A Neural Filter for Electrolocation in Weakly Electric Fish

摘要

The weakly electric fish have the electric organ to generate the electric field and electrosensory mechanism to read the change of electric field with their electroreceptors. The electric organ produce a waveform of electric field as their electric organ discharge (EOD). Their active electrolocation system can detect the distortion of the self-generated electric field, which is caused by a target object, and estimate the position of a target object. In this paper, we suggest a hypothesis that the periodic EOD signals are involved to extract localization features from noisy electrosensory signals and then provide a possible neural network to process the noise-filtering to obtain the accurate information of a target position. The neural network has sinusoidal weights to process a time series of sensor readings for each electroreceptor.