S J Jones1, N Perez. 1. The National Hospital for Neurology and Neurosurgery, Queen Square, WC1N 3BG, London, UK. sjjones@ion.ucl.ac.uk
Abstract
OBJECTIVE: To examine the hypothesis that auditory evoked potentials (AEPs) to pitch and timbre change of complex harmonic tones reflect a process of spectral envelope analysis. METHODS: AEPs were recorded to: (1) continuous tones of 'clarinet' timbre whose pitch abruptly rose or fell by 1 or 7 semitones every 0.5 or 1.5 s; (2) a cycle of 6 pitches changing every 0.5 s; (3) tones of constant pitch whose timbre (spectral envelope shape) changed periodically; (4) pitch change of high- and low-pass filtered 'clarinet' tones. RESULTS: The amplitudes of the 'change-N1' (CN1) potential peaking at ca. 90 ms and the following CP2 were influenced to a far greater degree by the time interval between changes, than by the magnitude of the change or by the time interval between occurrences of the same pitch. Amplitudes were also strongly dependent on the number of partials present, irrespective of whether they were increasing or decreasing in energy. The algebraic sum of the responses to pitch change of high- and low-pass filtered tones closely approximated the response to the unfiltered tone. CONCLUSION: The rate-sensitivity of the responses cannot be explained by the refractoriness of frequency-specific 'feature detector' neurones, but rather of a process (termed 'C-process') which analyzes amplitude modulations across the spectral envelope, the contribution of different frequency bands combining linearly in the scalp-recorded activity. On-going computation of the spectral envelope shape may be an important factor in maintaining the perceptual constancy of timbre.
OBJECTIVE: To examine the hypothesis that auditory evoked potentials (AEPs) to pitch and timbre change of complex harmonic tones reflect a process of spectral envelope analysis. METHODS: AEPs were recorded to: (1) continuous tones of 'clarinet' timbre whose pitch abruptly rose or fell by 1 or 7 semitones every 0.5 or 1.5 s; (2) a cycle of 6 pitches changing every 0.5 s; (3) tones of constant pitch whose timbre (spectral envelope shape) changed periodically; (4) pitch change of high- and low-pass filtered 'clarinet' tones. RESULTS: The amplitudes of the 'change-N1' (CN1) potential peaking at ca. 90 ms and the following CP2 were influenced to a far greater degree by the time interval between changes, than by the magnitude of the change or by the time interval between occurrences of the same pitch. Amplitudes were also strongly dependent on the number of partials present, irrespective of whether they were increasing or decreasing in energy. The algebraic sum of the responses to pitch change of high- and low-pass filtered tones closely approximated the response to the unfiltered tone. CONCLUSION: The rate-sensitivity of the responses cannot be explained by the refractoriness of frequency-specific 'feature detector' neurones, but rather of a process (termed 'C-process') which analyzes amplitude modulations across the spectral envelope, the contribution of different frequency bands combining linearly in the scalp-recorded activity. On-going computation of the spectral envelope shape may be an important factor in maintaining the perceptual constancy of timbre.