A woman holds a stuffed rabbit toy after it was found at her destroyed home where she said she had lost her three children, in Palu, Central Sulawesi, Indonesia.Jorge Silva / Reuters

When a 7.5-magnitude earthquake rocked the sea floor off the coast of Indonesia last week, the resulting tsunami devastated much of the city of Palu. The confirmed death count has soared to more than 1,700 and will almost certainly continue to rise. As of this writing, an estimated 70,000 people are displaced, with dwindling water supplies, in desperate need of help that might not arrive in time.

These numbers might sadden or alarm you; they might also leave you strangely unmoved. You wouldn’t be alone. For decades, social scientists have documented a troubling quirk in human empathy: People tend to care more about the suffering of single individuals, and less about the pain of many people. Such “compassion collapse” is morally backwards—dozens or hundreds of people, by definition, can lose more, fear more, and hurt more than any one of us; human concern should scale with the amount of pain in front of us. Instead, it dries up.

Compassion collapse may seem like just a (lack of) feeling, but its consequences extend further. Most important, it affects how and when people choose to help one another. In 2015, a 3-year-old Syrian refugee named Alan Kurdi, along with his brother and mother, drowned as his family tried to cross from Turkey to Greece along a narrow strait in the Mediterranean Sea. Images of his small body on the shore spread around the world. The tragedy, and his father’s anguish, moved millions of viewers, and donations to refugee-aid organizations poured in. Within days, and for a variety of other reasons, Angela Merkel made the fateful decision to open German borders to refugees. But within weeks, most people moved on, and the money stopped. Anti-migration politicians gained popularity across Europe; borders tightened again.

Millions of refugees, tens of thousands of children among them, continued to suffer. But their numbers, unlike a single vivid tragedy, left the world cold. Biases like this pop up in laboratory experiments as well. Across a number of studies, people donate more money to charity after learning about one person, as compared to many people, in need.

Compassion collapse is a dramatic psychological problem, but where does it come from? Researchers offer two competing answers to that question. Some suggest that people simply can’t care about others at a level warranted by a natural or man-made disaster. To some degree, this is built into the way humans feel emotions. The first $100 a restaurant makes is worth as much as the 100th $100, but it feels better to the owner. As good things pile up, the goodness of each individual thing is diluted. The same goes for suffering: As it compounds, people’s minds habituate, and the weight of additional pain wears off. In a way, it must. Think of the concern you’d feel for a close friend crying in front of you. Now imagine multiplying that feeling by two, or 10, or 5,000. The emotional load would quickly overpower anyone. The “can’t” camp also points out that human empathy has been built, over thousands of generations, to respond to certain triggers—for instance, a child’s cry or an anguished face. A single victim produces these signs of distress, which tug at us and inspire our help. Groups give us statistics, which land flat, triggering little and thus benefiting less from others’ compassion.   

A second group of psychologists argues that compassion—and, by extension, its collapse—is a choice. Sure, our feelings are more easily triggered by some cues than others, but we also have world-beating imaginations. A novel does not scream or cry; it silently displays the stories of people who don’t actually exist. And yet it can bowl us over with emotion. Likewise, we can deeply imagine multiple victims of a tragedy, letting the weight of their pain wash over us. But more often than not, we won’t.

The downside of compassion is clear. Given the volume of suffering in the world, empathizing with many victims could burn us out or leave us with constant guilt for not doing more. The upside is less obvious. We can make a real difference for one suffering person, but barely make a dent in the lives of thousands. If empathy hurts us more than it helps anyone else, why bother? Scientists in the “won’t” camp hold that people faced with mass suffering willfully turn down their compassion like the volume knob on a stereo. Consistent with this, people who can effectively control their feelings experience more compassion collapse than those who can’t.

Is compassion collapse a “can’t” or “won’t” problem? As with most debates like this, the answer is both. People do empathize more naturally with one person’s visible, heart-wrenching sorrow than with descriptions of massive tragedies, and human emotion does have a limited range. But even when people could extend their care toward a suffering group, they often shy away.

Understanding this is not merely an academic exercise; it can help us fight compassion collapse—because each type of collapse calls for different solutions. Imagine that someone is having difficulty connecting with a large number of sufferers—a “can’t” problem. Evidence suggests that focusing on one of the sufferers can jump-start empathy for the entire group, giving them a vivid case on which to hang their care. My collaborators and I have used virtual reality to do just that. In a recent study, we found that individuals who virtually experienced the plight of one homeless person reported more concern for homeless people overall, and more support for affordable-housing policy, even a month later.  

If someone worries that empathizing with many is useless—a “won’t” problem—merely giving them a way to connect might not work. Instead, they might need convincing that caring is worthwhile. “Empathic self-efficacy,” the idea that someone can truly help others without being overwhelmed, can drive generosity. Pointing people to the difference they can make might inspire them to dig into their empathy even amid great tragedy.

If people could empathize with mass suffering, philanthropy could become broader and more effective. But by no means would it fix all problems. The most generous charity after a disaster like Indonesia’s still must be delivered in order to help and, as in Palu, lack of good infrastructure can prevent that. Empathy generated by mass death can’t lower the toll. Charitable donations tend to be reactive, not proactive—it’s easier to care about the ongoing suffering of many than the potential suffering of future people that could still be prevented. In cases like these, aid and philanthropy should be driven by something else—for instance, objectively reasoned principles about which policies can make the biggest difference. But the fact remains that many of us will give only and most often to the causes that move us. We might as well move ourselves in the best direction we can.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.