unlimited access with print and download

Free

Continue searching

Coincidence detection, internal delay and neural coding in the gerbil medial superior olive

ProQuest Dissertations and Theses, 2009
Dissertation
Author: Mitchell L Day
Abstract:
The horizontal location of a sound source that contains low-frequency components is largely determined by the difference of arrival times of the sound reaching the two ears - the interaural time difference (ITD). The medial superior olive (MSO) in mammals is the primary central auditory site whose neurons are differentially responsive to ITDs. ITD sensitivity in the MSO is constructed via phase-locked inputs converging on coincidence-detecting neurons. Using a biophysical model of MSO, I show that the dynamic properties of the low-threshold potassium current increases the precision of both phase-locking and coincidence detection in response to realistic inputs. The pathways of input from each ear to the MSO are subject to differential internal delays, creating neurons tuned to different ITDs. Using data from single-unit recordings in gerbil MSO, I show evidence that slight frequency mismatches in the cochlear origin of inputs onto the MSO from each ear (cochlear disparity) contribute to internal delay, in addition to previously-studied delay mechanisms. Further, I show evidence that cochlear disparity produces MSO neurons that are not tuned to a consistent ITD across frequency (e.g., "trough-type" units). ITD tuning in MSO neurons was found to scale with sound level, even in cases where the firing rate decreased. I demonstrate that this observation is inconsistent with the current model of how inhibition contributes to internal delay. The distribution of internal delays across the MSO population has been proposed to be optimized for encoding sound azimuth by being organized as a spatial map across ITD (creating a place code) or broad tuning to one spatial channel (a slope code). I found no evidence for a spatial map of ITD in the gerbil MSO. The distribution of internal delays was not optimized for a place or slope code, although subpopulations existed which would be ideal for each code. Further, there were many units that would not be relevant to sound localization, but would be relevant to signal detection. MSO units were tuned to ITDs of contralateral-leading sounds at lower frequencies, similar to previously-published reports, however there were many ipsilaterally-tuned units at high phase-locking frequencies.

TABLE OF CONTENTS

ACKNOWLEDGMENTS

iii

PREFACE

iv

ABSTRACT

v

LIST OF FIGURES

xi

LIST OF TABLES

xv

LIST OF ABBREVIATIONS

xvi

CHAPTER 1: GENERAL INTRODUCTION

1

1.1 The early history of sound localization

1

1.1.1 The duplex theory of sound loca lization

1

1.1.2 The Jeffress model

4

1.2 The interaural time difference pathway

7

1.2.1 Anatomy

8

1.2.2 Physiology

9

1.3 Outline of the dissertation

11

CHAPTER 2: TEMPORAL PRECISION AND THE DYNAMICS

OF THE LOW - THRESHOLD POTASSIUM CURRENT

15

2.1 I ntroduction

15

2.2 Methods

18

2.2.1 Model

18

2.2.2 Stimuli

21

2.2.3 Pattern classification analysis

21

2.2.4 Physiological conductance stimuli

27

viii

2.3 Results

30

2.3.1 Differential voltage response to current ramps and

random stimuli of diffe rent timescales

30

2.3.2 Differences in STAs

33

2.3.3 Stimulus selection difference for narrowband stimuli

35

2.3.4 Stimulus selection difference for broadband stimuli

41

2.3.5 Stimulus selection difference for physiological

conductance stimuli

4 3

2.4 Discussion

50

2.4.1 Comparison to previous experimental studies

50

2.4.2 I KLT influence on spike encoding is largely bypassed

with broadband stimuli, especially for strong stimuli

52

2.4.3 Pattern classification on spike - triggered stimuli

53

2.4.4 Generality of results, across auditory brainstem

areas and models

55

CHAPTER 3: MECHANISMS OF INTERNAL DELAY

57

3.1 Introduction

57

3.2 Methods

62

3.2.1 Surgery

62

3.2.2 Electrophysiological recording

63

3.2.3 Histology

64

3.2.4 Acoust ic stimulation

64

3.2.5 Data analysis

66

3.2.6 Cochlear disparity model

68

3.3 Results

72

3.3.1 General description of the sample

72

3.3.2 Monaural response types

77

ix

3.3.3 Effects of sound level on monaural and binaural

responses

88

3.3.4 Re lationship between interaural best phase and tone

frequency

106

3.3.5 Monaural and binaural properties of nearby neurons

127

3.4 Discussion

135

3.4.1 Inhibition in the MSO

135

3.4.2 Cochlear disparity as a source of internal delay

140

3.4.3 Non - peak - type phase - frequency relations in the MSO

144

3.4.4 Sample characteristics and histological localization

145

3.4.5 Systematic delay lines

147

3.4.6 Frequency - independent delays

149

3.4.7 Anesthesia

150

CHAPTER 4: NEURAL CODING OF INTERAURAL TI ME

DIFFERENCES

152

4.1 Introduction

152

4.2 Methods

156

4.3 Results

157

4.3.1 Distribution of best delay across best frequency

157

4.3.2 Internal delay and the distribution of best delays

161

4.3.3 Maximal slope of the ITD tuning function

168

4. 3.4 Tonal composite ITD tuning functions

172

4.4 Discussion

187

4.4.1 Imprecise internal delays

187

4.4.2 Bilateral representation at higher best frequencies

193

4.4.3 Encoding sound source azimuth

196

CHAPTER 5: FUTURE DIRECTIONS

202

x

5.1 False dic hotomy between peak - type and trough - type units

202

5.2 Cochlear delays and fixed delays

204

5.3 The development of contralateral representation

207

5.4 Bilateral representation at high phase - locking frequencies

208

5.5 ITD tuning width and localization acuity

209

5.6 I KLT and ITD tuning width

210

5.7 The role of inhibition

211

5.8 Large best delays and signal detection

214

REFERENCES

217

xi

LIST OF FIGURES

FIG. 2.1

Current stimuli

22

FIG. 2.2

Schematic of Fisher linear discrimination on spike -

trigge red stimulus waveforms

24

FIG. 2.3

Comparison of model responses to stimuli with slow

and fast timescales

31

FIG. 2.4

Difference between dynamic and frozen KLT spike

encoding to narrowband stimuli at different center

frequencies

37

FIG. 2.5

The change i n the influence of I KLT on spike encoding

as the time constant of I KLT activation, τ KLT , is varied

40

FIG. 2.6

Difference in spike encoding between dynamic and

frozen KLT models to a stimulus with combined low -

and high - frequency content

42

FIG. 2.7

Dif ference in spike encoding between dynamic and

frozen KLT models to a broadband stimulus

44

FIG. 2.8

Difference between dynamic and frozen KLT model

responses to physiological conductance stimuli

46

FIG. 3.1

Electrolytic lesions

74

FIG. 3.2

Composite map of recording sites

76

FIG. 3.3

Responses to monaural stimulation

78

FIG. 3.4

Phase - locked excitatory activity underlies monaural

xii

unresponsiveness

84

FIG. 3.5

Maximal binaural spike rate is greater than monaural

spike rate

86

FIG. 3.6

Most MSO units with OE monaural response type are

located medial to the MSO column

87

FIG. 3.7

Rate - level functions

90

FIG. 3.8

The difference of monaural mean phases determines

the interaural best phase

92

FIG. 3.9

Monaural mean phases shift approximately equally

bilate rally with sound level to maintain a constant best

phase

94

FIG. 3.10

Interaural best phase changes little with sound level

97

FIG. 3.11

Interaural level differences shift the interaural best

phase

98

FIG. 3.12

Width of IPD tuning function changes little with

sound level

100

FIG. 3.13

Strength of monaural phase - locking changes little with

sound level

103

FIG. 3.14

The phase - frequency relation

107

FIG. 3.15

MSO units can have non - peak - type phase - frequency

relations

110

FIG. 3.16

Population distribution of characteristic phase

113

FIG. 3.17

Units with non - peak - type phase - frequency relations

xiii

and EE monaural response type

114

FIG. 3.18

Period histograms are unimodal and reasonably

symmetric

117

FIG. 3.19

Cochlear disparity model produces non - peak - type

ph ase - frequency relations

119

FIG. 3.20

Cochlear disparity model fits well to trough - type and

in - between - type phase - frequency relations

124

FIG. 3.21

Cochlear disparity model fits well to nonlinear phase -

frequency relations

126

FIG. 3.22

Cochlear disparit y model fits well to peak - type phase -

frequency relations with small, systematic nonlinearities

128

FIG. 3.23

Monaural mean phases of nearby units can be radically

different

130

FIG. 3.24

Best delay is not correlated with the anatomical axes

of MSO

134

FIG. 3.25

Inhibitory current is effectively tonic at most

stimulation frequencies

138

FIG. 4.1

The distribution of best delays

158

FIG. 4.2

Auditory space is represented bilaterally at high phase -

locking frequencies

160

FIG. 4.3

Cochlear disparity intro duces frequency - dependent

variance into the BD - BF distribution

162

FIG. 4.4

Internal delays that exceed the lower π - limit

164

xiv

FIG. 4.5

The best delays of units with peak - type phase - frequency

relations are not confined to a small range

167

FIG. 4.6

Large range in the distribution of fixed delays

169

FIG. 4.7

The majority of ITDs of maximal slope of ITD tuning

functio ns are within the physiological range

170

FIG. 4.8

The central peak of tonal composite ITD functions can

be ambiguous

174

FIG. 4.9

MSO units with the slope of the tonal composite ITD

function within the physiological ITD range

176

FIG. 4.10

Non - MSO SOC u nits with the slope of the tonal

composite ITD function within the physiological ITD

range

178

FIG. 4.11

Units with place code - like tonal composite ITD functions

181

FIG. 4.12

Tonal composite ITD functions with little to no firing

rate within the physio logical ITD range

183

FIG. 4.13

ITD tuning functions with constant firing rate within the

physiological ITD range

185

xv

LIST OF TABLES

TABLE 2.1

Summary of model properties

20

TABLE 3.1

Monaural response types

82

TABLE 3.2

Fit parameters of the cochlear d isparity model

123

xvi

LIST OF ABBREVIATIONS

AN

auditory nerve

AVCN

anteroventral cochlear nucleus

BD

best delay

BF

best frequency

BP

best phase

CD

characteristic delay

CF

characteristic frequency

CN

cochlear nucleus

CP

characteristic phase

DNLL

dorsal nucl eus of the lateral lemniscus

DTX

dendrotoxin

EE

monaural response type: excitatory (ipsi)/excitatory (contra)

EI

monaural response type: excitatory (ipsi)/inhibitory (contra)

EO

monaural response type: excitatory (ipsi)/unresponsive (contra)

EPSC

excitator y post - synaptic current

EPSP

excitatory post - synaptic potential

GABA

γ - aminobutyric acid

I KLT

low - threshold potassium current

xvii

I Na

high - threshold, fast - inactivating sodium current

IC

inferior colliculus

IE

monaural response type: inhibitory (ipsi)/excitator y (contra)

ILD

interaural level difference

IPD

interaural phase difference

IPSC

inhibitory post - synaptic current

IPSP

inhibitory post - synaptic potential

ITD

interaural time difference

K +

potassium ion

KLT

low - threshold potassium

LNTB

lateral nucleus of the trapezoid body

LSO

lateral superior olive

MNTB

medial nucleus of the trapezoid body

MSO

medial superior olive

Na +

sodium ion

NL

nucleus laminaris

OE

monaural response type: unresponsive (ipsi)/excitatory (contra)

OO

monaural response type: unresponsive ( ipsi)/unresponsive (contra)

PBS

phosphate - buffered saline

PSP

post - synaptic potential

PSTH

post - stimulus time histogram

xviii

RM03

Rothman and Manis (2003b) model

SD

standard deviation

SOC

superior olivary complex

SPL

sound pressure level

SPN

superior periolivar y nucleus

SSD

stimulus selection difference

STA

spike - triggered average

STC

spike - triggered covariance

STE

spike - triggered ensemble

VCN

ventral cochlear nucleus

VNLL

ventral nucleus of the lateral lemniscus

1

CHAPTER 1

General Introduction

1.1 THE EARLY HI STORY OF SOUND LOCALIZATION

1.1.1 The duplex theory of sound localization

The investigation into the use of two ears to localize sound began over a century ago. Several investigators were engaged in research on the subject in the late 1800s, but in parti cular the work of Lord Rayleigh stands out, who was to later formalize the “duplex theory” of sound localization in the horizontal (azimuthal) plane. In one paper (Rayleigh 1876) , Rayleigh recounts an experiment where an observer (likely Rayleigh, himself) stands in the middle of a large lawn while assistants move round about him mak ing various sounds. Rayleigh remarked on the ease with which the observer could precisely localize the sound. The easiest explanation for the ability to localize sound was the difference in the level of the sound arriving to the two ears (the interaural level difference, ILD). Rayleigh performed calculations of the ILDs that may be available to a human based on a spherical model of the head. He noted that the ILDs available depended on the frequency of the sound being localized – low - frequency sounds (h aving large wavelengths compared to the diameter of the head) would not cast a sufficient sound shadow to create an ILD. He supposed that one

2

would not be able to localize low - frequency sounds due to the absence of ILDs. Rayleigh tested this on the lawn with assistants striking tuning forks in front of resonator boxes, allowing more control over the pitch of the sound. To his surprise, right and left discriminations were easily made for both low and high frequencies. His dismay at the time was evident i n his writing: “I am obliged to leave the question [of sound localization] in rather an unsatisfactory state, for my calculations are very far from explaining the facts; in fact, they rather go to take away the force from what had hitherto been supposed to be the explanation. At one time I was almost inclined to suppose that we did not distinguish through our ears at all, but in some other mysterious way.” (Rayleigh 1876)

Despite the dismay, Rayleigh noted the theoretical possibility that small differences in arrival time of sound to the two ears (interaural time difference, ITD) co uld be used as a cue to localize sound (Rayleigh 1876) . Later, he experimentall y demonstrated the use of such cues (Rayleigh 1907) . Rayleigh showed that subjects could perceive lateral movement in a particul ar sound now called the “binaural beat” –

a sound stimulus used extensively in my studies of ITD sensitivity. The binaural beat consists of the presentation of pure tones to each ear that are slightly mismatched in frequency, creating a smoothly - changing interaural phase difference (IPD). Rayleigh created the binaural beat by taking identical tuning forks and applying a small amount of wax to one to slightly alter its frequency. The sound from the tuning forks was generated in separate rooms from the sub ject and led to the two ears through tubes,

3

thereby ensuring segregation of the sounds. If the subject detected the IPD, then the perception had to have been generated centrally where phase information from the two ears combined. Rayleigh demonstrated th at the subject could detect the binaural beat for low - frequency tones, but not for high - frequency tones. This led to the classical duplex theory of azimuthal sound localization: IPD cues (or ITD cues) were available to localize low - frequency tones where I LD cues were not, while ILD cues were available to localize high - frequency tones where IPD cues were not (Rayleigh 1907) .

The idea that the detection of IPDs arises centrally was controversial not only because the time differences involved were extremely small, but also because it implied that timing information was carried on electrically in the brain. As Rayleigh remarked at t he time: “It seems no longer possible to hold that the vibratory character of sound terminates at the outer ends of the nerves along which the communication with the brain is established. On the contrary, the processes in the nerve must themselves be vibr atory, not, of course, in the gross mechanical sense, but with preservation of the period and retaining the characteristic of phase.” (Rayleigh 1907)

In fact, another group at the time attempted to alternatively explain the ability to detect IPDs by a peripheral mechanism in which sound from the contralateral ear was conducted through the skull bones to interact with ipsilateral so und at the inner ear (Wilson and Myers 1908) . It wasn’t until the 1930s that electrophysiological recordings were made demonstrating that electrical impulses in the auditory nerve (AN) oscillated with the stimulus frequency, confirming the Rayleigh interpretation of

4

IPD detection (Derbyshire and Davis 1935; Galambos and Davis 1943; Wever and Bray 1930) .

Further support was given to the duplex theory with free - field sound psychophysical measurements in combination with physical measures of ITD and ILD cues: maximal errors in the localization of free - field azimuth were shown to occur in a range of tone frequencies where both ITD and ILD cues were minimal (Stevens and Newman 1936) . The same qualitative result of localization errors at the frequencies of minimal binaural cue overlap has also been reported for my experimental ani mal, the gerbil (Heffner and Heffner 1988) . The duplex theory applies to the localization of pure tone sound; recent psychophysical studies have demonstrated the particular importance of ITD cues in localizing more realistic sound with multiple frequency components. For instance, when bina ural sound over headphones is experimentally manipulated to counterpose ITD and ILD cues, the lateral direction of the ITD cue dominates for sounds containing low - frequency components, including broadband noise (Wightman and Kistler 1992) . Also, envelope ITD cues hav e been shown to be effective in judging the laterality of sounds consisting of only high - frequency components (Henning 1974; Leakey et al. 1958; Macpherson and Middlebrooks 2002; McFadden and Pasanen 1976) .

1.1.2 The Jeffress model

5

In 1948, Lloyd Jeffress proposed an elegant neural model of the localization of low - frequency sound that in many respects is still the basic framework used today (Jeffress 1948) . Jeffress, amongst others at the time, was influenced by the attractive idea that perceptual and cognitive processes were mapped out spatially in the brain. In fact , he began his article entitled “A place theory of sound localization” by quoting a passage from the book, The Physical Dimensions of Consciousness : “…all conscious distinctions are discriminative, that consciousness is, broadly speaking, discrimination. Such a view implies that all ‘knowledge’ is potentially spatial in a physiological sense… In a certain sense we are looking for a ‘place theory’ of every dimension.” (Boring 1933)

By 1948, several facts were kn own about sound localization and the general properties of the nervous system that Jeffress used to craft his place theory of sound localization: 1) ITD cues are used to localize low - frequency pure tones; 2) the impulses of the AN carry timing information of pure tones; 3) synaptic inputs can spatially summate to elicit a spike from the convergent neuron; and 4) the conduction of spikes is slower for axonal fibers with a smaller diameter. Jeffress proposed that some area along the auditory pathway after th e cochlear nuclei would act as a binaural comparator. He assumed that the timing information evident in spike trains of low - frequency AN fibers would be carried on in the output of the cochlear nuclei. The output from opposite cochlear nuclei (CN) would converge on the binaural nucleus, which was assumed to act as a coincidence detector based on the spatial summation of

6

synaptic inputs. The “place” concept derived from the proposed architecture of the binaural nucleus: collatateral axons would branch off of the main afferent fibers arriving from each side onto an array of binaural comparator neurons. The collaterals of one side would branch off systematically along some direction onto the array of neurons, creating a smooth progression of neural delay (e .g., short to long delays from top to bottom). The collaterals from the opposite side would branch off in the opposite direction creating counter delays (e.g., long to short delays from top to bottom). Being a coincidence detector, any given neuron in th e array would only spike when the ITD of the sound matched the difference in conduction delays from each side (i.e., the internal delay). Thus, the place theory of ITD – the activity of each neuron in the array would indicate a sound with a particular ITD , and therefore, a particular azimuthal location. Further, the topographic arrangement of neurons tuned to particular ITDs would not be haphazard, but would smoothly vary over the binaural array, forming a topographic map of ITD.

The fourth fact listed above (slower conduction with smaller axonal diameter) Jeffress used to make his model more physiologically plausible. The total range of ITDs accessible to humans (~1,400 µs) is relatively large, and would require a sizeable difference in collateral path lengths to create the necessary delays. By proposing that the collaterals were of smaller diameter, and therefore would conduct impulses more slowly, Jeffress argued that the necessary delays could be constructed from a more plausible, compact neural stru cture.

7

Based on anatomical considerations and the minimal amount of physiological data from the central auditory system in 1948, Jeffress speculated that the primary site of ITD sensitivity was the inferior colliculus (IC). Today, there exists much evid ence that the primary site is actually the medial superior olive (MSO) in mammals and the nucleus laminaris (NL) in birds. Much of the Jeffress model has been experimentally confirmed, specifically the precisely phase - locked input from the CN, and the bio physical function of MSO neurons as detectors of coincident synaptic input. The proposal that the nucleus of binaural comparison is constructed as a topographic map of ITD – which is really the essence of the Jeffress model – has received the least experi mental attention. There is strong evidence that a topographic map of ITD exists in the NL of the barn owl (Carr and Konis hi 1990) , and more recently, the chicken (Koppl and Carr 2008) . Further, auditory spatial maps in the barn owl are evident in other brain structures downstream from the NL (Konishi 2003) . However, there is scant evidence for a topographic map in mammals, and continued controversy over the existence of other coding schemes of ITD in the MSO other than a place code (Joris and Yin 2007; McAlpine and Grothe 2003; Schnupp and Carr 2009) . In particular, the gerbil has been implicated as a species whose existing data is most supportive of an alternative ITD code (for review, see (Joris and Yin 2007) .

1.2 THE INTERAURAL TIME DIFFERENCE PATHWAY

8

1.2.1 Anatomy

The main pathway to the MSO consists of sound being transduced in the cochlea to electrical impulses in the AN, which then synapse onto neurons of the anteroventral cochlear nucleus (AVCN), which then send axons to the ipsilateral a nd contralateral MSOs. Specifically, spherical bushy cells of the AVCN send axons that split and provide the only excitatory input to the ipsilateral and contralateral MSOs (Cant and Casseday 1986; Tolbert et al. 19 82) . Therefore, each MSO neuron receives excitatory input from the ipsilateral and contralateral AVCN. Spherical bushy cells are themselves innervated by one to three AN fibers that synapse onto the soma of the bushy cells through specialized terminals called the endbulbs of Held (Brawer and Morest 1975; Sento and Ryugo 1989) . MSO neurons have bipolar dendrites and are arranged in a sheet roughly orthogonal to the mediolateral axis; one dendrite protrudes mediall y and the other laterally. Inputs from opposite AVCN are segregated with the medial dendrite receiving input from the contralateral AVCN and the lateral dendrite receiving input from the ipsilateral AVCN (Lindsey 19 75; Smith et al. 1993; Stotler 1953) .

The MSO also receives inhibitory inputs. The medial nucleus of the trapezoid body (MNTB) is known to have a prominent glycinergic projection to the lateral superior olive (LSO) with collaterals splitting off and i nnervating the ipsilateral MSO (Adams and Mugnaini 1990; Banks and Smith 1992; Kuwabara and Zook 1992; Smith et al. 1998) . The MNTB is itself innervated by globular bushy cells of the

9

contralateral AVCN, which syna pse onto the soma of a MNTB neuron in a specialized, claw - like synapse called the calyx of Held (Friauf and Ostwald 1988; Guinan and Li 1990; Kuwabara et al. 1991; Smith et al. 1991; Sommer et al. 1993; Spirou et al. 1990; Warr 1972) . This calyx ensures fast and reliable transmission of AVCN input, thereby transmitting reliable contralateral inhibition onto the LSO and MSO. The lateral nucleus of the trapezoid body (LNTB) is also known to have a glycinergic project ion to the MSO (Ca nt and Hyson 1992) . The LNTB is itself innervated by globular bushy cells of the ipsilateral AVCN. Glycinergic synapses onto MSO neurons are originally dispersed across dendrites and soma, but have been shown to cluster specifically onto the soma throug h an experience - dependent process during development (Kapfer et al. 2002) .

Altogether, the MSO receives ipsilateral and contralateral excitatory input from the ipsilateral and contralateral AVCN, respectively; contralateral inhibitory input via the ipsilateral MNTB; and ipsilater al inhibitory input via the ipsilateral LNTB.

1.2.2 Physiology

The excitatory inputs to the MSO are strongly phase - locked to low - frequency tones; in fact, the spike trains of both spherical and globular bushy cells of the AVCN have been shown to synchr onize to tone frequency at a stronger level than their own input, the AN (Joris et al. 1994) . Further, the inhibitory inputs to MSO are also highly phase - locked – the response of MNTB cells has been shown to be similar to their

10

input, the globular bushy cells, due to the nearly one - to - one transmission conferred by the calyceal synapse (Guinan et al. 1972a; Smith et al. 1998; Tsuchitani 1997) . Therefore, the inputs onto MSO relay precise timing information as originally assumed in t he Jeffress model.

Electrophysiological recordings in brain slice have confirmed that the excitatory inputs onto MSO create glutamatergic excitatory post - synaptic potentials (EPSPs) (Smith 1995) . While excitation appears to be the dominant input to MSO, stimulus - evoked inhibitory post - synaptic potentials (IPSPs) could be elicited in 89% of MSO neurons, and increased enough at high stimulation amplitude to i nhibit excitation (Grothe and Sanes 1993; 1994) .

The first demonstration of response sensitivity to ITDs was in the IC (Rose et al. 1966) , which receives output from the MSO. Earlier, binaural sensitivity had been demon strated in the MSO, but not specifically ITD sensitivity (e.g., (Moushegian et al. 1964) . The first study of coincidence detection in the MSO was that of Goldberg and Brown (1969) in the dog. This study presented data from a sizeable number of single units that showed clear ITD sensitivity in response to pure tones with static ITDs. Further, they demonstrated that the binaural response maximum at the peak of the ITD tuning function was greater tha n the responses to monaural stimulation, and the binaural response minimum was often suppressed with respect to monaural stimulation. Importantly, Goldberg and Brown provided evidence for coincidence detection by demonstrating that the IPD of maximal resp onse was approximately

Full document contains 257 pages
Abstract: The horizontal location of a sound source that contains low-frequency components is largely determined by the difference of arrival times of the sound reaching the two ears - the interaural time difference (ITD). The medial superior olive (MSO) in mammals is the primary central auditory site whose neurons are differentially responsive to ITDs. ITD sensitivity in the MSO is constructed via phase-locked inputs converging on coincidence-detecting neurons. Using a biophysical model of MSO, I show that the dynamic properties of the low-threshold potassium current increases the precision of both phase-locking and coincidence detection in response to realistic inputs. The pathways of input from each ear to the MSO are subject to differential internal delays, creating neurons tuned to different ITDs. Using data from single-unit recordings in gerbil MSO, I show evidence that slight frequency mismatches in the cochlear origin of inputs onto the MSO from each ear (cochlear disparity) contribute to internal delay, in addition to previously-studied delay mechanisms. Further, I show evidence that cochlear disparity produces MSO neurons that are not tuned to a consistent ITD across frequency (e.g., "trough-type" units). ITD tuning in MSO neurons was found to scale with sound level, even in cases where the firing rate decreased. I demonstrate that this observation is inconsistent with the current model of how inhibition contributes to internal delay. The distribution of internal delays across the MSO population has been proposed to be optimized for encoding sound azimuth by being organized as a spatial map across ITD (creating a place code) or broad tuning to one spatial channel (a slope code). I found no evidence for a spatial map of ITD in the gerbil MSO. The distribution of internal delays was not optimized for a place or slope code, although subpopulations existed which would be ideal for each code. Further, there were many units that would not be relevant to sound localization, but would be relevant to signal detection. MSO units were tuned to ITDs of contralateral-leading sounds at lower frequencies, similar to previously-published reports, however there were many ipsilaterally-tuned units at high phase-locking frequencies.