How do the nervous systems of animals integrate feedback from multiple sensory systems to coordinate robust behaviors in complex sensory landscapes? Although the neurobiology of individual sensory systems has received much attention, the processing of multisensory information, from either an experimental or theoretical perspective, has received far less attention.
My goal is to contribute to the development of a theory of biological multisensory integration by studying the sensory control of flight by Drosophila melanogaster. The agile flight behaviors exhibited by flies are achieved by a flight control system that is highly specialized to rapidly combine information from multiple sensory modalities, including vision, olfaction, audition, and several mechanosensory systems. The challenges of multisensory integration constrain the architecture and encoding properties of individual sensory systems; explicitly studying the integration mechanisms may be the swiftest route to advancing our understanding of the essential logic of nervous systems.
Behavioral Investigation into Multisensory Control of Flight
Behavior is inherently the outcome of integrated multisensory information, but in the lab, it is possible to stimulate individual modalities in isolation or in combination. It is also possible to arrange experimental conditions that only rarely emerge in natural flight. In particular, we can study the behavioral response to conflicting cues, such as a visual stimulus corresponding to straight flight, and a simultaneous crosswind. Even with great care, investigations into the neural circuitry underlying multisensory integration can suffer from combinatorial explosion—all potentially relevant stimuli for each sensory system should be tested against relevant stimuli for all other systems. For this reason, my research program exploits quantitative behavior as the necessary, expeditious tool for establishing an algorithmic basis for the multisensory process, for which the contributing neuronal circuitry can be subsequently identified.
The fly's nervous system must solve a challenging inference problem—self-generated motion in varying landscapes combined with shifting winds generates a confounding and potentially ambiguous sensory percept. For this reason it is essential that the fly's nervous system makes use of information from multiple sensory systems. At Janelia Farm, I will attempt to disentangle the contributions of visual, wind, and inertial sensing to the flight control system of Drosophila. I will continue to use tethered flight experiments to quantify motor responses to controlled multisensory stimuli.
In the virtual-reality flight simulators, tethered flies are held stationary while flying, and motion is simulated by dynamically updating a panoramic visual display. To study sensory systems that are ideally stimulated only when the fly is in motion, I have developed a robotic system for moving a tethered fly in a controlled manner. I will continue to use computational models of the behavioral responses of flies to provide a necessary context for investigations into the supporting neural architecture.
Investigation of Neuronal Circuits Underlying Multisensory Processing
Once I have used behavioral methods to identify and characterize examples of multisensory fusion in wild-type flies, I plan to use the molecular tools that are available for Drosophila to pursue the neuronal circuits underlying these behaviors. One line of work will use genetically encoded optical reporters, such as those that measure intracellular calcium, to localize the integration of sensory information to specific brain regions. Another promising avenue for the study of brain circuitry in intact animals is the use of optically gated ion channels to activate or inhibit specific neurons in candidate regions. The ability to regulate neural activity in behaving flies will go a long way toward identifying the components of the circuitry that implement the sensor fusion process that generates robust flight control in flies. By combining detailed single fly behavioral measurements with the expression control provided by the GAL4-UAS system, I can begin to examine the role of specific brain regions in candidate behaviors.
I do not expect any single method to provide a magic bullet for elucidating structure-function relationships in the brain; a coordinated and collaborative effort that utilizes multiple strategies will be critical. During my time at Janelia, I plan to use both top-down (behavior) and bottom-up (imaging/genetics) approaches, combined with computational modeling, to study sensory fusion in Drosophila.
Development of Modular Visual Stimuli
I am convinced that understanding behavior requires a variety of experimental methods; the design of laboratory instruments is a large part of what I do. During work on my dissertation, I designed electronic displays for virtual-reality simulators of fly flight. I developed a modular system of LED panels that can be connected to "tile" an experimental environment with controllable displays. The panels allow for the construction of displays with arbitrary geometry and control of individual-pixel brightness, allowing experimentation over a broad range of behaviorally relevant stimuli conditions. The modularity of these displays makes it possible to combine a visual stimulus with any tethered fly experiment. The software components and hardware plans for this project are available online. I plan to continue to develop this system and to support the growing community of users.
System Theory Applications in Behavior
An ongoing effort in our lab will be to use the tools of system theory to understand Drosophila flight behavior. This effort will include studying the role of feedback in flight and using system identification methods to develop models of biological systems. Applying a systems perspective to the study of the neural control of fly behavior will be mutually beneficial: theory-inspired considerations motivate further experiments, and a study of the architecture of insect nervous systems may reveal fruitful theoretic problems that are unresolved.