"Understanding how cognitive functions are implemented by biophysical processes in the brain is one of the most important and challenging goals of neuroscience. In the recent years, significant progress has been achieved by developing theoretical models for neural correlates of decision making, working memory, attention and representation of emotional value. These models were based on experimental evidence and they provided us with the basic building blocks for the study of more complex cognitive tasks. Many of such cognitive tasks can be described as context-dependent behavior: while the outcome of most cognitive operations is an association between a sensory stimulus and a motor reaction, this association is in general not unique, but depends on the cognitive context. Indeed the response to the stimulus can depend on the recent history of experiences, the goal, the motivation of the subject, the environment and other factors. The aim of this project is to develop a biologically realistic theoretical framework for context dependent behavior and, in particular, for the formation of the neural representations of the context. We intend to characterize the dynamical properties of the patterns of neural activities that encode the context, understand how they emerge from the temporal contiguity between task relevant events and determine the scaling properties of networks that implement context dependent behavior. The theoretical framework will be used to build a novel model for the formation of context-dependent emotional value of sensory stimuli. The predictions of this model will be confronted with the available experimental data, and will allow us to validate the model. The proposed theoretical studies should lead to the understanding of the general principles underlying the formation of new mental states representing the most general form of context, that includes emotional states, intentionality and motivation."