One of our main methods is developing computational models of cognition. The basic approach is to take hard questions such as “How do people pay attention to only one source of perceptual input (e.g., the map of a city) and ignore the irrelevant ones (e.g., traffic noise)?”, and try to formulate it in terms we understand (a bit) better. This often means reformulating it as a (reinforcement) learning problem. Ideally, a model allows conceptual clarification, good experimental designs, and precise model comparison.
We frequently use electrophysiological (EEG) recordings to measure electrical brain activity from the scalp. Because of the high temporal resolution of EEG, we can look at changes within the brain on a millisecond scale. We have access to a Biosemi system (64 or 128 channels) and an ActiChamp system (64 channels). Our lab has experience with ERP, time-frequency and connectivity analysis. We also use neural decoding to unravel the dynamics of cognitive processes.