The Milgrim Experiment. A participant is in control of a machine that delivers an electric shock to another person in the next room, who is not seen but can be heard. There is a scientist in the room who is conducting the experiment who gives directions to the participant. The participant is instructed to deliver the shocks (just a press of a button) with increasing power. Eventually, the unseen person yells that it is too painful and wants to stop the experiment. The scientists tells the participant to continue regardless. In theory, only 1% should be willing to continue giving the shocks, but it was found that 65% continued and even completed the experiment. The experiment was designed to explore why average Nazi soldiers and citizens accepted Hitlers actions. While it is debated that this actually models Nazi Germany, it does raise interesting points about authority.
This is really interesting, although another huge reason that the citizens of nazi Germany obeyed such orders is the influence of the social acceptable, whose's huge reasons for existence are fear and pride, which were both hugely in that time. And when it comes to soldiers, they were simply "doing their job" of obeying any orders of authority without any reaction or even questioning. So a better way to use such an experiment would be to use more people at the same room, at the same time, some of which we would intensially put there to do as we say, just to do as we want, so we would also see different results influenced by the huge factor of social acceptance. So I see this thought as a personal motivation, to go soon and try this updated experiment on a different way obviously, as a way of research, and film it.
This reminds me of a different experiment where people were taking phone calls (sort of like a customer service or information type thing) when the person they were talking to had a heart attack or stroke. If it was just one person on the call, they would generally go gt help. If there were two or more people taking the call, they would wait to see what the other person would do, leading to deadlock and nothing happening. Unless one of them chose to act, in which case the others would also very quickly act themselves. Your mention of soldiers reminded me of Abu Ghraib, though people usually say that that is more relevant to the Stanford Prison Experiment.This is closer to what you were talking about: people will fill the social role they are expected to take.