Common approaches to using artificial intelligence techniques in legal reasoning have generally been based on the logical reasoning methods developed in AI. In most cases, such systems can be considered to be expert systems applied in the legal domain. Other aspects of AI technology that rely more heavily on human psychology or behavioural patterns have rarely been used. This paper aims to show that some of these techniques have a rightful place in legal reasoning, and in particular that the notion of modelling the attitudes of agents can be of great benefit in developing more realistic systems for re-enactment in investigation and proof.