A basic requirement for intelligent virtual agents is believability, which largely consists in demonstrating autonomous behavior, consistent with the agents' personality and world conditions. This requires a mechanism to dynamically generate agents' goals and behavior. When no external stimulus is present, the agent should still be able to function coherently, due to self-motivation, dependent on internal factors. The concept of motivation has already been investigated by research works on intelligent agents, trying to achieve autonomy. The current work presents an architecture and model for action selection and goal generation in intelligent virtual agents, using the concept of motivations. Based on biopsychological needs theories, we present a motivational approach to represent human needs and produce emergent behavior through motivation synthesis.