Neural Networks and Dynamical Systems
Prof. John Beggs
Cortical slices and cultures are prepared from rat and mouse brains. These simplified slice networks are placed on advanced microelectrode arrays, allowing up to hundreds of individual spiking neurons to be sampled at high temporal resolution. We borrow concepts from statistical physics (models of frustrated magnetic materials, models of avalanching systems, models of complex networks) to describe the activity we see in data sets recorded with the microelectrode arrays. Current projects on which undergraduates have worked include measuring information flow between neurons, modeling trajectories of network activity through a simplified state space, developing maximum entropy models to describe the probability distribution of network states, applying new measures of synergy between information flows to cellular automata models and to neurophysiological data.
Visual Information Processing
Prof. Rob de Ruyter
Vision in animals, including humans, is based on an ongoing interpretation of optical signals gathered by the eye, and the physical properties of these highly complex signals put fundamental limits on visual information processing. We study visually guided behavior in a specially developed flight tracker system. The fly uses adaptive computational strategies to cope with the complexity of the visual input data stream, and we try to understand these strategies on a quantitative basis, hoping to uncover fundamental principles of biological computation.