Principles of sensory integration and behavioral flexibility in Drosophila: symmetry, plasticity, and variability

Abstract

This dissertation is organized around three basic and interrelated functions of animal brains: integration of disparate sensory streams into unified but flexible internal representations, learning through integration of new sensory experiences with prior beliefs, and generation of behavioral diversity across individuals. By exploiting the powerful experimental capabilities of the fruit fly Drosophila melanogaster, I have dissected the neuronal substrates of these phenomena and obtained new insights into the biological principles that govern them. The first chapter explores interhemispheric integration in the larval olfactory system. By combining volumetric calcium imaging with asymmetric manipulations of the sensory periphery, I have identified the mushroom body (MB) as a key circuit node for integrating sensory input from the two sides of the body. I found that while the input to the MB is essentially independent between the left and right hemispheres, the output channels exhibit varying degrees of left–right symmetrization. Moreover, asymmetric activation of a subset of these output channels can impart a side-specific bias to the animal's chemotaxis behavior. The second chapter examines how the dopaminergic neurons (DANs) of the larval MB represent olfactory and gustatory stimuli. Through calcium imaging, I provide the first direct evidence that these neurons compute reward prediction errors rather than simply encoding the valence of sensory experiences. I discovered that unreinforced presentation of innately attractive odors activates punishment-associated DANs, while mixing these same odorants with sugar eliminates these responses. Furthermore, I found that the transient responses of these neurons to positive and negative reinforcers exhibit a derivative-like character consistent with models of temporal difference learning. The third chapter investigates neuroanatomical correlates of individual variation in locomotor handedness in adult flies. By combining high-throughput behavioral assays with neural network-based morphometric analysis of central brain neuropils, I discovered that the magnitude (but not direction) of turning bias correlates with the size of specific central complex structures, notably the first pair of noduli. A parallel experiment to characterize the anatomical correlates of artificial selection for extreme handedness bias converged on exactly the same brain structure. In both cases, the left-right symmetry of the observed morphological correlations suggests a model in which symmetric variation at key circuit nodes modulate the magnitude of individual differences by amplifying microscopic asymmetries in upstream processing layers. The final chapter places these findings into the broader context of recent developments in the field, suggesting potentially fruitful new lines of inquiry. In sum, this work advances our understanding of how brains transform sensory stimuli into coherent and flexible, yet idiosyncratic, patterns of behavior.