Responses of Cochlear Nucleus Units in the Chinchilla to Iterated
Rippled Noises: Analysis of Neural Autocorrelograms
WILLIAM P. SHOFNER
Parmly Hearing Institute, Loyola University Chicago, Chicago, Illinois 60626
Shofner, William P. Responses of cochlear nucleus units in the
chinchilla to iterated rippled noises: analysis of neural autocorrelograms. J. Neurophysiol. 81: 2662–2674, 1999. Temporal encoding of
stimulus features related to the pitch of iterated rippled noises was
studied for single units in the chinchilla cochlear nucleus. Unlike other
periodic complex sounds that produce pitch, iterated rippled noises
have neither periodic waveforms nor highly modulated envelopes.
Infinitely iterated rippled noise (IIRN) is generated when wideband
noise (WBN) is delayed (t), attenuated, and then added to (1) or
subtracted from (2) the undelayed WBN through positive feedback.
The pitch of IIRN[1, t, 21 dB] is at 1/t, whereas the pitch of
IIRN[2, t, 21 dB] is at 1/2t. Temporal responses of cochlear nucleus
units were measured using neural autocorrelograms. Synchronous
responses as shown by peaks in neural autocorrelograms that occur at
time lags corresponding to the IIRN t can be observed for both
primarylike and chopper unit types. Comparison of the neural autocorrelograms in response to IIRN[1, t, 21 dB] and IIRN[2, t, 21
dB] indicates that the temporal discharge of primarylike units reflects
the stimulus waveform fine structure, whereas the temporal discharge
patterns of chopper units reflect the stimulus envelope. The pitch of
IIRN[6, t, 21 dB] can be accounted for by the temporal discharge
patterns of primarylike units but not by the temporal discharge of
chopper units. To quantify the temporal responses, the height of the
peak in the neural autocorrelogram at a given time lag was measured
as normalized rate. Although it is well documented that chopper units
give larger synchronous responses than primarylike units to the fundamental frequency of periodic complex stimuli, the largest normalized rates in response to IIRN[1, t, 21 dB] were obtained for
primarylike units, not chopper units. The results suggest that if temporal encoding is important in pitch processing, then primarylike units
are likely to be an important cochlear nucleus subsystem that carries
the pitch-related information to higher auditory centers.
INTRODUCTION
A variety of complex sounds produce the perception of pitch
in human subjects (Fastl and Stoll 1979), and it is likely that
the neural mechanisms underlying the perception of pitch are
similar among the various types of complex sounds. Licklider
(1951) proposed that the nervous system extracts the pitch of a
complex stimulus by performing an autocorrelation analysis of
the information encoded in the auditory nerve. Similar conclusions concerning complex pitch and autocorrelation have been
made based on computational models of the auditory periphery
(Meddis and Hewitt 1991; Meddis and O’Mard 1997). Using a
wide variety of periodic complex sounds, Cariani and Delgutte
(1996a,b) have shown that autocorrelation analysis of auditory
The costs of publication of this article were defrayed in part by the payment
of page charges. The article must therefore be hereby marked ‘‘advertisement’’
in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
2662
nerve spike trains provides a robust representation of pitchrelated information.
Many of the neurophysiological investigations that have
studied the temporal representation of fundamental frequency
in the auditory nerve and cochlear nucleus have used periodic
stimuli having highly modulated stimulus envelopes. These
stimuli include sinusoidally amplitude modulated tones (Frisina et al. 1990; Javel 1980; Joris and Yin 1992; Khanna and
Teich 1989; Kim et al. 1990; Rhode 1995; Rhode and Greenberg 1994; Zhao and Liang 1995), complex tones (Greenberg
and Rhode 1987; Palmer and Winter 1992; Rhode 1994), and
synthetic vowels (Keilson et al. 1997; Kim and Leonard 1988;
Kim et al. 1986; Palmer 1992; Palmer and Winter 1992; Rhode
1998; Wang and Sachs 1993, 1994). One common observation
from several of these studies is that the synchronization to the
fundamental frequency measured for chopper and onset units
in the cochlear nucleus is larger than that observed for auditory
nerve fibers or primarylike units unit types in the cochlear
nucleus (Frisina et al. 1990; Keilson et al. 1997; Rhode 1998;
Rhode and Greenberg 1994; Wang and Sachs 1994). Although
not explicitly stated, the hierarchy of synchrony described in
many of the above studies implies that primarylike units may
play only a minor role in the temporal encoding of pitch-related
information in the cochlear nucleus. One proposed neuronal
circuit model of periodicity pitch involves components from
onset units, chopper units, and pauser units but does not include any input from primarylike units (Langer 1988). However, the enhanced synchronization of chopper and onset units
may only reflect strong temporal responses to highly modulated stimulus envelopes, while having lesser importance for
pitch per se.
In contrast to the complex sounds mentioned above, rippled
noises generate the perception of pitch, but do not have highly
modulated stimulus envelopes. There is a temporal regularity
in the waveform of rippled noise, but this regularity is not
repeated in a periodic manner. Rippled noise of one iteration is
generated when a wideband noise (WBN) is delayed (t) and
the delayed repetition of the WBN is added to the original
WBN; this type of rippled noise has been referred to as cosine
noise (Bilsen et al. 1975). This delay-and-add network can be
repeated N times to generate rippled noises of N iterations (see
Yost et al. 1996). Figure 1 shows the circuit used for generating
a rippled noise of infinite iterations, i.e., infinitely iterated
rippled noise (IIRN). WBN is delayed and attenuated, and the
delayed repetition of the WBN is added to the original WBN
through positive feedback. This type of rippled noise has a
spectrum with sharp peaks at integer multiples of 1/t (see Fig.
1) and has been referred to as comb-filtered noise (e.g., Raat-
0022-3077/99 $5.00 Copyright © 1999 The American Physiological Society
CN RESPONSES TO ITERATED RIPPLED NOISES
2663
FIG. 1. Circuit diagram for generating rippled noise having an infinite number of iterations. Wideband noise (WBN) is delayed,
attenuated, and added to the original undelayed WBN through positive feedback. Inset, left: spectrum of WBN. Inset, right:
spectrum of infinitely iterated rippled noise having a delay of 4 ms and a delayed noise attenuation of 21 dB.
gever and Bilsen 1992). If the delayed repetition is inverted
before it is added (i.e., subtracted), then the spectral valleys
occur at integer multiples of 1/t, and the spectral peaks occur
at odd integer multiples of 1/2t. For simplicity throughout this
paper, IIRN stimuli will be referred to as IIRN[6, t, att],
where 6 indicates whether the delayed noise was added (1) or
subtracted (2), t is the delay in milliseconds, and att is the
attenuation of the delayed noise in decibels. Thus an IIRN that
was generated by adding the delayed noise to the undelayed
noise with a t of 4 ms and a delayed noise attenuation of 21
dB will be referred to as IIRN[1, 4 ms, 21 dB].
Figure 2 shows examples of the autocorrelation functions of
the waveforms for WBN and IIRN. The waveform autocorrelation function of WBN is flat (Fig. 2A), whereas there are
positive correlations in the autocorrelation functions of
IIRN[1, 4 ms, 21 dB] at time lags of 4 ms and integer
multiples of 4 ms (Fig. 2B). The perceived pitch of the
IIRN[1, 4 ms, 21 dB] is at 1/t or 250 Hz (Fastl 1988;
Raatgever and Bilsen 1992; Yost 1996a), and if the pitch
strength of a tone complex is 100%, then the estimated pitch
strength for the IIRN[1, 4 ms, 21 dB] is ;50 – 60% (see Yost
1996b). For IIRN[2, 4 ms, 21 dB], there are negative corre-
FIG. 2. A–C: autocorrelation functions of the stimulus waveforms for WBN (A), infinitely iterated rippled
noise (IIRN)[1, 4 ms, 21 dB] (B), and IIRN[2, 4 ms,
21 dB] (C). 4 in B and C, time lag corresponding to the
perceived pitch of the stimulus. D–F: autocorrelation
functions of the stimulus envelope for WBN (D),
IIRN[1, 4 ms, 21 dB] (E), and IIRN[2, 4 ms, 21 dB]
(F). Envelope was obtained from the Hilbert transform of
the waveform. 4 in E and F, peak at time lag t for both
IIRN[1, 4 ms, 21 dB] and IIRN[2, 4 ms, 21 dB].
2664
W. P. SHOFNER
lations in the waveform autocorrelation function at time lags of
4 ms and at odd multiples of 4 ms but positive correlations at
even multiples of 4 ms (Fig. 2C). That is, these negative and
positive correlations alternate throughout the waveform autocorrelation function. Perceptually, the perceived pitch of
IIRN[2, 4 ms, 21 dB] is at 1/2t, or 125 Hz, i.e., one octave
lower than the pitch of IIRN[1, 4 ms, 21 dB] (Fastl 1988;
Raatgever and Bilsen 1992; Yost 1996a).
Figure 2 also shows examples of the autocorrelation functions for the envelopes obtained from a Hilbert transform of the
waveform for WBN and IIRN. The envelope autocorrelation
function for WBN is flat (Fig. 2D), whereas there are positive
correlations at t and integer multiples of t for both IIRN[1, 4
ms, 21 dB] and IIRN[2, 4 ms, 21 dB] (Fig. 2, E and F). The
alternation of negative and positive correlations observed in the
waveform autocorrelation function is not observed in the envelope autocorrelation function of IIRN [2, 4 ms, 21 dB].
Recent psychophysical studies have shown that the pitch and
pitch strength of iterated rippled noises as well as discrimination data can be accounted for using temporal processing
models based on autocorrelation (Patterson et al. 1996; Yost
1996a,b, 1997; Yost et al. 1996). We have shown previously
that chinchillas can discriminate behaviorally IIRN[1, t, att]
from WBN and have argued that the underlying neural mechanisms are common between chinchillas and human subjects
(Shofner and Yost 1995, 1997). Neurophysiological studies in
the auditory nerve and cochlear nucleus using rippled noise of
one iteration have shown that a temporal representation related
to pitch can be found in the discharge patterns of low-frequency auditory nerve fibers and primarylike units (Shofner
1991; ten Kate and van Bekkum 1988). The present study
examined the representation of t in the temporal discharge
patterns of cochlear nucleus units in the chinchilla in response
to IIRN stimuli. In this paper, the responses of primarylike and
chopper units are compared. If the hierarchy of synchrony
previously discussed reflects the relative importance of units in
the temporal encoding of pitch-related information, then primarylike units should give the smallest synchronous responses
to rippled noise stimuli compared with other unit types.
METHODS
The procedures have been reviewed and approved by the Institutional Animal Care and Use Committee of Loyola University Chicago. Twenty-three adult chinchillas weighing between 500 and 800 g
were anesthetized with intraperitoneal injections of pentobarbital sodium (70 mg/kg); supplemental injections were given to maintain
areflexia. Body temperature was maintained around 37°C with a DC
heating pad. Animals were tracheotomized, and the external auditory
canals were exposed and transected. Animals were placed in a modified headholder (Kopf Model 900), and the left bulla was exposed and
opened. The cerebellum was exposed by an opening in the temporal
bone in a manner similar to that described for gerbils by Frisina et al.
(1982). Animals were placed in a double-walled sound-attenuating
chamber (Tracoustics) during neurophysiological recording of unit
responses.
Indium-filled micropipettes (Dowben and Rose 1953) or tungsten
microelectrodes (Microprobe) were used to record single-unit activity.
Neural spikes were amplified and filtered using standard electrophysiological procedures. Electrodes were advanced through the cerebellum into the cochlear nucleus using an hydraulic microdrive system
(Kopf 650). Using this approach, it was not uncommon to hold an
isolated single unit for 1–2 h. However, the use of metal microelec-
trodes often made the isolation of clear single units difficult, because
multiple-unit activity or large neurophonic evoked potentials were
recorded frequently. Data regarding spike times were obtained from
neural spike trains that clearly were evoked from a single unit and
were not obtained by attempting to isolate single unit activity from
multiunit clusters of spike trains.
Data acquisition and stimulus presentation were under the control
of either a MassComp computer system or, in later experiments, a
Gateway2000 Pentium computer system with Tucker-Davis Technologies (TDT) modules. The sampling and conversion rates were set at
50 kHz for the A/D and D/A devices. The times of occurrences of
spikes were determined on-line relative to the onset of the stimulus;
the amplified neural spikes were digitized and averaged on-line to
examine the averaged spike waveform for the presence of prepotentials (Pfeiffer 1966). Acoustic stimuli were presented to the left ear
through a Sennheiser HD 414 SL earphone that was enclosed in a
brass housing; the housing also held a calibration microphone (Bruel
and Kjaer 4134). Acoustic search stimuli consisted of 100-ms bursts
of either wideband noise or tones at the best frequency (BF) of the
background neural activity.
When a unit was isolated, its BF first was determined using audiovisual cues, and data were collected to classify the unit physiologically. Classification of unit types was based on poststimulus time
(PST) histograms, interspike interval (ISI) histograms, and regularity
analysis (Bourk 1976; Young et al. 1988) as well as the presence or
absence of a prepotential. BF tones were generated by the computer
and presented through a 16-bit D/A converter (MassComp DA04H or
TDT DA3–2 module). Rate-level functions were generated over a
100-dB range in 1-dB steps for either 200- or 400-ms BF tone bursts
with rise/fall times of 10 ms presented once per second. One tone
burst was presented at each level, and the rate-level function was
smoothed using a 5-bin triangular moving window average. Threshold
was defined as the level that first evoked an increase in discharge rate
.2 SDs above the estimated spontaneous discharge rate, provided that
the next three levels also evoked firing rates .2 SDs above spontaneous rate. PST histograms were typically generated at 20 – 40 dB
above threshold for 250 presentations of a 50-ms BF tone with 2-ms
rise/fall times presented once every 250 ms. If the characteristic
discharge pattern was obscured in the PST histogram due to strong
phase-locking, then additional PST histograms were generated in
which each of the 250 presentations of the BF tone had a random
starting phase. The use of random starting phases sometimes allowed
the characteristic discharge pattern to become apparent.
After the data were collected for unit classification, the responses to
WBN and IIRN were studied. A set of WBN and IIRN stimuli were
generated using the same equipment and parameter settings that were
used previously to measure discrimination thresholds in chinchillas
(Shofner and Yost 1995). For WBN and each IIRN, 5 s of the
waveform was sampled; the waveform amplitudes were adjusted so
that all WBN and IIRN stimuli had equal root-mean-square (rms)
amplitudes. The 5-s samples of each noise then were stored on disk as
stimulus files. Rate-level functions first were generated over a 100-dB
range in 1-dB steps for 500-ms WBN bursts with rise/fall times of 10
ms presented once per second, and threshold was estimated as defined
above for BF tone rate-level functions. The responses of the isolated
unit then were studied for WBN and IIRN stimuli at a fixed overall
level, generally at 20 dB above threshold as determined from the
WBN rate-level function. A total of 100 separate samples of WBN or
IIRN were presented. Consequently, the waveform of the noise was
not the same for each presentation. Each sample had a duration of 500
ms with 10-ms rise/fall times and were presented once every second;
these are the same parameters used in the previous behavioral study
(Shofner and Yost 1995). Typically, the value of t was first chosen to
be 4 ms, followed by ts of 2 and 8 ms. These values of t were chosen,
because these were the ts at which psychometric functions were
obtained for chinchillas (Shofner and Yost 1995). The delayed noise
attenuation was fixed at 21 dB; an attenuation of 21 dB generated an
CN RESPONSES TO ITERATED RIPPLED NOISES
2665
FIG. 3. Discharge characteristics of a primarylike
unit having a best frequency (BF) of 0.85 kHz. A:
poststimulus time (PST) histogram obtained in response to 250 50-ms BF tone bursts presented with
random starting phases at 32 dB above threshold.
Binwidth is 200 ms. B: expanded version of PST
histogram showing the discharge pattern at the onset.
C: interspike interval (ISI) histogram generated from
the same spike trains used to generate the PST histogram in A; spike times between 20 and 50 ms are used
for generating the ISI histogram. Binwidth is 100 ms.
D: regularity analysis of the spike train showing the
mean subsequent ISI during the time of the PST
histogram. —, mean ISI (m); z z z , standard deviation
(s). Functions were smoothed using a 5-bin moving
window average. Inset: coefficient of variation as a
function of time for the regularity analysis and was
smoothed using a 9-bin moving window average.
IIRN having the greatest pitch strength without causing the positive
feedback circuit (see Fig. 1) to oscillate.
The spike trains obtained in response to WBN and IIRN were
analyzed as PST histograms, spike count distributions, ISI histograms,
and neural autocorrelograms (i.e., all-order ISI histograms). The ordinates of the autocorrelograms are scaled in terms of firing rate as
described by Abeles (1982), and each autocorrelogram displays three
horizontal reference lines. The middle horizontal line shows the
average firing rate obtained from the spike count distributions; the
upper and lower horizontal lines show the average rate 62 SDs. These
horizontal lines only serve as visual aids and are not meant to imply
statistical significance. For comparison, temporal discharge properties
were occasionally obtained in response to harmonic tone complexes
consisting of the fundamental frequency and all harmonic frequencies
#10 kHz. This produced tone complexes having bandwidths similar
to the bandwidths of the IIRNs. Tone complexes were synthesized by
adding individual components in cosine phase, sine phase, or random
phase. The rms amplitude of each tone complex was adjusted to equal
the rms amplitudes of the WBN and IIRN stimuli.
among the various subcategories of chopper units was made,
all chopper units were grouped together for the present analysis. A total of nine units were grouped broadly as onset units.
Onset units in this sample typically showed some low rate of
discharge during the duration of the tone stimulus and showed
PST histograms similar in shape to on-A or on-P units described by Bourk (1976). Units were grouped as onset if the
firing rate between 20 and 50 ms was ,100 spike/s (i.e., mean
interspike interval was .10 ms) similar to the scheme described by Blackburn and Sachs (1989). An additional nine
units having BFs ,450 Hz were grouped as phase-locked;
these units showed strong BF phase-locking, and a characteristic discharge pattern could not clearly be observed when BF
tones were presented with random starting phases. Finally, a
total of seven additional units could not be classified easily into
any of the preceding groups and are grouped as unusual.
Prepotentials were not observed in the spike waveform for any
nonprimarylike unit.
RESULTS
Classification of unit types
The responses to IIRN stimuli were studied in a total of 86
single units. A total of 26 units were grouped broadly as
primarylike; this sample includes both primarylike (n 5 20)
and primarylike with notch (n 5 6) subcategories of units. Of
these units, a total of 10 units were observed to have a prepotential in the averaged spike waveform. A total of 35 units were
grouped broadly as chopper units.1 Although a distinction
1
Of these 35 chopper units, 20 units were classified initially as chopper and
15 were classified initially as on-G (i.e., onset-gradual). The PST histograms of
these on-G units did not show clear multiple peaks characteristic of chopping,
but rather were more similar in shape to PST histograms of primarylike or
onset-enhanced primarylike units (Palmer et al. 1986). However, the interspike
interval (ISI) histograms of these on-G units were more similar to ISI histograms of chopper units (see Fig. 8, C and D). A single-factor ANOVA showed
that there was a significant difference across the mean coefficients of variation
(CV) of the ISI histograms for primarylike, chopper (n 5 20), and on-G (n 5
15) units; the mean CVs were 0.66, 0.42, and 0.41, respectively. Paired
Temporal responses to IIRN stimuli
The effect of t on the temporal discharge pattern of a
primarylike unit (Fig. 3) in response to IIRN[1, t, 21 dB] is
illustrated in Fig. 4, A–C. There are peaks in the neural autocorrelograms that occur at time lags of t and integer multiples
of t for each of the IIRN[1, t, 21 dB] conditions. In response
to IIRN[1, 2 ms, 21 dB], there are peaks in the autocorrelogram at time lags of 2 ms and at integer multiples of 2 ms (Fig.
4A); in response to IIRN[1, 4 ms, 21 dB], there are peaks in
the autocorrelogram at time lags of 4 ms and at integer multiples of 4 ms (Fig. 4B); in response to IIRN[1, 8 ms, 21 dB],
comparisons subsequently were made using the Tukey’s HSD test. At the a 5
0.01 level, there was a significant difference in the means between primarylike
versus chopper units and a significant difference in the means between primarylike and on-G units. There was no significant difference in the means
between chopper and on-G units. Therefore, it was concluded that the on-G
units were not a separate group of units but should be classified as chopper
units.
2666
W. P. SHOFNER
FIG. 4. Neural autocorrelograms for the primarylike unit in Fig. 3 obtained in response to IIRN[1, 2
ms, 21 dB] (A), IIRN[1, 4 ms, 21 dB] (B), IIRN[1,
8 ms, 21 dB] (C), IIRN[2, 2 ms, 21 dB] (D),
IIRN[2, 4 ms, 21 dB] (E), and IIRN[2, 8 ms, 21
dB] (F). Spike times between 20 and 500 ms are used
for generating the autocorrelograms. Binwidth is 100
ms. Ordinate is expressed in terms of firing rate by
dividing the number of occurrences within a given
bin by the product of the total number of spikes and
binwidth. Three horizontal lines in each autocorrelogram indicate the average firing rate 6 2 SDs estimated between 20 and 500 ms. Arrows indicate the
peaks occurring at time lags of t (A–C) and 2t (D–F).
Dotted vertical line in D–F indicates the delays of the
IIRN stimuli.
there are peaks in the autocorrelogram at time lags of 8 ms and
at integer multiples of 8 ms (Fig. 4C). Note that these peaks are
narrow and that there are oscillations in firing rate that occur
around t and each integer multiple of t.
Figure 4, D–F, shows the autocorrelograms of the primarylike unit obtained in response to IIRN[2, t, 21 dB] with
delays of 2, 4, and 8 ms. In contrast to the autocorrelograms
obtained for IIRN[1, t, 21 dB], there are nulls in the IIRN[2,
t, 21 dB] autocorreolograms at time lags corresponding to t
and at odd-integer multiples of t. For example, when t is 4 ms,
there is a null in the autocorrelogram at a time lag of 4 ms (Fig.
4E). Note that there is also a pair of positive peaks at time lags
of ;3.5 and 4.5 ms that flank these nulls. The pair of peaks
around the null at t reflect the tuning properties of the unit and
are not directly related to the pitch of the stimulus. The peaks
are further apart for low BF units and are closer together for
high BF units. In addition to this pair of peaks, there is also a
peak in the autocorrelogram at a time lag of 8 ms (i.e., at 2t).
Consequently, there are nulls at time lags of t and odd-integer
multiples of t but peaks at time lags of even-integer multiples
of t. Similar results are observed in the autocorrelograms in
response to IIRN[2, 2 ms, 21 dB] (Fig. 4D) and IIRN[2, 8
ms, 21 dB] (Fig. 4F). Comparison of these neural autocorrelograms to the stimulus autocorrelograms (Fig. 2) suggests that
the temporal discharge properties of the primarylike unit are
driven by the waveform fine structure.
The effect of t on the temporal discharge pattern of a high
BF primarylike unit in response to WBN and IIRN[6, t, 21
dB] is illustrated in Fig. 5. Note that this unit has a prepotential
in its spike waveform (Fig. 5A, inset). Similar to the preceding
primarylike unit, the autocorrelograms of this high BF primarylike unit show peaks at t and integer multiples of t in
response to IIRN[1, t, 21 dB] (Fig. 5, C and D). However, in
contrast to the previous primarylike unit, the autocorrelograms
in response to IIRN[2, t, 21 dB] also show peaks at t and
integer multiples of t (Fig. 5, E and F). That is, nulls at time
lags of t and odd-integer multiples of t are not observed for the
high BF primarylike unit. Comparison of these neural autocorrelograms with stimulus autocorrelograms (Fig. 2) suggests
that the temporal discharge properties of this primarylike unit
are driven by the stimulus envelope.
The effect of t on the temporal discharge pattern of a
transient-chopper unit (Fig. 6) in response to IIRN[1, t, 21
dB] is illustrated in Fig. 7, A and B. The temporal discharge
pattern of the chopper unit is similar to that described above for
the primarylike unit in that there are peaks at t and integer
multiples of t in response to IIRN[1, t, 21 dB]. When t is 4
ms, there are peaks in the autocorrelogram at time lags of 4 ms
and at integer multiples of 4 ms (Fig. 7A); when t is 8 ms, there
are peaks in the autocorrelogram at time lags of 8 ms and at
integer multiples of 8 ms (Fig. 7B). The temporal discharge of
the chopper unit in response to IIRN[2, t, 21 dB] is shown in
Fig. 7, C and D. The autocorrelograms in response to IIRN [2,
t, 21 dB] are similar to those obtained for IIRN [1, t, 21 dB]
when t is 4 and 8 ms. In particular, in response to IIRN [2, 4
ms, 21 dB], there are no nulls observed at a time lag of 4 ms,
CN RESPONSES TO ITERATED RIPPLED NOISES
2667
FIG. 5. Responses of a primarylike unit
having a BF of 4.63 kHz. A: PST histogram
obtained in response to BF tone bursts at 20
dB above threshold. Inset: averaged spike
waveform obtained from 1,011 spikes. 4,
presence of a prepotential; —, 0.5 ms. B–F:
autocorrelograms obtained in response to
WBN (B), IIRN[1, 8 ms, 21 dB] (C),
IIRN[1, 4 ms, 21 dB] (D), IIRN[2, 8 ms,
21 dB] (E), and IIRN[2, 4 ms, 21 dB] (F).
4, peaks occurring at time lags of t.
but rather there are peaks at a time lag of 4 ms and integer
multiples of 4 ms (Fig. 7C). Similar peaks at time lags of 8 ms
and integer multiples of 8 ms also are observed in the autocorrelograms in response to IIRN[2, 8 ms, 21 dB] (Fig. 7D).
Example autocorrelograms in response to IIRN[1, t, att] and
IIRN[2, t, att] stimuli obtained from 1 of the 15 units classified as a chopper1 based on its ISI histogram but having a PST
histogram more similar to a primarylike pattern (Fig. 8) are
shown in Fig. 9. Again there are peaks in the autocorrelograms
at time lags of 8 ms and at integer multiples of 8 ms in response
to both IIRN[1, 8 ms, 21 dB] and IIRN[2, 8 ms, 21 dB]
(Fig. 9, A and C); there are peaks in the autocorrelogram at
time lags of 4 ms and at integer multiples of 4 ms in response
to both IIRN[1, 4 ms, 21 dB] and IIRN[2, 4 ms, 21 dB]
(Fig. 9, B and D). Comparison of these neural autocorrelograms with stimulus autocorrelograms (Fig. 2) suggests that
the temporal discharge properties of the chopper units are
driven by the stimulus envelope.
The temporal responses to IIRN[1, 4 ms, 21 dB] and to
harmonic tone complexes having a fundamental frequency
of 250 Hz (i.e., period of 4 ms) for an onset unit are shown
in Fig. 10. It can be observed clearly that neither IIRN[1, 4
ms, 21 dB] (Fig. 10A) nor the random phase harmonic tone
complex (Fig. 10B) evoked an excitatory response from the
unit. In contrast, a large excitatory response can be observed
to both the harmonic tone complex added in cosine and sine
phase (Fig. 10, C and D). Although 4-ms ISIs were not
obtained, the autocorrelograms (Fig. 10, E and F) show that
the temporal discharge pattern is related to the periodicity of
the tone complex. Figure 11 shows the autocorrelograms of
another onset unit in response to WBN, IIRN[1, 4 ms, 21
dB] and harmonic tone complexes. Although this unit gave
an excitatory response, the autocorrelograms show no temporal features at a time lag of 4 ms in response to either
IIRN[1, 4 ms, 21 dB] (Fig. 11C) or a cosine phase harmonic tone complex with a period of 4 ms (Fig. 11D). In
response to a random phase harmonic tone complex with a
period of 20 ms, there is a weak peak at a time lag of 20 ms
in the autocorrelogram (Fig. 11E). Note the general similarity in the shape among the autocorrelograms for WBN,
IIRN[1, 4 ms, 21 dB] and these two harmonic tone complexes. In contrast, the autocorrelogram shows that there is
strong phase-locking to the period of 20 ms obtained
in response to a cosine phase harmonic tone complex with a
period of 20 ms (Fig. 11F). Thus a strong synchronous
response from this unit is driven by a stimulus having
a highly modulated envelope but a low fundamental frequency.
2668
W. P. SHOFNER
FIG. 6. Discharge characteristics of a chopper unit
having a BF of 3.1 kHz. A: PST histogram obtained in
response to 250 50-ms BF tone bursts presented at 20
dB above threshold. B–D, inset: same as Fig. 3.
Quantitative analysis of temporal responses to
IIRN[1, 4 ms, 21 db]: peak height
The temporal properties of IIRN stimuli can be quantified by
measuring the heights of the peaks at a time lag of t in the
stimulus autocorrelation functions. The temporal discharge
patterns of units displayed in neural autocorrelograms can be
quantified in an analogous manner by estimating the heights of
the peaks at a time lag corresponding to t. Note that in all of
the neural autocorrelograms, the peak heights are expressed in
units of firing rate; consequently, the firing rates at a particular
time lag in the autocorrelogram will be referred to as instantaneous firing rates. The instantaneous firing rate at t is defined
as the maximum firing rate in the autocorrelogram in a 1-ms
window centered at a time lag of t. To compare across units
having different average firing rates, the instantaneous firing
rates at a time lag of t were measured in terms of normalized
rate at a particular time lag as given by
Normalized Rate 5
(R t 2 R ave )
R ave
(1)
where Rt is the instantaneous firing rate at time lag t and Rave
is the average firing rate. If the firing rate at t is equal to the
average rate, then the normalized rate at t is 0; normalized rate
is positive if the firing rate at t is above the average rate and is
negative if the firing rate at t is less than the average rate.
Normalized rate is 21 when the firing rate at t is 0.
Figure 12 compares the normalized rates at a time lag of 4
ms for WBN and IIRN[1, 4 ms, 21 dB] for the primarylike
FIG. 7. Autocorrelograms for the chopper unit
in Fig. 6 obtained in response to IIRN[1, 4 ms, 21
dB] (A), IIRN[1, 8 ms, 21 dB] (B), IIRN[2,
4 ms, 21 dB] (C), and IIRN[2, 8 ms, 21 dB]
(D). 4, peaks occurring at time lags of t.
CN RESPONSES TO ITERATED RIPPLED NOISES
2669
FIG. 8. Discharge characteristics of a chopper unit
having a BF of 2.43 kHz. A: PST histogram obtained
in response to 250 50-ms BF tone bursts presented at
40 dB above threshold. B–D, inset: same as Fig. 3.
Note that there is no clear chopping pattern in the PST
histogram (A and B), but there are clear differences in
the regularity analysis (D) between this unit and the
primarylike unit in Fig. 3D.
and chopper units sampled in this study. The normalized rates
for WBN also were measured as the maximum firing rate in a
1-ms window centered around the time lag of 4 ms and are
shown for comparison. Note that for primarylike units (Fig.
12A) the normalized rates obtained in response to IIRN[1, 4
ms, 21 dB] are generally larger than the normalized rates for
WBN, whereas this does not appear to be the case for chopper
units (Fig. 12B). Although there is scatter in the data, it can be
observed that for the primarylike/phase-locked units, there are
no normalized rates measured at a 4-ms lag having negative
values for IIRN[1, 4 ms, 21 dB], whereas there are normalized rates having negative values for the chopper units (see Fig.
12B, ). In response to IIRN[1, 4 ms, 21 dB], there were 0/51
(0%) negative normalized rates observed for primarylike/
phase-locked group (Fig. 12A), there were 14/55 (25.4%) negative normalized rates observed for chopper group (Fig. 12B),
there were 7/15 (46.7%) negative normalized rates observed
for onset group, and there were 6/10 (60%) negative normalized rates observed for unusual group.
The normalized rates obtained in response to IIRN[1, 4 ms,
21 dB] for a sample of 15 autocorrelograms from 11/20
primarylike units and 12 autocorrelograms from 5/6 primarylike with notch units were ranked. These primarylike and
primarylike with notch units had a similar range of BFs between 0.85 and 5.44 kHz. Based on a Mann-Whitney U test,
the normalized rates for the primarylike and primarylike notch
units were the same. Therefore the data for primarylike and
primarylike with notch units were pooled. Figure 13 shows the
average normalized rates obtained for both WBN and IIRN[1,
4 ms, 21 dB] for primarylike and chopper units having BFs
between 0.64 and 1.6 kHz (left) and BFs .1.6 kHz (right). A
single-factor ANOVA showed that there was a significant
FIG. 9. Autocorrelograms for the chopper unit in
Fig. 8 obtained in response to IIRN[1, 8 ms, 21 dB]
(A), IIRN[1, 4 ms, 21 dB] (B), IIRN[2, 8 ms, 21
dB] (C), and IIRN[2, 4 ms, 21 dB] (D). 4, peaks
occurring at time lags of t.
2670
W. P. SHOFNER
FIG. 10. PST histograms for an onset unit having a BF of 4.4 kHz obtained in response to
IIRN[1, 4 ms, 21 dB] (A), a 40-component random
phase harmonic tone complex with a 250-Hz fundamental (B), a 40-component cosine phase harmonic tone complex with a 250-Hz fundamental
(C), and a 40-component sine phase harmonic tone
complex with a 250-Hz fundamental (D). E and F:
autocorrelograms in response to 40-component harmonic tone complex with a 250-Hz fundamental
with components added in cosine phase (E) and sine
phase (F). Inset: PST histogram obtained in response to BF tone bursts at 20 dB above threshold.
difference across the eight groups of WBN and IIRN conditions for primarylike and chopper units [F(7,166)515.86; P ,
0.0005]. Onset and unusual units were not included in the
analysis because there were relatively few responses in each of
the above groups. Paired comparisons were subsequently made
using the Tukey’s HSD test and are summarized in Table 1.
For the IIRN[1, 4 ms, 21 dB] conditions, there is a significant
difference between the mean of the primarylike and the mean
of the chopper units at a 5 0.01 for units with BFs between
0.64 and 1.6 kHz as well as for units with BFs .1.6 kHz.
Paired comparisons between the responses to WBN and
IIRN[1, 4 ms, 21 dB] within a given unit type show that there
is a significant difference at a 5 0.01 between the means for
WBN and IIRN[1, 4 ms, 21 dB] for primarylike units with
BFs between 0.64 and 1.6 kHz and for primarylike units with
BFs .1.6 kHz.
Quantitative analysis of temporal responses to
IIRN[6, t, 21 dB]: peak location
As previously described, there appears to be a tendency that
the autocorrelograms of low BF primarylike units show a peak
at a time lag of t in response to IIRN[1, t, 21 dB] but show
a peak at 2t in response to IIRN[2, t, 21 dB], whereas the
autocorrelograms of chopper units and high BF primarylike
units appear to show a peak at a time lag of t in response to
both IIRN[1, t, 21 dB] and IIRN[2, t, 21 dB]. To quantify
this, it was determined whether the time lag of the first or
largest peak in the neural autocorrelogram occurred at t or 2t
for each response to IIRN stimuli, provided that there was a
clear temporal response to the IIRN and that data were obtained for both IIRN[1, t, 21 dB] and IIRN[2, t, 21 dB] for
the same unit.
Figure 14 shows the number of responses occurring at a time
lag of either t or 2t for primarylike and chopper units. For
primarylike units, 28/29 (96.6%) of the peaks were at a time
lag of t in response to IIRN[1, t, 21 dB]. Only one response
was observed at a time lag of 2t; no peak at t was observed in
this autocorrelogram. In response to IIRN[2, t, 21 dB], 8/29
(27.6%) of the peaks were located at a time lag of t; the ISI
histograms of these primarylike units did not show any evidence of phase-locking to BF tones. In contrast, 20/29 (70%)
were observed at 2t in response to IIRN[2, t, 21 dB]; the ISI
histograms of these primarylike units showed phase-locking to
BF tones. One response was observed at 4t; this was the same
unit that showed a peak at 2t for IIRN[1, t, 21 dB] described
above. For chopper units, 12/16 (75%) of the peaks were
located at a time lag of t, whereas 4/16 (25%) were located at
2t in response to IIRN[1, t, 21 dB]. In response to IIRN[2,
CN RESPONSES TO ITERATED RIPPLED NOISES
2671
FIG. 11. Responses of an onset unit having a BF
of 1.75 kHz. A: PST histogram obtained in response
to BF tone bursts at 20 dB above threshold. B–F:
autocorrelograms obtained in response to WBN (B),
IIRN[1, 4 ms, 21 dB] (C), a 40-component random
phase harmonic tone complex with a 250-Hz fundamental; i.e., 4-ms period (D), a 200-component random phase harmonic tone complex with a 50-Hz
fundamental; i.e., 20-ms period (E), and a 200-component cosine phase harmonic tone complex with a
50-Hz fundamental (F).
t, 21 dB], 11/16 (68.8%) were located at time lags of t and
5/16 (31.2%) were at time lags of 2t. To determine whether
there was a significant difference in the location of the peaks
FIG.
12. Scatter diagrams showing normalized rate as a function of BF
obtained in response to IIRN[1, 4 ms, 21 dB] () and WBN (E) for phaselocked and primarylike units (A) and chopper units (B). Normalized rate was
estimated at a time lag of 4 ms.
for primarylike and chopper units, contingency tables were
generated and a x2 analysis was carried out. The location of the
peak in the neural autocorrelogram is not the same in response
to IIRN[1, t, 21 dB] and IIRN[2, t, 21 dB] for primarylike
units (x2 5 29.3; P , 0.001). In contrast, the location of the
peak in the autocorrelogram is the same in response to IIRN[1,
t, 21 dB] and IIRN[2, t, 21 dB] for chopper units (x2 5 0;
not significant at a 5 0.05).
FIG. 13. Bar graphs showing the average normalized rate of units having
BFs between 0.64 and 1.6 kHz (left) and BFs .1.6 kHz (right) for primarylike
units (pri), chopper units (chop) in response to IIRN[1, 4 ms, 21 dB] (h) and
WBN (■). Error bars indicate the 95% confidence intervals. Numbers above
the bars indicate the number of normalized rates.
2672
W. P. SHOFNER
TABLE 1. Summary of the paired comparisons based on Tukey’s
HSD test
Unit Type, BF, Stimulus Condition
Primarylike, 0.64–1.6 kHz, WBN
vs. Chopper, 0.64–1.6 kHz, WBN
Primarylike, .1.6 kHz, WBN vs.
Chopper, .1.6 kHz, WBN
Primarylike, 0.64–1.6 kHz, IIRN vs.
Chopper, 0.64–1.6 kHz, IIRN
Primarylike, .1.6 kHz, IIRN vs.
Chopper, .1.6 kHz, IIRN
Primarylike, 0.64–1.6 kHz, IIRN vs.
Primarylike, .1.6 kHz, IIRN
Chopper, 0.64–1.6 kHz, IIRN vs.
Chopper, .1.6 kHz, IIRN
Primarylike, 0.64–1.6 kHz, IIRN vs.
Primarylike, 0.64–1.6 kHz, WBN
Primarylike, .1.6 kHz, IIRN vs.
Primarylike, .1.6 kHz, WBN
Chopper, 0.64–1.6 kHz, IIRN vs.
Chopper, 0.64–1.6 kHz, WBN
Chopper, .1.6 kHz, IIRN vs.
Chopper, .1.6 kHz, WBN
Significance
Not significant
Not significant
Significant at a 5 0.01
Significant at a 5 0.01
Not significant
Not significant
Significant at a 5 0.01
Significant at a 5 0.01
Not significant
Not significant
The particular paired comparison is indicated in bold italics. WBN, wideband noise; IIRN, infinitely iterated rippled noise.
DISCUSSION
The responses of primarylike and chopper units in the chinchilla cochlear nucleus were measured to the same IIRN stimuli that have been studied previously in psychophysical experiments using chinchillas (Shofner and Yost 1995) and human
listeners (Fastl 1988; Raatgever and Bilsen 1992). Temporal
responses of single units were studied using autocorrelograms
or all-order ISI histograms. Neural autocorrelograms show the
probability of discharge after a given spike; that is, they show
the average firing pattern of a unit after a spike. Autocorrelation analysis provides a robust representation of the pitch
related information encoded in the spike trains of auditory
nerve fibers (Cariani and Delgutte 1996a,b; Palmer 1992). The
present results show that a temporal representation of t can be
obtained in neural autocorrelograms in response to IIRN[6, t,
21 dB] from both primarylike and chopper units. To gain
some insight into the relative importance of primarylike and
chopper units, it is informative to make comparisons between
the neurophysiological data and behavioral results with respect
to IIRN pitch and coloration discrimination.
correlation functions. Autocorrelograms of some chopper
units, as well as nonphase-locked primarylike units show peaks
at time lags of t and integer multiples of t in response to both
IIRN[1, t, 21 dB] and IIRN[2, t, 21 dB]. That is, the
temporal discharge patterns of these units are more similar to
the stimulus envelope autocorrelation function than the waveform autocorrelation function.
x2 analysis of the location of the first or largest peak in the
autocorrelogram shows that for primarylike units, there is a
difference in the responses to IIRN[1, t, 21 dB] and IIRN[2,
t, 21 dB], but the difference does not exist for chopper units.
Thus a discrimination between IIRN[1, t, 21 dB] and
IIRN[2, t, 21 dB] can be made based on temporal discharge
patterns of primarylike units, whereas a discrimination between IIRN[1, t, 21 dB] and IIRN[2, t, 21 dB] cannot be
made on the basis of the temporal discharge properties of
chopper units. Current models that account for the pitch and
pitch strength of iterated rippled noises are based on temporal
processing of the waveform fine structure (Patterson et al.
1996; Yost 1996a,b, 1997; Yost et al. 1996), and the present
study shows that the temporal discharge patterns of primarylike
units are driven by the waveform fine structure. More recent
psychophysical data argue that the pitch perception of iterated
rippled noises cannot be based on stimulus envelope (Yost et
al. 1998), and the present results show that the temporal discharge patterns of chopper units are driven by the stimulus
envelope of IIRN. This finding is consistent with the idea that
chopper units are principal subsystems that encode stimulus
envelope in their temporal discharge (see Greenberg and
Rhode 1987).
Coloration discrimination
At the present time, it is not known whether chinchillas can
discriminate IIRN[1, t, 21 dB] from IIRN[2, t, 21 dB].
IIRN pitch
The perceived pitch of IIRN[1, 4 ms, 21 dB] corresponds
to 1/t or 250 Hz, whereas the pitch of IIRN[2, 4 ms, 21 dB]
corresponds to 1/2t or 125 Hz (Fastl 1988; Raatgever and
Bilsen 1992). In other words, human listeners can discriminate
IIRN[1, 4 ms, 21 dB] from IIRN[2, 4 ms, 21 dB] based on
pitch. Autocorrelograms of some primarylike units show peaks
at time lags of t and integer multiples of t in the neural
autocorrelograms in response to IIRN[1, t, 21 dB], but nulls
at times lags of t followed by a peak at 2t in the autocorrelograms in response to IIRN[2, t, 21 dB]. These primarylike
units typically show phase-locking to BF tones. Thus the
temporal discharge patterns of this phase-locked group of
primarylike units are similar to the stimulus waveform auto-
FIG. 14. Bar graphs showing the number of occurrences that the peak in the
neural autocorrelograms occurred at a time lag of t or 2t for primarylike and
phase-locked units (A) and chopper units (B). Responses to IIRN[1, t, 21 dB]
are shown by ■; responses to IIRN[2, t, 21 dB] are shown by h.
CN RESPONSES TO ITERATED RIPPLED NOISES
Behavioral studies using chinchillas (Shofner and Yost 1995,
1997) have been based on coloration discrimination paradigms
(see Bilsen and Ritsma 1970). In the coloration discrimination
experiment, the listener discriminates IIRN from WBN, and
chinchillas can easily discriminate IIRN[1, 4 ms, 21 dB] from
WBN (Shofner and Yost 1995, 1997). In the present study, the
magnitudes of the temporal responses were compared between
IIRN[1, 4 ms, 21 dB] and WBN. The response to WBN
provides insights into the temporal response to a noise that
does not generate a pitch and then can be compared with the
temporal response to a noise that does generate a relatively
salient pitch. For human listeners, the pitch strength of
IIRN[1, 4 ms, 21 dB] is ;50 – 60% of the pitch strength of a
harmonic tone complex (see Yost 1996b).
The magnitude of the neural response was estimated as the
normalized rate at t milliseconds from the neural autocorrelograms. The peak at t is generally the largest t-related peak in
the autocorrelogram. Yost (1996b) has shown that the pitch
strength of IIRN[1, t, att] can be accounted for using a model
based on the height of the first peak in the autocorrelation
function. The average normalized rate at a time lag of 4 ms in
response to IIRN[1, 4 ms, 21 dB] is larger for the sample of
primarylike units than that of the chopper units, whereas the
average normalized rate at a time lag of 4 ms is the same for
primarylike and chopper units in response to WBN. Moreover,
the average normalized rates at 4 ms are larger for primarylike
units in response to IIRN[1, 4 ms, 21 dB] than in response to
WBN, whereas the average normalized rates at 4 ms for
chopper units in response to IIRN[1, 4 ms, 21 dB] are not
different from those obtained in response to WBN. However,
for some units the largest peak in the neural autocorrelogram is
observed at a time lag of 2t rather than at t (e.g., see Fig. 9, B
and D). Peaks at 2t also should convey information about pitch
strength. A single-factor ANOVA and subsequent paired comparisons based on either the normalized rates at 2t or an
average of the normalized rates at t and 2t also showed that the
responses of primarylike units are larger than chopper units.
Moreover including the data at 2t does not make the difference
between IIRN and WBN statistically significant for chopper
units. Thus the largest t-related temporal responses for IIRN
are those associated with primarylike units not chopper units.
Chopper units do show an enhancement of synchronization
compared with primarylike units in response to the fundamental frequency of SAM tones, complex tones and synthetic
vowels (Frisina et al. 1990; Keilson et al. 1997; Rhode 1998;
Rhode and Greenberg 1994; Wang and Sachs 1994). Although
the periodic stimuli that have been used in the preceding
studies do generate the perception of pitch in human listeners,
these stimuli also possess highly modulated stimulus envelopes. The enhanced synchronization to the fundamental frequency observed for chopper units in response to periodic
complex sounds having highly modulated envelopes is not
found for their temporal responses to IIRN stimuli.
Conclusion
The results of the present study show that primarylike units
give larger temporal responses than chopper units to IIRN, and
the temporal responses of primarylike units can account for the
differences in pitch between IIRN[1, t, 21 dB] and IIRN[2,
t, 21 dB], whereas those of chopper units cannot. However, it
2673
cannot yet be concluded that primarylike units are the principal
cochlear nucleus subsystem that encodes the pitch related
information in their temporal discharge. The present study did
not sample onset-type-I (on-I) and onset-chopper units, which
have been suggested to be important cochlear nucleus subsystems that encode pitch-related information in their temporal
discharge patterns (Greenberg and Rhode 1987; Kim and Leonard 1988; Kim et al. 1986). Recent studies have shown that
on-I units can show strong synchronization to the fundamental
frequency of a tone complex in which components were added
in cosine phase but weak or no synchronization to random
phase tone complexes (Evans and Zhao 1998), suggesting that
on-I units may not give strong temporal responses to IIRN
stimuli. Nevertheless, the results of the present study suggest
that the contribution of primarylike units in the temporal encoding of pitch-related information is not insignificant. Although primarylike units may not be the principal subsystem,
they are likely to be an important cochlear nucleus subsystem
underlying the perception of pitch.
The author thanks R. Patka for assistance in the data collection of some of
these experiments and Drs. W. A. Yost and R. H. Dye, Jr., and two anonymous
reviewers for comments on the manuscript.
This research was supported by National Institute of Deafness and Other
Communications Disorders Program Project Grant (P01 DC-00293).
Address for reprint requests: Parmly Hearing Institute, Loyola University
Chicago, 6525 N. Sheridan Rd., Chicago, IL 60626.
Received 1 September 1998; accepted in final form 9 February 1999.
REFERENCES
ABELES, M. Quantification, smoothing, and confidence limits for single-units’
histograms. Hear. Res. 5: 317–325, 1982.
BILSEN, F. A. AND RITSMA, R. J. Some parameters influencing the perceptibility
of pitch. J. Acoust. Soc. Am. 47: 469 – 475, 1970.
BILSEN, F. A., TEN KATE, J. H., BUUNEN, T.J.F., AND RAATGEVER, J. Responses
of single units in the cochlear nucleus of the cat to cosine noise. J. Acoust.
Soc. Am. 58: 858 – 866, 1975.
BLACKBURN, C. C. AND SACHS, M. B. Classification of unit types in the
anteroventral cochlear nucleus: PST histograms and regularity analysis.
J. Neurophysiol. 62: 1303–1329, 1989.
BOURK, T. R. Electrical Responses of Neural Units in the Anteroventral
Cochlear Nucleus of the Cat (PhD thesis). Cambridge, MA: MIT, 1976.
CARIANI, P. A. AND DELGUTTE, B. Neural correlates of the pitch of complex tones. I. Pitch and pitch salience. J. Neurophysiol. 76: 1698 –1716,
1996a.
CARIANI, P. A. AND DELGUTTE, B. Neural correlates of the pitch of complex
tones. II. Pitch shift, pitch ambiguity, phase invariance, pitch circularity, rate
pitch, and the dominance region for pitch. J. Neurophysiol. 76: 1717–1734,
1996b.
DOWBEN, R. M. AND ROSE, J. E. A metal-filled microelectrode. Science 118:
22–24, 1953.
EVANS, E. F. AND ZHAO, W. Periodicity coding of the fundamental frequency
of harmonic complexes: physiological and pharmacological study of onset
units in the ventral cochlear nucleus. In: Psychophysical and Physiological
Advances in Hearing, edited by A. R. Palmer, A. Rees, A. Q. Summerfield,
and R. Meddis. London: Whurr Publishers, 1998, p. 186 –194.
FASTL, H. Pitch and pitch strength of peaked rippled noise. In: Basic Issues in
Hearing, edited by H. Duifhuis, J. W. Horst, and H. P. Wit. San Diego:
Academic, 1988, p. 370 –379.
FASTL, H. AND STOLL, G. Scaling of pitch strength. Hear. Res. 1: 293–301,
1979.
FRISINA, R. D., CHAMBERLAIN, S. C., BRACHMAN, M. L., AND SMITH, R. L.
Anatomy and physiology of the gerbil cochlear nucleus: an improved
surgical approach for microelectrode studies. Hear. Res. 6: 259 –275, 1982.
FRISINA, R. D., SMITH, R. L., AND CHAMBERLAIN, S. C. Encoding of amplitude
modulation in the gerbil cochlear nucleus. I. A hierarchy of enhancement.
Hear. Res. 44: 99 –122, 1990.
2674
W. P. SHOFNER
GREENBERG, S. AND RHODE, W. S. Periodicity coding in cochlear nerve and
ventral cochlear nucleus. In: Auditory Processing of Complex Sounds, edited
by W. A. Yost and C. S. Watson. Hillsdale, NJ: Lawrence Erlbaum, 1987,
p. 225–236.
JAVEL, E. Coding of AM tones in the chinchilla auditory nerve: implications
for the pitch of complex tones. J. Acoust. Soc. Am. 68: 133–146, 1980.
JORIS, P. X. AND YIN, T.C.T. Responses to amplitude-modulated tones in the
auditory nerve of the cat. J. Acoust. Soc. Am. 91: 215–232, 1992.
KEILSON, S. E., RICHARDS, V. M., WYMAN, B. T., AND YOUNG, E. D. The
representation of concurrent vowels in the cat anesthetized ventral cochlear
nucleus: evidence for a periodicity-tagged spectral representation. J. Acoust.
Soc. Am. 102: 1056 –1071, 1997.
KHANNA, S. M. AND TEICH, M. C. Spectral characteristics of the responses of
primary auditory-nerve fibers to amplitude-modulated tones. Hear. Res. 39:
143–158, 1989.
KIM, D. O. AND LEONARD, G. Pitch-period following response of cat cochlear
nucleus neurons to speech sounds. In: Basic Issues in Hearing, edited by H.
Duifhuis, J. W. Horst, and H. P. Wit. San Diego: Academic, 1988, p.
252–260.
KIM, D. O., RHODE, W. S., AND GREENBERG, S. R. Responses of cochlear
nucleus neurons to speech signals: neural encoding of pitch, intensity and
other parameters. In: Auditory Frequency Selectivity, edited by B.C.J. Moore
and R. D. Patterson. New York: Plenum, 1986, p. 281–288.
KIM, D. O., SIRIANNI, J. G., AND CHANG, S. O. Responses of DCN-PVCN
neurons and auditory nerve fibers in unanesthetized decrebrate cats to AM
and pure tones: analysis with autocorrelation/power-spectrum. Hear. Res.
45: 95–113, 1990.
LANGER, G. Physiological properties of units in the cochlear nucleus are
adequate for a model of periodicity analysis in the auditory midbrain. In:
Auditory Pathway-Structure and Function, edited by J. Syka and R. B.
Masterton. New York: Plenum, 1988, p. 207–212.
LICKLIDER, J.C.R. A duplex theory of pitch perception. Experientia 7: 128 –
134, 1951.
MEDDIS, R. AND HEWITT, M. J. Virtual pitch and phase sensitivity of a
computer model of the auditory periphery. I. Pitch identification. J. Acoust.
Soc. Am. 89: 2866 –2882, 1991.
MEDDIS, R. AND O’MARD, L. A unitary model of pitch perception. J. Acoust.
Soc. Am. 102: 1811–1820, 1997.
PALMER, A. R. Segregation of the responses to paired vowels in the auditory
nerve of the guinea-pig using autocorrelation. In: The Auditory Processing
of Speech. From Sounds to Words, edited by M.E.H. Schouten. New York:
Mouton de Gruyter, 1992, p. 115–124.
PALMER, A. R. AND WINTER, I. M. Cochlear nerve and cochlear nucleus
responses to the fundamental frequency of voiced speech sounds and harmonic tone complexes. In: Auditory Phsyiology and Perception, edited by
Y. Cazals, K. Horner, and L. Demany. New York: Pergamon, 1992, p.
231–239.
PALMER, A. R., WINTER, I. M., AND DARWIN, C. J. The representation of
steady-state vowel sounds in the temporal discharge patterns of the guinea
pig cochlear nerve and primarylike cochlear nucleus neurons. J. Acoust. Soc.
Am. 79: 100 –113, 1986.
PATTERSON, R. D., HANDEL, S., YOST, W. A., AND DATTA, A. J. The relative
strength of the tone and noise components in iterated rippled noise. J.
Acoust. Soc. Am. 100: 3286 –3294, 1996.
PFEIFFER, R. R. Anteroventral cochlear nucleus: wave forms of extracellularly
recorded spike potentials. Science 154: 667– 668, 1966.
RAATGEVER, J. AND BILSEN, F. A. The pitch of anharmonic comb filtered noise
reconsidered. In: Auditory Physiology and Perception, edited by Y. Cazals,
K. Horner, and L. Demany. New York: Pergamon, 1992, p. 215–222, 1992.
RHODE, W. S. Temporal encoding of 200% amplitude modulated signals in the
ventral cochlear nucleus of the cat. Hear. Res. 77: 43– 68, 1994.
RHODE, W. S. Interspike intervals as a correlate of periodicity in the cat
cochelar nucleus. J. Acoust. Soc. Am. 97: 2414 –2429, 1995.
RHODE, W. S. Neural encoding of single-formant stimuli in the ventral cochlear
nucleus of the chinchilla. Hear. Res. 117: 39 –56, 1998.
RHODE, W. S. AND GREENBERG, S. Encoding of amplitude modulation in the
cochlear nucleus of the cat. J. Neurophsyiol. 71: 1797–1825, 1994.
SHOFNER, W. P. Temporal representation of rippled noise in the anteroventral
cochlear nucleus of the chinchilla. J. Acoust. Soc. Am. 90: 2450 –2466, 1991.
SHOFNER, W. P. AND YOST, W. A. Discrimination of rippled-spectrum noise
from flat-spectrum noise by chinchillas. Aud. Neurosci. 1: 127–138, 1995.
SHOFNER, W. P. AND YOST, W. A. Discrimination of rippled-spectrum noise
from flat-spectrum noise by chinchillas: evidence for a spectral dominance
region. Hear. Res. 110: 15–24, 1997.
TEN KATE, J. H. AND VAN BEKKUM, M. F. Synchrony-dependent autocorrelation
in eighth-nerve-fiber response to rippled noise. J. Acoust. Soc. Am. 84:
2092–2102, 1988.
WANG, X. AND SACHS, M. B. Neural encoding of single-formant stimuli in the
cat. I. Responses of auditory nerve fibers. J. Neurophysiol. 70: 1054 –1075,
1993.
WANG, X. AND SACHS, M. B. Neural encoding of single-formant stimuli in the
cat. II. Responses of anteroventral cochlear nucleus units. J. Neurophysiol.
71: 59 –78, 1994.
YOST, W. A. Pitch of iterated rippled noise. J. Acoust. Soc. Am. 100: 511–518,
1996a.
YOST, W. A. Pitch strength of iterated rippled noise. J. Acoust. Soc. Am. 100:
3329 –3335, 1996b
YOST, W. A. Pitch strength of iterated rippled noise when the pitch is ambiguous. J. Acoust. Soc. Am. 101: 1644 –1648, 1997.
YOST, W. A., PATTERSON, R. D., AND SHEFT, S. A time domain description for
the pitch strength of iterated rippled noise. J. Acoust. Soc. Am. 99: 1066 –
1078, 1996.
YOST, W. A., PATTERSON, R. D., AND SHEFT, S. The role of envelope in
processing iterated rippled noise. J. Acoust. Soc. Am. 104: 2349 –2361,
1998.
YOUNG, E. D., ROBERT, J.-M., AND SHOFNER, W. P. Regularity and latency of
units in ventral cochlear nucleus: implications for unit classification and
generation of response properties. J. Neurophysiol. 60: 1–29, 1988.
ZHAO, H.-B. AND LIANG, Z.-A. Processing of modulation frequency in the
dorsal cochlear nucleus of the guinea pig: amplitude modulated tones. Hear.
Res. 82: 244 –256, 1995.