From an external perspective, cognitive agent behavior can be described by specifying (temporal) correlations of a certain complexity between stimuli (input states) and (re)actions (output states) of the agent. From an internal perspective the agent's dynamics can be characterized by direct (causal) temporal relations between internal and mental states of the agent. The latter type of specifications can be represented in a relatively simple, executable format, which enables different types of analysis of the agent's behavior. In particular, simulations of the agent's behavior under different (environmental) circumstances can be explored. Furthermore, by applying verification techniques, automated analysis of the consequences of the agent's behavior can be carried out. To enable such types of analysis when only given an external behavioral specification, this has to be transformed first into some type of executable format. An automated procedure for such a transformation is proposed in this paper. The application of the transformation procedure is demonstrated for a number of cases, showing examples of the types of analysis as mentioned for different forms of behavior. © 2007 Marta Olivetti Belardinelli and Springer-Verlag.