||The limits of human performance in computer use are determined jointly by the properties of the user interface (UI) and the human perceptual, motor, and cognitive systems. Recent technological advances have vastly expanded the means for constructing UIs, but we still see limited progress in overcoming the traditional interfaces in user performance. The group's mission is to identify the optima of interactive performance. The scientific approach is based on 1) information theoretical measurement of skilled motor performance to identify candidates for highest throughput, 2) formal analysis of UI design spaces, 3) predictive modeling of user performance, and 4) computational search for UIs that maximize user performance. Whereas previous work in human-computer interaction (HCI) has been largely based on trial and error, this approach allows aggressive exploration of user interfaces. The outcomes are demonstrated as novel user interfaces targeted to two domains: 1) classic interactive tasks, such as target acquisition, text entry, information retrieval, and visual search, and 2) "embodied" tasks where the environment mediates interaction, such as in mixed reality applications.