When Emotional Machines are Intelligent Machines: The Tangled Knot of Affective Cognition

Research in neurobiology has provided evidence that emotions pervade human intelligence at many levels. However, “emotion” and “cognition” are still largely conceptualized as separate notions that “interact”, and untangling and modeling those interactions remains a challenge, both in biological and artificial systems. My research focuses on modeling in autonomous robots how “cognition”, “motivation” and “emotion” interact in what we could term embodied affective cognition, and particularly investigating how affect lies at the root of and drives how agents apprehend and interact with the world, making them “intelligent” in the sense of being able to adapt to their environments in flexible and beneficial ways. In this chapter, I discuss this issue as I illustrate how my embodied model of affect has been used in my group to ground a broad range of affective, cognitive and social skills such as adaptive action selection, different types of learning, development, and social interaction.