Sound frequency representation in primary auditory cortex is level tolerant for moderately loud, complex sounds

The distribution of neuronal characteristic frequencies over the area of primary auditory cortex (AI) roughly reflects the tonotopic organization of the cochlea. However, because the area of AI activated by any given sound frequency increases erratically with sound level, it has generally been proposed that frequency is represented in AI not with a rate-place code but with some more complex, distributed code. Here, on the basis of both spike and local field potential (LFP) recordings in the anesthetized cat, we show that the tonotopic representation in AI is much more level tolerant when mapped with spectrotemporally dense tone pip ensembles rather than with individually presented tone pips. That is, we show that the tuning properties of individual unit and LFP responses are less variable with sound level under dense compared with sparse stimulation, and that the spatial frequency resolution achieved by the AI neural population at moderate stimulus levels (65 dB SPL) is better with densely than with sparsely presented sounds. This implies that nonlinear processing in the central auditory system can compensate (in part) for the level-dependent coding of sound frequency in the cochlea, and suggests that there may be a functional role for the cortical tonotopic map in the representation of complex sounds.