Why we are less likely to help someone when others are around
In an emergency, we often look to others to decide how to react. If everyone else is doing nothing, we assume help isn't needed. This is the bystander effect. In a classic 1968 study, when a participant thought they were the sole witness to a medical emergency, 85% took action. But when they believed four others were present, that number dropped to just 31%.
The Origin of Bystander Research
In March 1964, a young woman named Kitty Genovese was attacked and murdered near her apartment building in Queens, New York. Shortly after the crime, widespread news reports claimed that thirty-eight neighbors had witnessed the protracted assault from their windows, yet failed to intervene or contact the police. Although subsequent historical and journalistic reviews revealed that the original media accounts significantly exaggerated the number of eyewitnesses and the extent of their inaction, the narrative shocked the public and captured the attention of social psychologists.
Researchers Bibb Latané and John Darley questioned the popular explanation that the neighbors suffered from modern urban apathy or moral decay. Instead, they hypothesized that the sheer presence of other people altered individual decision-making. They proposed that individuals are not necessarily indifferent to suffering, but rather that the social environment introduces subtle cognitive and behavioral pressures that inhibit intervention when multiple observers are present.
The Core Psychological Mechanisms
Latané and Darley identified three primary psychological mechanisms that drive the bystander effect: diffusion of responsibility, pluralistic ignorance, and evaluation apprehension. Diffusion of responsibility occurs when an individual assumes that someone else will take action, call for help, or handle the crisis. When an observer is alone, the full moral and psychological burden of responding rests entirely on their shoulders. When multiple observers are present, that burden is distributed among the crowd, lowering each person's felt obligation to act.
Pluralistic ignorance arises in situations where an event is ambiguous. When people are unsure whether a situation constitutes a genuine emergency, they look to the reactions of those around them for cues. Because most people strive to maintain public composure and avoid appearing panicked, everyone observes everyone else acting calm. Consequently, observers mutually conclude that because no one is reacting with alarm, the situation must not be an emergency after all.
Evaluation apprehension, sometimes referred to as audience inhibition, involves the fear of social embarrassment or negative judgment. A potential helper may worry about misinterpreting the situation, overreacting, interfering where assistance is not wanted, or botching the intervention in front of peers. This anxiety about being judged by other onlookers creates a powerful deterrent to taking public initiative.
Classic Laboratory Demonstrations
To test their theories under controlled conditions, Darley and Latané conducted a series of landmark laboratory experiments in the late 1960s. In one study, college students were placed in individual cubicles and asked to discuss personal problems over an intercom system to preserve anonymity. During the discussion, one participant—an actor—staged a severe, audible medical seizure, pleading for help before falling silent.
Participants who believed they were the only person listening reacted swiftly and decisively, with 85% leaving their cubicle to seek help. When participants believed one other person was listening, the response rate fell to 62%. When they believed four other people were listening, only 31% took action to alert the experimenter. The presence of passive, unseen peers dramatically suppressed the likelihood and speed of intervention.
In another classic experiment, known as the smoke-filled room study, participants completed a questionnaire in a room that gradually began filling with non-toxic smoke through a wall vent. When participants were alone, 75% reported the smoke to the experimenters within minutes. When three naive participants were placed in the room together, only 38% reported the smoke. When two confederates instructed to ignore the smoke sat with a single naive participant, only 10% took action, demonstrating how social cues directly override sensory evidence.
The Five-Step Model of Helping
Building on their experimental findings, Latané and Darley outlined a sequential, five-stage cognitive model that an individual must navigate before intervening in an emergency. A failure or hesitation at any single stage prevents the bystander from providing assistance.
First, the bystander must notice that an unusual event is occurring, which can be impeded by distraction or sensory overload. Second, they must interpret the event as an emergency, a step frequently derailed by pluralistic ignorance. Third, the bystander must accept personal responsibility to act, which is vulnerable to diffusion of responsibility. Fourth, they must know how to help, possessing the relevant skills or resources. Finally, they must decide to implement the action, overcoming evaluation apprehension, legal fears, or physical risks.
Conditions That Weaken the Effect
Decades of follow-up research have revealed that the bystander effect is not absolute and varies considerably based on situational characteristics. When an emergency is unambiguous, severe, or physically dangerous—such as an ongoing assault or an active fire—the bystander effect is significantly reduced. In dangerous situations, cues of distress are clear, eliminating pluralistic ignorance and accelerating action.
Social relationships among bystanders also alter the outcome. When bystanders know one another, belong to a cohesive group, or share a salient social identity, they are far more likely to intervene collectively than when surrounded by total strangers. Similarly, bystanders who possess specialized skills—such as off-duty medical personnel, police officers, or trained lifeguards—routinely bypass diffusion of responsibility because they recognize their unique competence to manage the crisis.
Surveillance Evidence and Real-World Nuance
In recent years, researchers have examined the bystander effect outside laboratory settings using real-world closed-circuit television (CCTV) footage of public conflicts and street violence. Cross-national analyses of surveillance video in cities across different countries have revealed that in the vast majority of real-world public conflicts—often exceeding 90% of cases—at least one bystander intervenes to de-escalate, separate combatants, or comfort a victim.
This empirical finding clarifies a crucial statistical nuance. While an individual bystander's probability of intervening decreases as crowd size grows, the collective probability that at least one person in a large crowd will intervene often increases. Having more people present creates more opportunities for someone possessing low inhibition, high empathy, or relevant training to step forward.
Key takeaways
•The bystander effect is driven by diffusion of responsibility, pluralistic ignorance, and fear of social evaluation.
•Classic experiments showed that participants were far less likely to report an emergency or help a seizing peer when they believed others were present.
•The bystander effect weakens substantially in dangerous, unambiguous emergencies and among groups of acquaintances or trained professionals.
•Real-world surveillance studies show that while individual likelihood of helping drops, large crowds almost always produce at least one person who intervenes.