[Consolation]

It’s for an undergrad’s first independent research project so we don’t mind if they try something a bit flakey. Not my choice but…
However, there has been some good research done recently, you just have to be a lot more subtle with your measures. For example, threat images can reliably show increased cortisol levels in saliva etc… This is for social cognition so we are talking about small but statistically significant changes in social judgment tasks. So no, we can’t get subjects to kill their flatmates, but we can get them to say… rate people as being more likely to be criminals etc.