Bystanders are more likely to rise to the challenge and come to someone's defence if the person appears to be at serious risk, a new study suggests.
The research, conducted at Munich's Ludwig-Maximilians University, looked at how onlookers respond in different situations.
Researchers found that the more bystanders there were, the less likely any would intervene, a phenomenon dubbed the "bystander effect."
The researchers also found, however, that - even when accompanied by another person - a bystander would enter the fray despite the risk if it appeared the victim in the incident being observed was in serious danger.
"The classical bystander effect was replicated when the situation involved low potential danger, but not when the situation involved potential danger," lead researcher Peter Fischer said.
"The good news is that when people are in real trouble, they have a good chance of receiving help, even when more than one bystander is present."
The findings were published latest edition of the European Journal of Social Psychology.
To test people's response, researchers recruited 54 women and 32 men and told them they were going to observe how two people, a man and a woman, who had never met communicated.
The two people were actually actors and within a few minutes the interaction became violent, with the male first verbally attacking and then grabbing the woman. Researchers were looking to see for how long it took someone to step in and try to stop the fight.
They also tried to vary the perceived threat of the situation by changing the sizes of the two actors at the centre of the scenario. In some of the experiments, a third person was also used as an observer. That person was told not to intervene.
According to the findings, about 50 per cent of observers tried to help the victim if they were watching alone. That percentage dropped to 6 per cent when another bystander was present.
When the threat to the victim was increased, 44 per cent of observers tried to help when they were alone, and that percentage did not drop off nearly as dramatically when another observer was introduced to the situation. In that scenario, 40 per cent of observers tried to help, even if someone else was also watching.
"Why do dangerous emergencies seemingly reduce the bystander effect?" the researchers asked in Monday's paper.
"We assume that dangerous emergencies are recognized as real emergencies more clearly and thus increase the costs for not helping."
As a result, they suggested, the observers' "empathetic arousal" increases, leading to more people helping regardless of whether they were accompanied by someone else or not.