Ambient Intelligence (AmI) is an ubiquitous computing paradigm whose main objective is to provide living environments with personalized services for improving quality of people life. In spite of their technological definition, AmI systems are more than a straightforward integration among computer equipments. Indeed, AmI frameworks design strongly depends upon psychology and social sciences whose exploitation allows artificial systems to analyze the human being status and define the most suitable services collection for achieving users' satisfaction. Our work meets AmI objectives by using a formal method for representing human moods and introducing the so-called cognitive agents whose inference capabilities enable the emotional services distribution for enhancing users' comfort and simplifying the human/systems interactions. The agents reasoning engine is based on a novel extension of Fuzzy Cognitive Maps (FCMs) benefiting on the theory of Timed Automata: Timed Automata based FCMs. As will be shown in experimental results, our proposal improves the system's usability in terms of efficiency, accuracy and emotional response.