Computer simulation of chopper neurons

The aim of this work was to test a new model for oscillating neurons (chopper neurons) in the cochlear nucleus of the auditory system. In the beginning, it is shown that multiples of 0.4 ms are apparent in intrinsic oscillations in the auditory system and in pitch shift experiments. The existence of a time constant of 0.4 ms is explained by the assumption of a minimum chemical synaptic delay of this size between chopper neurons. The large dynamic range of periodicity coding, the small dynamic range of pure tone response, and the sharp frequency tuning of chopper neurons can be explained as a functional result of simultaneous projections from both the auditory nerve fibers and onset neurons to chopper neurons. As a consequence, the topology of the simulation of chopper neurons is as follows: To ensure the preference for multiples of 0.4 ms as observed in physiological and psychophysical experiments, chopper neurons are arranged in a circular network. The minimum number of two chopper neurons in this network results in a chopper period of 0.8 ms which corresponds to the proposed minimum refractory period of 0.8 ms. In the topology, chopper neurons receive input from both auditory nerve fibers and onset neurons. Simulations of the model show that in contrast to previous models, the present model can explain the preference for multiples of 0.4 ms. The model has also the advantage of explaining their large dynamic range of periodicity encoding of chopper neurons in spite of their narrow frequency tuning. Like the models investigated previously by other authors, the present model is able to simulate interspike intervals of spike trains of the chopper responses with high precision. Moreover, the simulation can explain essential properties of real chopper neurons by input from onset neurons. Simulations of the chopper neurons show that variation of the integration widths of onset neurons results in a corresponding variation of the spectral resolution of chopper neurons with smaller widths resulting in a higher resolution of frequency components. Variation of the integration widths of onset neurons also results in variation of the periodicity encoding of chopper neurons. Narrow integration widths lead to better periodicity encoding at low levels. At high levels broader integration widths lead to better periodicity encoding. Therefore it is a conflicting demand at high levels for to adapt the width of the integration to tuning or to encoding periodicity. The observed pitch dichotomy of individual preferences of human subjects for either periodicity pitch or the pitch of low spectral components of harmonic sounds (Schneider et al., 2005) can be explained by assuming adaptations of the width of the integration to either spectral or temporal coding. In contrast to physiological data, Hodgkin-Huxley(HH)-like models of single chopper neurons (e.g. from Rothman and Manis, 2003c) show a strong dependency of their interspike intervals when changing the input strength and do not show any preference for multiples of 0.4 ms. My simulations show that networks of HH-like chopper neurons with a synaptic delay of 0.4 ms do exhibit this preference and their chopper intervals are independent of changing the input strength. The HH-like model of chopper neurons by Rothman and Manis (2003c) does not account for short oscillating intervals of real chopper neurons. The model has been modified with genetic algorithms to generate oscillating intervals as short as 0.8 ms. This "fast Rothman chopper" has been successfully integrated in simulation topology. An enlarged network which is synchronized by a circuit of two "fast Rothman choppers" can account for a preference of ISIs for multiples of 0.4 ms as found in physiological data