Electrophysiological and Psychophysical Measures of Temporal Pitch Sensitivity in Normal-hearing Listeners

Abstract

To obtain combined behavioural and electrophysiological measures of pitch perception, we presented harmonic complexes, bandpass filtered to contain only high-numbered harmonics, to normal-hearing listeners. These stimuli resemble bandlimited pulse trains and convey pitch using a purely temporal code. A core set of conditions consisted of six stimuli with baseline pulse rates of 94, 188 and 280 pps, filtered into a HIGH (3365-4755 Hz) or VHIGH (7800-10,800 Hz) region, alternating with a 36% higher pulse rate. Brainstem and cortical processing were measured using the frequency following response (FFR) and auditory change complex (ACC), respectively. Behavioural rate change difference limens (DLs) were measured by requiring participants to discriminate between a stimulus that changed rate twice (up-down or down-up) during its 750-ms presentation from a constant-rate pulse train. FFRs revealed robust brainstem phase locking whose amplitude decreased with increasing rate. Moderate-sized but reliable ACCs were obtained in response to changes in purely temporal pitch and, like the psychophysical DLs, did not depend consistently on the direction of rate change or on the pulse rate for baseline rates between 94 and 280 pps. ACCs were larger and DLs lower for stimuli in the HIGH than in the VHGH region. We argue that the ACC may be a useful surrogate for behavioural measures of rate discrimination, both for normal-hearing listeners and for cochlear-implant users. We also showed that rate DLs increased markedly when the baseline rate was reduced to 48 pps, and compared the behavioural and electrophysiological findings to recent cat data obtained with similar stimuli and methods.

Keywords: Auditory change complex; Emporal pitch perception; Frequency following response; Psychophysics.

Fig. 1

Schematic of the stimuli used in both experiments. Part A shows the spectrum of a pulse train filtered into the HIGH region plotted in black, with the pink noise plotted in pink. Parts B and C show the time waveform of the pulse train, without the pink noise, plotted on a coarse and on a fine time scale, respectively. The waveforms on the right of the plot illustrate pulse trains generated by summing harmonics in sine and in alternating phase

Fig. 2

A Example average spectrum of the FFT for a 94-pps sine-phase pulse train (blue) and for a 280-pps alternating-phase pulse train (red) obtained with the contralateral montage. The pulse rate (H1) and harmonics 2–4 for each stimulus are indicated by labelled arrows. B Amplitude of the FFR component at the pulse rate (H1), harmonics 2–4, and the composite amplitude as a function of pulse rate. The left- and right-hand plots are for stimuli filtered into the HIGH and VHIGH regions respectively, both obtained with the contralateral montage. C Amplitude of the composite FFR peak as a function of pulse rate for the contralateral (black) and ipsilateral (red) montage. Data from the cat [27] are shown in blue

Fig. 3

Composite FFR as a function of stimulus pulse rate. Data from the HIGH and VHIGH regions are shown in red and teal bars respectively. For this and all other box-and-whisker plots in this article, the solid horizontal lines show the median, and each box extends from the median to plus-and-minus the inter-quartile range (IQR)

Fig. 4

A Unwrapped phase-vs-frequency plots for the FFRs obtained with the contralateral montage and in the HIGH and VHIGH regions. B Group delays derived from the phase plots in the HIGH and VHIGH region and for the contralateral and ipsilateral montages

Fig. 5

Parts A and B show the ACC averaged across all pulse rates and for increasing and decreasing rate changes for conditions 1–6 of experiment 1 and for the HIGH and VHIGH regions respectively. Individual data are shown by the faint coloured lines and mean data are shown by the thick black lines. Parts C and D are analogous to parts A and B but show the RMS calculated over a 200-ms running window, and with the baseline period and window of interest shown by the red and blue shaded areas, respectively

Fig. 6

A The RMS ACCs measured during the window of interest are shown as a function of pulse rate, separately for rate decreases and increases, in the HIGH and VHIGH regions by the solid red and teal boxes, respectively. The RMS values measured during the baseline period are shown by the white boxes. B ACCs from the region of interest re-plotted from part A so as to aid a visual comparison between HIGH and VHIGH regions. Data are shown for conditions 1 to 6 for which the rate change was 36%. C RMS ACCs in target (solid bars) and baseline (open bars) intervals for conditions 1 and 9, which differed only in switch rate

Fig. 7

ACC RMS amplitude for 36% and 66% rate changes and for 188- and 280-pps baseline rates. Stimuli were filtered into the VHIGH region

Fig. 8

Schematic of the trial structure in the change detection (A) and the sequential comparison (B) conditions of experiment 2

Fig. 9

Rate DLs for the main part of experiment 2 as a function of baseline pulse rate and for the HIGH and VHIGH regions in the left- and right-hand plots respectively. Data for individual participants are shown by the faint coloured lines while mean data are shown by the thick lines. Error bars in Figs. 9, 10 and 11 show plus and minus one standard deviation

Fig. 10

Percent-correct performance in the main (change detection) part of experiment 2 as a function of block number. The left- and right-hand panels show data for stimuli filtered into the HIGH and VHIGH regions respectively. Baseline pulse rates are indicated by the colours of the lines. All data were averaged over percentage rate differences and both directions of rate change

Fig. 11

Performance on experiment 2 as a function of the baseline rate on the previous trial, with baseline rate on the present trial indicated by the colour of each line. The left- and right-hand plots show data for trials in which the first switch in the signal interval was downward or upward, respectively. Data are averaged over all different percent rate differences and are shown only for stimuli filtered into the HIGH region

Fig. 12

Results of the additional parts of experiment 2 showing the effects of task type (left-hand panel) and the duration of each segment of the stimulus (right-hand panel) for each participant