The behavioral science people are actually starting to answer the ancient question raised by Aristotle, is viewed (stage) violence cathartic or stimulative? That is, does viewing stage violence (as in plays, video games, movies) cathartic (relieving inner tension to be violent (lowering the probability of actually being violent)) or is it stimulative (increasing the probability of being violent). As a statistician I have to tell you that (at least in clinical studies) the issue of causal vs correlation is very well understood to be extremely difficult to tease out of data. But I have read studies that indicate that viewing violence reduces the thresholds that hold us back by making the behavior seem more prevalent and therefore less wrong. Myself, I think error on the side of caution is wise, a position that puts me at odds with my otherwise science-loving causitive-denier libertarian friends, who, to a person, will argue that THEY aren't affected (gotta love those sample sizes of n=1). This sort of "I'm exceptional" is pretty well understood, and seems to be a factor in the poor risky decision making processes of most males through at least 25yrs life experience.
Maybe if we used electrical shock to punish people who make poor choices in video games we could train them out of it, oh, wait, that's the science in "Terminal Man".
Lots of science says "don't let kids play video games" and lots of kids deny the effect. Which do you trust?