Tatiana Engel

The brain’s activity is in constant motion: it ebbs and flows in big waves when we are in a deep slumber, turns into small ripples when we reawaken, and flows in orchestrated streams when we perceive, decide and remember. These complex dynamics are driven by intricate networks of microscopic interactions between hundreds thousands neurons and thus are only vaguely glimpsed in spike-trains of single neurons. Fortunately, recent advances in recording techniques enable us to monitor the activity of large neural populations in behaving animals, offering the opportunity to investigate how dynamic variations of collective neural-activity states translate into behavior. To gain insights from these large-scale recordings, we develop and apply computational methods for discovering collective neural dynamics from sparse, high-dimensional spike-train data. We also develop models and theory to explain how collective neural dynamics support specific network computations and how these dynamics are constrained by biophysical properties of neural circuits. In these endeavors, we employ and extend tools and ideas from diverse fields, including statistical mechanics, machine learning, dynamical systems theory, and information theory. Our work benefits from close collaborations with experimental neuroscience laboratories that are collecting neurophysiological data in animals engaged in sophisticated tasks, such as attention, decision making and learning.