Exploring a neural circuit for estimating ambient wind direction in flight
Neuroscience ePoster by Christina May, John Crimaldi, Floris van Breugel and Katherine Nagel
Part of COSYNE 2023 posters, NYU Langone Health; Neuroscience Institute; University of Colorado Boulder; University of Nevada, Reno
Sunday
Abstract
To navigate in a natural environment, a flying fly must estimate and contend with external forces such as wind. Estimating wind direction is difficult because the fly cannot directly sense ambient wind in flight. Instead, the fly experiences a vector sum of self-generated motion and the motion imposed by wind, which produces both mechanosensory (airflow) and visual (optic flow) experience. How the fly brain integrates these sensory inputs to estimate ambient wind direction is not known. Here we explore how a circuit in the Drosophila Central Complex (CX) might integrate airflow and optic flow experience to estimate wind direction during flight. We first developed a physical model to estimate the airflow and optic flow a flying fly experiences during wind gusts. This model demonstrates that wind causes airflow and optic flow directions to diverge, with transient changes in airflow direction followed by sustained changes in optic flow direction. We next imaged calcium activity from a set of CX inputs, called PFNs, that had previously been shown to encode airflow and optic flow. We found that one population (PFNd) encodes both airflow and optic flow direction with dynamics that mimic those seen in our physical model (transient airflow and sustained optic flow responses). In response to coherent airflow and optic flow, PFNd neurons exhibit nonlinear amplification, suggesting they encode confidence about the fly’s direction of travel. Finally, we examined calcium activity in a set of downstream local neurons called h∆J. While PFNd neurons respond reliably to the same stimulus across trials, h∆J neurons show divergent responses to the same stimulus. h∆J neurons instead accumulated stable patterns of activity over minutes, suggesting a role in estimating wind direction from noisy sensory input. Together, these models and experiments illuminate circuit principles for estimating the direction of an external force from complex multi-sensory experience.
Topics
Details
- Session
- Poster Session III
More in Poster Session III
All posters- Mar 12, 2023Reidar Riveland et al.
Accounting for visual cortex variability with distributed neural activity states
Mar 12, 2023Anna Li et al.Alignment of ANN Language Models with Humans After a Developmentally Realistic Amount of Training
Mar 12, 2023Eghbal Hosseini et al.Arousal dynamics: diverse measurements of a universal manifold
Mar 12, 2023Ryan Raut et al.
More with Floris van Breugel
All their workA computational framework for decoding active sensing
ePoster · Mar 27, 2025COSYNE 2025Deep reinforcement learning trains agents to track odor plumes with active sensing
ePoster · Mar 27, 2025COSYNE 2025Emergent behavior and neural dynamics in artificial agents tracking turbulent plumes
ePoster · Mar 19, 2022COSYNE 2022