Cortical Adaptive Filtering in Bioacoustic Signal Classification

摘要

Frequency receptive fields (RF) in the auditory cortex are systematically alteredas a result of a brief training experience. Responses to training signals (conditioned stimulus, CS) are increased whereas responses to other frequency, including the pretraining best frequency (BF) are decreased. These changes can be sufficiently large so that the frequency of the CS becomes the new BF. Frequency RFs describe the frequency filtering characteristics of auditory system neurons. Systematic changes of these RF may then be seen to indicate adaptive filtering; i.e., adaptive filtering may be said to have occurred when training with a selected acoustic signal results in plasticity of the receptive field that is highly specific to that signal. The goal of this project is to determine the underlying processes of adaptive filtering both neurophysioogically and computationally. Of particular, we are applying findings from olfactory (paleo) cortex and training rules derived from the hippocampus (archicortex) to the architecture of the thalamo-neocortical auditory system to determine computational implications for bioacoustic signal classification. An initial stage of this project has been to determine fundamental charactertisics of adaptive filtering which are essential for computational analysis. These findings provide a basis for experiments to test predictions which have been generated by the computational work.