Meso-scale modelingHow do multiple interconnected brain regions interact to collectively shape behavior? How is information organized in neural circuits with substantial recurrence between components? We aim to investigate how sensory, internal, and behavioral signals are encoded by neural populations from multiple brain regions, such as the nuclei of the hypothalamus and basal ganglia. We use this tuning analysis to constrain multi-region neural network models of circuit dynamics, and model perturbations to circuit activity produced by experience, by changes in internal state, and by neurological disorders. |
Cell type heterogeneityNeurons express a tremendous diversity of neuromodulators, neuropeptides, receptors, and ion channels, whose expression is organized into a set of cell types. What computational implications does this neuronal cell type heterogeneity have for the brain? And how can neural circuits best take advantage of the diversity of available signaling molecules to encode information? We investigate these questions in biological neural network models, to uncover computational principles that can inform future experimental study of cell types in the brain. |
Quantitative behavior analysisTo understand how the brain gives rise to behavior, we must first define what that behavior is. We use techniques from machine learning to study the structure of animals’ actions, and to create highly detailed, data-driven phenotypic profiles of animals as they interact with conspecifics and with their environment. These profiles in turn provide new, detailed insight into the structure of animal behavior, and how behavior is modified by past experiences, by the animal’s own internal motivational state, and by diseases and disorders of the nervous system. |