Abstract
Evidence for and against classical theories of ‘place’ and ‘period’ mechanisms for the coding of frequency, and the modifications of the theories invoked to account for the pitch of ‘residue’ and other types of stimuli, are examined in the light of physiological data. These include new data on the temporal discharge patterns of cochlear nerve fibres under stimulation with two-tone complexes, harmonic and inharmonic three-tone complexes, and five-tone complexes of differing relative phase. They show, in particular, that certain arguments against ‘period’ coding of ‘residue’ pitch are invalid. The interspike intervals in the discharge patterns of cochlear fibres under these conditions are consistent with the pitches heard. On the other hand, the classical ‘period’ theory needs to be modified to take into account the normally relatively sharp frequency selectivity of cochlear fibres, and requires certain inefficiencies on the part of the central processor for pitch.
Comparison of measures of cochlear fibre frequency selectivity with analogous psychophysical data in man, including those on the ‘existence region’ of ‘residue’ pitch, suggests that ‘residue’-type stimuli judged to be tonal in quality could both: (a) be sufficiently resolved spectrally at the cochlear fibre level to serve as input to any of the current spectral ‘pattern recognition’ mechanisms proposed for the pitch extraction of complex signals, and also, (b) could generate patterns of temporal discharge reflecting enough waveform interaction between the harmonics to convey the pitch heard, because of the shape of the cochlear filters. (This conclusion might have to be qualified in the light of further physiological experiments on the ‘second effect’ of pitch shift.)
The present evidence, both psychophysical and physiological, suggests the following synthesis: musical interval recognition and relatively crude frequency discrimination can be accomplished by trained observers on signals where the frequency appears to be coded exclusively in terms of temporal information. However, the pitch quality of these signals is judged to be poor or absent. Likewise, signals, apparently coded exclusively by ‘place’ mechanisms, while having tonality, allow relatively crude frequency discrimination and judgment of musical intervals. With the possible exception of psychophysical data on the phenomenon of diplacusis, the present evidence cannot exclude the possibility that the central pitch extractor mechanism utilizes both the ‘place’ and ‘period’ cues produced by pure-tone signals (below 5 kHz) and ‘residue’-type signals, both signals evoking strong pitch and fine acuity of frequency discrimination. The degree of salience of a signal's pitch could well depend on the coherence of the two types of cue.
However, the greatest obstacle to the acceptance of ‘place’ coding mechanisms for frequency, particularly of the frequency components of a complex sound, is the restricted dynamic range of the peripheral elements of the auditory nervous system. Because of this, it is not clear how differences in the spectral energy distribution in signals at medium to high sound levels can be established in terms of patterns of mean discharge rate across the cochlear fibre array. At high sound levels, physiological evidence suggests that the discharge rates of the majority of stimulated fibres will be saturated, whereas psychophysical evidence suggests that the coding of the frequency and the relative level of even single-component signals can be carried out over a wide dynamic range in the absence of cues derived from spread of activity across the fibre array. Some new data, however, indicate that this problem may be circumvented at the cochlear nucleus level, but the coding mechanisms involved at the primary neurone level are obscure. One intriguing possibility exists that the auditory nervous system may utilize the fine temporal structure of cochlear fibre discharge patterns for the transmission of ‘place’ information.