Understanding statistical properties of cognitive systems is one of the main goal of complex systems physics. The automata gas is a statistical system whose particles perform information based interactions and use a decision mechanism (cognitive behavior). The collective nature of such interactions and the existence of a "free will" play a fundamental role for the existence of self-organized dynamical states, emergent properties and critical phenomena. In some cases it is possible to study the emergence of these states, by using an adiabatic separation between the single particle time scale and the particle distribution time scale. Then the emergent properties can be related to the solutions of a nonlinear diffusion equation. The considered models have a wide range of application in biology and social sciences to describe the self-organization properties observed in the experiments. Our main interest is to develop models for the pedestrian dynamics in a urban space, both to understand crowd dynamics phenomena and to plan a mobility governance during great touristic events. We have analyzed data from video movies during the Venetian Carnival by studying the individual microdynamics to point out the effects directly related to decision strategies or cognitive behaviors and their relation with the macroscopic properties.