Good fences make good neighbours. The “naturalistic fallacy” is a particularly solid fence. It not only claims that the normative disciplines need not worry about facts. It would actually be a category error if they care. Although at its core the law is a normative discipline, it has not always been that hostile towards facts. But it used to have a rather loose attitude towards empirics. Often, any factual statement “that matters” has been accepted as a legitimate element of legal reasoning. A social scientist would likely disagree. A factual claim is not the same as empirical evidence, most importantly since correlation is not causation. Criminology defines itself as the social science concerned with crime and punishment. This invites a division of labour: to the extent that a normative conclusion in the area of criminal law rests on an empirical claim, it is for criminology to deliver the evidence. Yet what counts as evidence? To a relevant degree, this is not only a question of epistemology. It also is a matter of disciplinary culture. Historically, criminology defines itself as applied sociology. This historical root is a productive one. But it has led the discipline to largely ignore an important source of evidence: experiments under the controlled conditions of the lab. Interestingly, in empirical legal studies, this method is much more established than in criminology, despite the fact that empirical legal studies are much more recent. The talk will explain why lab experiments could be helpful for criminology, and accordingly for the empirical grounding of criminal law, and will illustrate the power with experiments on crime and punishment.