This dissertation will use neural imaging, eye-tracking, machine learning, and system development to elucidate the process of visual decision-making in environments that simulate important elements of a human's natural experience. This "naturalistic visual decision-making" represents a relatively unexplored space in neuroscience: while the simplest reductions of visual decision-making are well studied, many of the complexities of natural environments---rich visual scenes, dynamic views, and subject agency---are absent in all but a few experiments. In this dissertation, we first characterize the effects of discrete evidence accumulation, an important element of processing complex stimuli, on visual decision-making. Next, we construct an experimental design environment to facilitate controlled studies of naturalistic visual decision-making. Finally, we develop a system that can apply our newfound understanding of naturalistic visual decision-making, test it in the experimental design environment, and leverage it into a practical BCI system. Taken together, these studies explore new avenues in neuroscience, machine learning, and application development. |