The helping experiment was a social-psychological study in which an apparently distressed participant requested assistance while other participants listened. Its decisive finding was that the presence of potential helpers did not increase intervention: it diluted each individual’s sense of personal responsibility. The experiment therefore turned a morally uncomfortable outcome into evidence about a causal feature of the situation rather than a simple verdict on the participants’ character.

Experimental Setting and Participants

The experiment was conducted by Richard Nisbett and his student Eugene Borgida at the University of Michigan, using an earlier helping experiment carried out at New York University as its model. Participants were placed in individual booths and communicated over an intercom about their personal lives and problems. Each person was supposed to speak for roughly two minutes in turn, and only one microphone was active at a time. Six people belonged to each group, but one was a stooge following a script prepared by the experimenters.

The stooge spoke first. He described the difficulty of adjusting to New York and, with evident embarrassment, disclosed that he was prone to seizures, particularly when stressed. The disclosure established a plausible vulnerability before the emergency began. Every participant then took a turn, creating a shared situation in which all of them had heard the same account and knew that several other people were also listening.

The Apparent Emergency

When the microphone returned to the stooge, he became agitated and incoherent. He said that he felt a seizure coming on and asked whether someone could help him. His final transmission sounded increasingly desperate and then stopped, leaving the other participants with no confirmation that he was safe. Because each person occupied a separate booth, none could see what the others were doing; each had to decide whether to leave and respond.

The scene placed two interpretations against each other. On one level, a person seemed to be experiencing a medical crisis and had directly requested help. On another, every listener knew that several other people had heard the same request. That shared knowledge transformed the emergency from an apparently private obligation into a collective one, allowing each participant to assume that someone else might respond.

Results and Diffusion of Responsibility

The response was strikingly limited. Of fifteen participants, only four reacted immediately. Six never left their booths, while five others emerged only after a substantial delay, when the supposed seizure victim apparently had already choked and fallen silent. The figures showed that hearing an urgent appeal did not reliably produce immediate intervention when responsibility was distributed across a group.

The experiment’s interpretation was not that the participants were uniquely callous. Normal, decent people may hesitate when they expect others to take on the unpleasantness and uncertainty of dealing with a seizure. The presence of witnesses reduced personal responsibility more than people commonly anticipate, producing inaction without requiring an unusual personality trait or an explicit decision to refuse assistance.

What the Experiment Reveals About Judgment

The study challenged the everyday belief that decent people, including oneself, would automatically rush to help in such circumstances. People often imagine their likely behavior in isolation: if they alone encountered a seizure victim, they believe they would respond. The helping experiment showed that the social structure of the situation can alter that response. The same person may behave differently when several others have an equal opportunity to act.

Nisbett and Borgida used the finding to teach a broader lesson about causal base rates. A surprisingly low rate of helping should not merely be treated as evidence that the participants lacked compassion. It can instead reveal a potent situational feature: diffusion of responsibility. The logic parallels the inference that an unusually high failure rate on an examination may indicate an exceptionally difficult test. In both cases, an aggregate pattern points toward a cause embedded in the circumstances.

Wider Significance

The experiment also illustrates the limits of learning psychological principles abstractly. Readers and students may sincerely predict that they would intervene immediately, yet their prediction can ignore the effect of other available helpers. Even when people are told the procedure and results in advance, they may initially resist applying the lesson to themselves. The study’s uncomfortable force lies in making the observer part of the population being described: the possibility of inaction is not confined to conspicuously selfish people but may extend to ordinary individuals in a shared emergency.

In the book’s larger argument, the helping experiment is an example of how behavior that looks like a stable personal trait may actually be shaped by context. The apparent victim’s seizure is the dramatic surface event, but the deeper subject is the invisible redistribution of obligation among witnesses. What seems to be a failure of individual decency becomes intelligible as a predictable consequence of a social arrangement in which everyone can wait for someone else to move.