The Bystander Effect is about more than the diffusion of responsibility

Inspired by the shocking murder of a woman in New York in 1964, reportedly in front of numerous witnesses who did nothing to help (although this was exaggerated), the Bystander Effect is a well-researched phenomenon that describes the diminishing likelihood that any one person will help as the number of other people available to help increases.

The most popular and widely researched explanation is that people experience a diffusion of responsibility when in the company of other bystanders. We don't help the person who is being assaulted in a busy street because we assume that someone else will.

But a new study in the Journal of Experimental Psychology: General provides evidence that there is much more to it than this – in particular, when deciding whether to help, we take into account the perspective of the other bystanders – whether they know that help is needed, whether they know that we know that help is needed, whether they know that we know they know that help is needed, and so on.

Kyle Thomas at Harvard University and his colleagues, including the psychologist and best-selling author Steven Pinker, say their findings show that "…the bystander effect has a major strategic component, rather than being an irrational anomaly in our moral psychology, and that it is driven by mentalising … about the knowledge of other bystanders, rather than only being sensitive to their presence or absence".

The researchers recruited over 2,000 volunteers on Amazon's Mechanical Turk survey website to take part in an economic game designed to simulate and simplify a bystander situation. In groups of two or five people, the participants were asked to imagine that they rented stalls at a market, earning $1 a day. Occasionally the market manager needed help from one merchant to get supplies and whoever helped would suffer a loss of half their earnings for that day. If no one helped, the manager would fine all merchants $1 for the day. Players knew the scenarios were fictional but that the money at stake was real.

The researchers varied how the participants were informed about the manager's need for help, and what they were told about the knowledge of the other participants in their group. For example, in one condition, the participants were told that the need for help had been announced by loudspeaker, so everyone knew. In another, they alone were told by private messenger. In other variations, the private messenger told the participant and told the other merchants, but whether the other merchant(s) in their group knew that the participant had been told was varied, so too was the fact of whether the other merchants knew that the participant had been told that they knew. Thinking about what other people know about what we know they know, and so on, is the "recursive mentalising" referred to in the title of the study.

In the loudspeaker condition, in which everyone knew that help was needed (and everyone knew that everyone else knew), the classic "diffusion of responsibility" finding was replicated – players were less likely to help when in a group of 5 than a group of 2. But in the other conditions, the outcomes were more complex and depended on participants' judgments about how likely it was that other people would help based on what they knew. For example, when participants knew that help was needed but that other merchants were ignorant, they were more likely to help. By contrast, when other merchants knew help was needed, and the other merchants didn't know that the participant knew help was needed, then the participant was less likely to offer to help.

The use of economic games in psychology research has many advantages, making it easy to control conditions and measure outcomes, but of course the cost is in a loss of realism. The researchers acknowledge this point, but they argue that if the intuitive principles they've demonstrate carry over to real life as they predict, then there are practical implications for encouraging or discouraging helping. "For instance," they write, "people who email requests for help can strategically use 'cc' and 'bcc' options to create the recursive knowledge states that promote helping", or "Witnesses to an emergency can feign ignorance to encourage other bystanders to help instead, or they can make it clear that they know another bystander has also witnessed the emergency."

Further reading

—Recursive mentalising and common knowledge in the bystander effect