Uncovering population dynamics in spinal circuitry
Abstract
Though low-dimensional dynamics are a ubiquitous feature of population activity in neural circuits, our field has yet to precisely and causally delineate their role in performing computations and driving motor behaviors. With our combination of ideal model system (recording population activity from interneurons in the mammalian spinal cord), cutting-edge AI tools (deep learning-based dynamical systems models), and experimental capabilities (real-time state estimation and optogenetic perturbation), we are uniquely positioned to uncover population dynamics at unprecedented temporal and spatial resolution, and further, to directly and causally test their link to motor output. Neural circuits in the motor system consist of thousands to millions of neurons which often exhibit low-dimensional, dynamic features. These dynamics - with dimensionality that is orders of magnitude lower than the total number of neurons - are hypothesized to be the primary mechanism through which neural circuits implement computations and produce behavioral output. However, it remains unclear whether such dynamics are simply useful for intuition-building, or whether they explain movement control at the level of spatial and temporal precision achieved by the motor system. Further, it is unclear how a system s internal dynamics are shaped by inputs to adjust its output. We hypothesize that low-D population dynamics precisely control features of motor behavior including muscle activation patterns and timing on a moment-by-moment basis and at millisecond timescale. We will test this hypothesis through population recordings from spinal interneurons in mice, along with intramuscular EMG, high-resolution behavioral tracking, and optogenetic perturbation of afferent inputs. We will also use our AI tools to estimate dynamics underlying spinal population activity and muscle activity with unprecedented temporal precision. This setup will allow us to test, for the first time, whether low-dimensional population dynamics precisely link to motor output, and further, whether and how those low-D population dynamics are modified by sensory inputs.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Mar 14, 2024
- Source ID
- FA95502310727
Entities
People
- Chethan Pandarinath
Organizations
- Air Force Office of Scientific Research
- Emory University
- United States Air Force