Theodore Dalrymple has a short article in the TLS on the implications of neuro-determinism. It is often supposed that if your neurons made you do it, then you shouldn't be held fully responsible for a crime. Some take this (or related deterministic views) to reduce the need for punishment of, or interference with, the perpetrator. For instance, an Italian court reduced a murderer's sentence by one year because the culprit "has genes linked to violent behaviour." Dalrymple challenges this move. He points out that if science could offer reliable precitions about who is likely to commit crimes (or about who has these genes linked to violent behaviour), then the state may want to screen people and try to fix or quarantine or simply eliminate those with the menacing genes. In short, we get a Clockwork Orange view of how to handle wrongdoers.
This was brought home to me in class after I told the students about the above Italian case. Several students argued that the court should have increased the murderer's sentence since he was probably beyond rehabilitation and would only threaten others again in the future.
In the end, Dalrymple is pessimistic about the prospects of genetics to yield reliable information about whether someone is sufficiently likely to commit crimes. This topic would be interesting to study in an entire criminology or legal-studies related class. Here are Warren Brown and Nancey Murphy on related issues.
Update (Feb. 14, 2010): Here's a WSJ article from last fall about Jim Fallon, a neuroscientist who studies biological features that predispose one to a life of crime. He found that he has some of these traits. Also, there's this brief entry on a related topic at the LA Times.
Update (March 2, 2010): Still more on this topic, this time concerning a Chicago court case.
Update (March 25, 2010): More about the Chicago case.