A statistical study of cochlear nerve discharge patterns in response to complex speech stimuli

Area of Science:

• Neuroscience
• Auditory Neuroscience
• Computational Neuroscience

Background:

• Cochlear nerve fibers encode auditory information through their discharge patterns.
• Previous models, like the Poisson process, have limitations in explaining the full complexity of neural responses.
• Auditory nerve fiber responses exhibit history-dependent refractory properties.

Purpose of the Study:

• To investigate cochlear nerve discharge patterns in response to synthesized speech stimuli.
• To evaluate the suitability of the inhomogeneous Poisson counting process model.
• To adopt and validate the Markov process model for auditory nerve responses.

Main Methods:

• Collected cochlear nerve discharge data from 223 auditory nerve fibers in a cat.
• Analyzed post-stimulus time histograms and Fourier transforms of neural responses.
• Employed maximum-likelihood and minimum description length algorithms to estimate stimulus and recovery functions.
• Simulated Markov point processes using estimated functions.

Main Results:

• Discharge statistics were inconsistent with the inhomogeneous Poisson counting process model.
• Synchronized components showed variances up to 3 times lower than predicted by the Poisson model.
• Simulated Markov processes closely matched the measured first- and second-order statistics of auditory nerve fibers.

Conclusions:

• The Markov point process model effectively accounts for correlations in auditory nerve discharges.
• History-dependent refractory properties are crucial for accurate modeling of auditory nerve responses.
• This study validates the Markov model's utility in understanding auditory coding.