characters such as “Beth” from North Dakota or “Salma” from Egypt. They were later asked questions such as whether they would provide U.S. visas to Arabs and donate to an Arab charity. In every instance, parochial empathy was the more significant predictor of the outcome. The higher it was, the less altruism was displayed. General empathetic concern predicted nothing.

This research highlights the complexity of using empathy to improve relationships between distrustful groups. “If you bring kids from either side of a conflict together, and you manage to increase their global empathy, that won't have any effect on how they treat the out-group,” Bruneau says. “And if they form close ties with members of their own group, you might have actually increased empathy for the in-group more than for the out-group.” To truly improve such a situation, he notes, requires precision: when such a method is proposed, say a camp bringing together Catholic and Protestant kids from Northern Ireland, for instance, “what you really mean is you want to decrease the gap in empathy between in-group and out-group.”

A force for good?

Improving a situation is, of course, the goal of empathy interventions. But will the sports fan made to feel the hurt of misogynist tweets change his own behavior in the future? Not necessarily. In one of the earliest and most well-known social psychology experiments, C. Daniel Batson and John Darley, both then at Princeton University, proved this outcome in spectacularly ironic fashion. In 1973 they assigned some seminary students to give a talk about the parable of the Good Samaritan and others to give one about a topic unrelated to altruism. Then they arranged things so that the seminarians had to rush from one building to another to give the talk. Along the way, each passed a miserable figure moaning on the sidewalk. The study counted which seminarians stopped to help. Being well versed in the story of the Good Samaritan made no difference in the likelihood of offering help, but being in a hurry markedly decreased the inclination to do so. Decades later, however, Batson, now a professor emeritus at the University of Kansas, established that people who feel compassion help more often than those who are upset by others’ distress.

Since her pioneering work in 2004, Singer has shifted her interest to focus squarely on compassion. In several studies published since 2012, her laboratory examined the neural effects of training for compassion, which it defined as a feeling of concern that includes the motivation to help. The training consisted of a contemplative technique that extends the caring feelings people usually feel for close loved ones to other human beings. The researchers found that contemplative training increased positive emotional experiences, even when witnessing others in distress. It also raised activity in parts of the brain associated with perspective taking. In a 2014 study, Singer and her colleagues concluded that this form of compassion training might be “a new coping strategy to overcome empathic distress and strengthen resilience.”

Zaki speaks of motivational empathy rather than compassion, although they are essentially the same thing. He has been exploring the extent to which intervention can enhance the desire to be empathetic. At Stanford, he joined forces with psychologist Carol Dweck, who is known for her work on how a person's mindset can affect performance. Dweck found that people with a fixed mindset about, say, intelligence believe they are powerless to change how well they perform, whereas those with growth mindsets—say, a “can-do” attitude—believe performance can be improved with effort. In a series of 2014 studies, Zaki, Dweck and Karina Schumann of Stanford discovered that similar mindsets exist for empathy. Participants who believed that effort could change one's level of empathy were more likely to try to take the perspective of someone from a social out-group than those who thought of empathy as a stable, unchanging trait. Future interventions, the researchers argued, should emphasize empathy's malleability.

In another series of studies, published in 2016, Zaki also showed that group norms can inspire people to be more helpful. In one of the studies, for example, participants had to choose how much out of $1 to donate to charity before learning whether others had been generous or stingy. Initially the average donation was nine cents, but those who then viewed generous behavior significantly increased their giving, ultimately giving nearly twice as much as those who observed stingy behavior.

Psychologist Jason Okonofua of the University of California, Berkeley, is applying these kinds of findings to schools. In a 2016 study, he examined and sought to change teachers’ mindsets about discipline. He and his colleagues first randomly assigned teachers to read one of two brief articles: One reminded them of the importance of good teacher-student relationships in helping students learn self-control. The other stated that punishment was critical for teachers to take control of the classroom. When teachers were subsequently presented with examples of disciplinary incidents and asked how they would handle the situation, their responses were less punitive if they had read the empathetic mindset article. A second experiment instructed college students to imagine themselves as middle schoolers in trouble with a teacher for repeatedly disrupting class by walking to the trash can. The participants were asked how levels of respect for the teacher were affected by whether a teacher responded by assigning detention (punitive) or by asking questions and moving the trash can closer to the student's desk (empathetic). As predicted, the students reported more respect for empathetic teachers.

Finally, Okonofua set up a randomized trial to test whether a brief online module that encouraged empathetic discipline would make a difference across an academic year. Math teachers at five diverse middle schools in three districts across California participated, and students whose teachers received the empathetic intervention rather than a control one were half as likely to be suspended. Okonofua is now expanding the study to 20 schools.

It is important to note that Okonofua stipulates what his intervention does not do: It does not require the teachers to share the students’ view of the situation. Rather it emphasizes understanding and valuing the students’ perspective. The goal, as he and his colleagues wrote in the 2016 paper, is discipline administered “in a context of mutual understanding and trust.” Is that still empathy? Okonofua thinks so. It is empathy writ large. And it is helping.