Intrinsic oscillations in spike trains indicate non-renewal statistics due to convergence of inputs in dorsal cochlear nucleus neurons.

The occurrence of intrinsic oscillations (IOs) in a unit's discharge is reflected by a prominent peak in the power spectrum (i.e., Fourier transform of the autocorrelation function) of spike trains obtained from single-unit discharge, at a frequency independent of stimulus spectral characteristics. IOs have been reported by researchers in the dorsal cochlear nucleus (DCN) of both the cat and the Mongolian gerbil. It has been hypothesized that IOs are related to inter-spike interval (ISI) regularity (e.g., [Hear. Res. 58 (1992) 153]). This hypothesis is tested in this paper. Responses to multiple presentations of 50-300 ms duration tone bursts, at and near the unit's best frequency (BF) at 20-60 dB re threshold were recorded from DCN units of barbiturate-anesthetized (30 units), as well as decerebrate (53 units) Mongolian gerbils. IOs in the recordings were then compared with the IOs in simulations of spiking-neuron models. The models were selected because: (1) their ISI regularity characteristics follow those of experimental data and (2) their IO properties are completely determined by their ISI regularity. Such comparison reveals that Ghoshal's hypothesis fails for a fraction of the units. These results suggest a re-evaluation of the purported relationship between IOs, ISI regularity, and SAM response. Alternate hypotheses are proposed here using computational models that are based on convergence of multiple neural inputs onto the unit under study. These models produce non-renewal statistics that resemble those of the experimental data, as is evident from IO-based analysis.