If you're going to fall, injure yourself and need help, where is a good place to do it? Should you choose a busy thoroughfare or a deserted backstreet?
Statistics and experiments in social psychology will tell you that if you need help, you should avoid dropping in a busy street, even if hundreds of people are passing through.
This is because of a phenomenon known as the bystander effect. The more individuals gathered in one place, the less chance there is of one of them coming to the aid of a person in need. When an emergency situation occurs, it's more likely that someone will come to the rescue if there are fewer or almost no witnesses.
And in the 21st century, when our thoroughfares are online and on social networks, millions of people are effectively passing each other by every minute.
The presence of other people has long been shown to give rise to confusion about responsibility. Since there are other people around, each individual is less pressed to take action, given that the responsibility for taking action is assumed to be evenly distributed amongst those present.
One may be tempted to think that when a single person realises that the others aren't acting on their responsibility, then the entire burden of responsibility falls to them. But if a situation is ambiguous and it's not clear whether help is needed or not, observers may take the inaction of other people as a sign that there is no reason to intervene.
Updating the research
The bystander effect has been well researched over the past 50 years but most experiments have taken place in controlled situations and focused on small populations.
My group at the University of Copenhagen has run a number of experiments in the centre of the city to kick off a new approach to the problem.
The experiments involved getting three actors to take a tumble on one of the main thoroughfares in central Copenhagen – a young woman, a young man and an elderly man. Each actor had two different outfits – a business suit and a shabby-looking costume. It turns out that the costumes alone elicited strong public signals.
If the young woman or man or the elderly gentlemen were wearing business suits and fell, the time it took for someone to intervene could be as short as a few seconds. For the actors wearing shabby outfits, it could take up to around four minutes for anyone to intervene, even though plenty of people witnessed the accident.
More surprising is that the presence of certain people transmitted such a powerful public signal that entire groups would refrain from intervening even though they all witnessed the accident. When a tour guide passed by without acting, for example, none of his group did either. He sent a strong public signal that there was no need to intervene and his group followed like lemmings.
What's more, objects can be used to transmit public signals that deter people from intervening. If a young man falls holding a can of beer, a clear public signal for others to walk on by might be sent because of the social stigma attached to humans with beer cans. If the beer can is removed, help is offered much more quickly.
The information we gathered will now be fed into a computer simulation to test out whether the bystander phenomenon is robust over, say, thousands of runs.
We have isolated simple features like awareness vicinity, attention time, the passer-by's line of vision and other characteristics enough to generate the bystander effect in virtual environments.
An accident appears in the virtual environment, the passers-by have a "radar" screen of awareness, an angle of vision and devote a certain amount of time to the incident. Whether they choose to observe, intervene or walk on turns out to be acutely sensitive not only to the way in which the parameters are set but also to whether or not others are in the immediate vicinity.
The 21st century bystander
The simulation is a work in progress but how the phenomenon plays out in this context could have interesting implications for how we look at our online behaviour. The bystander effect occurs because people observe each other before assisting. And the more people observing each other the stronger the signal that help is neither required nor appropriate. Once you take such public signals to social media in terms of, say, aggregated likes, you may just reinforce bystander behaviour even more.
In the case of cyberbullying, by innocently "liking" you may be part of movement to establish a strong public signal about what the correct collective response is. You register your "like" with no obligation to actually intervene and non-intervention may just become the norm. Thousands and even millions of people witness terrible human behaviour online every day but how many of us actually do anything about it?
We may have a serious problem about the meaning and power of public signal. These could severely influence rational deliberation, decision and action, both online and offline.
Explore further: Mapping the world's linguistic diversity—scientists discover links between your genes and the language you speak