The Bystander Effect: an Information Cascade
It seems that it is a basic instinct that you should help someone if they are lying injured on the ground. However, this does not always occur, especially when there are many people around.
If you were in an isolated area and saw someone injured, you would go help them, because that is your basic instinct. Being by yourself, you cannot make your decision based on what others are doing or have done.
Put other passersby into the picture and your reasoning changes. If you see that no one has helped the person, then you can assume that the other people have some information that you do not have, that says that any action is unnecessary. Since no one else has reacted, it must not be an emergency. This is an example of an information cascade; you base your decision off of the actions of other people by assuming that their decisions convey information.
Also, if there were many people in the general vicinity, you would assume that at least one other person has either intervened or at least called 911. Thus, your action would be redundant. This is an example of the Bystander Effect or Genovese Theorem. As the number of witnesses increases, the likeliness that any person will intervene decreases, because there is a larger pool of people from which someone could have intervened.
The example given in this article is of a Chinese toddler Wang Yue who was lying injured on the ground. A total of 18 people walked passed the girl and no one offered any help. When one of the people was asked why they did not intervene, she responded that she would have offered help if someone else was.
When someone witnesses something in which they should get involved, they usually do not do so. This is because they do not believe it to be an emergency, because no one else has acted, and they also do not want to be the first person involved.
http://www.cbc.ca/news/canada/story/2011/10/22/f-bystander-effect.html
http://en.wikipedia.org/wiki/Bystander_effect
-DTM