The virtual bystander effect

Ever hear of the ‘bystander effect’? In essence, it’s the observation that the more people present to witness a violent incident, the less likely it is that any one of them will intervene to help the victim. Mel Slater of University College London and his colleagues used immersive virtual reality to study this phenomenon.

The authors hypothesized that the bystander effect would be minimized if the victim were a member of the same group as the witness. Since actually mugging people in order to study the reactions of bystanders would be unethical, the researchers resorted to virtual attacks. They recruited 40 ardent supporters of the Arsenal Football Club, a football (or soccer) club based in Holloway London, and immersed them in a life-sized virtual bar. Within that environment, each participant was approached by a virtual human (V for victim) who spent a few minutes talking to the subject about football. In some cases, V was also a huge Arsenal fan. Next, a second virtual person (P for perpetrator) started an argument with V that ended violently.

Four combinations of factors were recorded: whether V was a fellow Arsenal fan (in group versus out group) and whether V made eye contact with the volunteers during the attack. Ten participants were tested for each of these combinations. Any attempt by the volunteers to intervene verbally or physically was recorded. As expected, there were more intervention attempts on the part of the test subjects when V was part of the same in-group. Whether or not V looked to the subjects for help did not seem to make much difference.

I’m not sure this says all that much about human psychology. It’s telling that the number one thing the participants said would have gotten them more involved would have been if the virtual characters had been more reactive to them. In other words, it was probably clear that the little vignette was going to be played out regardless of what the test subjects did. Besides, the bystander effect is supposed to involve multiple witnesses not just one. That said, I still think this was an interesting experiment. With a little fine-tuning, virtual environments might very well turn out to be excellent ways to study human behavior.


Mel Slater, Aitor Rovira, Richard Southern, David Swapp, Jian J. Zhang, Claire Campbell, & Mark Levine (2012). Bystander Responses to a Violent Incident in an Immersive Virtual Environment PloS ONE : doi:10.1371/journal.pone.0052766.