Meeting Abstract

45.3  Monday, Jan. 5  Reconciling Open- and Closed-Loop Experiments in Sensorimotor Control of Drosophila ROTH, E.*; REISER, M.B.; COWAN, N.J.; Johns Hopkins University; Howard Hughes Medical Institute; Johns Hopkins University eatai@jhu.edu

In optomotor yaw regulation experiments, a rigidly tethered fly (Drosophila) modulates yaw torque to frontally fixate a moving vertical stripe. In closed-loop experiments, the measured yaw torque stabilizes the error signal (the angular displacement of the stripe) via real-time feedback. In open loop, torque is measured without this feedback. Heisenberg and Wolf (1988) observed that, when presented with stimuli oscillating at low frequency, flies exhibited qualitatively different responses to the same error signals in closed- and open-loop trials. They concluded that flies distinguish open- from closed-loop conditions and employ different sensorimotor transformations (controllers) accordingly. We present evidence that, for stimuli with faster dynamics (higher frequency content), closed- and open-loop responses are comparable. Further, we address the question of whether a fly knows that it is flying under closed-loop conditions through the analysis of a candidate model of the sensorimotor transform: a standard PID (proportional, integral, derivative) controller with a biologically feasible nonlinear saturation to the integral term. We demonstrate that even this simple controller can capture the categorical differences in behavior previously observed as well as the similarities seen in high-frequency trials. In the model, the internal states of the open-loop system exhibit sensitivities to biases in noise or initial conditions. Feedback mitigates these factors which, in open loop, drive the state of the system to regimes not typically encountered in closed loop. Implementing a single controller, the behavior transitions between categorically different responses as governed by the internal state of the system, not mediated from a higher center.