Affective Computing & Intelligent Interactive Agents

Research Leads: Jonathan Gratch, Gale Lucas, Mohammad Soleymani

The Affective Computing Group advances research into the role of emotion in human and human-machine interaction. This includes automatic analysis and understanding of multimodal affective signals, computational models of how emotions arise from and shape social cognition, and the application of these methods to human-machine teaming, interpersonal skills training and advancing theories of human emotion.

Broadly, affective computing means trying to get computers to reason about emotions, recognize emotion in people, and understand how emotion shapes our decisions, in both adaptive and maladaptive ways, and possibly use techniques to shape people’s emotions. Why do we want to give machines the ability to reason about emotions? The real answer is that machines are increasingly interacting with us, and there to help us in our everyday lives. It’s important for machines to understand emotions, perhaps to help people regulate their emotional state. For example, people with autism, who have difficulty in perceiving emotional cues, might benefit from software which could augment those social deficits in their ability to process emotional information in real time. Or, you could set your playlist to whatever mood or emotion you want to feel today, and have your music playlist automatically adjust and play songs to regulate for that state. There’s a whole host of reasons why emotion is important to how we feel our well-being and the sense that machines can reason about that can help us achieve our goals. Our team engages in cutting edge interdisciplinary research around these topics. This includes published research across a wide range of fields including computer science, human-computer and human-robot interaction, education on this topic with application to training and organizational behavior.

– Jonathan Gratch, Director, Affective Computing, USC Institute for Creative Technologies