Pitch is a key perceptual attribute of natural sounds, but its neural basis remains under active debate. The pattern of average neural firing rates along the tonotopic axis, called the rate-place code, adequately explains human pitch perception for most stimuli, including harmonic complex tones (HCTs). However, spectrally denser stimuli, such as mixtures of multiple simultaneous HCTs (i.e., chords) have limited rate-place information due to auditory filtering. Some pitch tasks remain feasible under such conditions, but it is unclear whether listeners rely on weak residual rate-place cues or turn to other cues. To adjudicate between these possibilities, we generated synthetic stimuli we call rate-place metamers (META), which have near-identical rate-place representations to original pitch-evoking stimuli (ORIG) according to an established model of auditory nerve firing (Zilany, Bruce, and Carney, 2014), but are otherwise unconstrained. Temporal representations of ORIG and META stimuli, evaluated via autocorrelation (Meddis and O’Mard, 1997) or neural fluctuations (Vecchi et al., 2022), differed according to F0 and frequency region. In human listeners, we measured F0 difference limens (F0DLs) and chord discrimination performance for ORIG and META across a range of F0s and spectral regions. The behavioral results follow broadly similar patterns for ORIG and META, suggesting that rate-place cues are sufficient for multiple pitch perception in most circumstances. However, significant differences did emerge between the two stimulus types in some conditions, indicating contribution of cues other than the rate-place code. Timing-based models also failed to fully explain the behavioral results. Rate-place metamers are a tool that may allow us to measure the proportion of pitch perception attributable to rate-place cues, and so to better understand the neural mechanisms underlying pitch perception.

Pitch is a key perceptual attribute of natural sounds, but its neural basis remains under active debate. The pattern of average neural firing rates along the tonotopic axis, called the rate-place code, adequately explains human pitch perception for most stimuli, including harmonic complex tones (HCTs). However, spectrally denser stimuli, such as mixtures of multiple simultaneous HCTs (i.e., chords) have limited rate-place information due to auditory filtering. Some pitch tasks remain feasible under such conditions, but it is unclear whether listeners rely on weak residual rate-place cues or turn to other cues. To adjudicate between these possibilities, we generated synthetic stimuli we call rate-place metamers (META), which have near-identical rate-place representations to original pitch-evoking stimuli (ORIG) according to an established model of auditory nerve firing (Zilany, Bruce, and Carney, 2014), but are otherwise unconstrained. Temporal representations of ORIG and META stimuli, evaluated via autocorrelation (Meddis and O’Mard, 1997) or neural fluctuations (Vecchi et al., 2022), differed according to F0 and frequency region. In human listeners, we measured F0 difference limens (F0DLs) and chord discrimination performance for ORIG and META across a range of F0s and spectral regions. The behavioral results follow broadly similar patterns for ORIG and META, suggesting that rate-place cues are sufficient for multiple pitch perception in most circumstances. However, significant differences did emerge between the two stimulus types in some conditions, indicating contribution of cues other than the rate-place code. Timing-based models also failed to fully explain the behavioral results. Rate-place metamers are a tool that may allow us to measure the proportion of pitch perception attributable to rate-place cues, and so to better understand the neural mechanisms underlying pitch perception.