Auditory Interneurones in the Metathoracic Ganglion of the Grasshopper Chorthippus Biguttulus: I. Morphological and Physiological Characterization

Auditory interneurones of the locust (Locusta migratoria) have been thoroughly investigated (e.g. Adam, 1969; Kalmring, 1975a,b; Rehbein, 1976; Römer and Rheinlaender, 1983; Römer and Marquart, 1984; Boyan and Altman, 1985). However, no acoustic behaviour of the locust has been described that has any relevance for intraspecific communication. To compare neuronal and behavioural responses, therefore, it is more promising to investigate grasshoppers with an elaborate system of acoustic communication (see D. von Helversen, 1972; O. von Helversen, 1979; D. von Helversen and O. von Helversen, 1975a,b, 1983; Elsner, 1974, 1975; Hedwig, 1986, for a review; see also Elsner and Popov, 1978). For Chorthippus biguttulus, a strong correlation has been established between the response of an interneurone and certain properties of the female’s songrecognizing system (Ronacher and Stumpner, 1988).

This study describes metathoracic interneurones of C. biguttulus that show distinct and reliable responses to acoustic stimuli (Boyan, 1984). Interneurones originating and branching in the metathoracic ganglion are of special interest, since lesion experiments with males of the same species indicated that the first important processing of auditory information takes place within this ganglion (Ronacher et al. 1986; see also Römer et al. 1981, 1988; Römer and Marquart, 1984). Furthermore, males and females with one tympanic nerve severed and one thoracic connective cut on the contralateral side between the mesothoracic and metathoracic ganglia were still able to recognize the species-specific song (Ronacher et al. 1986; B. Ronacher, unpublished results). In this experiment, the only functioning structures of the auditory receptors were in the metathoracic hemiganglion with the intact tympanic nerve. Since axons of auditory receptors only ascend ipsilaterally (Römer, 1985; Römer et al. 1988; Halex et al. 1988; Stumpner, 1988), the ascending collaterals of auditory receptors are not necessary for pattern recognition. Behavioural experiments with selective heating of ganglia in a related species (C. parallelus), however, suggest that the head ganglia make the final decision about whether a song is from a potential mate (Bauer and von Helversen, 1987). For song recognition, therefore, the filtering circuit in the brain must get its main input from auditory interneurones ascending from the thoracic ganglia.

This paper on the metathoracic auditory interneurones includes a morphological description and a physiological characterization. A second paper will describe the filtering characteristics of these interneurones for temporal parameters of the male’s song (Stumpner et al. 1991).

The animals used in the experiments were female C. biguttulus L., caught in the field in southern Germany. They were briefly anaesthesized with CO₂ and attached to a free-standing holder (thickness 4 mm) either by a wax-resin mixture or with minute insect pins. The head, legs, wings and gut were removed. The thorax was opened dorsally and the metathoracic ganglion was exposed. In some experiments the ganglion was partially desheathed; in most experiments it was stabilized with a NiCr spoon. The whole torso was filled with locust Ringer (Pearson and Robertson, 1981).

The experiments were performed in an anechoic Faraday cage at room temperature (22–26°C). This must be kept in mind when comparing these results with behavioural ones, which are usually obtained at higher temperatures (30-35°C). The recorded signals were amplified with a List LM-1 electrode amplifier and stored on magnetic tape (with a Racal store 4DS or a Blaupunkt video recorder with Bio-Logic PCM-adapter). The stimuli were delivered via two Motorola speakers (PH10, 2.5–40 kHz) located 35 cm from the preparation on the left and right sides. The amplitudes of the white-noise (WN; Fa. Noizeg, 100 Hz-100 kHz) or sine-wave stimuli (5 or 20 kHz) were modulated by a computer (AIM 65, Rockwell). Sound intensities were adjusted with a Brüel & Kjäer condenser microphone (1/2 inch) located at the site of the preparation and with a Brüel & Kjäer measuring amplifier (type 2602), and are given in dB re 2×10⁻⁵Nm⁻² SPL. The standard stimulation sequence consisted of WN stimuli, 100 ms in duration, 50–90dB SPL in 10 dB increments on the left and right sides of the preparation (see Figs 1, 3, 5 and 7, left-hand column) and sine-wave stimuli, 23ms in duration (1ms rise and fall times), 50–90dB SPL, usually tested on the side with the lower threshold (Figs 1, 3, 5 and 7, middle column). Each stimulus was repeated five times at a rate of 2s⁻¹. The intensity range (50-90 dB) for WN stimuli was shifted to lower values (30–70 dB) when necessary. The sine-wave stimuli were, for technical reasons, usually not tested at lower intensities. Therefore, we present the neurones’ responses to pure tones between 50 and 90 dB only. This is the intensity range where both low-frequency receptors and high-frequency receptors stimulate the interneurones (see Fig. 1). The low-frequency background noise (<1 kHz) was around 30–35 dB. The data were evaluated on a Data General Nova 4X with a ‘spike-detector’ interface (Zarnack and Mohl, 1977).

Intracellular or quasi-intracellular recordings were made with thin-walled borosilicate glass microelectrodes, whose tips were filled with a 3–5% solution of Lucifer Yellow (Aldrich) in 0.5 mol l⁻¹ LiCl or distilled water. After an experiment, the thoracic ganglia were fixed in 4% paraformaldehyde, dehydrated, and cleared in methylsalicylate. The whole mount with the stained cell was viewed under a fluorescence microscope, photographed, and drawn via a drawing tube. The relative depth of the observed structure in the ganglion was monitored with a measuring device (1 μm resolution) and revealed the three-dimensional structure of the cells. For each neurone type described in this study, at least three specimens were recorded and stained. When the first stained cell in an experiment was clearly identified by its physiology, in several cases a second cell was recorded and stained in the same preparation.

More than 450 identified neurones were investigated in approximately 400 female C. biguttulus. Most of them had their soma located in the metathoracic ganglion and were classified into 27 distinct neurone types. For reasons discussed below, the nomenclature introduced by Römer and Marquart (1984) for L. migratoria has been adopted for C. biguttulus. (SN, segmental neurone, only located in the metathoracic ganglion complex; BSN, bisegmental neurone, branching in the meta-and mesothoracic ganglia; TN, T-fibre with an ascending and descending axon; AN, neurone with an ascending axon). (For synonymity with other nomenclatural systems see Boyan, 1986; Stumpner, 1988; Robert, 1989.) Neurones that have not been described for L. migratoria are named according to the same system.

There are no obvious differences between the morphology of tympanic receptor fibres of C. biguttulus and of L. migratoria as far as the rough branching patterns in the metathoracic ganglion are concerned (see Römer, 1985; Halex et al. 1988). Fig. 1 shows the physiological characteristics of a low-frequency receptor and a high-frequency receptor. The intensity-response functions have a dynamic range of 20–25 dB. In the neurophysiological preparation, contralateral stimulation reduced the receptor’s sensitivity for WN by about 6 dB (range 3–8 dB) compared with ipsilateral stimulation. In behavioural tests a sensitivity difference of 8–9 dB was measured between the two ears (von Helversen, 1984); a 2dB difference evoked 100% correct turns in the males (see D. von Helversen and O. von Helversen, 1983).

The local auditory neurones SN1, SN4, SN5 and SN6 and BSN1 (which ascends to the mesothoracic ganglion) have several morphological features in common (Fig. 2). The somata of these cells lie in the frontal part of the metathoracic ganglion, in a lateral or ventral location. Dense dendritic processes with smooth endings predominate in the frontal auditory neuropile (fNP) ipsilateral to the soma. Branches in the contralateral fNP are less dense, have beaded endings, and are often restricted to the anterior part of the neuropile. The segment connecting these two branching areas differs among the neurone types: the crossing segment of SN1 and BSN1 is located near the anterior border of the fNP, the crossing segment of SN4 and SN5 runs through the caudal half of the fNP, and SN6 has a deep, ventral crossing segment. SN4, SN5, SN6 and BSN1 have descending branches on the contralateral side, which reach the region of the caudal neuropile (SN4, SN5) or the second or third abdominal neuromere (SN6, BSN1). Medially directed processes of these branches have beaded endings. The axon of BSN1 ascends to the mesothoracic ganglion, where it ends in two or three medially directed collaterals with a beaded appearance (in one out of 36 stainings, the axon ascended further than the mesothoracic ganglion).

The morphology of SN7 corresponds to that of DUM-type neurones (dorsal unpaired median, Evans and O’Shea, 1977), with a dorsomedially located soma and a symmetrical branching pattern in both frontal auditory neuropiles (Fig. 2). One auditory, non-spiking DUM neurone with very similar morphology has been described in the locust (SN3, Marquart, 1985b); SN7, however, is a spiking neurone.

Several local and bisegmental neurones exist as ‘twins’ (Römer and Marquart, 1984; Stumpner, 1989), i.e. two (or more) neurones of very similar morphology are found on each side of the ganglion. SN1 twins and BSN1 twins were demonstrated in double stainings.

SN1, SN4, SN5, SN6 and BSN1 show some similarities in their physiological properties. These neurones are more sensitive to ipsilateral stimulation (the terms ipsilateral and contralateral are used with respect to the soma of the neurones). Ipsilateral recordings reveal postsynaptic potentials (PSPs), whereas contralateral recordings appear to be similar to axonal penetrations. For SN7, the terms ‘ipsilateral’ and ‘contralateral’ cannot be defined relative to the soma position; this neurone shows identical reponses to stimulation from the left and from the right (Fig. 3D).

SN1 responds tonically and its intensity-response curve (Fig. 3A) is similar to that of receptors, though its spiking rate is typically lower (about 200 Hz at maximum), and a slight reduction in spike numbers may occur at high intensities. SN4, SN5, SN7 (Fig. 3B–D) and BSN1 (see Fig. 5A,B), in contrast, are inhibited to a certain degree at intensities more than 20 dB above threshold. This results in a phasic response or even total suppression of spikes (SN7, some examples of BSN1). The responses of different BSN1 neurones, however, depend quite differently on stimulus intensity: whereas some BSN1 cells can be called tonically responding neurones with a (sometimes only slight) reduction of spike numbers in response to louder stimuli (see Fig. 5A), others have a phasic-tonic to phasic spiking pattern, especially at higher stimulus intensities (see Fig. 5B). Differences were found not only in different preparations but also between the twins in one preparation (Stumpner, 1989).

The directionality of SN1 and SN4 is comparable to that of auditory receptors. Most BSN1 cells, however, exhibited reduced responses to contralateral stimulation, corresponding to an attenuation of 20–30dB (see Fig. 5A,B).

The lowest thresholds of SN1, SN4, SN7 and BSN1 lie in the low-frequency range (below 10kHz), while SN5 neurones are most sensitive to stimuli above 20 kHz (compare middle diagrams in Fig. 3). Intensity-response curves for different frequencies shift along the x-axis for SN1 and for some SN4 cells. Other SN4 cells as well as SN7 and BSN1 respond with a phasic pattern to low-frequency stimuli at any intensity, while high-frequency stimuli elicit tonic responses.

SN6 is usually spontaneously active; this activity is suppressed by auditory stimuli. A tonic hyperpolarization can be seen in ipsilateral recordings; often, this inhibition is interrupted by a single action potential a few milliseconds after the onset of the hyperpolarization (arrows in Fig. 4). With high-frequency stimulation, some SN6 cells exhibit a pure excitation near threshold, and, at high intensities, a postinhibitory rebound (Fig. 4).

Two metathoracic T-fibres, TNI and TN4, have their soma in a dorsolateral location of the first abdominal neuromere (TNI) or the metathoracic neuromere (TN4). The ascending and descending axons run ipsilateral to the soma (Fig. 2), unlike those of the ascending neurones (see Fig. 6). Both neurones show extensive branching: branches of TN4 cover nearly all the metathoracic ganglion complex, while the most prominent processes of TNI are positioned in both frontal auditory neuropiles with a very thick, ventrally situated segment crossing the midline. This branch of TNI seems to be mainly presynaptic, because recordings in this position never showed PSPs (see Peters et al. 1986; see also Römer and Marquart, 1984, who reported a slight hyperpolarization induced by high-frequency tones in TNI in L. migratoria). PSPs were visible in recordings of the more posterior, ipsilateral structures of TNI. One of many smaller dendrites in this posterior half crosses the midline and branches in the region of the caudal neuropile. The ascending axon was not stained beyond the prothoracic ganglion; in one TNI cell the axon clearly ended in the mesothoracic ganglion.

TNI and TN4 respond tonically. TNI cells spike (Fig. 5C) in a similar way to receptors: the tonic discharge shows a dynamic range from 50 dB SPL (threshold) to 70 or 80 dB SPL. The threshold for contralateral stimulation is 5–9 dB higher than for ipsilateral stimulation.

TN4 (Fig. 5D) is usually less sensitive than TN1 and its response is more variable; most TN4 cells show irregular spontaneous activity. In addition to acoustic stimuli, vibrations and air currents elicit suprathreshold responses in TN4 and some TN1 neurones.

The threshold of both neurones lies below 5 kHz and is thus lower than in other interneurones. 20 kHz stimuli elicit spikes only at intensities of 80 dB SPL or more.

Neurones with an ascending axon form the most prominent group of metathoracic auditory interneurones; at least 17 different types have been identified so far. The physiology of 10 neurones will be described in some detail. The remaining neurones (AN7, AN10, AN16-20) showed either a weak response to acoustic stimuli or were recorded only once or twice. The responses of AN1 (=B-neurone) have been described for C. biguttulus by Wolf (1986) in an extracellular preparation, and will be mentioned here only briefly.

The soma of ascending neurones can be found in a frontal lateral location (AN2, AN11, AN12, AN13), in a dorsolateral location near to the entrance of the sympanic nerve (AN3), in a more dorsal location (AN1, AN15) or in a ventral location near the midline (AN4, AN6, AN14) (Fig. 6). The axon ascends on the contralateral side to the brain; this has been proved for AN1, AN3, AN4, AN11 and AN12, and probably also holds for the other ascending neurones (see Hedwig, 1985, for Omocestus viridulus, and Eichendorf and Kalmring, 1980, for L. migratoria). The smooth dense dendrites in the frontal auditory neuropile originate ipsilaterally, except in AN4 where they originate contralaterally, and may cross the midline (AN3, AN4); AN11-AN15 have additional contralateral dendrites of the same structure. Recordings from regions with smooth dendrites show clear PSPs. Where beaded branches exist, they lie contralaterally in the metathoracic ganglion (AN1, AN6, in the region of the fNP; AN2, in the frontal dorsal one-third of the ganglion; AN3, only sparse). Single beaded branches can regularly be found in anterior thoracic ganglia.

Only one ascending neurone, AN6, responds ionically to WN stimuli at all intensities (Fig. 7F). This response does not saturate at the highest intensities tested. The highest spiking rates only exceptionally exceed 200 Hz. It is typical of this neurone that the intensity-response curves for ipsilateral and contralateral stimulation cross at approximately 60dB.

Some other ascending neurones also show tonic discharges in response to WN stimuli. However, these responses occur in the range near threshold (AN1, AN3, AN11: Fig. 7A,C,D,G) or at high intensities (AN2, AN3, AN4; Fig. 7B,D,E). At intermediate intensities these neurones respond phasic-tonically (e.g. Fig. 7D) or with scattered spikes (Fig. 7B). The spiking pattern of most ascending neurones adapts strongly. Furthermore, in some neurones, especially in AN2, the response is highly variable: the spiking pattern may be tonic, phasic or irregular for the same stimulus in one individual.

AN12 is an ascending neurone with a predominantly phasic response. From 5 to 10 dB above threshold (45–50 dB SPL) to 90 dB SPL, a phasic burst of 3-6 spikes is produced at the onset of a stimulus (Fig. 7H) with very short interspike intervals (sometimes less than 2 ms). At high intensities, this phasic burst is followed by additional spikes.

A characteristic of several neurones, an initial inhibitory postsynaptic potential (IPSP) before the excitatory response, can best be seen in dendritic recordings of AN4, AN3 and, to a lesser extent, AN12 and some AN6 cells (Fig. 7C,E,H). This IPSP is most clearly triggered by the onset of white-noise stimuli (see Ronacher and Stumpner, 1988) and low-frequency stimuli (less than 10kHz), but also occurs in response to high-frequency tones.

Three ascending neurones (AN13–AN15) were inhibited by acoustic stimuli. As an example, the intensity-response functions of an AN13 are shown in Fig. 8A. All inhibited cells were spontaneously active (Fig. 8B), though some of them had a rather low spiking rate (less than 5 Hz). The inhibition was detectable around 40–50 dB SPL; contralateral stimuli were slightly more effective than ipsilateral ones. The duration of the inhibition in all cases was diminished by adaptation.

The influence of stimulus direction on the spiking response differs substantially between ascending neurones. Nevertheless, all those described here exhibit the lowest threshold with contralateral stimulation, and the majority show a reduced response to ipsilateral stimuli. Note, however, that most of these neurones are also influenced by intensity, and this influence usually exceeds the dependence on stimulus direction (see Fig. 7C,G). AN2, however, is completely suppressed by ipsilateral stimuli up to 80 dB SPL (Fig. 7B); only the loudest ipsilateral stimuli evoke suprathreshold activity. AN1 shows a clear directionality, too, but is usually slightly excited by ipsilateral stimuli up to about 10-20 dB above threshold (Fig. 7A, see also Wolf, 1986).

Ascending neurones of C. biguttulus are most sensitive below 10 kHz. The responses to low-frequency stimuli are similar to the responses to white noise. High-frequency sounds elicit more uniform responses: the discharges are tonic over a broad intensity range. Only at the highest intensities can a decrease in spike number per stimulus be seen (Fig. 7A,C,E). Adaptation is less conspicuous with high frequencies than with low frequencies or white noise.

In grasshoppers, the metathoracic ganglion accommodates the first important level of auditory processing (see Introduction). The head ganglia must perform further filtering steps on the basis of information ascending from the thoracic ganglia. The diversity of information is obviously delimited by the set size and by the properties of ascending auditory neurones. Thus, it is important to have as complete a survey as possible of the thoracic auditory pathway. Several identified neurones (Stumpner, 1988) are not included in this study because of low excitability to acoustic stimuli or fragmentary physiological data.

The main objective of this study is to further the understanding of the auditory pathway of a grasshopper with elaborate acoustic communication. For a comparison of the neuronal responses with behavioural data one has to know that model songs composed of several WN syllables (as used here) are as effective as natural songs, provided that the temporal pattern is correct. The intensity of the male’s song reaches 76dB at 10cm distance. In most behavioural tests with females, intensities between 64 and 76dB were used (O. von Helversen, 1979). The behaviourally effective intensities of model songs (WN) ranged from approximately 45–50 to 80dB SPL (D. von Helversen, 1984, and personal communication). Stimuli lacking the high-frequency component are less effective. In both respects (intensity and spectrum) the females exhibit large interindividual variability.

The short response latencies and the location of the input regions (where PSPs can be recorded) on the side of greater sensitivity suggest that the local neurons get a least part of their input directly from ipsilateral receptors. In addition to low-frequency excitation, most local interneurones receive an inhibitory input at higher intensities (some of them perhaps from TNI, see Sokoliuk et al. 1989). BSN1 seems to be the only one of these thoracic interneurones that receives contralateral inhibition (Fig. 5A,B). Furthermore, BSN1 is obviously excited not only by the most sensitive low-frequency receptors but also by high-frequency receptors. The same might be true for SN7, while SN5 is exclusively excited by high-frequency input. The spiking patterns of local and bisegmental neurones in response to WN stimuli range from tonic respones (SN1), phasic-tonic responses (most BSN1), phasic responses (though usually only at higher intensities) to predominant inhibition at all intensities (SN6).

Ascending neurones, too, exhibit the whole range of responses from tonic (AN6) to distinct phasic activity (AN12). Obviously, most ascending neurones are excited by sensitive low-frequency receptors -probably via the local interneurones mentioned above. The most sensitive neurones show a response peak approximately 10–20 dB above threshold. WN stimuli above 50–60 dB SPL evoke complex patterns of excitation and inhibition, resulting in reduced spiking activity (Fig. 7). This inhibition usually begins at intensities at which the high-frequency receptors are not yet excited by the WN stimuli. Therefore, this inhibition must be caused by low-frequency receptors, probably mediated by interneurones like TNI (see Römer et al. 1981; Sokoliuk et al. 1989). In some neurones (AN2, AN3, AN4) a further excitation can be seen at intensities above 70 dB SPL. The intensityresponse functions at 5 and 20kHz in Fig. 7B,D,E suggest that this response can be interpreted as high-frequency excitation which overcomes the intensitydependent low-frequency inhibition.

BSN1, AN1 and AN2 are the neurones whose responses are most influenced by the direction of the sound (for AN1 see Wolf, 1986; for the locust see also Rheinlander and Mörchen, 1979). The activity of these neurones is rather effectively suppressed by contralateral (BSN1) or ipsilateral (AN1, AN2) stimuli. This would be in accordance with the BSN1 neurones being presynaptic to AN1 and AN2. In simultaneous recordings in L. migratoria, BSN1 has been demonstrated to excite AN1 (Marquart, 1985b). The response difference for ipsilateral and contralateral stimuli becomes smaller at middle intensities owing to the intensity-dependent inhibition mentioned above. Other auditory interneurones receive nearly equal excitation from both ears, as is most obvious for AN6 and AN12. A neuronal summation of auditory inputs from both ears has also been postulated from behavioural results (D. von Helversen, 1984; D. von Helversen and O. von Helversen, 1990).

AN3 and AN4 show a conspicuous IPSP which precedes the excitation at most intensities (see above); similar IPSPs can be seen also in AN12 and in some AN6 cells. What could be the function of an initial IPSP? In the first place, its effect is to delay the first action potential, which might be important in coding directional information (Römer et al. 1981; Rheinlaender, 1984). In AN3 and AN4 the latency is usually shorter for a contralateral stimulus than for the same stimulus delivered ipsilaterally. Another effect might be to trigger the first action potential more precisely; AN12, for example, shows a very constant latency (with less than 0.5 ms standard deviation in most cases).

There is extensive congruence of the morphological characteristics of thoracic auditory neurones in L. migratoria and in C. biguttulus. Only 5 out of 22 neurones described for the locust (Römer and Marquart, 1984; Marquart, 1985a) have not yet been found in C. biguttulus (SN3, TN2, AN5, AN8, AN9). 10 ‘new’ neurones have been identified in C. biguttulus (SN6, SN7, AN13-AN20, Stumpner, 1988); 6 of these could be stained in the locust (SN6, SN7, AN15, AN16, AN17, AN20). The morphological similarity extends to other acridid species, especially Omoces-tus viridulus (Hedwig, 1985,1986; see also Römer et al. 1988). The responses of the neurones to the stimuli used here are also very similar in both species. Interspecific differences exist, of course, in the characteristic frequencies (higher in the respective neurones of C. biguttulus) and in the sensitivity, with the locust being about 10-15 dB more sensitive to white-noise stimuli at the same temperature. At least for one auditory interneurone (TNI), the interspecific similarity also extends to the putative transmitter; in both L. migratoria and C. biguttulus the TNI neurone shows GABA-like immunoreactivity (Sokoliuk et al. 1989; see also Robertson and Wisniowski, 1988).

Uncertainty about homology exists only with the sister cells of the AN11/AN12 group and with SN5, which, in C. biguttulus, possesses a contralaterally descending branch like SN4; this branch, however, is missing in the locust (Marquart, 1985a).

The information flow in the auditory pathway of C. biguttulus seems to be the same as in the locust (see Römer and Marquart, 1984; Römer et al. 1988). Local interneurones probably receive direct input from auditory receptors (see above). All local neurones (except the DUM-type SN7) are assumed to have their main input region (smooth, dense dendrites) on the side ipsilateral to the soma, and their main output region (less-dense, beaded dendrites) on the side contralateral to the soma. In ascending neurones, too, the input regions are largely located on the soma side. The lower thresholds, however, are found on the axon side. As far as is known from L. migratoria, ascending neurones do not get direct input from auditory receptors (except AN10, see Pearson et al. 1985). Thus, local and bisegmental neurones seem to be interposed between receptors and ascending neurones. Of course, other interactions are to be expected in addition to this basic connectivity scheme.

In conclusion, there is little doubt that corresponding neurones of C. biguttulus and L. migratoria are homologous (criteria der Lage and der spezifischen Qualität, Remane, 1952; see also Rowell, 1989). However, stridulation has obviously been developed independently in the subfamilies Gomphocerinae (Chorthippus) and Oedipodinae (Locusta), since the stridulatory pegs are on the hind femur in Chorthippus but on the forewing in Locusta. Therefore, we conclude that the ancestral common set of auditory interneurones has been modified only slightly during the radiation of these grasshoppers and can be interpreted as a preadaptation for the evolution of acoustic communication. Consequently, we expect that in different species the mechanisms for recognizing conspecific songs (innate releasing mechanisms) will reflect in many details the properties of these common local and ascending neurones of the auditory pathway.

We want to thank Alfred Schmiedl and Erwin Schreier for many years of excellent technical assistance; Otto von Helversen and two anonymous referees gave valuable comments on the manuscript.