Multisensory integration and processing for decision making in Drosophila flight
Abstract
This is currently a turncated MURI - the portion of the MURI being worked on is in the attachments. When more funds become available we will continue to add back sections of the original proposal. the full and complete abstract follows- Sensory processing and control hardware in modern robots boast gigabits-per-second sampling and gigahertz computation with negligible delays. In contrast, animal wetware suffers orders of magnitude slower throughput, compounded by closed-loop delays of approximately 10-100 ms. How do animals so radically outperform their engineered analogs when performing tasks requiring complex sensory integration for control and decision making? We contend this performance gap stems from profoundly restrictive engineering designs including- (1) modular and isolated sensors, (2) sensation performed independent of control, and (3) central computation for control commands. We propose to discover new design paradigms by studying neuromuscular control of behavior in animals.We will implement ground-breaking experiments and analyses to develop a seminal computational theory of Drosophila sensory integration and decision making for flight by elucidating-1) Interdependencies between the visual (eyes), gyroscopic (halteres), and airflow (antennae)detection systems. Using an array of physical and optogentic perturbation experiments, we willdetermine how each sensor modality triggers and controls the neuromuscular pathways altering sensor function in the other two sensor modalities. Next, we will elucidate general principles for organizing sensor interactions by conducting sensory conflict experiments. Finally, we will develop control theory models and test these in tethered and freely flying animals to understand the pairwise and emergent ternary interactions.2) Sensory integration and active sensation. We will determine how sensory information fromthe antennae, halteres, and eyes are integrated by compass neurons in the fly brain to generate a sense of direction in flying flies. First, we will use results from neural pathway reconstruction from electron microscopy data and 2-photon imaging studies in tethered flies to develop a theoretical framework, grounded in state estimation theory, that elucidates how this process takes place in the ellipsoid body, a sub region of the central complex. Second, we will determine how flies use this representation to generate motor commands that enhance sensation in one modality when another sensing modality is suppressed - the equivalent of using your hands to feel your way in the dark - by testing predictions to elucidate the principles of sensory integration and active sensation. 3) Integration of layered decision pathways. In flies, sensorimotor transformation and decisionmaking is layered rather than centralized, including slow (central complex), medium (visual and antennal reflexes) and fast (gyroscopic reflex) pathways. Here, we will determine how different decision pathways are organized-combined to generate a specific motor command. Specifically, we will use available electron microscopy data, state of the art behavioral and neurophysiological experiments, circuit reconstruction, and theoretical modeling, to determine how each neural pathway implements wing motor commands. Next we will elucidate how flies integrate these pathways by pitting them against each other in a behavior we call ‘interrupted pursuit , where a fly decides to abandon flying towards a previous target (e.g. a bump in the slow central complex pathway) in favor of a new one (e.g. a visual stripe in the medium speed vision pathway). Elucidating the organizational principles responsible for superior decision-making in Drosophila flight, using the language of control theory, will enable new paradigms for designing sensor integration and distributed decision-making in human engineered systems.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Mar 14, 2024
- Source ID
- FA95502310722
Entities
People
- Itai Cohen
Organizations
- Air Force Office of Scientific Research
- Cornell University
- United States Air Force