Efficiently learning the costs and benefits of different behaviors is necessary for making informed choices and critical to the success of adaptive systems, both natural and artificial. Our group is interested in understanding this learning and decision making process in humans, animals, and groups using theoretical models to guide behavioral and neural experiments. Our primary research focus is on how choices are made in mammalian brains using reinforcement learning as a theoretical framework. Reinforcement learning is a general theoretical framework that describes how an animal or artificial system should (or could) solve the problem of choosing the ‘best’ behavior in any given situation. By comparing behavior and neural activity to the predictions of specific reinforcement learning models, we hope to both improve the models and better understand the computations of the brain. Our group also uses neuroscience to inform and develop other computational and theoretical approaches, like deep reinforcement learning neural networks and behavioral economic models. We are now exploring group learning and decision making, leveraging the computational experimental approach used to study individual behavior.
We believe in collaborative, cooperative science and work with many groups in Champalimaud Research as well as in other institutions worldwide.