Meeting Abstract
As an insect flies through its environment, it integrates sensory information across modalities to guide its trajectory; a broad repertoire of behaviors emerges from this confluence of parallel sensory processes as well as the physical interaction with the environment. In the laboratory, we often aim to isolate the sensorimotor computations that give rise to behaviors, measuring locomotor responses to individual sensory stimuli via constrained preparations. And at times, these laboratory findings do not jibe with naturalistic behavior, confounding our expectations and intuitions (e.g. tethered flies fixate a focus of expansion akin to flying backwards and arista-ablated flies recover from wind gust perturbations more rapidly than their intact counterparts). In his seminal cybernetics text “Vehicles”, Braitenberg illustrates via simple robots (vehicles) that the interaction of reflexive sensorimotor circuits can yield seemingly complex or volitional behaviors. In this work, we leverage this bottom-up synthesis, drawing upon the literature of visual and antennal sensing to populate a vehicle model for flying insects. The model comprises a simulated sensory environment (visual scenes and airflow landscapes), visual and antennal sensorimotor computations that transform percepts into motor commands, sensory interactions between modalities (e.g. visually mediated head orientation and antennal positioning), and a physics-based flight model which integrates both self-generated flight forces and exogenous perturbations. Without any high-level decision making, this model demonstrates how the interaction between reflexive circuits proffers a parsimonious reconciliation for a broad behavioral repertoire and numerous (and sometimes confounding) experimental observations.
