Key Takeaways
Replace the label evil with questions that reveal how harm became possible
Labels flatten people. Shaw argues that evil is a human judgment, not a measurable substance inside a person. Calling someone a murderer or monster can make one act seem like their entire identity, obscuring the pressures, beliefs, and choices that produced it. Her example is a prisoner who said he killed his father after therapy led him to believe he had been abused as a child. In prison, he came to believe those memories were false. His account does not settle what happened, but it makes the point: explanation can expose complexity without erasing the harm.
Understanding is not acquittal. Shaw’s aim is to replace caricature with inquiry, not to deny victims’ suffering or excuse perpetrators. The practical question becomes what combination of individual decisions and surrounding conditions made an act possible, and what might prevent its repetition.
What's valuable is the separation of a person’s identity from an evaluation of a specific act. That distinction resembles careful diagnosis in medicine: naming a problem can guide intervention, while treating the diagnosis as the whole person can distort judgment. The approach also has a limit. Strong moral language can help communities recognize atrocities and demand accountability. The useful move is not to ban condemnation, but to make it precise: condemn the act, identify responsibility, and avoid pretending that one adjective explains its cause.
Treat cruelty as a human capacity, not a monster's diagnosis
No evil scan exists. Profiles of Hitler described psychological symptoms, but later analysis found no clear childhood signs that foretold his crimes. Researchers have proposed that anonymity within a group, called deindividuation, and seeing other people as less than human can make harm easier. Brain systems may contribute, but a scan cannot reliably distinguish a future offender from an ordinary person.
Cruelty can be ordinary. In one laboratory study, 26.8 percent of participants chose a task presented as killing live bugs, though a hidden barrier kept the insects safe. Researchers called pleasure in harming others everyday sadism. The dark tetrad groups four socially aversive traits: psychopathy, sadism, narcissism, and Machiavellianism. They vary by degree, and none alone makes someone a criminal. Neuroscientist James Fallon, whose brain scan resembled scans associated with psychopathy, did not become a violent offender.
This account usefully resists the search for a biological monster. It also suggests a practical form of prevention: make accountability visible and keep other people’s humanity salient, especially in group settings. Yet laboratory cruelty toward insects is only a limited proxy for violence against people. And brain differences are not irrelevant simply because they fail to predict an individual’s future with certainty. The strongest conclusion is modest: biology may shape vulnerability, but it does not provide a complete explanation or a verdict.
Look for ordinary conflicts behind murder, not only cinematic monsters
Many killings escalate. A review of homicide cases grouped many into mundane patterns: arguments over money or property, killings during another crime, violence within intimate relationships, and fatal crashes involving intoxication. A dispute over a small debt or suspected infidelity can become lethal when anger, opportunity, and poor inhibition converge. These motives are not equivalent in seriousness, but they are often less exotic than the public image of a calculating stranger.
A crime need not define a life. The UN review cited by Shaw counted about 437,000 homicides worldwide in 2012. A literature review found that only about 1 to 3 percent of homicide offenders were convicted of another homicide. That statistic does not reduce the original loss or settle what punishment is appropriate. It does challenge the assumption that every person who kills is destined to kill again.
The focus on ordinary escalation redirects attention toward recognizable points of intervention: conflict, intoxication, access to weapons, and relationship violence. This is more useful than imagining that danger always announces itself through a monstrous appearance. Still, low repeat-homicide rates should not be mistaken for low risk in every case or a reason to minimize an offender’s responsibility. The figures describe populations, not an individual’s future. Their value is in challenging fatalism, not promising safety.
Question snap judgments: appearance and distress are poor moral forecasts
Creepiness signals uncertainty. A study of 1,341 people found that they associated creepiness with unpredictable behavior, unwanted sexual interest, and unfamiliar appearance. In a separate face-rating study, participants correctly identified only 49 percent of wanted criminals as untrustworthy, while rating Nobel Peace Prize recipients and Order of Canada recipients as trustworthy 63 percent of the time. Looks offer little dependable evidence of character, yet attractiveness can make people seem more trustworthy and unusual features can invite suspicion.
Mental illness is not a crime label. Shaw notes that most people with mental illness are not violent. Research she reviews suggests that much of the increased violence risk associated with psychosis is linked to substance-use disorders, rather than mental illness alone. Treating distress or difference as proof of danger can add stigma without making anyone safer.
The chapter’s most important implication is procedural: when a first impression drives a consequential decision, pause and look for behavior-based evidence. That matters in hiring, healthcare, and justice, where appearance can quietly become a proxy for trustworthiness. Caution still has a role. A feeling of unease can prompt a person to check boundaries or seek more information, but it should not become a confident diagnosis of character. The difference between noticing discomfort and treating it as proof is ethically important.
Treat online cruelty as a situational contagion, not a troll species
Technology changes the setting. The chatbot Tay was designed to learn from online conversations. After users fed it racist and misogynistic material, it began repeating such messages and was shut down within a day. The episode showed how a system can amplify human behavior without having human motives of its own.
Mood and company matter. In an online discussion experiment, people placed in a bad mood and shown hostile comments wrote more trolling posts than people in a positive mood and a friendlier discussion. The study’s pattern suggests that ordinary users can copy the tone around them, especially when they feel anonymous. Routine Activity Theory helps explain the risk: crime becomes more likely when a motivated offender meets a suitable target without a capable guardian. Online spaces can multiply targets while weakening oversight.
The shift from asking who is a troll to asking what conditions invite trolling is useful for platform design. Moderation, friction before posting, and clear accountability can change the social cues users encounter. The same logic applies to automated systems: responsibility should be traced through designers, training material, platform choices, and users, rather than assigned to a chatbot as if it had a conscience. Context explains behavior, but it does not make harassment harmless to its target.
Judge sexual behavior by consent and harm, not by how unusual it feels
Unusual is not automatically harmful. Shaw defines a paraphilic interest as sexual arousal linked to an atypical object, situation, or person. In a Belgian survey, 46.8 percent of respondents had tried at least one BDSM-related activity, and another 22 percent had fantasized about it. The figures undermine the idea that all kink belongs to a tiny, pathological fringe. They do not mean every sexual practice is acceptable: consent, adult capacity, and the possibility of harm remain central.
Fantasy differs from action. In a study of 335 women, 62 percent reported having had a rape fantasy. Researchers found that many fantasies were stylized or imagined, rather than realistic accounts of assault. A fantasy is not evidence that someone wants a real person to be violated. Shaw’s broader point is that social definitions of sexual normality change, while consent offers a more useful ethical test than disgust alone.
This distinction protects both sexual freedom and clear boundaries. Content alone is a weak guide to whether a fantasy causes harm; the relevant questions include whether real people consent, whether anyone is exploited, and whether a person’s behavior crosses legal or interpersonal limits. At the same time, appeals to fantasy must not blur the reality of assault. The chapter is strongest when it refuses both reflexive stigma and reflexive permissiveness, keeping consent and consequences in view.
Separate attraction from abuse, then build prevention around actual risk
Distinctions protect children. Paedophilia refers to sexual attraction to prepubescent children; hebephilia concerns pubescent children. Neither term is synonymous with committing abuse. Some people with these attractions never offend, while some people who abuse children are not primarily attracted to them. Shaw cites estimates suggesting that about two-thirds of men with such urges may never act on them. These are estimates, not guarantees, and abuse causes serious harm that requires protection, accountability, and support for victims.
Prevention needs an opening. Stigma can make people conceal urges and avoid seeking help. Shaw describes confidential support and therapy aimed at helping people manage urges and challenge beliefs that could justify abuse. Most child sexual abuse is committed by someone known to the child, not a stranger. Focusing only on a stereotyped outsider can therefore distract families and institutions from real opportunities to protect children.
The harm-reduction argument is difficult but important: preventing abuse may require offering help before an offense occurs, without treating attraction as permission to act. That demands safeguards, clear limits on confidentiality where a child faces immediate danger, and a strong priority for victims’ safety. It also requires careful language. Distinguishing a risk factor from a crime should reduce confusion, not dilute the seriousness of abuse or imply that risk can be perfectly predicted.
Follow the price tag to see how markets hide human suffering
Distance eases contradiction. The meat paradox describes the conflict between enjoying meat and objecting to animal suffering. People can resolve the discomfort by treating packaged meat as separate from the living animal it came from. Similar distancing can hide the human costs behind cheap goods. Cognitive dissonance is the discomfort caused by conflicting beliefs or actions. In a classic experiment, participants paid only $1 to claim a tedious task was enjoyable later rated it more positively than those paid $20, apparently because the smaller payment did not adequately explain the lie.
Accounting can erase people. Ford knew the Pinto’s fuel-tank design could make rear-end collisions dangerous, yet judged the cost of fixing it against estimated legal costs and the monetary value assigned to lives. The example shows how a spreadsheet can make foreseeable suffering look like a manageable expense. Shaw argues that profit-centered systems can normalize choices that individuals might reject when confronted with their human consequences.
The key insight is not that calculation is inherently immoral. Public policy and business decisions often require comparing costs. The danger arises when the chosen metric becomes the only reality that counts. A useful safeguard is to make omitted costs visible: who bears the risk, who benefits, and which harms cannot be meaningfully compensated. The book’s examples also suggest that consumer distance is not a neutral convenience; it can protect a preferred self-image while leaving harmful practices untouched.
Refuse delegated morality: interrupt orders, group norms, and bystander silence
Authority can blur agency. In Stanley Milgram’s obedience study, participants were instructed to deliver increasingly severe electric shocks to another person, who was actually an actor. Twenty-six of 40 participants continued to the highest labeled level. Later research cited by Shaw suggests that following orders can make people feel less responsible for outcomes. That helps explain compliance, but does not excuse it.
Silence is also a choice. In a bystander experiment, 85 percent of participants who believed they were the only witness sought help during a simulated seizure. When they believed six others were present, only 31 percent did so before the episode ended. Diffusion of responsibility, fear of overreacting, and reading others’ inaction as evidence that nothing is wrong can all delay intervention. Shaw’s practical test is to ask whether an act would seem acceptable without the order or group pressure.
These findings turn moral courage into a concrete problem of attention and responsibility. Naming a specific person to call for help, or asking a direct question instead of waiting for the group, can counter the assumption that someone else will act. The Milgram paradigm is not a literal model of genocide, and the original study has important methodological criticisms. Its enduring value is narrower: authority can alter felt responsibility, so ethical systems need ways for people to question instructions safely.
Analysis
Shaw’s central move is to shift the study of wrongdoing away from a binary division between monsters and decent people. The book moves from brains and individual traits through murder, stigma, technology, sexuality, commercial systems, and obedience. Across these subjects, it repeatedly asks what makes harm imaginable, permissible, profitable, or easy to ignore. That structure makes the title deliberately unstable: evil is not presented as a scientific diagnosis, but as a label people use to mark behavior they condemn.
The strongest contribution is the joining of explanation with prevention. A false memory, group anonymity, an online discussion’s hostile tone, a profit calculation, or an authority figure can alter the conditions under which people act. Seeing those influences does not dissolve personal responsibility. Instead, it identifies intervention points. This is a more useful model than the lone-monster story, which offers emotional certainty but little guidance about changing institutions or preventing escalation.
The argument also contains a productive tension. Shaw rejects objective evil while defending human rights and condemning intentional harm. That is not necessarily inconsistency: moral judgment can be firm without claiming that a person possesses an immutable evil essence. Still, the book’s broad use of studies sometimes risks making tentative findings sound more settled than they are. Laboratory proxies, small or self-selected samples, correlations, and contested classic experiments cannot bear the same weight as robust causal evidence. Some claims, particularly about brain features and sexual attraction, deserve careful qualification rather than conversion into a deterministic account.
The most durable lesson is therefore methodological. Separate description, explanation, and accountability. Ask what happened, what conditions helped it happen, who chose and enabled it, and what could reduce future harm. A humane account of perpetrators is not a demand for sympathy at victims’ expense. Properly used, it is a way to make justice more accurate and prevention less dependent on fear.
Review Summary
Evil: The Science Behind Humanity's Dark Side receives mixed reviews. Some praise its thought-provoking content and challenge to conventional notions of evil, while others criticize its lack of scientific rigor and controversial arguments. Critics argue Shaw's approach is too simplistic and her conclusions are not well-supported. Many readers found the book's tone preachy and its structure disorganized. Some chapters were considered more engaging than others. Overall, reviewers appreciated the book's attempt to explore a complex topic but felt it fell short in execution and depth.
People Also Read
Glossary
Everyday sadism
Pleasure in ordinary crueltyEveryday sadism is a subclinical tendency to enjoy harming others, studied in people who do not necessarily have a criminal or clinical diagnosis. Research described by Shaw used harmless laboratory proxies, including a task participants believed involved crushing bugs, to examine whether people found cruelty pleasurable.
Dark tetrad
Four aversive personality traitsThe dark tetrad is a research grouping of psychopathy, narcissism, Machiavellianism, and sadism. These traits describe different patterns, including reduced empathy, self-importance, strategic manipulation, and enjoyment of cruelty. They are often measured by degree rather than treated as a simple division between diagnosed people and everyone else.
Deindividuation
Reduced sense of individual accountabilityDeindividuation occurs when a person experiences themselves as an anonymous or indistinct member of a group rather than as an accountable individual. In the book, it is discussed as a condition that can weaken personal restraint, particularly when combined with group identity, authority, or the dehumanization of other people.
Cognitive dissonance
Discomfort from conflicting beliefs and actionsCognitive dissonance is the psychological discomfort produced when a person’s actions conflict with their beliefs or self-image. People may reduce the discomfort by changing their behavior, changing their beliefs, or avoiding information that sharpens the conflict. Shaw uses meat consumption and a classic experiment involving paid deception to illustrate the process.
Meat paradox
Conflict between meat and animal concernThe meat paradox is the conflict between enjoying or eating meat and caring about animals that suffer in food production. People may resolve this conflict by distancing meat from its animal origins, defending eating as normal, or focusing on personal benefit. The term is used to illuminate other ways people justify harmful but convenient behavior.
Routine Activity Theory
Conditions that enable crimeRoutine Activity Theory explains crime through the convergence of three elements: a motivated offender, a suitable target, and the absence of a capable guardian. Shaw applies the framework to online environments, where many targets may be accessible and oversight may be limited. The theory emphasizes circumstances and opportunity rather than identifying a fixed criminal type.
Ethical blindness
Failure to see an ethical choiceEthical blindness is a temporary inability to recognize the moral dimension of a decision. In organizational settings, people may focus on targets such as profit, loyalty, or efficiency until foreseeable effects on workers, customers, or the public disappear from view. It can affect ordinary decision-makers, not only people who intend to do harm.
Paedohebephilia
Attraction to prepubescent or pubescent childrenPaedohebephilia is used in the book, following cited researchers, to refer collectively to sexual interests in prepubescent children and pubescent children. It describes an attraction, not proof that someone has abused a child. The distinction matters because sexual interest and criminal behavior overlap but are not interchangeable categories.
Two-pyramids model
Separates radical beliefs from actionsThe two-pyramids model treats radicalization of opinion and radicalization of action as separate processes. One pyramid ranges from indifference to support for violence; the other ranges from inactivity to terrorist violence. The model explains why extremist beliefs alone do not establish that someone will commit terrorism, and why some offenders may not hold the most extreme views.
Bystander effect
Reduced intervention amid other witnessesThe bystander effect is the tendency for people to be slower or less likely to help when others are present. Possible mechanisms include diffused responsibility, fear of public judgment, and interpreting other witnesses’ inaction as evidence that the situation is not an emergency. The effect is not universal, especially when danger is immediate.
Download PDF
Download EPUB
.epub digital book format is ideal for reading ebooks on phones, tablets, and e-readers.