The role of neuronal feedback in the detection of transient signals: a computational approach

Abstract

This study investigates the role of neuronal feedback in the detection of small amplitude transient signals. We focus specifically on how the weakly electric fish Apteronotus leptorhynchus capture prey such as Daphnia in a noisy environment, by means of its electric sense. Using the electrosensory network as a template, we build a computational model that allows us to evaluate detection performance in two different scenarios: without neuronal feedback (open-loop) and with neuronal feedback (closed-loop). For each network scenario, spike count distributions across realizations are computed in the absence and presence of the prey-related signal, and compared using ROC (Receiver Operating Characteristic) curves analyses. The area under the ROC curve (AUC) and the equal error rate (EER) are used to quantify the performance of the different network configurations. For body-object distances < 20 mm, the closed loop model results in a more robust and reliable signal detection than the open loop configuration. For larger distances, there are no differences between open and closed loop. These results depend on the exact choice of parameters for the model, in particular those controlling feedback inhibitory input strength; increasing inhibition, to a certain extent, further improves the closed-loop model. This study shows, in a simplified model framework that can be applied to a variety of sensory systems, how feedback can enhance the detection of weak transient signals.