The Medial Superior Olivary Nucleus: Meeting the Need for Speed
Abstract and Keywords
It doesn’t get any more precise than this: Neurons in the medial superior olive (MSO) in many mammals, including humans, are sensitive to temporal differences between their synaptic inputs of only a few microseconds. These neurons are the basis for our ability to resolve even a 5-μs disparity in the time of arrival of a sound at each ear. This capacity enables humans to discriminate between temporally overlapping sounds based on spatial segregation, as for instance, at a cocktail party or a poster session at a scientific conference. This chapter aims at providing a comprehensive summary on the current state of research regarding the MSO, ranging from cellular and circuit anatomy to subcellular and channel physiology to spatial coding and perception. Consequently, the chapter is subdivided according to these thematic aspects. Nonetheless, while such subdivisions are helpful for providing structure to the reader, they can also convey the inter-dependencies between these topics: for example, when studying the spatial sensitivity of MSO neurons on the mechanistic level, it is crucial to also consider its anatomical specializations (on the subcellular and circuit level), as well as corresponding perceptional phenomena. Finally, the chapter suggests a complete picture of the MSO only emerges by including evolutionary considerations, that is, the phylogenetic origin of the many fascinating specializations that can be observed within the MSO circuit.
Preface: Definition and Computations
Traditionally, neuronal structures are defined as regions or nuclei that can be anatomically delineated and, ideally, unambiguously recognized in all species of a clade (here: in all mammals). Usually, it is irrelevant whether the structure in question contains different neuronal types and/or serves one or more functions. The cochlear nucleus (CN), for instance, contains multiple neuronal types that clearly serve different functions, but these cell types can be identified in all mammals and always permit unambiguous delineation of the CN. In the case of the MSO, matters are more complex, and its delimitation has been guided largely by the morphology of a specific cell type, which is arranged in a particular way and performs an explicit function, rather than by comparative means. Specifically, the textbook definition of the MSO is as follows: A structure that consists of bipolar neurons whose somata are located in one parasagittal plane, whose dendrites are oriented orthogonally to this plane, point medially and laterally, and serve only one function. This “defining” function is the detection of microsecond differences in the time of arrival of inputs from the two ears (interaural time difference, ITD), which allows differences in the bilateral time of arrival of sounds to be utilized for sound localization in the horizontal plane. This straight-forward view of the MSO goes back to early studies in carnivores (e.g., cats and dogs), animals in which the MSO attracted attention because of the sheer beauty of its unique anatomical organization (Goldberg & Brown, 1969; Ramón y Cajal, 1909; Figure 1). Moreover, the MSOs of dogs and cats were among the first shown to be sensitive to ITDs (Caird & Klinke, 1983; Goldberg & Brown, 1969; Yin & Chan, 1990). Although its cytoarchitecture (Kapfer et al., 2002; Rautenberg, Grothe, & Felmy, 2009) and its ITD sensitivity (Brand et al., 2002; Franken, Bremen, & Joris, 2014; Pecka et al., 2008; Spitzer & Semple, 1995; van der Heijden et al., 2013) have since been confirmed in other ITD-using mammals, reducing the MSO to a single neuron type and a single function is problematic for the following reasons:
1. The MSO includes neurons that deviate from the strictly bipolar type and their unique biophysical properties (Smith, 1995).
2. Even the predominant bipolar neurons are not always arranged within one parasagittal plane, as they are in dogs, cats, or gerbils. In small mammals that do not use ITDs for sound localization, like various bats (reviewed in Grothe, 2000), mice (Fischl et al., 2016), rats, and short-tailed opossums (Kapfer et al., 2002), these cells are distributed within a more spherical MSO. Indeed, this diversity has given rise to the erroneous claim that the MSO is missing in these mammals.
3. Synaptic input patterns of MSO bipolar neurons differ across mammals, although they receive the same set of inputs from both sides of the brain (to be discussed in “The MSO Circuit and Anatomical Adaptations for Speed”). In mammals that use ITDs, such as cats, gerbils, and chinchillas, the four major axonal projections to MSO are arranged in an orderly fashion, with ipsilateral glutamatergic inputs targeting only lateral, and contralateral glutamatergic inputs targeting only medial dendrites. In addition, glycinergic inputs from both sides synapse only in the somatic region of these neurons (Clark, 1969; Kapfer et al., 2002; Kuwabara & Zook, 1992; Perkins, 1973; Stotler, 1953; Werthat et al., 2008). In contrast, in mammals that do not exploit ITDs, the inhibitory and excitatory inputs are intermingled along the somata and among the dendrites (Kapfer et al., 2002).
4. Finally, bipolar neurons in the MSO of bats can be highly sensitive to temporal features of sounds (like amplitude modulations; Grothe, 1994), but there ITD sensitivity seems to be too crude to serve sound localization for these small animals (Grothe & Park, 1998; Harnischfeger, Neuweiler, & Schlegel, 1985) or show predominantly monaural response properties (Covey, Vater, & Casseday, 1991; Grothe et al., 1992).
A phylogenetic framework that can explain these differences may be found in Grothe (2000) and (Grothe & Pecka, 2014). In this chapter, we follow a strategy that avoids problems relating to the anatomical “delineation” of nuclei/structures and simplifies matters, without losing sight of the organizational and functional variations found across different species. This can be done by simply addressing specific cell types with definable functional and/or computational properties independently of their exact location or arrangement within the SOC, rather than focusing on structures that contain multiple neuronal types. Emphasizing cell identity is the only way to disentangle complexities associated with the definition of a nucleus across species, and it may also help clarify the definitions of the superior periolivary/paraolivary nucleus (SPN/SPON); for example, with regard to the OFF cells that have been assigned to different nuclei in different species (compare to chapter by Magnusson & Gómez-Álvarez, this volume; see also Behrend et al., 2002).
With regard to MSO cell types, rather little is known about the properties of the “non-principal” neurons, except that they are multipolar, exhibit low thresholds, and fire repetitively upon current injections in vitro, which contrasts with the high threshold phasic responses seen in bipolar neurons (Smith, 1995). Nothing is known about their responses or function in vivo. We will therefore focus on the principal MSO neurons in ITD-using mammals only. These neurons are tuned for execution of the temporally most precise computations known in mammals, which rely on their anatomical and synaptic arrangement, as well as on some peculiar features of their inputs and their rather unique biophysical properties.
We will address and summarize in turn (1) the anatomical arrangement of principal cells, including synaptic inputs and the morphological adaptations for speed; (2) their biophysical properties, which permit processing of information on microsecond scales; (3) the computational tasks involved in ITD processing; and (4) the functional significance of the coding properties of these cells and their implications for auditory spatial perception.
The MSO Circuit and Anatomical Adaptations for Speed
A major feature of the MSO and its input pathways in mammals that use ITDs (Figure 2) is the structural organization that enables accurate registration of the arrival times of inputs. The specializations begin at the level of the two classes of bushy cells and their inputs (see chapter by Oertel, Cao, & Recio-Spinoza, this volume), which provide precise information relating to the temporal features of stimuli, such as envelope onsets and the fine structure of ongoing components (e.g., phase-locking to low-frequency sounds). Spherical bushy cells (SBCs) on both sides provide the binaural excitatory, glutamatergic inputs to the MSO, with ipsilateral inputs targeting the lateral and contralateral inputs the medial dendrites. Hence, as first described in detail by Stotler (1953), ipsi- and contralateral inputs are kept spatially separated at the level of the MSO neurons. It follows that owing to the longer path length, excitation from the contralateral side arrives slightly later than that from the ipsilateral ear. Because spike generation by the MSO neurons depends on coincidence detection (see under the following heading, “Biophysical Properties”), this configuration of inputs may itself generate a “preferred ITD”; that is, provide an external delay between incoming signals from the two ears that precisely offsets the difference in axonal path length between the excitatory inputs from the two sides. This “preferred delay” corresponds to an ITD that favors sounds from the contralateral hemisphere, since slightly earlier arrival of the sound at the contralateral ear will compensate for the internal delay. Indeed, this slight contralateral bias has been experimentally observed in the MSO (Brand et al., 2002; Franken, Bremen, & Joris, 2014; Pecka et al., 2008; to be discussed in “The Nature of Internal Delays” and Figure 6c).
In contrast to its avian analogue the nucleus laminaris (Seidl, Rubel, & Harris, 2010), in which the excitatory inputs are arranged as delay lines that provide different combinations of internal delays to different binaural neurons (in accordance with the scenario proposed by L. Jeffress [Jeffress, 1948]), there is no evidence for systematic variation in axonal length in the MSO inputs (Grothe, 2000; Karino et al., 2011).
In addition to bilateral excitation, the MSO receives bilateral inhibitory synaptic inputs and at least one additional inhibitory feedback input that acts via volume transmission (Figure 2). The latter originates from a monosynaptic MSO-SPN-MSO feedback loop. We will address the function of this GABAergic volume transmission later on in the section “The Nature of the Spatial Code and Dynamics of Spatial Processing”. The direct synaptic inhibition is glycinergic (Grothe & Sanes 1993, 1994) and comes from the medial and, less prominently, from the lateral nucleus of the trapezoid body (MNTB, LNTB, respectively; (Kuwabara & Zook, 1992; Spirou & Berrebi, 1997; Spirou, Rowland, & Berrebi, 1998). In contrast to the case in other mammals, in which these inputs target both dendrites and somata of MSO neurons, experience-dependent developmental pruning of glycinergic inputs from the dendrites takes place in ITD-using mammals, which restricts large inhibitory synapses to the somata and the most proximal dendrites (Couchman, Grothe, & Felmy, 2012; Kapfer et al., 2002; Werthat et al., 2008; Figure 3). This arrangement correlates with the fact that strong inhibitory currents can be elicited by only a small number of input fibers (Couchman, Grothe, & Felmy, 2010). Perturbation of this developmental removal of dendritic inhibitory inputs (Kapfer et al., 2002) disrupts the normal development of ITD tuning in gerbils (Seidl & Grothe, 2005). How such inhibition may alter ITD tuning will be discussed as well.
The inhibitory pathway starts with globular bushy cells (GBCs), which excite a monaural (ipsilateral) subpopulation of LNTB neurons (Roberts, Seeman, & Golding, 2014) via large endbulb of Held-like synapses (Spirou, Rowland, & Berrebi, 1998) and (contralateral) MNTB neurons via the extensively studied calyx of Held synapse (Borst & Soria van Hoeve, 2012). LNTB and, more prominently, MNTB then provide glycine-mediated hyperpolarizing inhibition to the MSO somata (Grothe & Sanes, 1993; Kapfer et al., 2002; Roberts, Seeman, & Golding, 2014).
The GBC-MNTB-MSO pathway in particular exhibits unique adaptations for fast conduction of action potentials and temporally precise transmission. First, GBCs receive strong synaptic inputs from auditory nerve (AN) fibers (Xu-Friedman & Regehr, 2005; see Oertel et al., this volume) that cause action potentials to follow the temporal structure of inputs with the highest possible temporal precision. For instance, the precision of phase-locking in GBCs surpasses that of single AN fibers, as shown for cats and gerbils (Joris et al., 1994; Wei et al., 2017). Second, GBC fibers have the largest axon diameter known in the auditory brainstem (Ford et al., 2015; Morest, 1968), which partly explains their higher conduction velocity. Moreover, GBC axons tuned to low frequencies have recently been shown to exhibit a unique myelination pattern: although they are even thicker than GBCs tuned to higher frequencies (and much thicker than SBC axons), their internodes are significantly shorter. Although at first sight paradoxical, this pattern in fact results in an even higher conduction velocity (Ford et al., 2015; Figure 2, Figure 6B). Notably, this unusual myelination pattern is not found on GBC axons of mice, which do not utilize ITDs; hence, it appears to be specifically related to ITD processing (Stange-Marten et al., 2017). Third, each MNTB neuron receives input from a GBC fiber via a single calyx of Held synapse, a type known for its exceptional speed and reliability (Borst & Soria van Hoeve, 2012). Fourth, as has recently been shown in gerbils, a significant proportion of such synapses tuned to low sound frequencies do not prolong synaptic delays during ongoing activity, a feature that is highly unusual for chemical synapses (Stange-Marten et al., 2017). Hence, the additional synapse not only expedites conduction speed, but in these cases also ensures absolute time constancy despite the additional synapse. Finally, the inputs are restricted to the MSO cell somata (Clark, 1969; Kapfer et al., 2002), which results in precise and strong inhibition (Couchman, Grothe, & Felmy, 2012; Grothe & Sanes, 1994; Roberts, Seeman, & Golding, 2013).
These unusual properties of the main inhibitory pathway to the MSO relate to a number of indirect and direct observations that demonstrate that this inhibition, despite the additional synapse in the MNTB, precedes the direct excitation from contralateral SBGs (Brand et al., 2002; Goldwyn et al., 2017; Grothe & Pecka, 2014; Roberts, Seeman, & Golding, 2013) and acts in a phase-locked manner (Pecka et al., 2008; see also figure 1d in Franken et al., 2015; compare figures 2 and 5 in Goldwyn et al., 2017). Less is known about the weaker LNTB-mediated inhibition, but in vitro recordings suggest that its timing may be less reliable than that of MNTB-driven inhibition (Roberts, Seeman, & Golding, 2014). Since strong inhibitory input is consistently found in in vitro MSO preparations, it is clear that inhibition at least contributes to cellular ITD processing. Its particular role is nonetheless still under debate, as we will discuss. But first we turn to biophysical aspects of cellular coincidence detection of excitatory and inhibitory inputs.
To achieve cycle-by-cycle ITD sensitivity of their phase-locked inputs, MSO neurons must be able to limit temporal integration of synaptic currents to the oscillation period of the fine structure of their bushy cell inputs. For example, if we assume the phase-locking limit of bushy cells to be about 2 kHz, the memory time scale of the coincidence mechanism must be at least 1/(2 kHz) = 500 µs or lower. Such microsecond-scale operation places strong constraints on the biophysical properties of the neural elements involved, including dendritic filtering, synaptic dynamics, and transmembrane channel kinetics. In the following paragraphs, we will review these aspects in the mature MSO cell and consider how they contribute to ITD processing.
Dendrite Morphology and Synaptic Distributions
As mentioned earlier, a defining anatomical property of MSO principal neurons is their bipolar form, with one lateral and one medial dendrite originating from the soma ( Henkel & Brunso-Bechtold, 1990; Kiss & Majorossy, 1983; Rautenberg, Grothe, & Felmy, 2009; Stotler, 1953). The two major dendrites are relatively thick (~4 µm diameter) close to the soma, such that the border between soma and dendrite is not clearly demarcated anatomically, and is largely a matter of definition (Rautenberg, Grothe & Felmy, 2009). The primary dendrites are relatively short (100 to 150 µm) and tapered, terminating in filopodia-like structures. Often one also observes dendrites branching at an intermediate distance from the soma and leading to two major dendritic trunks. The marked bipolar morphology serves two major functional purposes:
1. Excitatory synaptic inputs from the two input pathways remain anatomically segregated, with inhibitory synapses from either hemisphere being restricted to the soma (Clark, 1969; Kapfer et al., 2002; Kuwabara & Zook, 1992; Stotler, 1953).
2. Owing to the thickness of the proximal dendrite its axial resistance is very low, and it therefore has little temporal filtering capacity. The dendrites also act as large current sinks, which results in additional leakage from the soma (Agmon-Snir, Carr, & Rinzel, 1998; Dasika, White, & Colburn, 2007; Rall, 1962). The relatively small diameters of the distal dendrites result in locally higher input resistances and thereby amplify voltage responses to distal inputs relative to more proximal inputs (Mathews et al., 2010).
It has been argued that the separation of inputs on dendrites facilitates binaural coincidence detection via a ceiling effect imposed by the synaptic reversal potential (Agmon-Snir, Carr, & Rinzel, 1998; Dasika, White, & Colburn, 2007), which leads to an effective reduction of the threshold conductance for binaural inputs relative to monaural inputs. This threshold effect, however, is particularly strong for thin dendrites and might thus be more relevant to the avian binaural coincidence detector in the nucleus laminaris (Smith & Rubel, 1979) than to the mammalian MSO with its thick dendrites.
Thus, despite the clear anatomical allocation of the binaural input pathways to the lateral and medial dendrites first described by Stotler (1953), its functional role remains unclear. It may have an important developmental function, for example in selecting the composition of inputs that provide precise ipsi- or contralateral excitation/inhibition appropriately tuned to the timing of the other inputs. In this context, a detailed description of how individual fibers make synaptic contacts along the dendrite is of the utmost importance, because this pattern defines the temporal profile of dendritic depolarization upon unilateral excitation. Unfortunately, very little is known about the properties of single fibers. Based on single-fiber stimulation, Couchman et al. (2010) estimated that only 1 or 2 active excitatory SBC fibers per cochlear nucleus are necessary to evoke an action potential from the resting state. However, this result was obtained using single stimuli applied to slices during blockade of inhibitory inputs. When repetitive stimulation (pulse trains of few hundred Hz) is employed under similar conditions, these numbers increase by more than twofold due to synaptic adaptation (N. Kladisios and F. Felmy, personal communication, January 26, 2018). But even the latter figures are surprisingly small, given the large number of excitatory synaptic contacts that have been described anatomically (Stotler, 1953). The available evidence thus suggests that individual fibers may make multiple contacts on one dendrite. An anatomical verification of this assertion is, however, still lacking.
Synaptic Kinetics and Integration
Synaptic transmission at MSO cells mainly relies on two ligand-gated ion channels, amino-3-hydroxyl-5-methyl-4-isoxazole propionate (AMPA) receptor-mediated excitation and glycine receptor-mediated inhibition. Only a relatively small (< 5%) N-methyl-D-aspartate (NMDA) receptor-mediated current has been reported to contribute to EPSCs in the adult MSO (Couchman, Grothe, & Felmy, 2010; Smith, Owens, & Forsythe, 2000). MSO AMPA signaling is spectacularly fast. In contrast to the AMPA decay time constants of sometimes 10 and more milliseconds known from cortical synapses (see e.g., (Salin et al., 1996)), voltage clamp recordings from MSO neurons reveal decay time constants of about 250 µs (Couchman, Grothe, & Felmy, 2010; Figure 4A). At least synaptically, these short decay time constants would allow to detect the coincidence of excitatory currents for inputs up to about 4 kHz at least. Synaptic inhibition on the other hand exhibits much slower kinetics, decaying within about 1.5 ms (Couchman, Grothe, & Felmy, 2010; Myoga et al., 2014). Based on these studies, combined with other in vitro and in vivo data (Grothe & Sanes, 1994; Myoga et al., 2014; Pecka et al., 2008), the cycle-by-cycle action of inhibition is assumed to be restricted to frequencies up to approximately 1000 Hz (Figure 4B, C).
The mechanism of integration of the four synaptic input pathways has been controversially discussed in the literature, particularly with regard to the ability of the glycinergic currents to contribute to cycle-by-cycle coincidence detection (Franken et al., 2015; Myoga et al., 2014; Pecka et al., 2008; Roberts, Seeman, & Golding, 2013). There is general agreement that excitatory coincidence detection constitutes the backbone of ITD processing (Franken et al., 2015; Jeffress, 1948; Joris, Smith, & Yin, 1998; Leibold, 2010; Myoga et al., 2014; Pecka et al., 2008) and also that the spike rate of MSO cells can largely be accounted for by a linear combination of the binaural inputs (Joris et al., 2006; van der Heijden et al., 2013; Plauška, Borst, & van der Heijden, 2016); but see (Franken et al., 2015). However, both the anatomical evidence for strong and relatively fast-acting (Figure 4B) and temporally preceding (Figure 4D) glycinergic inhibitory synapses (see section 1) and the effects of pharmacological blockade of glycinergic signaling on ITD tuning (Brand et al., 2002; Pecka et al., 2008) suggest a fundamental role of the two inhibitory pathways in shaping the ITD tuning curves, particularly in determining the position of the preferred ITD (Figure 4C). Explaining such inhibition-induced peak shifts in models in which inhibitory and excitatory currents act on the same time scale is quite straightforward (Brand et al., 2002; Leibold and & Hemmen, 2005), if the contralateral inhibition precedes excitation. Incorporation of the measured, slower inhibitory synaptic kinetics generally does not produce shifts in the preferred ITD (Day & Semple, 2011; Leibold & van Hemmen, 2005; Roberts, Seeman, & Golding, 2013; Zhou, Carney, & Colburn, 2005) for all delays between contralateral excitation and inhibition, but nevertheless shows such effects for a wide range of delays (Leibold, 2010; Myoga et al., 2014). The question of the functional role of inhibition for ITD processing will be discussed in detail later.
The main determinant of the short memory time scale of MSO neurons is their short membrane time constant of 300 µs and sometimes even less (Couchman, Grothe, & Felmy, 2010), which is attained during early development (Figure 4E). For comparison, typical membrane time constants of cortical neurons are in the range of 10 milliseconds (Stuart & Sakmann, 1995). Apart from the anatomical leakage contribution imposed by the thick proximal dendrites as described earlier, the time constant is mainly determined by the fact that transmembrane channels continue to conduct at rest. Two such channels have been described in great detail in the literature: (1) a hyperpolarization-activated, cyclic nucleotide-gated (HCN) channel (Baumann et al., 2013; Khurana et al., 2011), which generates a cationic inward current, and (2) a dendrotoxin (DTX)-sensitive voltage-gated potassium (Kv1) channel (Mathews et al., 2010; Svirskis et al., 2002), which generates an outward current. The two channels balance each other and thereby fix both the resting potential at about –67 mV (Couchman, Grothe, & Felmy, 2010) and the input resistance of 2–8 mohm in mature neurons (Couchman, Grothe, & Felmy, 2010; Scott, Mathews, & Golding, 2005; ). Apart from their directionality, the channels mainly differ in their time constants. Kv1 channels are fast opening (τ ~ 1.5 ms), whereas the HCN channel kinetics is much slower, with a reaction time of ~100 ms, far longer than the dynamics typically induced by the fine structure of the stimulus.
Functionally, the fast Kv1 kinetics has been shown to shorten EPSPs (Mathews et al., 2010; Scott, Mathews, & Golding, 2005; Svirskis et al., 2002; Svirskis, Dodla, & Rinzel, 2003), thereby further reducing the time window for coincidence detection of the AMPA-mediated synaptic potentials. The same argument, however, also applies to the additional sharpening of glycine-mediated inhibitory potentials (Baumann et al., 2013). Furthermore, sharpening can be achieved not only by fast Kv1 kinetics, but also by the HCN reversal potential of about –35 mV, which speeds up the decay of IPSPs (Baumann et al., 2013). Thus, the membrane potential dynamics results from an intricate non-linear interplay between all synaptic and channel kinetics.
In addition to temporal sharpening, fast Kv1 activation has been shown to underlie membrane potential resonances in MSO cells (Remme et al., 2014; Mikiel-Hunter, Kotak, & Rinzel, 2016; Fischer, Leibold, & Felmy, 2018), which would amplify input fine structure in the range of several hundred Hz. The functional role of these resonances is not fully understood. Modeling and in vitro physiology suggest that these resonances permit a frequency-dependent reduction of the spike threshold (Lehnert et al., 2014; Remme et al., 2014; Figure 4F).
The Kv1 kinetics has also been shown to affect preferred ITDs in vitro (Myoga et al., 2014) and in modeling studies (Jercog et al., 2010; Myoga et al., 2014). Since Kv1 opens with a short time constant of only ~1.5 ms, it will be sensitive to the slope of the subthreshold PSCs. Fast-rising PSCs will not be affected by Kv1 opening, whereas slow depolarization will be counterbalanced by Kv1 opening. This effective high-pass filtering will differentially affect the inputs from the two ears (and hence shift preferred ITD) if they have different slopes, either because of differential temporal jitter or differences in the relative delay between excitation and inhibition (Jercog et al., 2010; Myoga et al., 2014).
Based on computational modeling, Kv1 and HCN1 channels have been proposed to be non-isotropically distributed along the dendrites, exhibiting a higher density at the soma (Mathews et al., 2010). However, antibody staining of HCN also shows strong immunofluorescence at the dendrites(Khurana et al., 2011). The putative somatic colocalization of Kv1 channels and glycinergic synapses argues in favor of a functional role for both currents (Figure 4G) rather than cooperation between Kv1 channels and (spatially segregated) glutamatergic synapses (Baumann et al., 2013; Myoga et al., 2014). Moreover, since Myoga et al. (2014) have shown that temporal broadening of the EPSPs by adding jitter actually amplifies inhibition-induced peak shifts, the idea that Kv1 mainly sharpens the already ultrabrief EPSCs would not seem necessary, at least from the perspective of the inhibition model.
While bath application of DTX suppresses a large fraction of the potassium currents, a background of DTX-insensitive outward currents with slower kinetics still remains (Scott, Mathews and Golding, 2005), which argues for the presence of additional yet unidentified potassium channels with as yet unknown functional roles.
Action Potential Generation
MSO cells exhibit unusually small action potential (AP) amplitudes of 10 mV or less (Yin & Chan, 1990; Scott, Hage, & Golding, 2007; Couchman, Grothe, & Felmy, 2010; Figure 4F, H). For practical reasons, it has even been argued that in vitro APs can be more reliably detected by measuring the associated after-hyperpolarization (Figure 4H) rather than their upstroke and peak (Couchman, Grothe, & Felmy, 2010). The large somatic leak currents described in the previous section are the most obvious reason for tiny APs. But if most of the synaptic currents leave the cell through subthreshold channels, how can a sufficiently large proportion of the ions remaining actually reach the spike-generating zone of the thin axon and sustain excitability? In a modeling study based on measured axonal morphology, Lehnert et al. (2014) showed that the large somatic leak may indeed prevent the initiation of APs in the axon initial segment in response to fast fluctuating inputs, and shift the spike initiation zone to the proximal nodes of Ranvier. This result is consistent with the finding that sodium channels in the axons of high-frequency neurons of the avian nucleus laminaris are located further distally than those in low-frequency neurons (Kuba, Ishii, & Ohmori, 2006). Thus, the initial portion of the axon isolates the somatic leak from the spike-generating zone and the efficacy of the isolation needs to be larger the higher the input frequency, that is, the more difficult it is for the stimulus to elicit an AP.
In addition to the many circuit, synapse and membrane elements and mechanisms that have been proposed to determine the optimal ITD of a neuron (reviewed earlier), asymmetries in axonal morphology have also been considered. If the axon originates from the lateral dendrite, somatic inhibition would effectively delay the contralateral input, leading to contralaterally leading best ITDs (Zhou, Carney, & Colburn, 2005). Anatomical evidence for systematic axonal asymmetry, however, is absent (Rautenberg, Grothe, & Felmy, 2009).
Since changes in ITD of only 10 µs have marked effects on MSO response rates, biophysical processes occurring on scales longer than about 5 ms cannot directly interact with the mechanism for coincidence detection of features in the fine structure of the stimulus. Nevertheless, such slow pathways have been described for MSO cells and play a role in controlling the excitability of the neuron.
HCN channels found in the axon initial segment decrease MSO spiking probability by adding a leak current (Ko et al., 2016). Serotonin released during states of increased attention shifts the membrane potential required for half-maximum activation of HCN to more negative voltages, thereby reducing leak and increasing spike probability (Ko et al., 2016).
Presynaptic GABAB receptors reduce the transmission of both the glutamatergic and the glycinergic MSO inputs (Stange et al., 2013). The GABA is supplied by a feedback loop which includes GABAergic neurons in the superior periolivary nucleus that are driven by the MSO, project back to it, and release GABA from extrasynaptic varicosities. Implications of this negative feedback on ITD encoding will be discussed next.
The Computation of Interaural Time Differences
As we have seen, principal MSO neurons fundamentally serve as coincidence detectors for their binaural inputs and each responds maximally to a specific range of ITDs. Although some MSO neurons are responsive (i.e., fire action potentials) to monaural stimulation of either ear alone, the sum of the two monaurally evoked responses is generally far below the response to binaural stimulation at favorable ITDs. Moreover, the preferred ITD can be predicted from the phase delay between the two monaural responses (Yin & Chan, 1990). Thus, to achieve maximal coincidence between ipsi- and contralateral inputs, the external delay (i.e., ITD) must be compensated for by the internal delay within the brain. Conversely, the effective internal delay determines the preferred ITD and thus the range of sensitivity (i.e., ITD tuning) of a given MSO neuron (and for a given stimulation frequency to be further discussed under “The Nature of Internal Delays”).
Generally, MSO neurons are very broadly tuned; that is, response rates are modulated over a large range of ITDs that exceeds the range of ITDs that are generated by the shape of the skull under anechoic conditions, the so-called “physiological range” (Figure 5). For the majority of recorded ITD functions—irrespective of species—the highest neuronal spatial sensitivity (the slope of the ITD rate function that conveys most information about changes in location) is positioned around the midline (Harper & McAlpine, 2004), and preferred ITDs mostly correspond to contralateral positions at the edge or even outside of the physiological range. More specifically, the preferred ITD of a cell depends on the best frequency (BF), irrespective of the head size of the species studied, and on average preferred ITDs increase with decreasing BF (Figure 5) (Middlebrooks et al., 1994; McAlpine, Jiang, & Palmer, 2001; Hancock & Delgutte, 2004; Pecka et al., 2008; Werner-Reiss & Groh, 2008). These data thus refuted the long-standing idea of the MSO as a distributed labeled-line encoder of azimuthal space (as had been found in owls and chicks) in which preferred ITDs in each frequency band are distributed within the physiological range (or even clustered around the midline).
What then is the nature of the neuronal ITD code—and hence of the internal delay—that underlies the BF dependency of ITD tuning in MSO neurons? In vivo recordings from the MSO are technically very challenging, and these questions are thus still subject of ongoing debate and investigation. A number of recent studies have included enormously difficult experiments that yielded exciting, but unfortunately contradictory, findings regarding the mechanism(s) used to create internal delays. We will first briefly review the current hypotheses in light of the known anatomy and physiology of the MSO circuit, and then turn to recent advances in our understanding of spatial coding in the MSO and mammals in general.
The Nature of Internal Delays
Important insights into possible mechanisms for the generation of internal delays come from the tuning of MSO neurons for sound frequency: by definition, the highest spike rates (and the greatest modulation of the response) are evoked by pure tones presented at a neuron’s BF. Shifting the pure-tone frequency away from the BF in either direction results in a systematic reduction in the maximum spike rates elicited, and in a reduction in the dynamic range of the spike rate elicited by different ITDs (Figure 5). More importantly, by testing ITD sensitivity in MSO neurons at multiple frequencies (or using broadband noise) one can determine their characteristic delay (CD), that is, the ITD for which the relative discharge rate is identical for different stimulus frequencies. Theoretically, for neurons with a pure axonal conduction delay from one or both ears (i.e., a pure time delay as generated by delay lines), the CD corresponds to the ITD at which response peaks are aligned for all stimulus frequencies to which the neuron is sensitive. As mentioned before, the observed dependency of best ITDs on BF (i.e., internal delays) in the MSO rules out a Jeffress-like topographic arrangement of best ITDs. Nonetheless, theoretically this BF dependency could still be achieved by systematic pattern of variation of axonal length from the two ears, analogous to the delay-line system found in birds (compare Figure 6). However, while the inhibitory pathway within the MSO circuit indeed exhibits anatomical adaptations to achieve superior conduction speed and timing relative to excitation (see first section and Figure 6), current anatomical data do not support the existence of any systematic arrangement of axon length within the excitatory inputs themselves (Karino et al., 2011). Moreover, CDs of MSO neurons are typically not found at response peaks, but tend to align along the slopes of the ITD functions. This suggests that ITD sensitivity in these neurons is not generated by a simple time delay mechanism and that some form of phase delay must be present (“CP,” the characteristic phase).
Given the anatomical and physiological evidence for inhibitory inputs to MSO neurons, it may not be surprising that CPs are indeed observed. In fact, it is well established that the constellation of ipsilateral excitation and contralateral inhibition from the MNTB for coincidence detection in the lateral superior olive (LSO) results in a CP of half a cycle. It is further well documented that the temporal sensitivity for the input timing between inhibition and excitation in the LSO lies in the range of tens to hundreds of microseconds (Tollin, 2003; Tollin & Yin, 2005; Beiderbeck et al., 2018). Inhibition is therefore likely to contribute to tuning internal delays to the relevant ITDs (Brand et al., 2002; Seidl & Grothe, 2005), reviewed in (Grothe, Pecka, & McAlpine, 2010). This, however, does not mean that other factors like axonal length are not important. It only means that there is no indication for systematic variations in other parameters and that such factors on their own cannot explain overall delays (see Figure 6).
Nevertheless, phase delays could also be generated by other mechanisms. Two recent studies were able to obtain in vivo whole-cell recordings from MSO cells, thus enabling direct examination of the changes in membrane polarization in MSO neurons during ITD processing.
Based on data from whole-cell and attached-cell recordings, van der Heijden and colleagues (van der Heijden et al., 2013) concluded that inhibition is not detectable during—and hence does not contribute to—ITD tuning of MSO cells. This apparent lack of discernible inhibition may be due to the fixed temporal relationship between contralateral excitation and inhibition, resulting in a net PSP which, at least without further manipulations, may not always allow one to disentangle the different synaptic components. Physiological studies of the MNTB-derived inhibitory input, however, have consistently found that contralateral inhibition precedes contralateral excitation (Roberts, Seeman, & Golding, 2013; Goldwyn et al., 2017; see Figure 4) and thereby changes the timing of the composite contralateral input (Myoga et al., 2014). Anatomical findings of thicker axons and unusual myelination geared for fast conduction times specifically within the inhibitory pathway to the MSO support these physiological results (Ford et al., 2015; Stange-Marten et al., 2017) that were already predicted based on theoretical considerations (Brand et al. 2002). Another in vivo whole-cell study of the MSO carried out additional pharmacological interventions during recordings, a feat requiring utmost technical ability (Franken et al., 2015). Unlike van der Heijden and colleagues (2013), these authors indeed observed prominent inhibition in their MSO recordings, albeit not consistently in all cells (which could be related to the relatively high CFs of the recorded neurons). Interestingly, they also consistently observed phasic hyperpolarizations prior to supra-threshold excitatory events. The authors concluded that this hyperpolarization might be caused by the action of additional, excitatory inputs that inactivate low-threshold potassium channels (Franken et al., 2015). However, the origin of this additional input is unclear.
Another, perhaps complementary, hypothesis for the generation of internal delays is based on the possible presence of small differences in spectral tuning between inputs from the two ears, and has been investigated in two other recent studies. The so-called stereausis hypothesis arose from the application of theories developed to account for binaural pitch detection based on coincidence detection (Loeb, White, & Merzenich, 1983) and later adapted for sound localization (Shamma, Shen, & Gopalaswamy, 1989). Analyses of monaural spike trains from auditory nerve recordings confirm the feasibility of this scenario (Joris et al., 2006), and one study indicates that such a mechanism may contribute to ITD tuning under some conditions involving complex stimuli (Benichoux et al., 2015). In contrast, studies using simple dichotic stimuli indicate no or, at best, only a minor and non-systematic contribution of stereausis to ITD tuning (Pecka et al., 2008; Plauška, van der Heijden, & Borst, 2017).
In summary, the nature of the internal delay in MSO neurons is still not clear. Quite possibly, many, if not all of the mechanisms mentioned previously (including conduction delay) may contribute. Nonetheless, one factor often neglected in considerations of the role of inhibitory inputs is the intriguing possibility that the characteristic delay is adjusted and/or refined during ontogeny. Anatomical refinement of glycinergic inhibition has been shown to be at least partially dependent on auditory experience during early development, and auditory deprivation can alter the distribution of best ITDs. Thus, while other mechanisms (axonal length, cochlear delays etc.) can establish an approximate BITD early in development, an experience-dependent process could selectively optimize inhibitory inputs following the onset of hearing to achieve and maintain the desired tuning.
The Nature of the Spatial Code and Dynamics of Spatial Processing
The fact that ITD functions of MSO neurons show very broad tuning (essentially linear modulation between -90° and 90°) that is stereotypical within a given spectral band stimulated the idea of hemispheric, oppositely coding channels on each side of the brain that might be compared at later stages of the pathway (Figure 5) (McAlpine, Jiang, & Palmer, 2001; Stecker & Middlebrooks, 2003; Hancock & Delgutte, 2004; Harper & McAlpine, 2004). This concept relies on the idea that similar activity levels in both channels should encode sound-source position at the midline, such that a relative increase in activity in one of the brain hemispheres would indicate a corresponding contralateral location with respect to the more active brain hemisphere. This strategy might well optimize coding efficiency in animals with small head sizes, but might be sub-optimal for animals with larger heads and high frequency hearing, owing to the increasing ambiguity of spike-rate-to-ITD mapping (Harper & McAlpine, 2004; Lüling et al., 2011; Goodman, Benichoux, & Brette, 2013; Harper et al., 2014). Such ambiguity could, however, be eliminated by making use of additional information from other ITD-sensitive channels (such as the low-frequency neurons in the Lateral Superior Olive, see next paragraph) during the decoding (Lingner et al., 2018). One interesting observation in this regard is that CD and BF, as well as CD and CP, are negatively correlated (Figure 5); (Agapiou & McAlpine, 2008; Lüling et al., 2011). This correlation could be neuronally exploited to acquire frequency-invariant information about ITDs, either in the context of population average coding (Lüling et al., 2011) or a pattern recognition approach (Goodman, Benichoux, & Brette, 2013; Benichoux et al., 2015).
While the models for possible ITD representation are manifold (see preceding discussion and Day & Delgutte, 2013), elegant adaptation paradigms with human subjects not only strongly imply that a population code underlies sound localization in humans/primates, but have also highlighted the fact that prior stimulation influences subsequent spatial perception (Phillips & Hall, 2005; Vigneault-MacLean, Hall, & Phillips, 2007). Traditionally, such adaptive coding strategies have not been associated with sound localization, yet recent findings from both human psychophysics (Getzmann, 2004; Phillips & Hall, 2005; Vigneault-MacLean, Hall, & Phillips, 2007; Maier et al., 2009; Dahmen et al., 2010; Lingner et al., 2018) and animal physiology (Magnusson et al., 2008; Park et al., 2008; Dahmen et al., 2010; Stange et al., 2013) have clearly demonstrated a dependency of spatial tuning, and even spatial perception, on the preceding stimulus (specifically, on the properties of the stimulus and their temporal profile). It has become apparent that dynamic adaptation mechanisms act on spatial processing in the brainstem, including the MSO and LSO (Magnusson et al., 2008; Stange et al., 2013). The response levels of individual neurons are negatively correlated with their prior spiking activity, which is typical of gain modulation. Notably, this modulation, which acts on time scales of tens of milliseconds and can last for seconds, is mediated by GABAB receptor signaling, even though MSO cells are not themselves GABAergic. Instead, as already described, a di-synaptic feedback-loop exists, with MSO neurons innervating GABAergic neurons in the nearby Superior Paraolivary Nucleus, which subsequently feed back onto the MSO (Figure 5; Stange et al., 2013).
The resulting activity-dependent rate adaptations do not shift the preferred ITDs of MSO neurons directly; that is, best ITDs are not altered. However, these modulations have significant consequences at the level of hemispheric population coding, because of the activity-dependent nature of the adaptation. For example, a strongly lateralized sound source will generate unequal adaptation in the two hemispheres, with pronounced rate adaptation only in the contralateral channel. Accordingly, this hemispheric asymmetry should shift the perceived location of a subsequently presented sound source (Figure 5). As noted earlier, Phillips and colleagues first tested this hypothesis in a number of psychophysical paradigms, and were able to demonstrate a pronounced shift in the perceived location of sound sources after prior presentation of a lateralized adapter in human listeners (Phillips & Hall, 2005; Vigneault-MacLean, Hall, & Phillips, 2007). Stange et al. (2013) have demonstrated that GABAB-mediated rate adaptation in the MSO is sufficient to explain these shifts in human perception. Moreover, we have recently developed a new decoding model in which sound location is initially computed in both brain hemispheres independently between MSO and LSO and then combined to yield a hemispherically balanced code. This model not only closely captures the observed absolute localization errors caused by stimulus history, but also predicts a selective dilation and compression of perceptional space (Lingner et al., 2018). These model predictions are confirmed by improvement and degradation of spatial resolution in human listeners. shown that the primary function of these perceptual shifts seems to be the relative segregation of the adapting and the subsequent sound source. Specifically, as the reported shifts in location are directed away from the adapter location, the perceived distance between the sound sources is increased relative to the actual distance (Lingner et al., 2018).
Taken together, these findings suggest that ITD coding in the MSO serves to encode the relative separation of concurrent or subsequent sound sources, rather than to provide an absolute representation of position in space.
Agapiou, J. P., & McAlpine, D. (2008). Low-frequency envelope sensitivity produces asymmetric binaural tuning curves. Journal of Neurophysiology, 100(4), 2381–2396. doi: 10.1152/jn.90393.2008.Find this resource:
Agmon-Snir, H., Carr, C. E., & Rinzel, J. (1998). The role of dendrites in auditory coincidence detection. Nature, 393(6682), 268–272. doi: 10.1038/30505.Find this resource:
Baumann, V. J. et al. (2013) Tonotopic organization of the hyperpolarization-activated current (Ih) in the mammalian medial superior olive. Frontiers in Neural Circuits, 7, 117. doi: 10.3389/fncir.2013.00117.Find this resource:
Behrend, O. et al. (2002). Auditory response properties in the superior paraolivary nucleus of the gerbil. Journal of Neurophysiology, 87(6), 2915–2928. doi: 10.1152/jn.2002.87.6.2915.Find this resource:
Beiderbeck, B. et al. (2018). Precisely timed inhibition facilitates action potential firing for spatial coding in the auditory brainstem. Nature Communications, 9(1), 1771. doi: 10.1038/s41467-018-04210-y.Find this resource:
Benichoux, V. et al. (2015). Neural tuning matches frequency-dependent time differences between the ears. eLife, 4. doi: 10.7554/eLife.06072.Find this resource:
Borst, J. G. G., & Soria van Hoeve, J. (2012). The calyx of Held synapse: From model synapse to auditory relay. Annual Review of Physiology, 74, 199–224. doi: 10.1146/annurev-physiol-020911-153236.Find this resource:
Brand, A. et al. (2002). Precise inhibition is essential for microsecond interaural time difference coding. Nature, 417(6888), 543–547. doi: 10.1038/417543a.Find this resource:
Caird, D., & Klinke, R. (1983). Processing of binaural stimuli by cat superior olivary complex neurons. Experimental Brain Research, 52(3), 385–399.Find this resource:
Clark, G. M. (1969). Vesicle shape versus type of synapse in the nerve endings of the cat medial superior olive. Brain Research, 15(2), 548–551.Find this resource:
Couchman, K., Grothe, B., & Felmy, F. (2010). Medial superior olivary neurons receive surprisingly few excitatory and inhibitory inputs with balanced strength and short-term dynamics. Journal of Neuroscience, 30(50), pp. 17111–17121. doi: 10.1523/JNEUROSCI.1760-10.2010.Find this resource:
Couchman, K., Grothe, B., & Felmy, F. (2012). Functional localization of neurotransmitter receptors and synaptic inputs to mature neurons of the medial superior olive. Journal of Neurophysiology, 107(4), 1186–1198. doi: 10.1152/jn.00586.2011.Find this resource:
Covey, E., Vater, M., & Casseday, J. H. (1991). Binaural properties of single units in the superior olivary complex of the mustached bat. Journal of Neurophysiology, 66(3), 1080–1094. doi: 10.1152/jn.19188.8.131.520.Find this resource:
Dahmen, J. C. et al. (2010). Adaptation to stimulus statistics in the perception and neural representation of auditory space. Neuron, 66(6), 937–948. doi: 10.1016/j.neuron.2010.05.018.Find this resource:
Dasika, V. K., White, J. A., & Colburn, H. S. (2007). Simple models show the general advantages of dendrites in coincidence detection.” Journal of Neurophysiology, 97(5), 3449–3459. doi: 10.1152/jn.00669.2005.Find this resource:
Day, M. L., & Delgutte, B. (2013). Decoding sound source location and separation using neural population activity patterns. Journal of Neuroscience, 33(40), 15837–15847. doi: 10.1523/JNEUROSCI.2034-13.2013.Find this resource:
Day, M. L., & Semple, M. N. (2011). Frequency-dependent interaural delays in the medial superior olive: Implications for interaural cochlear delays. Journal of Neurophysiology, 106(4), 1985–1999. doi: 10.1152/jn.00131.2011.Find this resource:
Dietz, M., Marquardt, T., Stange, A., Pecka, M., Grothe, B., & McAlpine, D. (2014) Emphasis of spatial cues in the temporal fine structure during the rising segments of amplitude-modulated sounds II: single-neuron recordings. Journal of Neurophysiology, 111(10), 1973–1985. doi: 10.1152/jn.00681.2013.Find this resource:
Fischer, L., Leibold, C., & Felmy, F. (2018). Resonance properties in auditory brainstem neurons. Frontiers in Cellular Neuroscience, 12, 8. doi: 10.3389/fncel.2018.00008.Find this resource:
Fischl, M. J. et al. (2016). Physiology and anatomy of neurons in the medial superior olive of the mouse. Journal of Neurophysiology, 116(6), 2676–2688. doi: 10.1152/jn.00523.2016.Find this resource:
Ford, M. C. et al. (2015). Tuning of Ranvier node and internode properties in myelinated axons to adjust action potential timing. Nature Communications, 6, 8073. doi: 10.1038/ncomms9073.Find this resource:
Franken, T. P. et al. (2015). In vivo coincidence detection in mammalian sound localization generates phase delays. Nature Neuroscience, 18(3), 444–452. doi: 10.1038/nn.3948.Find this resource:
Franken, T. P., Bremen, P., Joris, P. X. (2014). Coincidence detection in the medial superior olive: mechanistic implications of an analysis of input spiking patterns. Frontiers in Neural Circuits, 8, 42. doi: 10.3389/fncir.2014.00042.Find this resource:
Getzmann, S. (2004). Spatial discrimination of sound sources in the horizontal plane following an adapter sound. Hearing Research, 191(1–2), 14–20. doi: 10.1016/j.heares.2003.12.020.Find this resource:
Goldberg, J. M., & Brown, P. B. (1969). Response of binaural neurons of dog superior olivary complex to dichotic tonal stimuli: Some physiological mechanisms of sound localization. Journal of Neurophysiology, 32(4), 613–636.Find this resource:
Goldwyn, J. H. et al. (2017). Signatures of somatic inhibition and dendritic excitation in auditory brainstem field potentials. Journal of Neuroscience, 37(43), 10451–10467. doi: 10.1523/JNEUROSCI.0600-17.2017.Find this resource:
Goodman, D. F., Benichoux, V., & Brette, R. (2013). Decoding neural responses to temporal cues for sound localization. eLife, 2, e01312. doi: 10.7554/eLife.01312.Find this resource:
Grothe, B. (1994). Interaction of excitation and inhibition in processing of pure tone and amplitude-modulated stimuli in the medial superior olive of the mustached bat. Journal of Neurophysiology, 71(2), 706–721.Find this resource:
Grothe, B. (2000). The evolution of temporal processing in the medial superior olive, an auditory brainstem structure. Progress in Neurobiology, 61(6), 581–610.Find this resource:
Grothe, B. et al. (1992). Monaural interaction of excitation and inhibition in the medial superior olive of the mustached bat: an adaptation for biosonar. Proceedings of the National Academy of Sciences of the United States of America, 89(11), 5108–5112.Find this resource:
Grothe, B., & Park, T. J. (1998). Sensitivity to interaural time differences in the medial superior olive of a small mammal, the Mexican free-tailed bat.” Journal of Neuroscience, 18(16), 6608–6622.Find this resource:
Grothe, B., & Pecka, M. (2014). The natural history of sound localization in mammals—a story of neuronal inhibition. Frontiers in Neural Circuits, 8, 116. doi: 10.3389/fncir.2014.00116.Find this resource:
Grothe, B., Pecka, M., & McAlpine, D. (2010). Mechanisms of sound localization in mammals. Physiological Reviews, 90(3), 983–1012. doi: 10.1152/physrev.00026.2009.Find this resource:
Grothe, B., & Sanes, D. H. (1993). Bilateral inhibition by glycinergic afferents in the medial superior olive. Journal of Neurophysiology, 69(4), 1192–1196.Find this resource:
Grothe, B., & Sanes, D. H. (1994). Synaptic inhibition influences the temporal coding properties of medial superior olivary neurons: An in vitro study. The Journal of Neuroscience, 14(3 Pt. 2), 1701–1709.Find this resource:
Hancock, K. E., & Delgutte, B. (2004). A physiologically based model of interaural time difference discrimination. Journal of Neuroscience, 24(32), 7110–7117. doi: 10.1523/JNEUROSCI.0762-04.2004.Find this resource:
Harnischfeger, G., Neuweiler, G., & Schlegel, P. (1985). Interaural time and intensity coding in superior olivary complex and inferior colliculus of the echolocating bat Molossus ater. Journal of Neurophysiology, 53(1), 89–109.Find this resource:
Harper, N. S. et al. (2014). The neural code for auditory space depends on sound frequency and head size in an optimal manner. PloS One, 9(11), p. e108154. doi: 10.1371/journal.pone.0108154.Find this resource:
Harper, N. S., & McAlpine, D. (2004). Optimal neural population coding of an auditory spatial cue. Nature, 430(7000), 682–686. doi: 10.1038/nature02768.Find this resource:
van der Heijden, M. et al. (2013). Directional hearing by linear summation of binaural inputs at the medial superior olive.” Neuron, 78(5), 936–948. doi: 10.1016/j.neuron.2013.04.028.Find this resource:
Henkel, C. K., & Brunso-Bechtold, J. K. (1990). Dendritic morphology and development in the ferret medial superior olivary nucleus. Journal of Comparative Neurology, 294(3), 377–388. doi: 10.1002/cne.902940307.Find this resource:
Jeffress, L. A. (1948). A place theory of sound localization. Journal of Comparative and Physiological Psychology, 41(1), 35–39.Find this resource:
Jercog, P. E. et al. (2010). Asymmetric excitatory synaptic dynamics underlie interaural time difference processing in the auditory system. PLoS Biology, 8(6), e1000406. doi: 10.1371/journal.pbio.1000406.Find this resource:
Joris, P. X. et al. (1994). Enhancement of neural synchronization in the anteroventral cochlear nucleus. I. Responses to tones at the characteristic frequency. Journal of Neurophysiology, 71(3), 1022–1036.Find this resource:
Joris, P. X. et al. (2006). Binaural and cochlear disparities. Proceedings of the National Academy of Sciences of the United States of America, 103(34), 12917–12922. doi: 10.1073/pnas.0601396103.Find this resource:
Joris, P. X., Smith, P. H., & Yin, T. C. (1998). Coincidence detection in the auditory system: 50 years after Jeffress. Neuron, 21(6), 1235–1238.Find this resource:
Kapfer, C. et al. (2002). Experience-dependent refinement of inhibitory inputs to auditory coincidence-detector neurons. Nature Neuroscience, 5(3), 247–253. doi: 10.1038/nn810.Find this resource:
Karino, S. et al. (2011). Axonal branching patterns as sources of delay in the mammalian auditory brainstem: a re-examination. Journal of Neuroscience, 31(8), 3016–3031. doi: 10.1523/JNEUROSCI.5175-10.2011.Find this resource:
Khurana, S. et al. (2011). Dynamic interaction of Ih and IK-LVA during trains of synaptic potentials in principal neurons of the medial superior olive. Journal of Neuroscience, 31(24), 8936–8947. doi: 10.1523/JNEUROSCI.1079-11.2011.Find this resource:
Kiss, A., & Majorossy, K. (1983). Neuron morphology and synaptic architecture in the medial superior olivary nucleus. Light- and electron microscope studies in the cat. Experimental Brain Research, 52(3), 315–327.Find this resource:
Ko, K. W. et al. (2016). Serotonin modulates spike probability in the axon initial segment through HCN channels. Nature Neuroscience, 19(6), 826–834. doi: 10.1038/nn.4293.Find this resource:
Kuba, H., Ishii, T. M., & Ohmori, H. (2006). Axonal site of spike initiation enhances auditory coincidence detection. Nature, 444(7122), 1069–1072. doi: 10.1038/nature05347.Find this resource:
Kuwabara, N., & Zook, J. M. (1992). Projections to the medial superior olive from the medial and lateral nuclei of the trapezoid body in rodents and bats. Journal of Comparative Neurology, 324(4), 522–538. doi: 10.1002/cne.903240406.Find this resource:
Lehnert, S. et al. (2014). Action potential generation in an anatomically constrained model of medial superior olive axons. Journal of Neuroscience, 34(15), 5370–5384. doi: 10.1523/JNEUROSCI.4038-13.2014.Find this resource:
Leibold, C. (2010). Influence of inhibitory synaptic kinetics on the interaural time difference sensitivity in a linear model of binaural coincidence detection. Journal of the Acoustical Society of America, 127(2), 931–942. doi: 10.1121/1.3282997.Find this resource:
Leibold, C., & van Hemmen, J. L. (2005). Spiking neurons learning phase delays: How mammals may develop auditory time-difference sensitivity. Physical Review Letters, 94(16), 168102. doi: 10.1103/PhysRevLett.94.168102.Find this resource:
Lingner, A. et al. (2018). .A novel concept for dynamic adjustment of auditory space. Scientific Reports, 8(1), 8335. doi: 10.1038/s41598-018-26690-0.Find this resource:
Loeb, G. E., White, M. W., & Merzenich, M. M. (1983). .Spatial cross-correlation: A proposed mechanism for acoustic pitch perception. Biological Cybernetics, 47(3), 149–163.Find this resource:
Lüling, H. et al. (2011). Frequency-invariant representation of interaural time differences in mammals. PLoS Computational Biology, 7(3), e1002013. doi: 10.1371/journal.pcbi.1002013.Find this resource:
Magnusson, A. K. et al. (2008). Retrograde GABA signaling adjusts sound localization by balancing excitation and inhibition in the brainstem. Neuron, 59(1), 125–137. doi: 10.1016/j.neuron.2008.05.011.Find this resource:
Maier, J. K. et al. (2009). Context effects in the discriminability of spatial cues. Journal of the Association for Research in Otolaryngology, 11(2), 319–328. doi: 10.1007/s10162-009-0200-0.Find this resource:
Mathews, P. J. et al. (2010). Control of submillisecond synaptic timing in binaural coincidence detectors by K(v)1 channels. Nature Neuroscience, 13(5), 601–609. doi: 10.1038/nn.2530.Find this resource:
McAlpine, D., Jiang, D., Palmer, A. R. (2001). A neural code for low-frequency sound localization in mammals. Nature Neuroscience, 4(4), 396–401. doi: 10.1038/86049.Find this resource:
Middlebrooks, J. C. et al. (1994). A panoramic code for sound location by cortical neurons. Science, 264(5160), 842–844.Find this resource:
Mikiel-Hunter, J., Kotak, V., & Rinzel, J. (2016). High-Frequency Resonance in the Gerbil Medial Superior Olive. PLoS Computational Biology, 12(11), e1005166. doi: 10.1371/journal.pcbi.1005166.Find this resource:
Morest, D. K. (1968). The collateral system of the medial nucleus of the trapezoid body of the cat, its neuronal architecture and relation to the olivo-cochlear bundle. Brain Research, 9(2), 288–311.Find this resource:
Myoga, M. H. et al. (2014). Glycinergic inhibition tunes coincidence detection in the auditory brainstem. Nature Communications, 5, 3790. doi: 10.1038/ncomms4790.Find this resource:
Park, T. J. et al. (2008). Dynamic changes in level influence spatial coding in the lateral superior olive. Hearing Research, 238(1–2), 58–67. doi: 10.1016/j.heares.2007.10.009.Find this resource:
Pecka, M. et al. (2008). Interaural Time Difference Processing in the Mammalian Medial Superior Olive: The Role of Glycinergic Inhibition. Journal of Neuroscience., 28(27), 6914–6925. doi: 10.1523/JNEUROSCI.1660-08.2008.Find this resource:
Perkins, R. E. (1973). An electron microscopic study of synaptic organization in the medial superior olive of normal and experimental chinchillas. The Journal of Comparative Neurology, 148(3), pp. 387–415. doi: 10.1002/cne.901480306.Find this resource:
Phillips, D. P., & Hall, S. E. (2005). Psychophysical evidence for adaptation of central auditory processors for interaural differences in time and level. Hearing Research, 202(1–2), 188–199. doi: 10.1016/j.heares.2004.11.001.Find this resource:
Plauška, A., Borst, J. G., & van der Heijden, M. (2016). Predicting binaural responses from monaural responses in the gerbil medial superior olive. Journal of Neurophysiology, 115(6), 2950–2963. doi: 10.1152/jn.01146.2015.Find this resource:
Plauška, A., van der Heijden, M., & Borst, J. G. G. (2017). A test of the stereausis hypothesis for sound localization in mammals. Journal of Neuroscience, 37(30), 7278–7289. doi: 10.1523/JNEUROSCI.0233-17.2017.Find this resource:
Rall, W. (1962). Theory of physiological properties of dendrites. Annals of the New York Academy of Sciences, 96, 1071–1092.Find this resource:
Ramón y Cajal, S. (1909). Histologie du système nerveux de l’homme & des vertébrés. Paris , France: Maloine. Available at: http://archive.org/details/histologiedusyst01ram (Accessed: 30 May 2018).Find this resource:
Rautenberg, P. L., Grothe, B. and Felmy, F. (2009). Quantification of the three-dimensional morphology of coincidence detector neurons in the medial superior olive of gerbils during late postnatal development. The Journal of Comparative Neurology, 517(3), 385–396. doi: 10.1002/cne.22166.Find this resource:
Remme, M. W. H. et al. (2014). Subthreshold resonance properties contribute to the efficient coding of auditory spatial cues. Proceedings of the National Academy of Sciences of the United States of America, 111(22), E2339–2348. doi: 10.1073/pnas.1316216111.Find this resource:
Roberts, M. T., Seeman, S. C., & Golding, N. L. (2013). A mechanistic understanding of the role of feedforward inhibition in the mammalian sound localization circuitry. Neuron, 78(5), 923–935. doi: 10.1016/j.neuron.2013.04.022.Find this resource:
Roberts, M. T., Seeman, S. C., & Golding, N. L. (2014). The relative contributions of MNTB and LNTB neurons to inhibition in the medial superior olive assessed through single and paired recordings. Frontiers in Neural Circuits, 8, 49. doi: 10.3389/fncir.2014.00049.Find this resource:
Salin, P. A. et al. (1996). Distinct short-term plasticity at two excitatory synapses in the hippocampus. Proceedings of the National Academy of Sciences of the United States of America, 93(23), 13304–13309.Find this resource:
Schroeder, M. R. (1977). New viewpoints in binaural interactions. In: E. F. Evans, J. P. Wilson, (Eds), Psychophysics and physiology of hearing. New York: Academic Press; 455–467.Find this resource:
Scott, L. L., Hage, T. A., & Golding, N. L. (2007). Weak action potential backpropagation is associated with high-frequency axonal firing capability in principal neurons of the gerbil medial superior olive. Journal of Physiology, 583(Pt 2), 647–661. doi: 10.1113/jphysiol.2007.136366.Find this resource:
Scott, L. L., Mathews, P. J., & Golding, N. L. (2005). Posthearing developmental refinement of temporal processing in principal neurons of the medial superior olive. Journal of Neuroscience, 25(35), 7887–7895. doi: 10.1523/JNEUROSCI.1016-05.2005.Find this resource:
Seidl, A. H., & Grothe, B. (2005). Development of sound localization mechanisms in the Mongolian gerbil is shaped by early acoustic experience. Journal of Neurophysiology, 94(2), 1028–1036. doi: 10.1152/jn.01143.2004.Find this resource:
Seidl, A. H., Rubel, E. W., & Harris, D. M. (2010). Mechanisms for adjusting interaural time differences to achieve binaural coincidence detection. Journal of Neuroscience, 30(1), 70–80. doi: 10.1523/JNEUROSCI.3464-09.2010.Find this resource:
Seidl, A. H., & Rubel, E. W. (2016). “Systematic and differential myelination of axon collaterals in the mammalian auditory brainstem,” Glia, 64(4), 487–494. doi: 10.1002/glia.22941.Find this resource:
Shamma, S. A., Shen, N. M., & Gopalaswamy, P. (1989). Stereausis: Binaural processing without neural delays. Journal of the Acoustical Society of America, 86(3), 989–1006.Find this resource:
Smith, A. J., Owens, S., & Forsythe, I. D. (2000). Characterisation of inhibitory and excitatory postsynaptic currents of the rat medial superior olive. Journal of Physiology, 529(Pt 3), 681–698.Find this resource:
Smith, D. J., & Rubel, E. W. (1979). Organization and development of brain stem auditory nuclei of the chicken: Dendritic gradients in nucleus laminaris. Journal of Comparative Neurology, 186(2), 213–239. doi: 10.1002/cne.901860207.Find this resource:
Smith, P. H. (1995). Structural and functional differences distinguish principal from nonprincipal cells in the guinea pig MSO slice. Journal of Neurophysiology, 73(4), 1653–1667. doi: 10.1152/jn.19184.108.40.2063.Find this resource:
Spirou, G. A., & Berrebi, A. S. (1997). Glycine immunoreactivity in the lateral nucleus of the trapezoid body of the cat. Journal of Comparative Neurology, 383(4), 473–488.Find this resource:
Spirou, G. A., Rowland, K. C., & Berrebi, A. S. (1998). Ultrastructure of neurons and large synaptic terminals in the lateral nucleus of the trapezoid body of the cat. Journal of Comparative Neurology, 398(2), 257–272.Find this resource:
Spitzer, M. W., & Semple, M. N. (1995). Neurons sensitive to interaural phase disparity in gerbil superior olive: Diverse monaural and temporal response properties. Journal of Neurophysiology, 73(4), 1668–1690. doi: 10.1152/jn.19220.127.116.118.Find this resource:
Stange, A. et al. (2013). Adaptation in sound localization: From GABAB receptor-mediated synaptic modulation to perception. Nature Neuroscience, 16(12), 1840–1847. doi: 10.1038/nn.3548.Find this resource:
Stange-Marten, A. et al. (2017). Input timing for spatial processing is precisely tuned via constant synaptic delays and myelination patterns in the auditory brainstem.” Proceedings of the National Academy of Sciences of the United States of America. 114(24), E4851–E4858. doi: 10.1073/pnas.1702290114.Find this resource:
Stecker, G. C., & Middlebrooks, J. C. (2003). Distributed coding of sound locations in the auditory cortex. Biological Cybernetics, 89(5), 341–349. doi: 10.1007/s00422-003-0439-1.Find this resource:
Stotler, W. A. (1953). An experimental study of the cells and connections of the superior olivary complex of the cat. Journal of Comparative Neurology, 98(3), 401–431.Find this resource:
Stuart, G., & Sakmann, B. (1995). Amplification of EPSPs by axosomatic sodium channels in neocortical pyramidal neurons. Neuron, 15(5), 1065–1076.Find this resource:
Svirskis, G. et al. (2002). Enhancement of signal-to-noise ratio and phase locking for small inputs by a low-threshold outward current in auditory neurons. Journal of Neuroscience, 22(24), 11019–11025.Find this resource:
Svirskis, G., Dodla, R., & Rinzel, J. (2003). Subthreshold outward currents enhance temporal integration in auditory neurons. Biological Cybernetics, 89(5), pp. 333–340. doi: 10.1007/s00422-003-0438-2.Find this resource:
Tollin, D. J. (2003). The lateral superior olive: A functional role in sound source localization. Neuroscientist, 9(2),. 127–143.Find this resource:
Tollin, D. J., & Yin, T. C. T. (2005). Interaural phase and level difference sensitivity in low-frequency neurons in the lateral superior olive. Journal of Neuroscience, 25(46), 10648–10657. doi: 10.1523/JNEUROSCI.1609-05.2005.Find this resource:
Vigneault-MacLean, B. K., Hall, S. E., & Phillips, D. P. (2007). The effects of lateralized adaptors on lateral position judgements of tones within and across frequency channels. Hearing Research, 224(1–2), 93–100. doi: 10.1016/j.heares.2006.12.001.Find this resource:
Wei, L. et al. (2017). Enhancement of phase-locking in rodents. I. An axonal recording study in gerbil. Journal of Neurophysiology, 118(4), 2009–2023. doi: 10.1152/jn.00194.2016.Find this resource:
Werner-Reiss, U., & Groh, J. M. (2008). A rate code for sound azimuth in monkey auditory cortex: Implications for human neuroimaging studies. Journal of Neuroscience, 28(14), 3747–3758. doi: 10.1523/JNEUROSCI.5044-07.2008.Find this resource:
Werthat, F. et al. (2008). Experience-dependent refinement of the inhibitory axons projecting to the medial superior olive. Developmental Neurobiology, 68(13), 1454–1462. doi: 10.1002/dneu.20660.Find this resource:
Xu-Friedman, M. A., & Regehr, W. G. (2005). Dynamic-clamp analysis of the effects of convergence on spike timing. II. Few synaptic inputs. Journal of Neurophysiology, 94(4), 2526–2534. doi: 10.1152/jn.01308.2004.Find this resource:
Yin, T. C., & Chan, J. C. (1990). Interaural time sensitivity in medial superior olive of cat. Journal of Neurophysiology, 64(2), 465–488.Find this resource:
Zhou, Y., Carney, L. H., & Colburn, H. S. (2005). A model for interaural time difference sensitivity in the medial superior olive: interaction of excitatory and inhibitory synaptic inputs, channel dynamics, and cellular morphology. Journal of Neuroscience, 25(12), pp. 3046–3058. doi: 10.1523/JNEUROSCI.3064-04.2005.Find this resource: