neural coding of periodicity in marmoset auditory cortex

Neural Coding of Periodicity in Marmoset Auditory Cortex

Daniel Bendor and Xiaoqin WangLaboratory of Auditory Neurophysiology, Department of Biomedical Engineering, Johns Hopkins University School of Medicine,Baltimore, Maryland

Submitted 30 March 2009; accepted in final form 5 February 2010

Bendor D, Wang X. Neural coding of periodicity in marmosetauditory cortex. J Neurophysiol 103: 1809–1822, 2010. First pub-lished February 10, 2010; doi:10.1152/jn.00281.2009. Pitch, our per-ception of how high or low a sound is on a musical scale, cruciallydepends on a sound’s periodicity. If an acoustic signal is temporallyjittered so that it becomes aperiodic, the pitch will no longer beperceivable even though other acoustical features that normally co-vary with pitch are unchanged. Previous electrophysiological studiesinvestigating pitch have typically used only periodic acoustic stimuli,and as such these studies cannot distinguish between a neural repre-sentation of pitch and an acoustical feature that only correlates withpitch. In this report, we examine in the auditory cortex of awakemarmoset monkeys (Callithrix jacchus) the neural coding of a peri-odicity’s repetition rate, an acoustic feature that covaries with pitch.We first examine if individual neurons show similar repetition ratetuning for different periodic acoustic signals. We next measure howsensitive these neural representations are to the temporal regularity ofthe acoustic signal. We find that neurons throughout auditory cortexcovary their firing rate with the repetition rate of an acoustic signal.However, similar repetition rate tuning across acoustic stimuli andsensitivity to temporal regularity were generally only observed in asmall group of neurons found near the anterolateral border of primaryauditory cortex, the location of a previously identified putative pitchprocessing center. These results suggest that although the encoding ofrepetition rate is a general component of auditory cortical processing,the neural correlate of periodicity is confined to a special class ofpitch-selective neurons within the putative pitch processing center ofauditory cortex.

I N T R O D U C T I O N

An essential function of the auditory system is to extract andencode the time-varying features of acoustic signals. Acousticparameters like modulation frequency and envelope repetitionrate are important information-bearing components for vocalcommunication across many species (DiMattina and Wang2006; Rosen 1992; Singh and Theunissen 2003; Suga 1994).Moderately complex acoustic signals like sinusoidally ampli-tude modulated (sAM) tones and acoustic pulse trains havebeen commonly used in previous studies of temporal process-ing in the auditory system (Eggermont 1998; Gaese and Ost-wald 1995; Joris et al. 2004; Phan and Recanzone 2007;Schreiner and Urbas 1988). In response to sAM tones, neuronsin the auditory cortex of awake primates represent modulationfrequency using a temporal and/or rate representation, the lattertypically taking the form of a band-pass-tuned modulationtransfer function (Bendor and Wang 2008; Liang et al. 2002;Malone et al. 2007). Here we refer to neurons that covary their

firing rate with the modulation frequency or envelope repeti-tion rate of the stimulus as “modulation sensitive.” Single-unitrecordings in the primary auditory cortex (AI) of marmosetshave identified two types of modulation sensitive responsesinvolved with encoding the repetition rate of an acoustic pulsetrain (Lu et al. 2001; Wang et al. 2003). The first neuronalpopulation encodes repetition rates within the perceptual rangeof acoustic flutter (subpitch sounds with pulse rates of 10–45Hz) (Besser 1967; Miller and Taylor 1948) using both stimulussynchronization (Lu et al. 2001) and a discharge rate basedneural representation (Bendor and Wang 2007). A secondneuronal population referred to in previous studies as the“nonsynchronizing population” (Lu and Wang 2004; Lu et al.2001) encodes faster repetition rates that are within the per-ceptual range of pitch (Krumbholz et al. 2000; Plack et al.2005). The term nonsynchronizing is used to indicate that thispopulation does not synchronize its neuronal firing to theacoustic signal’s envelope and as such encodes repetition rateby only using its discharge rate.

For temporally regular sounds, like unipolar click trains andsine phase harmonic complex tones, the repetition rate gener-ally matches the perceived pitch (Plack et al. 2005). If temporalregularity is degraded, the pitch-salience decreases eventuallymaking the pitch undetectable even though the average repe-tition rate has not been changed. A putative pitch-processingcenter in the low frequency region of auditory cortex hasrecently been identified in both humans and monkeys (Bendorand Wang 2005, 2006; Patterson et al. 2002; Penagos et al.2004; Schneider et al. 2005; Schönwiesner and Zatorre 2008).The firing rates of pitch-selective neurons in this cortical regionincrease with pitch strength and are tuned to fundamentalfrequency. Pitch-selective neurons respond to both pure tonesat their best fundamental frequency and missing fundamentalsounds (composed of higher order harmonics) that have thesame pitch but lack spectral energy at their fundamental fre-quency. As such, these neurons provide a neural correlate ofcertain acoustical features that covary with pitch perception(Bendor and Wang 2005). Behavioral evidence that theseneurons are necessary for pitch perception still remains to bedemonstrated.

If neurons throughout auditory cortex can encode the repe-tition rate of complex sounds using their discharge rate, thenwhat is the difference between pitch-selective neurons andmodulation sensitive neurons? Modulation sensitive neuronsencode repetition rates for acoustic signals with carrier fre-quencies near their best frequency, whereas pitch-selectiveneurons are capable of responding to complex tones the spec-trum of which does not overlap with their best frequency whensuch a tone is modulated at the neuron’s preferred repetitionrate. (Bendor and Wang 2005). Here we compare the repetition

Present addres and address for reprint requests and other correspondence: D.Bendor, Picower Institute for Learning and Memory, Department of Brain andCognitive Sciences, Massachusetts Institute of Technology, Bldg. 46, Rm.5233, 43 Vassar St., Cambridge, MA (E-mail: [email protected]).

J Neurophysiol 103: 1809–1822, 2010.First published February 10, 2010; doi:10.1152/jn.00281.2009.

18090022-3077/10 $8.00 Copyright © 2010 The American Physiological Societywww.jn.org

on Septem

ber 16, 2010 jn.physiology.org

Dow

nloaded from

http://jn.physiology.org

rate tuning of neurons for acoustic stimuli that differ in spectralbandwidth and envelope shape and rise time. We find thatpitch-selective neurons have a more consistent best repetitionrate across different acoustic stimuli than modulation sensitiveneurons outside of the putative pitch center, indicating thatsome modulation sensitive neurons are tuned to additionalacoustic features (e.g., envelope rise time).

Another potential difference between modulation sensitiveand pitch-selective neurons is their sensitivity to changes intemporal regularity, an acoustic feature that covaries with pitchsalience (Patterson et al. 1996; Pollack 1968). Human imagingstudies have reported a neural correlate of pitch salience inlateral Heschl’s gyrus, the anatomical location of a putativepitch center in human auditory cortex (Penagos et al. 2004). Incontrast to pitch-selective neurons, neurons that encode repe-tition rate but not pitch salience may be only encoding theaverage repetition rate and lack sensitivity to temporal regu-larity. We tested this hypothesis using acoustic pulse trains, asequence of brief sounds that can be temporally jittered para-metrically between a periodic temporal structure (“regular”)and a highly random temporal structure (“irregular”). A regularpulse train generates a salient pitch percept, and the pitchsalience degrades with increasing temporal irregularity eventhough the average repetition rate remains unchanged (Kaern-bach and Demany 1998; Yost et al. 2005). Here we show thatpitch-selective neurons are sensitive to temporal regularity andthat the firing rates of modulation sensitive neurons outside theputative pitch center generally do not change between regularand irregular acoustic pulse trains with the same averagerepetition rate.

M E T H O D S

General experimental procedures

Details of experimental procedures can be found in recent publica-tions from our laboratory (Wang et al. 2005). Single-unit recordingswere conducted in four awake marmosets [M36N (right hemisphere),M2P (left and right hemisphere), M41O (left hemisphere), and M32Q(left hemisphere)]. For each experiment, the marmoset was sitting

quietly in a semi-restraint device with its head immobilized. Experi-ments were carried out within a double-walled soundproof chamber(Industrial Acoustics, Bronx, NY) with an interior covered by 3-inacoustic absorption foam (Sonex, Illbruck. High-impedance tungstenmicroelectrodes (AM Systems, 2–5 M�) were mounted on a micro-manipulator (Narishige) and advanced by a manual hydraulic micro-drive (Trent Wells). Typically 5–15 electrode penetrations were madewithin a miniature recording hole (diameter: �1 mm), after which thehole was sealed with dental cement, and another hole was opened fornew electrode penetrations. Action potentials were detected on-lineusing a template based spike sorter (MSD, Alpha Omega Engineer-ing), which was continuously monitored during the experiment. Neu-rons were recorded from all cortical layers but most commonly fromsupragranular layers. All experimental procedures were approved bythe Johns Hopkins University Animal Use and Care Committee.

Generation of acoustic stimuli

Digitally generated acoustic stimuli were played from a free-fieldspeaker located 1 m in front of the animal. All sound stimuli weregenerated at a 100 kHz sampling rate and low-pass filtered at 50 kHz.Harmonic artifacts were �43 dB lower than the fundamental at 80 dBSPL. This difference grew as the sound level of the fundamentaldecreased. The maximum sound level of individual frequency com-ponents used in this study was 80 dB SPL. The animal was awake andsemi-restrained in a custom-made primate chair but was not perform-ing a task during these experiments.

After each neuron was isolated, we measured its basic responseproperties, such as best frequency (BF) and sound level threshold.Neurons that were responsive to pure tones were tested with sAMtones, harmonic complex tones, and/or acoustic pulse trains (see Fig. 1).Complex tone, sAM tone, and acoustic pulse train stimuli were 500ms in duration. All intertrial intervals for these stimuli were �1 slong. Modulation frequency tuning functions for sAM stimuli spanned4–512 Hz in octave or half-octave steps.

ACOUSTIC PULSE TRAINS. Each pulse was generated by windowingthe preferred carrier signal (pure tones) by a Gaussian envelope. In alimited number of cases, we used other types of acoustic pulse trains,including rectangular clicks and Gaussian windowed acoustic pulseswith a broadband noise carrier. Pulse widths ranged from 0.1 to 1 msfor rectangular clicks and Gaussian pulses had a � ranging from 0.89to 2.53 ms. Pulse train repetition rates between 13 and 500 Hz were

10 15 20 25 30 35 40 45

10 15 20 25 30 35 40 45 2.6 2.8 3.0 3.2 3.4

0

-20

-40

0

-20

-40

Am

plitu

de (d

B)

Am

plitu

de (d

B)

Am

plitu

de

Am

plitu

de

Time (ms)

Time (ms)

Frequency (kHz)

Frequency (kHz)

2.6 2.8 3.0 3.2 3.4

acoustic pulse train

harmonic complex tone (cosine phase)

10 ms 100 Hz

10 15 20 25 30 35 40 45Time (ms)

Am

plitu

de sinusoidal

amplitude modulated tone

0

-20

-40Am

plitu

de (d

B)

Frequency (kHz)2.6 2.8 3.0 3.2 3.4

FIG. 1. Acoustic stimulus. Acoustic wave-form (left) and spectrum (right) of a Gaussiannarrowband acoustic pulse train (top), a miss-ing fundamental harmonic complex tone (mid-dle), and a sinusoidally amplitude modulated(sAM) tone (bottom). The acoustic pulse trainhas a 3 kHz carrier, 100 Hz repetition rate, and� � 0.89. Inset: an enlarged view of a singleGaussian pulse. The missing fundamental har-monic complex tone is composed of harmon-ics 28–32 and has a 100 Hz fundamentalfrequency (repetition rate). The sAM tone hasa modulation frequency of 100 Hz and a 3 kHzcarrier.

1810 D. BENDOR AND X. WANG

J Neurophysiol • VOL 103 • APRIL 2010 • www.jn.org

on Septem


Dow

nloaded from


used. The sound intensity level of each acoustic pulse train wasgenerally at the preferred sound level for modulation sensitive neu-rons with nonmonotonic rate-level functions and 10–30 dB abovethreshold for modulation sensitive neurons with monotonic rate-levelfunctions. For pitch-selective neurons, sound levels for acoustic pulsetrains were at or near the threshold for a tone at the neuron’s BF.Regular acoustic pulse trains had repetition rates near the preferredfundamental frequency for pitch-selective and nonmonotonic modu-lation sensitive neurons while for positive monotonic modulationsensitive neurons a repetition rate between 200 and 500 Hz waschosen. For modulation sensitive neurons, the carrier frequency wasequal to the BF, while the carrier frequency for pitch-selectiveneurons was an integer multiple of the acoustic pulse train’s repetitionrate, somewhere within the range of 1–4 kHz. Temporally regularacoustic pulse trains had interpulse intervals equal to the inverse of therepetition rate. Irregular acoustic pulse trains were created by ran-domly jittering each interpulse interval (IPI) by a random numbergenerated from a uniform distribution between [Ja, Jb], where Ja �IPI � IPI*(maximum jitter) and Jb � IPI � IPI*(maximum jitter).The maximum jitter was varied between 5 and 50% in 5% steps. Thusfor a mean interpulse interval of 10 ms and a maximum jitter of 10%,each interpulse interval would be chosen from a uniform distributionspanning values between 9 and 11 ms. For a given jitter amount, thetemporal pattern of aperiodic acoustic pulses was the same across alltrials.

HARMONIC COMPLEX TONES. The most commonly used harmoniccomplex tone had three components in cosine phase, and each com-ponent was played at the neuron’s sound level threshold measured atits BF. Sound levels 10 dB or more below BF threshold were also usedin roughly one-third of pitch-selective neurons (25/74) to evokesignificant missing fundamental responses. In a few cases we usedharmonic complex tones composed of more than three componentswith harmonics most commonly in Schroeder phase (Schroeder andStrube 1986). Neurons failed the criteria of pitch selectivity if they didnot respond to missing fundamental sounds with the individual har-monics presented at 10 dB above the neuron’s sound level thresholdat its BF (or lower sound levels). Components of the missing funda-mental harmonic complex tone (MF) were considered outside theneuron’s excitatory frequency response area if each harmonic com-ponent, when played individually at 0 and �10 dB relative to itssound level within the harmonic complex, did not evoke a significantexcitatory response. Also for neurons tested with harmonic complextones composed of greater than three components, no response toindividual components �20 dB above threshold was also required.Sound levels were varied in 10 dB steps.

Typically 10 repetitions of each acoustic pulse train, complex tone,and rate level stimulus set were presented; however, data with aminimum of five repetitions per stimulus were included in the anal-ysis. Frequency tuning curves and rate-level functions were typicallygenerated using 200 ms long pure tone stimuli and interstimulusintervals �500 ms.

Data analysis

Discharge rates �2 SD above the mean spontaneous rate and �1spike for 50% of the trials were considered significant. Vectorstrengths �0.1 with a Rayleigh statistic �13.8 (P � 0.001) wereconsidered statistically significant (Mardia and Jupp 2000). Theweighted best repetition rate, was the spike rate weighted average ofall repetition rates (in octaves) near the best repetition rate (peak firingrate) that had firing rates �50% of the peak firing rate.

CRITERIA FOR PITCH-SELECTIVE NEURONS. Pitch selectivity wasdefined as a neuron that responded to pure tones, responded to missingfundamental sounds with a fundamental frequency near its BF and didnot respond significantly to components of the missing fundamentalsound when they were played individually and the sound level of the

missing fundamental sound (measured relative to the individual com-ponents) did not need to be �10 dB above the neuron’s BF soundlevel threshold to drive the neuron (Bendor and Wang 2005). Missingfundamental sounds are complex tones containing only harmonicsabove the fundamental frequency and have the same pitch as a puretone played at the fundamental frequency. Sound levels were limitedto �10 dB above the neuron’s BF sound level threshold to avoidcombination tones (nonlinear distortion products produced by thecochlea when multiple harmonics are simultaneously played). Thecombination tone present at the fundamental frequency is �20 dBbelow the sound level of a single harmonic component (Pressnitzerand Patterson 2001).

Firing rates were calculated using the entire stimulus duration. It isimportant to note that sensitivity to temporal regularity was not usedas part of the criteria for pitch selectivity. Pitch-selective neurons werefound in each of the four marmosets used in these experiments. Aprevious publication (Bendor and Wang 2005) presented data frompitch-selective neurons recorded from three of the four marmosets(M36N, M2P, M41O) used in these experiments.

CRITERIA FOR MODULATION SENSITIVE NEURONS. Neurons that hadsignificant firing rates for at least a subset of repetition rates within therange of pitch but responded to pure tones at higher frequencies wereconsidered modulation sensitive neurons. A subset of repetition rateswas defined as at least three sequential repetition rates tested between30 and 500 Hz, spanning at least one-half octave. Discharge rateswere calculated over the duration of the acoustic stimulus plus 100 msfollowing the stimulus. In addition, modulation sensitive neuronswere also required to have significant firing rates for a subset ofrepetition rates when both the onset (1st 100 ms) and offset (100 msfollowing the end of the stimulus) were not included in the calculationof discharge rate. Without removing the onset and offset, we cannotdistinguish between sustained responses (observed in nonsynchroniz-ing neurons) and onset responses (observed in synchronizing neu-rons). Synchronizing neurons typically produce an onset response(followed by suppression) when the repetition rate is above theirsynchronization limit. In addition because some neurons had very lowspontaneous firing rates, significant responses were also required tohave one or more spikes on �50% of trials.

Our analysis only focused on neurons with significant firing ratesduring the acoustic stimulus. Neurons with purely onset responses(and nonsignificant sustained responses) to repetition rates in therange of pitch were not included in the analysis (n � 66). We did notfind any evidence of jitter sensitivity in this population of neuronsgiven that normalized firing rates between regular and irregularacoustic pulse trains were not significantly different (Wilcoxon ranksum test, P � 0.05 uncorrected, n � 10). We also did not find anyevidence of similarity in repetition rate tuning between sAM tones andacoustic pulse trains (Spearman correlation coefficient r � 0.16, P �0.46, n � 24).

The population of modulation sensitive neurons was further subdi-vided into nonsynchronized and mixed response neurons according toeach neuron’s ability to synchronize its firing to acoustic pulses in thestimulus (Rayleigh statistic �13.8, P � 0.001). Mixed responseneurons synchronized within the range of acoustic flutter (subpitchsounds with repetition rates below 40–45 Hz) while nonsynchronizedneurons did not. It is important to note that nonsynchronized neuronsdiffer from unsynchronized neurons (Bendor and Wang 2007) in thatunsynchronized neurons have significant firing rates in the range ofacoustic flutter while nonsynchronized neurons have significant firingrates in the range of pitch (�30 Hz).

In addition to a classification based on synchronization, modulationsensitive neurons were also classified according how they co-variedtheir firing rate with the repetition rate of pulses within the acousticpulse train. Positive monotonic neurons had higher discharge rates foracoustic pulse trains with higher repetition rates over the range of30–512 Hz. A nonparametric rank order correlation between dis-

1811CORTICAL PROCESSING OF PERIODICITY


on Septem


Dow

nloaded from


charge rate and repetition rate (Spearman correlation coefficient, r �0, P � 0.05) was used to determine statistically significant positive-sloped monotonicity. Because positive monotonic responses were themost commonly observed, all remaining tuning curve shapes wereclassified collectively as “other” in our analyses. This group containedboth negative monotonic and multiple types of nonmonotonic tuningcurves (see Table 1).

Five modulation sensitive neurons (2 mixed response, 3 nonsyn-chronizing) showed either a large or insignificant change in dischargerate between regular and irregular pulse trains depending on either thecarrier frequency or repetition rate used. These five neurons wereexcluded from the analysis in Fig. 14 and 15.

A normalized map of the location of pitch-selective neurons andmodulation sensitive neurons was created using data from four hemi-spheres. We did not adequately sample the fifth hemisphere (M2Pright hemisphere) that we recorded from to define the borders of eachauditory field. Frequency reversals were used to define the borders ofthe three core fields: primary auditory cortex (AI), rostral field (R),and rostrotemporal field (RT) (Bendor and Wang 2008; Kaas andHackett 2000; Morel and Kass 1992; Petkov et al. 2006). Differencesin the spatial distribution of pitch-selective neurons and modulationsensitive neurons were quantified using a Kolmogorov-Smirnov test(P � 0.05). We compared both the caudal-to-rostral or medial-to-lateral axis of auditory cortex in this analysis.

Statistical significant differences in population responses (normal-ized discharge rate and average vector strength) between two repeti-tion rates was determined using a Wilcoxon rank sum test (P � 0.05,Bonferroni corrected for multiple comparisons).

The change in discharge rate between regular and irregular acousticpulse trains (Fig. 14D), was calculated by performing a linear inter-polation on the firing rates at all tested jitter values (0–50%). Theslope obtained from this calculation was used to determine the changein firing rate between acoustic pulse trains with 0 and 50% maximumtemporal jitter.

R E S U L T S

The data presented in this report include a total of 254neurons recorded from the auditory cortex of four marmosets(5 hemispheres) that responded significantly to moderatelycomplex acoustic stimuli such as sAM tones, acoustic pulsetrains, or harmonic complex tones (Fig. 1).

In auditory cortex, the spectrum of an acoustic signal musttypically overlap with a neuron’s frequency response area toevoke a response. For such neurons, we centered the spectra ofthese acoustic stimuli at the neuron’s BF. Given that a modu-lation frequency is lower than its carrier frequency, theseneurons were generally tuned to modulation frequencies thatwere outside of their frequency response area. We use the term“modulation sensitive neurons” in this report to refer to thispopulation of neurons.

Responses of modulation sensitive neurons

The example modulation sensitive neuron shown in Fig. 2 istuned to repetition rates within the range of pitch (for an

acoustic pulse train) with a peak response at a repetition rate of200 Hz (A and B). This neuron does not synchronize to theacoustic pulse train (Fig. 2C) and therefore only provides aneural representation of the repetition rate with its dischargerate. This neuron has a BF of 10.2 kHz (Fig. 2A, inset), whichis substantially higher than its response range of repetition rates

B

12.5 25 50 100 200 400

0

10

20

30

40

50

60

Repetition rate (Hz)

8 9 10 11 12

0

20

40

Frequency (kHz)

Dis

char

ge R

ate

(spk

/s)A

12.5 25 50 100 200 4000

5

10

13.8


-500 0 500 1000

500333133

80503325201714

Time (ms)

Dis

char

ge R

ate

(spk

/s)

Ray

leig

h st

atis

tic v

alue

Rep

etiti

on ra

te (H

z)

C

FIG. 2. Response from an individual modulation sensitive neuron. A: dis-charge rate of a neuron [unit M2P-921.1, primary auditory cortex/rostral field(AI/R) border] to acoustic pulse trains (� � 0.89). Inset: the response of theneuron to pure tones varying in frequency. Error bars indicate the SE. B: rasterplot of the neuron shown in A responding to acoustic pulse trains at differentrepetition rates (13–500 Hz). The horizontal line underneath the raster plotindicates the time period that the acoustic stimulus was played. C: Rayleighstatistic for neuronal response shown in A. None of these responses reached thecriteria for statistical significance (13.8, P � 0.001) which is indicated on theplot with a dashed horizontal line.

TABLE 1. Types of modulation sensitive neurons

Mixed Nonsynchronized Total

Positive monotonic 17 56 73Negative monotonic 10 6 16Other tuning curve shape 10 16 26Total 37 78 115



on Septem


Dow

nloaded from


(in frequency). Across the population of modulation sensitiveneurons, tuning for modulation frequency (frequency of peri-odicity in sAM tones) and repetition rate (frequency of peri-odicity in acoustic pulse trains) spanned the entire range ofmodulation frequencies producing a missing fundamental pitch(�800 Hz; Fig. 3) (Ritsma 1962).

Repetition rate tuning in modulation sensitive neurons tooka variety of forms, the most common being a positive mono-tonic trend (n � 73) for which the firing rate increasedmonotonically with repetition rate (Fig. 4, A and B). We alsoclassified 16 neurons as having negative monotonic responses

where discharge rate increased with a decreasing repetitionrate. The remaining 26 neurons failing the criteria for mono-tonicity (see METHODS) had all-pass, band-pass, or notch/mul-tipeak tuning.

The majority of modulation sensitive neurons were not ableto stimulus synchronize to the individual pulses of acousticpulse trains in our stimulus set (Fig. 4C). In this report, we willrefer to neurons that were unable stimulus synchronize as“nonsynchronized,” whereas neurons showing some degree ofstimulus synchronization (while also covarying their firing ratewith repetition rate) will be referred to as having a “mixed

B

A

4 8 16 32 64 128 256 512

0

0.5

1

Modulation frequency (Hz)

Nor

mal

ized

rate

neuron 1neuron 2neuron 3neuron 4neuron 5neuron 6

16 32 64 128 256 512

0

0.5

1


Nor

mal

ized

rate

neuron 1neuron 2neuron 3neuron 4neuron 5neuron 6

FIG. 3. Modulation frequency and repetition rate tuning inmodulation sensitive neurons. A: modulation frequency tuningto sAM tones for 6 neurons in auditory cortex: neuron 1 (unitM41O-242.1), neuron 2 (unit M2P-109.1), neuron 3 (unit M32Q-348.1), neuron 4 (unit M41O-248.2), neuron 5 (unit M41O-241.1),neuron 6 (unit M2P-56.1). Neuron 3 was recorded in field AI. Theremaining neurons were recorded from field R. B: repetition ratetuning to acoustic pulse trains for 6 neurons in auditory cortex:neuron 1 (unit M2P-799.1, field AI), neuron 2 (unit-M32Q-79.1,field R), neuron 3 (unit M2P-30.1, field AI), neuron 4 (unitM2P-843.1, field AI), neuron 5 (unit M2P-411.1, field R), neuron6 (unit M2P-921.1, AI/R border). All neurons have significantdischarge rates over a subset of repetition rates in the range of pitchperception (�30 Hz). (Note that these neurons are not the sameneurons shown in Fig. 3A).

A

12.5 25 50 100 200 4000

0.2

0.4

0.6

0.8


Nor

mal

ized

resp

onse

n=73

12.5 25 50 100 200 400

0

0.2

0.4

0.6


Aver

age

vect

or s

treng

th

mixed response (n=37)non-synchronized (n=78)

*

* max repetition rate (P<0.05)

D

<10 10 20 40 80 160 320 6400

20

40

60

Synchronization limit (Hz)

Frac

tion

of p

opul

atio

n (%

)

B

n=115

-1 -0.5 0 0.5 10

20

40

Monotonicity index

Num

ber o

f uni

ts

not monotonic (n=26)positive monotonic (n=73)

C

negativemonotonic (n=16)

FIG. 4. Rate coding and synchronization ability of modula-tion sensitive neurons. A: distribution of monotonicity tuning inmodulation sensitive neurons. The monotonicity index is theSpearman correlation coefficient for firing rates between 30 and500 Hz. Statistically significant Spearman correlation coeffi-cients (P � 0.05) are indicated. B: normalized discharge rate ofmodulation sensitive neurons with positive monotonic re-sponses. Neurons monotonically increased their discharge ratewith increasing repetition rate over the range 25–333 Hz.Population responses at repetition rates between 40 and 200 Hzand at 400 Hz were significantly different (Wilcoxon rank sum,P � 0.05 Bonferonni corrected) from the population dischargerates at both 500 and 13 Hz. C: distribution of the synchroni-zation limit in modulation sensitive neurons. D: mean vectorstrength of modulation sensitive neurons with nonsynchronizedand mixed responses. Only the mixed response population hadan average vector strength significantly different from zero, andonly for repetition rates equal to and �133 Hz (Wilcoxon ranksum test, P � 0.05 Bonferonni corrected).



on Septem


Dow

nloaded from


response” (Fig. 4D). The median synchronization limit ofmixed response neurons was 100 Hz (25–75%: 48–156 Hz).Using a criterion of stimulus synchronization (see METHODS), 78nonsynchronized neurons and 37 mixed response neurons wereidentified (Table 1). For both nonsynchronized and mixedresponse neurons all types of repetition rate tuning was ob-served (see Table 1). The most common observed responseoccurring in roughly half of samples (56/115) was both posi-tive monotonic and nonsynchronized.

Pitch-selective responses

We have previously identified an additional class of neuronsthat have best repetition rates that are similar to their BFs(based on pure tones) (Bendor and Wang 2005). We refer tothese neurons as “pitch selective” because they respond tomissing fundamental sounds with a pitch near their BF andwith a spectrum outside of their excitatory frequency responsearea (Fig. 5, see METHODS).

Using these criteria, a total of 74 pitch-selective neuronswere identified in four marmosets. A subset of these pitch-selective neurons (53/74) have been reported in an earlierpublication (Bendor and Wang 2005) (see METHODS). Missingfundamental responses were measured near the sound levelthreshold of the neuron’s BF to avoid the effect of combinationtones. Combination tones are nonlinear distortion productsproduced by the cochlea when multiple harmonic tones arepresent. The combination tone produced at the fundamentalfrequency is �20 dB below the sound level of a singlecomponent (Pressnitzer and Patterson 2001). Therefore at suf-ficiently high sound levels, missing fundamental responsescould be the result of combination tones produced at theneuron’s BF. Neurons within the putative pitch center that didnot respond to near threshold missing fundamental soundscould still respond to missing fundamental sounds when soundlevels were 20 dB or more above threshold (Fig. 6). However,unlike pitch-selective neurons, these missing fundamental responsescould be blocked with the addition of a noise masker (Fig. 6).

Comparison pitch-selective and modulation sensitive neurons

Given that repetition rate tuning is observed in both modu-lation sensitive and pitch-selective neurons, how do these twoneuronal populations differ? As we have already discussed,modulation sensitive neurons have different best frequenciesand best repetition rates and respond to complex sounds withspectra overlapping their BF (Fig. 7). On the other hand,pitch-selective neurons have similar repetition rate and puretone tuning and respond to complex sounds with spectra that donot overlap with their BF (Fig. 7).

Although pitch-selective neurons generally have low frequencyBFs (Bendor and Wang 2005), we observed that the range of BFsin modulation sensitive neurons stretched across almost the entiretonotopic axis (Fig. 8, Wilcoxon rank sum test, P � 7.4 � 10–29).The lack of modulation sensitive neurons with very low BFs isdue to our criteria that the neuron’s BF needed to be higher thanthe repetition rates that the neuron encoded. Neurons that do notpass the criteria of pitch selectivity also exist within the putativepitch center (Bendor and Wang 2005); however, these are low BFneurons that do not encode the repetition rate of complex soundswith spectra at higher frequencies (outside of their excitatoryfrequency response area).

Spatial distribution of pitch-selective and modulationsensitive neurons

We next examined the spatial distribution of pitch-selectiveneurons and modulation sensitive neurons. An example of acortical map from one subject with the locations of these twoclasses of neurons is shown in Fig. 9A. We compared the spatial

100 200 300 400 -5

0

5

10

15

20

frequency (Hz)

Dis

char

ge R

ate

(spk

/s)

12-1410-128-10 6-8 4-6 1-3 0

5

10

15

20

harmonic composition

Dis

char

ge R

ate

(spk

/s)

2 3 4 5 6 7 8 9 10 11 12 -5

0

5

10

15

20

harmonic

Dis

char

ge R

ate

(spk

/s) 70 dB SPL

60 dB SPL50 dB SPL

A

B

C

806040200

0

10

20

30

Sound level (dB SPL)

Dis

char

ge R

ate

(spk

/s)

FIG. 5. Pitch-selectivity criteria in an example pitch-selective neuron. Alldata are shown is from unit M2P-157.1. Error bars in A–C indicate the SE.A: frequency tuning curve for tones played at 50 dB SPL. Inset: a rate-levelresponse at best frequency. B: response to harmonics of the best fundamentalfrequency (182 Hz) played individually at 0, �10, and �20 dB above thesound level threshold at the best frequency. C: response to harmonic complextones (each harmonic is played at 50 dB SPL). All acoustic stimuli are missingthe fundamental frequency except for the acoustic stimulus composed ofharmonics 1–3.



on Septem


Dow

nloaded from


distributions of pitch-selective neurons with modulation sensitiveneurons across four subjects on a normalized cortical map (Fig.9B). Pitch-selective neurons were found on the border of AI, R,and lateral belt, whereas modulation sensitive neurons were dis-tributed throughout auditory cortex. The spatial distributions weresignificantly different along the medial-to-lateral axis (Kolmogorov-Smirnov test, P � 7.4 � 10–9) and caudal-to-rostral axis ofauditory cortex (Kolmogorov-Smirnov test, P � 1.1 � 10–9).

Similarity in repetition rate tuning for differentacoustic signals

We next compared the similarity in repetition rate tuningacross different acoustic signals for these two neuronal popu-

lations. A total of 110 modulation sensitive neurons weretested with both sAM tones and acoustic pulse trains, whichdiffer in both their spectral bandwidth and envelope shape. Thetuning for repetition rate was similar for sAM tones andacoustic pulse trains for some neurons (Fig. 10A), whereasother neurons showed very different responses for the twotypes of acoustic signals (B). Nine pitch-selective neurons weretested with missing fundamental (MF) complex tones andeither sAM tones or acoustic pulse trains. We observed similartuning for repetition rate between the two acoustic signalstested (Fig. 10, C and D).

Overall, modulation sensitive neurons did not have a sig-nificant correlation (Spearman correlation coefficient, r ��0.0091, P � 0.93) between their weighted best repetition ratefor acoustic pulse trains and their weighted best modulationfrequency for sAM tones (Fig. 11A). Only 38 of 110 modula-tion sensitive neurons (35%) had best repetition rates withinone octave of each other for the two acoustic stimuli.

We next compared the monotonicity index (see METHODS) ofmodulation sensitive neurons for sAM tones and acoustic pulse

Non-pitch neuron (distortion product)

Pitch-selective neuronA

B

80 60 40 20 0

0

10

20

30

40


Dis

char

ge ra

te (s

pk/s

)

noise masker alone

MF alone pure tone alone

MF w/noise masker

806040200

0

5

10

15

20

25


Dis

char

ge ra

te (s

pk/s

)

noise masker alone

MF alone pure tone alone

MF w/noise masker

~20 dB

FIG. 6. Example of a pitch-selective neuron’s response and a distortionproduct response in a nonpitch neuron. Error bars indicate the SE. A: apitch-selective neuron’s rate-level response (unit M41O-294.1) to pure tonesand missing fundamental sounds (with and without a noise masker). Theneuron has a similar threshold for both pure tones and missing fundamentalsounds. At higher sound levels, the response to the missing fundamental isgreater than the pure tone response. With the addition of the noise masker, themissing fundamental response does not change. The noise masker itself did notevoke a significant response in the neuron. B: a nonpitch neuron’s rate-levelresponse (unit M41O-251.2) to pure tones and missing fundamental sounds(with and without a noise masker). The response to missing fundamentalsounds had a 20 dB higher sound level threshold. On the addition of the noisemasker, the neuron no longer responded to the missing fundamental sound.This neuron was recorded from within the pitch center.

spectral frequency (kHz)0.25 0.5 1 2

Spikerate

Modulation sensitive neuron

0.25 0.5 1 2

Spikerate

spectral frequency (kHz)

Pitch-selective neuron

Repetition rate or

or

Pure tone

Complex sound's spectrum

Complex sound

Repetition rate

FIG. 7. Comparison of modulation sensitive and pitch-selective neurons.Modulation sensitive neurons have different best frequencies (pure tonetuning) and best repetition rates (complex tone tuning) and respond to complexsounds with spectra overlapping their best frequency. Pitch-selective neuronshave similar pure tone and complex tone tuning and respond to complexsounds with spectra that do not overlap with their best frequency.

.125 .25 .5 1 2 4 8 16 320

10

20

30

40

Best frequency (kHz)

Per

cent

of s

ampl

es

pitch-selective (n=74)modulation sensitive (n=115)

FIG. 8. Best frequency (BF) distribution of pitch-selective and modulationsensitive neurons. Distribution of the BF of pitch-selective and modulationsensitive neurons across 4 subjects (5 hemispheres). Pitch-selective neuronshad a significantly lower BF (Wilcoxon rank sum test, P � 7.4 � 10�29).



on Septem


Dow

nloaded from


trains. Similarity in this index for two tuning curves indicatesa similar trend (firing rate increases or decreases with repetitionrate) even when peak values differ. We observed a significantcorrelation between the monotonicity indices of neurons forthese two acoustic stimuli (Spearman correlation coefficient,r � 0.54, P � 1.5 � 10�9). Even though repetition rate tuningcan have a similar trend across different acoustic stimuli, a bestrepetition rate generally appears to be stimulus specific. Spec-tral bandwidth and envelope shape changes with modulationfrequency for sAM stimuli (but is constant for acoustic pulsetrains). Thus modulation sensitive neurons that do not haveconsistent best repetition rates across different acoustic stimulimay be also sensitive to acoustic parameters such as spectralbandwidth and envelope shape.

In our population of pitch-selective neurons, we observedthat the best fundamental frequency was generally similar tothe best repetition rate (from an acoustic pulse train or sAMtone). We did not observe a significant correlation (Spearmancorrelation coefficient, r � 0.065, P � 0.067), but this may be

a result of the small sample size and a single outlier (Fig. 11B).Pitch-selective neurons had a significantly smaller difference intheir best repetition rates for different acoustic signals thanmodulation sensitive neurons (Wilcoxon rank sum test, P �2 � 10�4, Fig. 11C).

Pitch-selective neurons were found along the low-frequencyborder of fields AI and R, while modulation sensitive neuronswere distributed all over fields AI, R, and RT. Modulationsensitive neurons with large and small differences in bestrepetition rate between acoustic stimuli were found in all threecortical fields (Fig. 12).

Sensitivity to temporal regularity

We next tested pitch-selective and modulation sensitiveneurons with regular and irregular acoustic pulse trains. Irreg-ular pulse trains had the same average repetition rates ofregular pulse trains but were aperiodic. The temporal jitter ofthe interpulse interval was parametrically varied between 5 and50% in irregular pulse trains (Fig. 13, see METHODS).

0.125 kHz0.25 kHz0.5 kHz1 kHz2 kHz4 kHz8 kHz16 kHz

1 mmmodulation sensitive (n=31)pitch-selective (n=18)

highfreq

lowfreq

highfreq

lowfreq

2

1.5

1

0.5

0

Normalized distance along rostral-to-caudal axis

Dis

tanc

e fro

m la

tera

l sul

cus

(mm

)

lateral sulcus


RT R AI

0 10 20Number of neurons

0

20

40

Num

ber o

f neu

rons

M

L

rostral caudal

AIRT R

CR

A

B

Subject M32Q

FIG. 9. Spatial distribution of pitch-se-lective and modulation sensitive neurons.A: frequency map from 1 subject (M32Q-left hemisphere) with the location of pitch-selective neurons and modulation sensitiveneurons indicated. B: normalized corticalmap of locations of pitch-selective and mod-ulation sensitive neurons across four sub-jects (4 hemispheres).



on Septem


Dow

nloaded from


The majority of pitch-selective neurons (11/16) decreasedtheir firing rate more than 30% between regular and irregular(50% temporal jitter) acoustic pulse trains showing a prefer-ence for periodic acoustic signals (Fig. 14, A and C). Incontrast to this, �80% of modulation sensitive neurons (29/36)did not change their firing rates more than 30% betweenregular and irregular acoustic pulse trains (Fig. 14, B and C).

We further quantified temporal regularity tuning by measur-ing the change in firing rate between regular acoustic pulsetrains and irregular pulse trains with 50% jitter (maximumtested, see METHODS). Pitch-selective neurons had a negativemean change in firing rate between regular and maximallyirregular pulse trains (–48.2 38.0% change in firing rate,mean SD). This was not observed for modulation sensitiveneurons (1.5 23.3% change in firing rate). The difference insensitivity to temporal irregularity between the two populationswas statistically significant (Fig. 14D, Wilcoxon rank sum test,P � 3.6 � 10�5). There was also no difference in insensitivityto temporal irregularity for different types of modulation sen-sitive neurons. When modulation sensitive neurons were sub-divided into smaller populations, based on their location inauditory cortex, stimulus synchronization, or tuning curveshape, no significant differences were observed in the meanchange of firing rates for regular and irregular acoustic signals(Wilcoxon rank sum test, P � 0.05 uncorrected, Table 2). Thejitter sensitivity of a modulation sensitive neuron did notdepend on its location along the rostral-caudal axis of auditorycortex (Fig. 15).

D I S C U S S I O N

The temporal fidelity of the auditory nerve is remarkablyprecise with an ability to phase lock to the fine structure of anacoustic signal at rates �3–5 kHz (Johnson 1980). At eachsuccessive level in the pathway from the auditory nerve to audi-

tory cortex, this temporal precession progressively decreases (Jo-ris et al. 2004; Wallace et al. 2002). As a result, auditory cortexcannot encode an acoustic parameter such as envelope repetitionrate over the entire range of rates required for pitch perceptionusing an explicit temporal representation. To prevent the loss ofinformation from a degraded temporal representation, temporalparameters such as repetition rate and temporal regularity must bere-encoded using the discharge rates of neurons. Using theirdischarge rate, pitch-selective neurons found within the putativepitch center of auditory cortex encode temporal regularity andrepetition rate over the perceptual range of pitch (Figs. 11B and14, C and D). In contrast, modulation sensitive neurons foundthroughout auditory cortex encode the average repetition rate ofcomplex sounds but are generally insensitive to changes in tem-poral regularity (Fig. 14, C and D).

When a neuron is tuned to the repetition rate or fundamentalfrequency of a complex sound, can we consider this a sufficientcriterion for a neural representation of pitch (Bizley et al. 2009;Kalluri et al. 2008)? In human subjects, pitch discriminationworsens when an acoustic stimulus is temporally jittered (Pollack1968). Therefore if a neuronal population is insensitive to tempo-ral regularity, it would provide a similar neural threshold for pitchdiscrimination of periodic and aperiodic acoustic stimuli; this doesnot correlate with the decrease in the perceptual threshold ob-served. Thus it is crucial to demonstrate that “pitch-selective”neurons are sensitive to temporal regularity.

The responses of pitch-selective neurons are correlated withseveral acoustic features that co-vary with pitch, namely repetitionrate, fundamental frequency, and temporal regularity. However,because pitch is a percept rather than a physical sound property, itis important to measure how a subject’s perception of pitchcorrelates with a neural response. To fully demonstrate the neces-sity of pitch-selective neurons for the encoding of pitch percep-tion, a behavioral assay inactivating pitch-selective neurons wouldbe required.

4 8 16 32 64 128 256 512 1024

0

20

40

60


Dis

char

ge ra

te (s

pk/s

)

pulse trainsAM

4 8 16 32 64 128 256 512 10240

30

60

90

Repetition rate (Hz) D

isch

arge

rate

(spk

/s)

pulse trainsAM

BA

50 100 200 400 800

0

5

10

15

20

Dis

char

ge ra

te (s

pk/s

)

MFsAM

Repetition rate (Hz)100 150 200 300 400

0

5

10

15

20

25

30


Dis

char

ge R

ate

(spk

/s)

MFsAM

DC

example modulation sensitive neuron 1 example modulation sensitive neuron 2

example pitch-selective neuron 2example pitch-selective neuron 1

FIG. 10. Repetition rate tuning of modulation sensitive andpitch-selective neurons to 2 different acoustic stimuli. Errorbars in A–D indicate the SE. A: similar repetition rate tuning foracoustic pulse trains and sAM tones for a neuron in auditorycortex (unit M2P-901.2, AI/R border). B: dissimilar repetitionrate tuning for acoustic pulse trains and sAM tones for a neuronin auditory cortex [unit M2P-311.2, field rostrotemporal (RT)].C. similar repetition rate tuning for a missing fundamentalcomplex tone and an sAM tone for a pitch-selective neuron(unit M2P-233.1). D: similar repetition rate tuning for a missingfundamental complex tone and an sAM tone for a pitch-selective neuron (unit M41O-248.2).



on Septem


Dow

nloaded from


Implications for pitch-processing models

Given the complex nature of pitch perception, and the widevariety of acoustic stimuli that can generate a pitch percept, itis possible that regions of auditory cortex outside of the

putative pitch center are required for pitch extraction. How-ever, because firing rates were dependent on temporal regular-ity only in pitch-selective neurons and modulation sensitiveneurons were not able to stimulus synchronize at high enoughrates to encode the entire range of pitch perception, we mustconclude that pitch-selective neurons do not rely on corticalinputs (outside of the pitch center) for their sensitivity totemporal regularity. The temporal-to-rate coding transforma-tion representing temporal regularity likely happens in eitherthe inferior colliculus or thalamus, which then serves as one of theprimary inputs to pitch-selective neurons. Due to the putativepitch center’s proximity to both the core and belt regions ofauditory cortex (Fig. 9), it likely receives inputs from both thedorsal and ventral divisions of the medial geniculate body, andeither one of these nuclei may provide periodicity informationto pitch-selective neurons. Repetition rate (or modulation fre-quency) tuning in a subcortical nuclei that is based purely on arate code is not sufficient for a pitch-processing model. As wehave seen here, repetition rate tuning is a universal phenome-non throughout auditory cortex. However, repetition rate tun-ing that is dependent on temporal regularity generally occursonly within the putative pitch center. Identifying the neuralrepresentation of temporal regularity subcortically may providethe missing link between the representation of pitch in theperiphery and at the level of auditory cortex.

The putative pitch center was located in the low frequencyregion of auditory cortex, bordering fields AI, R, and the lateralbelt (Fig. 9), potentially a region anatomically homologous tolateral Heschl’s gyrus in humans (Bendor and Wang 2006;Formisano et al. 2003; Hackett et al. 2001). Although pitch-selective neurons were more abundant in AI compared withfield R, given that the size of AI is roughly double that of fieldR, the proportion of pitch-selective neurons found in each areawere comparable. Unlike pitch-selective neurons, modulationsensitive neurons were found across the whole tonotopic range(Fig. 8, 9) which suggests that they play a more general role intemporal processing, as the pitch processing of missing funda-mental sounds relies on harmonic frequencies below 5 kHz(Moore 2003; Oxenham et al. 2004). One prediction from these

AIRRT0

2

4

6


Pea

k re

petit

ion

rate

diff

eren

ce (o

ctav

es)

modulation sensitive (n=98)pitch-selective (n=9)

FIG. 12. Topographic distribution of repetition rate tuning similarity. Peakrepetition rate difference (from Fig. 11) of modulation sensitive and pitch-selective neurons plotted according to their position along the rostrocaudal axisof auditory cortex (in normalized coordinates). The dashed vertical linesindicate the boundaries between AI/R and R/RT.

A

B

C

64 128 256 512 1024 204864

128

256

512

1024

2048


Fund

amen

tal f

requ

ency

(Hz)

n=9

0 1 2 3 4 5 6 70

20

40

60

80

Difference between peak values (octaves)

Frac

tion

of p

opul

atio

n(%

) pitch-sensitive (n=9)modulation sensitive (n=110)

4 8 16 32 64 128 256 5124

8

16

32

64

128

256

512

Modulation frequency (Hz)

Rep

etiti

on ra

te (H

z)

FIG. 11. Similarity in repetition rate tuning for 2 acoustic signals. A: compar-ison of the weighted best modulation frequency (sAM tones) and weighted bestrepetition rate (acoustic pulse trains) for individual neurons within auditory cortex.The correlation was not statistically significant (Spearman correlation coefficient,r � �0.0091, P � 0.9252). The solid diagonal line indicates where y � x. Thedashed diagonal lines indicate the 1 octave boundaries surrounding the y � x line.B: comparison of the weighted best fundamental frequency (missing fundamentalcomplex tones) and weighted best repetition rate (sAM tone or acoustic pulsetrains) for individual pitch-selective neurons. The correlation did reach statisticalsignificance (Spearman correlation coefficient, r � 0.65, P � 0.067). C: ahistogram of the absolute difference in peak values for repetition rate tuning in the2 acoustic stimuli used in A and B. Pitch-selective neurons have a more similar bestrepetition rate for 2 spectrally different acoustic stimuli (Wilcoxon rank sum test,P � 2.0 � 10�4).



on Septem


Dow

nloaded from


results is that the inactivation of the putative pitch centerbilaterally would cause deficits in detecting changes in tempo-ral regularity or pitch salience, but discrimination of averagerepetition rate would still be possible. However, the thresholdfor detecting changes in repetition rate is lower in acousticsignals with a high pitch salience (Pollack 1968), and sopitch-selective neurons may be more sensitive to changes inrepetition rate than modulation sensitive neurons. Althoughrepetition rate discrimination would be possible with the inac-tivation of the putative pitch center, the discrimination thresh-olds could be significantly higher. The slope of the tuningcurve in positive monotonic modulation sensitive neurons isfairly constant up to �333 Hz (Fig. 4B), at which point itdecreases. One possible consequence of this is that rate dis-crimination thresholds (when pitch is not involved) wouldworsen when the slope of the tuning curve decreases. Thisobservation suggests a possible neural correlate for the limita-tion in rate discrimination in cochlear implantees (Baumann

and Nobbe 2004; Shannon 1983; Zeng 2002). If pitch-selectiveneurons are not driven by cochlear implants, and the perceptionof repetition rate is due to only modulation sensitive neurons,rate-coding constraints could have a perceptual effect notobserved in normal hearing subjects that have access to aworking population of pitch-selective neurons. According tothis hypothesis, the ability to discriminate the repetition rate ofirregular acoustic pulse trains should be similar between co-chlear implantees and normal hearing subjects as in this case,only the modulation sensitive population would be encodingthe repetition rate of this stimulus.

One difficulty in using multiunit or hemodynamic responsesto study pitch-selective neurons is the heterogeneous propertiesof neurons within the putative pitch center. Pitch-selectiveneurons only make up about 1/3 of the neurons in this region(Bendor and Wang 2005). If the responses of pitch-selectiveand modulation sensitive neurons in the putative pitch centerare combined together (using techniques such as LFP or

10% maximum jitter 50% maximum jitter 0% maximum jitter

0 10 20 30 40 50 60 70 80 90

0

Am

plitu

de

Time (ms)

2.4 2.6 2.8 3.0 3.2 3.4 3.6 -40

-20

0

Frequency (kHz)

Am

plitu

de (d

B)

0 10 20 30 40 50 60 70 80 90

0

Am

plitu

de

2.4 2.6 2.8 3.0 3.2 3.4 3.6 -40

-20

0

Frequency (kHz)A

mpl

itude

(dB

)

0 10 20 30 40 50 60 70 80 90

0

Time (ms)

2.4 2.6 2.8 3.0 3.2 3.4 3.6 -40

-20

0

Frequency (kHz)

Am

plitu

de (d

B)

Time (ms)

Am

plitu

deacoustic waveform

spectrum

FIG. 13. Acoustic stimuli used to test tem-poral regularity sensitivity. Acoustic waveform(top) and spectrum (bottom) of a Gaussiannarrowband acoustic pulse train with a 3 kHzcarrier, 100 Hz average repetition rate, and � �0.89 with 0% (left), 10% (middle), or 50%(right) temporal jitter.

0 10 20 30 40 50

5

15

25

35

Maximum temporal jitter (%)

Dis

char

ge ra

te (s

pk/s

)

Neuron 1Neuron 2

Neuron 3A B

DC

0 10 20 30 40 500

10

20

30

40

50


Dis

char

ge R

ate

(spk

/s)

Neuron 1 Neuron 3Neuron 2

0 10 20 30 40 50

0.3

0.5

0.7

0.9


Nor

mal

ized

dis

char

ge ra

te


* * ***

* P<0.05

-120 -80 -40 0 40 800

10

20

30

40

Change in discharge rate (%)

Per

cent

of s

ampl

es


modulation sensitive neuron examplespitch-selective neuron examples

FIG. 14. Comparison of sensitivity to temporalirregularity between pitch-selective and modulationsensitive neurons. A: individual examples of regularand irregular pulse train responses in pitch-selectiveneurons. Neuron 1 (unit M32Q-101.1, rectangularclicks), neuron 2 (unit M41O-276.1, acoustic pulsetrain with tone carrier), neuron 3 (unit M36N-523.1,acoustic pulse train with noise carrier). B: individualexamples of regular and irregular pulse train re-sponses in modulation sensitive neurons: neuron 1(unit M2P-357.2, field RT), neuron 2 (unit M32Q-117.1, field R), neuron 3 (unit M36N-418.1, fieldAI). C: normalized tuning in pitch-selective andmodulation sensitive neurons to acoustic pulse trainsvarying in temporal irregularity. For pitch-selectiveneurons, normalized responses for all jitter valuessignificantly different from regular click trains (P �0.05 Bonferonni corrected, Wilcoxon rank sum test)are indicated (*). Normalized responses of modula-tion sensitive neurons to irregular acoustic pulsetrains were not significantly different from regularacoustic pulse trains. D: a comparison betweenpitch-selective and modulation sensitive neurons intheir interpolated percent change in discharge ratebetween a regular and irregular (50% jitter) acousticpulse train. The 2 distributions are significantlydifferent (P � 3.6� 10�5, Wilcoxon rank sum test).



on Septem


Dow

nloaded from


fMRI), the results can be difficult to interpret. The criteria usedfor pitch-selective neurons (Bendor and Wang 2005) are de-pendent on a single neuron’s BF and sound level threshold andcannot be used if recording from a population of neurons.However, if distortion products are masked using noise, then acomparison between periodic and aperiodic acoustic pulsetrains can provide a much more rapid identification of theputative pitch center. Due to the high pitch salience andmatched spectral bandwidths of these acoustic stimuli, this isan optimal stimulus to use in LFP or fMRI based experiments.

Comparison with previous studies

We observed that mixed response neurons have a highersynchronization limit than reported for synchronizing corticalneurons (Lu et al. 2001) and a lower synchronization limit thansynchronizing thalamic neurons (Bartlett and Wang 2007):median synchronization limit- current study mixed neurons:100 Hz (25–75%, 48 Hz, 156 Hz); Lu et al. (2001), synchro-nizing neurons 46.9 Hz (25–75%, 18.4 Hz, 80.6 Hz); Bartlettand Wang (2007),thalamic synchronizing neurons 192 Hz(25–75%: 88–303 Hz). The majority of neurons in the auditorythalamus that can stimulus synchronize also have nonsynchro-nized responses at higher repetition rates, similar to the mixedresponse modulation sensitive neurons reported here (Bartlettand Wang 2007). As such, mixed response neurons mayreceive direct thalamic input and then project to both nonsyn-chronized and synchronized neurons within auditory cortex,acting as an intermediate stage in the temporal-to-rate codingtransformation occurring between thalamus and cortex (Wanget al. 2008).

In addition to marmosets, nonsynchronized responses havealso been reported in rats and cats (Anderson et al. 2006; Luand Wang 2000; Sakai et al. 2009). Failure to observe nonsyn-chronized responses in auditory cortex (Malone et al. 2007)could be due to several issues including species differences,criteria of nonsynchronized responses, acoustic stimuli (sAMvs. narrowband click trains), and the use of anesthesia (Ren-naker et al. 2007).

The positive monotonic responses we report here are differ-ent from those reported in flutter encoding neurons (Bendorand Wang 2007) for two reasons. First, they are encoding therange of repetition rates 30–500 Hz, within the range of pitchperception rather than the range of acoustic flutter perception(10–45 Hz). Second, roughly equal numbers of positive andnegative monotonic responses were observed in flutter encod-ing neurons (Bendor and Wang 2007), while in this study thevast majority of monotonic response in modulation sensitiveneurons were positive monotonic (increased their dischargerate with higher repetition rates). Nevertheless, neurons with

monotonic and nonmonotonic tuning curves both likely play arole in encoding temporal information in the range of flutterand pitch perception.

Comparison with the somatosensory system

The need to encode temporal information is not unique to theauditory system, and similar neural coding strategies may alsobe utilized in other sensory systems. In the somatosensorysystem, Pacinian receptors in the periphery and Pacinian neu-rons in primary somatosensory cortex encode tactile vibration(50–400 Hz) at frequencies above the range of tactile flutter(Hyvarinen et al. 1968). Although stimulus synchronizationexists at the periphery, Pacinian neurons in primary somato-sensory cortex use a positive monotonic nonsynchronized ratecode for vibration frequency (pulse repetition rate) similar tothe majority of modulation sensitive neurons in this study(Hyvarinen et al. 1968). A similar neural coding strategy fortemporal information may be used in primary auditory cortexand primary somatosensory cortex (Bendor and Wang 2007;Lemus et al. 2009; Romo and Salinas 2003). Slowly repeatedstimuli, perceived as a stream of discretely occurring sensoryevents (acoustic or tactile flutter), are encoded by stimulussynchronization across one neuronal population. Faster repeti-tion rates, perceived as a stream of fused sensory events (pitchor vibration), are encoded by a nonsynchronized positivemonotonic rate code within a second neuronal population.Thus the change in perception from flutter to fusion may be thedirect consequence of the lack of stimulus synchronization inthe neuronal population at higher repetition rates. The upperlimit of this positive monotonic rate code is where it is nolonger monotonic (saturation of the tuning curve), which basedon this model would correspond to the upper limit of ratediscrimination in each modality. Although we refer to the fusedauditory sensation of repetition rate here as pitch, we shouldpoint out that a fused percept can still be perceived when anacoustic signal has no pitch salience (e.g., an irregular clicktrain with an average repetition rate of 200 Hz). Unlike pitchperception, which has a fine grain discrimination of repetitionrate allowing for octave perception and recognition of melodiesand harmonies, modulation sensitive neurons may instead be

AIRRT

-100

-50

0

50


Jitt

er s

ensi

tivity

(%

cha

nge

in fi

ring

rate

)


FIG. 15. Topographic distribution of temporal regularity sensitivity. Tem-poral regularity sensitivity (from Fig. 14D) of modulation sensitive andpitch-selective neurons plotted according to their position along the rostro-caudal axis of auditory cortex (in normalized coordinates). The dashed verticallines indicate the boundaries between AI/R and R/RT.

TABLE 2. Comparison of sensitivity to temporal irregularitybetween subclasses of modulation sensitive neurons

Group 1 Group 2Wilcoxon RankSum P Value

Nonsynchronized (29) Mixed response (7) 0.40Positive monotonic (22) Other tuning curve shapes (14) 0.47Al neurons (18) R and RT neurons (18) 0.81

n values in parentheses. AI, primary auditory cortex; R, rostral field; RT,rostrotemporal field.



on Septem


Dow

nloaded from


representing the acoustic perceptual equivalent of roughness ortactile vibration. Thus we suggest that flutter and vibrationshare similar perceptual ranges and neural coding strategies inboth the auditory and somatosensory system. However, theperception of pitch and the sensitive encoding of periodicity bycortical neurons is unique to the auditory system.

A C K N O W L E D G M E N T S

We thank A. Pistorio, J. Estes, E. Issa, E. Bartlett, and Y. Zhou for assistancewith animal care.

G R A N T S

This work was supported by National Institute of Deafness and OtherCommunication Disorders Grants DC-03180 to X. Wang and F31 DC-006528to D. Bender and a Helen Hay Whitney Postdoctoral Fellowship D. Bender.

R E F E R E N C E S

Anderson SE, Kilgard MP, Sloan AM, Rennaker RL. Response to broad-band repetitive stimuli in auditory cortex of the unanesthetized rat. Hear Res213: 107–117, 2006.

Bartlett EL, Wang X. Neural representations of temporally modulated signalsin the auditory thalamus of awake primates. J Neurophysiol 97: 1005–1017,2007.

Baumann U, Nobbe A. Pulse rate discrimination with deeply inserted elec-trode arrays. Hear Res 196: 49–57, 2004.

Bendor D, Wang X. The neuronal representation of pitch in primate auditorycortex. Nature 436: 1161–1165, 2005.

Bendor D, Wang X. Cortical representations of pitch in monkeys and humans.Curr Opin Neurobiol 16: 391–399, 2006.

Bendor D, Wang X. Differential neural coding of acoustic flutter withinprimate auditory cortex. Nat Neurosci 10: 763–771, 2007.

Bendor D, Wang X. Neural response properties of primary, rostral, androstrotemporal core fields in the auditory cortex of marmoset monkeys.J Neurophysiol 100: 888–906, 2008.

Besser GM. Some physiological characteristics of auditory flutter fusion inman. Nature 214: 17–19, 1967.

Bizley JK, Walker KM, Silverman BW, King AJ, Schnupp JW. Interde-pendent encoding of pitch, timbre, and spatial location in auditory cortex.J Neurosci 29: 2064–2075, 2009.

DiMattina C, Wang X. Virtual vocalization stimuli for investigating neuralrepresentations of species-specific vocalizations. J Neurophysiol 95: 1244–1262, 2006.

Eggermont JJ. Representation of spectral and temporal sound features in threecortical fields of the cat. Similarities outweigh differences. J Neurophysiol80: 2743–2764, 1998.

Formisano E, Kim DS, Di Salle F, van de Moortele PF, Ugurbil K, GoebelR. Mirror-symmetric tonotopic maps in human primary auditory cortex.Neuron 40: 859–869, 2003.

Gaese BH, Ostwald J. Temporal coding of amplitude and frequency modu-lation in the rat auditory cortex. Eur J Neurosci 7: 438–450, 1995.

Hackett TA, Preuss TM, Kaas JH. Architectonic identification of the coreregion in auditory cortex of macaques, chimpanzees, and humans. J CompNeurol 441: 197–222, 2001.

Hyvarinen J, Sakata H, Talbot WH, Mountcastle VB. Neuronal coding bycortical cells of the frequency of oscillating peripheral stimuli. Science 162:1130–1132, 1968.

Johnson DH. The relationship between spike rate and synchrony in responsesof auditory-nerve fibers to single tones. J Acoust Soc Am 68: 1115–1122,1980.

Joris PX, Schreiner CE, Rees A. Neural processing of amplitude-modulatedsounds. Physiol Rev 84: 541–577, 2004.

Kaas JH, Hackett TA. Subdivisions of auditory cortex and processingstreams in primates. Proc Natl Acad Sci USA 97: 11793–11799, 2000.

Kaernbach C, Demany L. Psychophysical evidence against the autocorrela-tion theory of auditory temporal processing. J Acoust Soc Am 104: 2298–2306, 1998.

Kalluri S, Depireux DA, Shamma SA. Perception and cortical neural codingof harmonic fusion in ferrets. J Acoust Soc Am 123: 2701–2716, 2008.

Krumbholz K, Patterson RD, Pressnitzer D. The lower limit of pitch asdetermined by rate discrimination. J Acoust Soc Am 108: 1170–1180, 2000.

Lemus L, Hernandez A, Romo R. Neural codes for perceptual discriminationof acoustic flutter in the primate auditory cortex. Proc Natl Acad Sci USA106: 9471–9476, 2009.

Lu T, Liang L, Wang X. Temporal and rate representations of time-varyingsignals in the auditory cortex of awake primates. Nat Neurosci 4: 1131–1138, 2001.

Lu T, Wang X. Temporal discharge patterns evoked by rapid sequences ofwide- and narrowband clicks in the primary auditory cortex of cat. J Neu-rophysiol 84: 236–246, 2000.

Lu T, Wang X. Information content of auditory cortical responses to time-varying acoustic stimuli. J Neurophysiol 91: 301–313, 2004.

Malone BJ, Scott BH, Semple MN. Dynamic amplitude coding in theauditory cortex of awake rhesus macaques. J Neurophysiol 98: 1451–1474,2007.

Mardia KV, Jupp P E. Directional Statistics. New York: Wiley, 2000.Miller GA, Taylor WG. The perception of repeated bursts of noise. J Acoust

Soc Am 20: 171–182, 1948.Moore BC. An Introduction to the Psychology of Hearing. London: Academic, 2003.Morel A, Kaas JH. Subdivisions and connections of auditory cortex in owl

monkeys. J Comp Neurol 318: 27–63, 1992.Oxenham AJ, Bernstein JG, Penagos H. Correct tonotopic representation is

necessary for complex pitch perception. Proc Natl Acad Sci USA 101:1421–1425, 2004.

Patterson RD, Handel S, Yost WA, Datta AJ. The relative strength of thetone and noise components in iterated rippled noise. J Acoust Soc Am 98:1355–1364, 1996.

Patterson RD, Uppenkamp S, Johnsrude IS, Griffiths TD. The processingof temporal pitch and melody information in auditory cortex. Neuron 36:767–776, 2002.

Penagos H, Melcher JR, Oxenham AJ. A neural representation of pitchsalience in nonprimary human auditory cortex revealed with functionalmagnetic resonance imaging. J Neurosci 24: 6810–6815, 2004.

Petkov CI, Kayser C, Augath M, Logothetis NK. Functional imaging revealsnumerous fields in the monkey auditory cortex. PLoS Biol 4: e215, 2006.

Phan ML, Recanzone GH. Single-neuron responses to rapidly presentedtemporal sequences in the primary auditory cortex of the awake macaquemonkey. J Neurophysiol 97: 1726–1737, 2007.

Plack CJ, Oxenham AJ, Fay RR, Popper AN. Pitch: neural coding andperception. Springer Handbook of Auditory Research, 2005.

Pollack I. Discrimination of mean temporal interval within jittered auditorypulse trains. J Acoust Soc Am 43: 1107–1112, 1968.

Pressnitzer D, Patterson RD. Distortion products and the perceived pitch ofharmonic complex tones. In: Physiological and Psychophysical Bases of AuditoryFunction, edited by Breebart, DJ, Houtsma, AJM, Kohlrausch, A, Prijs, VF,Schoonoven, R. Maastricht, The Netherlands: Shaker Publishing BV. 2001, p.97–104.

Rennaker RL, Carey HL, Anderson SE, Sloan AM, Kilgard MP. Anes-thesia suppresses nonsynchronous responses to repetitive broadband stimuli.Neuroscience 145: 357–369, 2007.

Ritsma RJ. Existence region of the tonal residue. I J Acoust Soc Am 24:1224–1229, 1962.

Romo R, Salinas E. Flutter discrimination: neural codes, perception, memoryand decision making. Nat Rev Neurosci 4: 203–218, 2003.

Rosen S. Temporal information in speech: acoustic, auditory and linguisticaspects. Philos Trans R Soc Lond B Biol Sci 336: 367–373, 1992.

Sakai M, Chimoto S, Qin L, Sato Y. Neural mechanisms of interstimulusinterval-dependent responses in the primary auditory cortex of awake cats.BMC Neurosci 10: 10, 2009.

Schneider P, Sluming V, Roberts N, Scherg M, Goebel R, Specht HJ,Dosch HG, Bleeck S, Stippich C, Rupp A. Structural and functionalasymmetry of lateral Heschl’s gyrus reflects pitch perception preference. NatNeurosci 8: 1241–1247, 2005.

Schönwiesner M, Zatorre RJ. Depth electrode recordings show double dissoci-ation between pitch processing in lateral Heschl’s gyrus and sound onset processingin medial Heschl’s gyrus. Exp Brain Res 187: 97–105, 2008.

Schreiner CE, Urbas JV. Representation of amplitude modulation in the auditorycortex of the cat. II. Comparison between cortical fields. Hear Res 32: 49–63, 1988.

Schroeder MR, Strube HW. Flat-spectrum speech. J Acoust Soc Am 79:1580–1583, 1986.

Shannon R. Multichannel electrical stimulation of the auditory nerve in man.I. Basic psychophysics. Hear Res 11: 157–189, 1983.

Singh NC, Theunissen FE. Modulation spectra of natural sounds and ethologicaltheories of auditory processing. J Acoust Soc Am. 114: 3394–3411, 2003.



on Septem


Dow

nloaded from


Suga N. Processing of auditory information carried by species-specific com-plex sounds. In: The Cognitive Neurosciences, edited by Gazzanica MS.Cambridge, MA: MIT Press, 1994, p. 295–313.

Wallace MN, Shackleton TM, Palmer AR. Phase-locked responses to puretones in the primary auditory cortex. Hear Res 172: 160–171, 2002.

Wang X, Lu T, Liang L. Cortical processing of temporal modulations. SpeechCommun 41: 107–121, 2003.

Wang X, Lu T, Bendor D, Bartlett E. Neural coding of temporal informationin auditory thalamus and cortex. Neuroscience 157: 484–494, 2008.

Wang X, Lu T, Snider RK, Liang L. Sustained firing in auditory cortexevoked by preferred stimuli. Nature 435: 341–346, 2005.

Yost WA, Mapes-Riordan D, Shofner W, Dye R, Sheft S. Pitch strength ofregular-interval click trains with different length “runs” of regular intervals.J Acoust Soc Am 117: 3054–3068, 2005.

Yost WA, Patterson RD, Sheft S. A time domain description for the pitchstrength of iterated rippled noise. J Acoust Soc Am 99: 1066–1078, 1996

Zeng FG. Temporal pitch in electric hearing. Hear Res 174: 101–106,2002.



on Septem


Dow

nloaded from


neural coding of periodicity in marmoset auditory cortex

Documents