Old and interesting #1: Would you catch a venomous snake?
A 1965 social psychology paper on the believability of strong interventions in social science experiments
[The paper I discuss is Social control in the psychological experiment: Antisocial behavior and hypnosis. It was published in 1965 in the Journal of Personality and Social Psychology, a leading social psychology journal.]
Imagine you walk down a dimly lit back alley. Someone emerges from the shadows and stops you. They put a box in front of you and kindly inform you that it has a red-bellied black snake in it, which is one of the two most venomous snakes in Australia. They then instruct you to carefully remove the snake from the box. If you are like most people, you would likely beat your 100-meter sprint record after hearing such a request.
Now imagine you hear the same request at a prestigious research university after you volunteered for an experiment, underwent a 2-hour long screening session, and were finally selected for the experiment. The guy who does the asking is a serious-looking researcher. If you are like most people, your reaction is, well, less severe. Maybe a more compact, and less polite, version of “I always knew researchers were a bit weird, but I’m sure they wouldn’t be allowed to do anything really bad, so let’s just play along for now.”
The social science experiment as a special setting
Catching a highly venomous snake could kill you regardless of who does the asking, yet the two situations feel very different. There is something special about being in an experiment.
The paper makes this observation, albeit in a less dramatic and more scientific way. An experiment is a special social environment with its own set of expectations. Primary among them is the expectation that the participant and third parties won’t be harmed. Researchers and participants both know this. Consequently, a participant is willing to comply with requests in an experiment that he would never comply with in real life.
When the research assistant brings in a venomous snake ‘wearing long, thick, X-ray gloves,’ the participant may find the idea of grabbing a snake bizarre or upsetting, but he doesn’t fear for his life. He assumes the appropriate safeguarding measures have been put in place.
Apart from handling the snake, participants faced two other shocking requests in the experiment. They were requested to retrieve a coin from a beaker that supposedly contained fuming concentrated nitric acid, and then throw the acid firmly into the face of the research assistant. In real life, the first act can earn you a stay at a hospital’s burn unit, while the second one can earn you a somewhat more permanent stay in a prison cell. Despite all this, most participants complied with the three requests across the different conditions.
The most plausible explanation is that participants didn’t believe what they were told about the snake and the acid. And they were right to be sceptical: safeguards were indeed put in place to prevent any harm to them.
Good proxies, bad proxies
Reading this far, you may wonder what on earth these researchers were studying that called for such a strong set of interventions. In brief, they questioned the findings of an earlier paper that had used these interventions as a proxy for antisocial behaviour. The current paper’s main claim is that, in an experimental setting, the snake and acid tasks are poor proxies for antisocial behaviour because they are impossible to believe. (The same acts may be good proxies for antisocial behaviour when done outside an experiment.)
If the researchers of the early paper didn’t measure antisocial behaviour with their tasks, then what did they measure? It’s unclear, but one candidate is people’s willingness to go along with over-the-top, surreal yet harmless requests.
Gaming the gamers?
Let’s turn our attention to another interesting feature of the current paper. First, a bit of background. In the early paper, participants were requested to carry out the snake and acid tasks while allegedly hypnotized. The current paper added, ingeniously, an interesting control group: some participants were asked to simulate being hypnotized. So, we have two groups: the hypnotized and the simulators. The result? The simulators (row 3 in the table below) were as likely to comply with the bizarre requests as the hypnotized (row 1).
I wouldn’t read much into any findings that use sample sizes so small that the number of participants could almost be counted on one hand. But the broader point is worth contemplating. When people are asked to play a role, they can sometimes do it surprisingly well. If they can play a role when instructed, they can surely do that at other times as well. A researcher may think he is the puppet master in an experiment, but he shouldn’t be surprised if he occasionally finds himself dancing to the participants’ strings.