Frequency selectivity emerging from spike-timing-dependent plasticity
Neural Computation
| Hi-index | 0.00 |
Acoustic signals transmit information by temporal characteristics and envelope periodicity as well as by their frequency content. Many animals can extract the frequency content of a signal by means of specialized organs such as the cochlea but for the detection and identification of higher-order periodicity, e.g., amplitude modulations, this type of organ is useless. In addition, many animals do not have a cochlea but still depend on a reliable identification of different frequencies in the vast variety of acoustic signals they perceive in their natural environment. Hence, neural mechanisms to decode periodicity information must exist. We present a detailed mathematical analysis of a recurrent and a feedforward model of neuronal periodicity extraction and discuss basic constraints for neuronal circuitry performing such a task in a biological system. Both the recurrent and the feedforward model perform well using neuronal parameters typical for the auditory system. Performance is limited mainly by the temporal precision of the connections between the neurons.