A new model for the shapes of rate-level functions of auditory-nerve fibers

All acoustic information relayed to the central nervous system is encoded in the spiking patterns of auditory-nerve (AN) fibres. Here we re-examine and model the dependence of the spike rates of AN fibres on the amplitude of tonal stimuli, building upon the seminal study of Sachs and Abbas (1974). These authors modelled the spike rate vs. sound amplitude functions of AN fibres as the result of the interaction of a 'mechanical stage', describing basilar membrane displacement as a function of sound amplitude, with a 'transducer stage', converting displacement into AN fibre spike rate. The latter stage was modelled as a saturating power function, and spontaneous rate was assumed to simply add to the sound-driven rate. However, the 'transducer stage' of the model - though widely used - has several limitations. Here, we present a physiologically plausible modification of this stage. With this modification, spontaneous activity and its tight correlation with AN fibre sensitivity are emergent properties of the model. Furthermore, we show that for frequencies well below characteristic frequency (CF), where the mechanics are linear, the power which best accounts for all 154 measured cat AN fibre rate-level functions is 3, independent of spontaneous rate or CF. Since this power is the same as that obtained from analysis of absolute thresholds at the perceptual level (Heil and Neubauer, 2003; Neubauer and Heil, 2004), our model also unites AN fibres properties with psychophysics.