Paul Bloom on Cruelty
What this covers
Paul Bloom, a Yale psychologist, joins Russ Roberts to examine what actually drives cruelty and why ordinary people inflict harm while feeling morally righteous. Rather than accepting the common view that perpetrators dehumanize their victims, Bloom argues the opposite: the worst atrocities—slavery, the Holocaust, misogynistic violence—typically depend on perpetrators recognizing their victims' full humanity and wanting them to suffer as people. The conversation moves from historical atrocities to the mechanics of everyday cruelty, focusing on how technology and group dynamics let individuals commit small acts of harm (a social media pile-on, a snide remark) while each person feels blameless or vindicated, producing cumulative damage through what Bloom frames using Derek Parfit's harmless torture thought experiment.
The discussion spans several distinct territories. Bloom argues that resentment and grievance—not dehumanization—better explain cruelty, since you cannot resent a non-human being. He examines how internet shaming mobs operate without proportionality or mercy, how contempt in relationships signals fatal loss of regard, and why people committing harm typically believe themselves to be righteous enforcers. The conversation also takes up whether abusing robots constitutes genuine cruelty, invoking the problem of other minds and Descartes alongside questions of sentience. Roberts raises concerns about case-by-case moral reasoning and the role of Adam Smith's insight that humans crave not just love but the status of being "lovely"—worthy of respect. A notable thread disputes whether violent video games cause real violence, using evidence that gaming's rise coincided with declining youth violence.
Bloom argues that cruelty is typically not driven by dehumanization but by the full recognition of others' humanity—their capacity to be resented, dominated, and made to suffer as people—and that this same recognition, amplified by technology and group dynamics, lets ordinary people inflict cumulative harm while feeling virtuous.
- Atrocities like the Holocaust and misogynistic violence reflect recognition of victims' humanity, not its denial
- Small accumulated acts (social media mobs, microaggressions) cause great harm while each individual feels blameless or righteous
- Treating sentient-seeming beings cruelly may be wrong in itself and morally corrosive to how we treat real people
Cruelty typically requires recognizing the victim's humanity and perceived wrongdoing, not denying it.
- When people commit the worst atrocities—the Holocaust, slavery, misogynistic violence—it is often not because they dehumanize victims but because they fully recognize their humanity and want them to suffer as people; the humiliations and degradations only make sense if the perpetrators saw their victims as morally tainted people whose humanity they hated.
“Rather, what you see is a full recognition of the humanity of other people. So, Appiah points out that in genocide in, say, in Germany in World War II there are all sorts of humiliations and degradations done toward the Jews, and it's hard to make sense of this if you say the Germans didn't think of the Jews as people. Rather, they thought of them as people and they wanted them to suffer as people.”
- The psychology of cruelty is often better captured by grievance than by dehumanization: you can only have a grievance against a person, not a dog or a rock, and cases like Elliot Rodger—who killed because he felt rejected and resentful, not because he saw victims as non-human—show that cruelty flows from resentment toward recognized people.
“He had a grievance. That's a lovely term, 'grievance.' You don't have a grievance against a dog, or a rock. You have a grievance against a person... And I think that sort of instance captures the psychology of cruelty, maybe better, than a picture, which sometimes is true, of sometimes someone gunning down a bunch of people because he thinks they are monkeys.”
- A unifying theme of cruelty is that those who act cruelly do not think of themselves as villains or even morally neutral—they think of themselves as good people striking at someone doing something wrong, which is not irrational because sometimes punishing wrongdoers genuinely is good.
“the people who are being cruel don't think of themselves as villains. They don't even think of themselves as morally neutral. They think of themselves as good people-- ... to strike at somebody who is doing something wrong is easily seen as a good thing.”
- There is a distinction between the architects of atrocities, who target victims' explicit humanity, and the people implementing cruelty on the ground (camp guards, distant military operators), who often genuinely dehumanize—coding victims as combatants or non-persons—simply to make the work psychologically easier rather than out of animus.
“I think a lot of the stuff on the ground, as it were, really is dehumanization. These are often the people who do the most terrible things have no wish, have no animus to harm other people, so they just tell themselves, 'These are not people,' and that just makes it go easier.”
- Research on couples found that what predicted breakups was not whether partners were quiet, noisy, or even whether they screamed and fought, but whether they showed contempt for each other—because contempt is a dehumanizing emotion that signals you no longer regard the other as worthy, and it is fatal to a relationship.
“they wanted to predict what aspects in interactions predicted that couples would break up. And it turned out, in this story, that it wasn't whether or not they were quiet or noisy. Not even whether they screamed at each and fought. It was whether they showed contempt for each other.”
- Slave owners worked hard to frame slavery as a morally good relationship, arguing that ownership produced better treatment than renting and casting their domination as a service to people they deemed unable to run their own lives.
“There's a whole literature on this written by people trying to justify themselves... 'Certainly, if you own something you treat it better. You take care of it. And that's why the relationship between a master and a slave is a moral one.' And, these people are working very hard to frame what they are doing as morally good.”
Internet shaming mobs inflict disproportionate harm through technology-enabled collective action without mercy or justice.
- Derek Parfit's 'harmless torture' thought experiment—where each person infinitesimally increases a victim's pain so no individual perceptibly harms anyone, yet collectively all are tortured into agony—plays out in social media, where a single Facebook 'like' on a cruel post is no big deal but a thousand likes leave the target crushed, an accumulation of tiny acts producing terrible consequences through technology.
“as a result of everybody throwing a switch, each individual ends up in terrible agony. So, as a sort of group, you have all tortured all these people. But as an individual, you haven't increased the misery of not a single person.”
- Internet shaming mobs typically punch down at obscure people caught in the worst 30 seconds of their lives, often act on false or incomplete information, and—unlike legal systems, bureaucracies, or families that maintain proportionality and mercy—exhibit no sense of mercy, justice, or proportion even when the target is genuinely guilty.
“in the cases that I see so often: First, they are always punching down. They are always some obscure schmo who has done something stupid. Who has been caught in the worst 30 seconds of their life... It's often based on false information.”
- Technology drives online cruelty because, unlike a face-to-face accusation where you understand the consequences, tweeting an insult anonymously and as part of a thousand-person group means no individual intends to destroy a life, and humans are ill-equipped to grasp the magnitude of such collective harm.
“if I tweet that at you-- ... anonymously--and then a thousand other people do it, I'm part of a group. My own intention was never to destroy your life, perhaps... But, we are not--we are ill equipped to think about the magnitude of such things.”
- Because death by a thousand cuts could be justified against a true monster, the danger is that when people are infused with moral outrage, act as part of a crowd, and operate in a virtual world with no fixed system of evaluation, law, or justice, all their enemies become Hitler—making Hitler a terrible paradigm for everyday moral judgments.
“the problem is that when we are infused with moral outrage, acting as part of a crowd and operating in a virtual world with no fixed system of evaluation, law or justice, all our enemies are Hitler.”
- People jump into shaming mobs because of the pleasure they get—partly the pleasure of punishment and partly the pleasure of affiliation, like the imagined fun of being part of a crowd throwing rocks in the name of justice.
“people jumping into the mob because of the pleasure they get. In part it's a pleasure of punishment; I think in part it's a pleasure of affiliation.”
Sentient beings warrant moral protection from suffering regardless of substrate or origin.
- Robots of the sophistication and complexity depicted in Westworld are probably sentient—it beggars belief that they don't have feelings—and since most cognitive scientists hold that being made of flesh rather than silicon doesn't obviously matter, our best bet when something begs for mercy, converses, and has goals is to conclude there is somebody home and that we shouldn't treat them badly.
“these robots are of such sophistication, complexity, it beggars belief that they don't have feelings... if you have in front of me has a machine who begs for mercy, has a conversation, has goals--seems to have goals and desires--I think our very best bet is to say there's somebody home. And if there's somebody home, I shouldn't treat them badly.”
- Perfectly duplicating a brain is one way to build a sentient robot, but it is not the only way; there may be many routes to constructing intelligence and consciousness, just as conscious beings could plausibly evolve on a very different planet through entirely different processes.
“one way to build a sentient robot is to perfectly duplicate a brain... But it's not the only way. I could imagine there could be many ways to construct intelligence and consciousness... there may be many routes to consciousness.”
- It is as hard to understand how a lump of flesh above the shoulders produces feelings as how silicon could, and since our theory of science cannot explain consciousness, thinkers like David Chalmers argue our scientific framework may itself be fundamentally wrong and in need of re-evaluation.
“I find it equally hard to imagine how a lump of flesh and meat above my shoulders can have feelings, either. Apparently it does.”
- The morally pressing issue with sentient beings—whether a dog, animals raised for food, or Westworld robots—is not so much killing as suffering: it is probably not wrong to painlessly put down a dog with cancer, but it is wrong to let it suffer in agony, and there is every indication the Westworld robots can suffer.
“I don't think the issue is so much life or death. And I feel this way also regarding the manufacture of animals for food. It's not so much the killing. It's the suffering... And there's every indication that those things in Westworld can suffer.”
- Following Descartes, we can never truly know that a robot is conscious—but the same problem of other minds applies to a real spouse: it is possible the only consciousness one can be certain of is one's own, so our judgment of a robot's sentience would in practice depend on the complexity and human-likeness of its responses.
“we can never know, at a certain level, just like you could never know that for your real spouse... It's possible that I'm the only conscious person around.”
People possess competing moral appetites including domination, hierarchy, and sadistic pleasure alongside altruism.
- As a developmental psychologist, Bloom answers whether babies are innately good or bad with 'yes'—people are basically complicated, possessing both good appetites and bad ones, including a desire to dominate and not be on the bottom.
“I'm a developmental psychologist, and sometimes I get the question: 'Are babies innately good or innately bad?' And I always answer 'Yes.' We have both appetites.”
- Because humans are hierarchical creatures who want to be on top, people derive genuine pleasure and satisfaction from dominating others; while respect and admiration are the preferred route to the top, terrifying someone and making them fall before you is a go-to strategy for some, and this domination drive is a grotesque exaggeration of an appetite that exists in everyone.
“I think what it is, is we are hierarchical creatures, and we want to be on top. And there's all sorts of ways of being on top. There's to be respected and admired. But, failing that, terrifying somebody and making him fall before you is a go-to some individuals use.”
- All of us, or nearly all, have a little sadist inside—a capacity to take pleasure in cruelty that is not confined to honest-to-God sadists or psychopaths.
“I think all of us have a little sadist in us. Or many of us do. ... I think I'd say all of us.”
Moral judgment requires navigating tension between detached principle and emotional attachment to individuals.
- Because we love our family more than we like our colleagues, their rejection or failure to take us seriously bothers us far more, so we are paradoxically more likely to react with rage toward those we love—indeed we are far more likely to kill family than strangers—since deep love opens up the possibility of hatred just as it opens the possibility of love.
“because I love my family, their rejection, their failure to take me seriously... it really bothered me. While, my colleagues listening to my plan... And, of course, this is reflected more seriously. In fact, I'm far more likely to kill my family.”
- Far from being too cold-blooded, real-world policy errs by taking individual cases too much into account—being moved by particular victims (a murder, a gun death, a sexual assault) distorts policy in irrational ways, so the utilitarian's detachment is a feature, not a bug.
“I think that often our very worst decisions are made because we very much take into account individual cases, and we are moved by them--say, the murder of somebody by an immigration, a gun death, a sexual assault--and our feelings about the particular individuals there which will vary from person to person will often distort policies in all sorts of crazy ways.”
- Drawing on the anthropologist Fiske, there are different levels on which we interact: a market level based on autonomy and mutual consent (you pay money, you get a product) and a friendship model (you don't pay a host for a home-cooked meal), and much of life requires navigating both at once, which is difficult and a frequent source of embarrassment, especially when there are large gaps in inequality, gender, race, or class.
“this anthropologist, Fisk[?], talks about these different levels on which we interact with each other. And there's a market level, where, you know, you pay your money, you get your product... But there is also a friendship model. You know: If I go to your house and you make me a nice meal, I don't give you money afterwards. That would be disrespectful.”
Individual and cultural moral attitudes genuinely improve despite unchanged evolutionary human nature.
- Research by Molly Crockett and colleagues supports the intuition that people like deontologists—those who follow moral rules and favor friends and family—and have little patience for utilitarians or even effective altruists, leaving the utilitarian with few friends.
“A colleague of mine, Molly Crockett, has just published a paper with several other people supporting the idea which I think you and I know intuitively that people like deontologists. They like people who have moral rules; they like people who favor their friends and their family; and they have little patience for utilitarians, even for effective altruists.”
- The evidence that violent first-person shooter video games cause real violence is tremendously weak; in fact, the period in which games became more realistic and violent coincided with a large drop in violence among that age group and increased compassion toward minorities, so any causal argument from correlation would absurdly imply violent games make people nicer.
“people like Steve Pinker correctly point out that the period in which the games have become more realistic and more violent is the same period in which violence in that age group has dropped a lot... if you are going to make a big deal about these correlations, you would have to be forced into the view that violent video games make us much nicer.”
- While human nature as a product of evolution has not changed, individual and cultural psychologies have genuinely shifted in a positive way—attitudes toward sexual minorities like gay and trans people are authentically different and more accepting than a generation ago, not merely a matter of new laws constraining unchanged feelings.
“the attitudes of the people in our community toward sexual minorities, like gay people or trans people, are genuinely different than they used to be when I was a kid... I think the attitudes and feelings are genuinely different. And in a positive way.”
- An irony built into Westworld is that, unlike the characters in the show who comfortably treat sentient humanoid robots in terrible ways, viewers very quickly take the side of the robots and become repelled by the human guests.
“One of the ironies of the show, which is built into the show, is you very quickly take the side of the robots-- ... and begin to be repelled by the people.”
- Westworld functions as a philosophical thought experiment more powerful than a verbal description because having the robots played by brilliant actors makes viewers feel the force of the wrongness of harming them, illustrating how movies and shows like Memento can themselves be genuine works of philosophy.
“Westworld is itself a thought experiment. Having the robots played by such brilliant actors makes you feel the force of the fact that you shouldn't harm them... a lot of movies and movie shows--the movie Memento being another example, are actually themselves works of philosophy.”
- Treating someone as not fully a person is not necessarily bad: a surgeon treats a patient as a problem to be solved, and a utilitarian who reasons in mathematical terms about saving more lives can make the world much better precisely by not dwelling on each individual as a person.
“there's a wonderful phrase by Atul Gawande: A surgeon treats his or her patient as a problem to be solved. I think a utilitarian philosopher goes around to make the world a better place but doesn't actually think of each person as a person.”
- The idea that ceasing to view something as human removes all moral rules is undercut by the fact that people are often harsher toward those who harm innocent animals than toward those who harm people, even though animals are not human.
“In fact, I think we are harsher toward those who harm innocent animals than those who harm people. And so, even on the outset, the idea that once you don't think of somebody as human, all the moral rules go away, has its difficulties.”
Violence and bad behavior require informal and formal social punishment to sustain cooperative goodness.
- From an evolutionary and societal standpoint, goodness can only survive if there is some mechanism to punish or shun those who are not good; otherwise being good is a losing strategy that falls by the wayside, which is why we want both formal enforcers (cops) and informal social correction.
“the only way we could be good people to each other, as a society and as a species, the only way goodness could survive, is if we had some mechanism in place to punish or shun people who are not good. Otherwise goodness is a crappy strategy and will just fall by the wayside.”
- Evidence from Tom Tyler shows that what matters most to people in the workplace is the feeling of being respected and treated as a serious individual of dignity, and that this respect is worth a substantial amount of money.
“Tom Tyler, a colleague at Yale Law School... has long has evidence that what matters to us most in our workplace is the feeling of being respected. And, you know, being respected, being treated as a serious individual, worthy, a creature of dignity, is worth a lot of money.”
Treating humanoid robots as mere objects risks psychologically corrupting how one treats actual people.
- Even if humanoid robots are known to be mere machines like toasters, treating things indistinguishable from people in terrible ways—raping, beating, or torturing them—may be morally corrosive, lowering one's inhibitions and bleeding over into how one treats real people, a plausible worry even without empirical evidence.
“even if they aren't--even if you know they are like toasters--it might be morally corrosive to treat things that are indistinguishable from people in terrible ways. It might bleed over into your interactions with people.”
- There may be a movement to design service robots to look distinctly non-human, because making them too humanoid would trigger our social responses and make it psychologically difficult to use them in the demeaning ways we want, paralleling our discomfort with those who serve us.
“there will be a movement to make us look different from humans. That some of them will be specifically built so you don't treat them as if they are humanoid, because the humanoid part of them might make it very difficult for you to use them in certain ways.”