| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Zoology, University of Cambridge, Cambridge CB2 3EJ, United Kingdom; 2 Computational Neuroscience Group, Department of Theoretical Physics, Universidad Autónoma de Madrid, Madrid 28049 Spain
Submitted 27 June 2003; accepted in final form 27 June 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
, 2001). Neurophysiological studies on temporal processing in auditory pathways are generally based on the evaluation of spike times or average neuronal spike rates [e.g., Eggermont 2001; Langner and Schreiner 1988 (cat); Kuwada and Batra 1999 (rabbit); Grothe et al. 1997, 2001 (bat); Condon et al. 1991; Penna et al. 1997; Rose and Capranica 1985; (frog); Rheinlaender et al. 1976; Wohlers and Huber 1982 (cricket), Surlykke et al. 1988 (moth)]. An analysis of spike rates averaged over long time periods neglects the dynamics of the instantaneous neuronal spike rate as a coding principle for neuronal signal processing. The instantaneous spike rate, however, is crucial for temporal summation of postsynaptic potentials, and its importance is becoming more evident (Koch 1999). Moths evade bat cries when the afferent spike rate reaches 
500 Hz (Roeder 1964). Afferent axons in the auditory system of the grasshopper (Machens et al. 2003), neurons in the electro-sensory pathway of weakly electric fish (Wessel et al. 1996) and neurons in the primary visual cortex (Reich et al. 2000) encode information using high instantaneous spike rates of 
200 Hz. Here we take advantage of the simple behavior and auditory system of crickets and consider the instantaneous spike rate of auditory interneurons as a coding principle underlying pattern recognition in the auditory pathway.
As part of their mating behavior, male crickets produce loud calling songs and the females phonotactically walk or fly toward the singing males. Female phonotaxis is tuned to the temporal pattern of the species-specific song (Doherty 1985a; Pollack and Hoy 1979; Popov and Shuvalov 1977; Stout et al. 1983; Thorson et al. 1982). Behavioral results have led Thorson et al. (1982) to formulate a "30-Hz hypothesis", claiming that the 30-Hz syllable repetition rate is both sufficient and necessary for the release of phonotactic behavior in Gryllus campestris. However, because variations in chirp rate, syllable rate, and syllable number modify the attractiveness of the sound pattern in G. bimaculatus and Acheta domesticus, a more complex multicomponent analysis underlying pattern recognition has been suggested (Doherty 1985b; Stout and McGhee 1988; Stout et al. 1983).
The auditory pathway of crickets is well described (Ball et al. 1989; Michelsen 1998; Pollack 1998; Schildberger et al. 1989). From the cricket ears in the front tibiae, 60 primary afferents project into the prothoracic ganglion. The afferent input activates several different types of local, ascending, descending, and T-shaped auditory interneurons (Atkins and Pollack 1987; Popov and Markovich 1982; Popov et al. 1978; Wohlers and Huber 1978, 1982). Besides some evidence for low-level temporal filtering (Pollack 1986; Stabel et al. 1989; Wiese and Eilts 1985), it is assumed that these prothoracic interneurons are not involved in temporal filtering of the auditory pattern (Huber 1983; Schildberger et al. 1989; Wohlers and Huber 1982). The ascending interneurons are thought to copy the temporal structure of auditory stimuli and then forward this information to the brain. In the brain, serial processing by interneurons with low- and high-pass filter properties is thought to drive local brain neurons tuned to the species-specific sound pattern (Schildberger 1984; Schildberger et al. 1989). The conclusion that thoracic neurons just copy the sound pattern was based on evaluating the overall number of spikes elicited by different chirp patterns (Wohlers and Huber 1982). In our experiments, we challenged the preceding conclusion by analyzing the instantaneous spike rates of the local Omega neurons (ON1) (Casaday and Hoy 1977; Popov et al. 1978; Wohlers and Huber 1978) as the relevant coding principle for sound processing and conspecific pattern recognition.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Female G. bimaculatus were taken from the departmental colony at a minimum age of 7 days after final moulting. The colony was fed with dried dog food and water and kept at 24°C with a LD 12:12.
Dissection and recordings
Prior to dissection the animals were cooled for a maximum of 30 min at 4°C. The crickets were then fixed ventral side up on Plasticine. The hind and middle legs were tethered with metal clamps and the front legs stabilized in an upright position; care was taken not to obstruct the tympana. Experiments were performed at a temperature of 21–23°C.
For extracellular recording of the summed auditory afferent activity, the cuticle of the distal femur was opened. The dorsal branch of the leg nerve (Nocke 1972) was positioned on a pair of 125-µm silver wires and insulated with petroleum jelly. Signals were amplified and band-pass filtered (100–3,000 Hz).
Intracellular recordings were obtained from the prothoracic ganglion. The ventral connective membranes and the cuticular structures between the head and the mesothoracic sternites were removed. The prothoracic ganglion was positioned on a 0.5-mm-diam platform and stabilized by a minute ring pressing gently on its ventral surface. Intracellular recordings with micro capillaries of 80–120 M
resistance were obtained from an ON1 in its axonal or dendritic branches. Generally the axonal recordings were found to be more stable and less sensitive to hyperpolarizing current injections. The neurons were identified by iontophoretic staining with Lucifer yellow but could always easily be recognized by their typical response patterns (Wiese and Eilts 1985). After the experiments the ganglia were dissected, fixed in 4% para-formaldehyde, dehydrated in ethanol, and cleared in methyl-salicylate and finally the staining was checked under an epifluorescence microscope.
Sound stimulation
Sound patterns with a carrier frequency of 4.8 kHz were used that were identical to the patterns of previous phonotactic experiments (Thorson et al. 1982). Chirps were 250 ms in duration and were repeated at 500-ms intervals. Different chirp patterns had a different syllable period (SP), which increased in steps of 8 ms from 10 to 98 ms and had a 50% duty cycle (Fig. 1A). Thus at SP42, the syllable period was 42 ms and the syllable duration (SD) and silent period were 21 ms. All syllables had a rising and falling ramp of 2 ms. Sound was presented from two speakers housed in a 17-cm brass tube, the opening of which was positioned 20 mm away from the posterior tympanum of an ear. Intensity was calibrated to 75 dB SPL (RMS, sound pressure level relative to 20 µPa) at the position of the cricket ear with a Bruel&Kjaer microphone (Type 4191) and measuring amplifier (Type 2610). Twelve chirp patterns each with a different SP were consecutively presented for 5 s, and the whole paradigm with all 12 patterns was repeated in a continuous loop for 600 s. In some experiments, sound pulses of 250-ms duration were used instead of chirps. All acoustic stimuli were computer generated at a sampling rate of 22.05 kHz using Cool Edit 2000 and were presented using standard audio boards.
|
Data recording and analysis
Neuron recordings and sound patterns were sampled on-line to the hard disk of a PC using a National Instruments AD board (PC-MIO-16E-4) controlled by custom-made software written in LabView 5.01. Amplitude resolution was 12 bit, corresponding to increments 0.03 mV for the intracellular recording. The sampling rate was set to 10 kHz per channel. Data analysis was done off-line using the software NEUROLAB (Hedwig and Knepper 1992). As a measure of the auditory afferent response, the absolute voltage change of the summed nerve recording (
|dV|/dt) was calculated within a time window of 1 ms, which was continuously sliding in steps of 100 µs along the recording (Meyer and Hedwig 1995). This algorithm is not sensitive to DC offsets of the recording and takes into account the full amplitude of spikes even when they cross the baseline. The instantaneous spike rate, given by the number of spikes in a time interval divided by the length of the interval when this interval is small, can be calculated either fixing the time bin length and counting the number of spikes in these bins or fixing the number of spikes per bin and then using variable bin lengths. Both procedures give analogous results when the time bin lengths are comparable. We use the second method with a single spike per bin, also known as instantaneous discharge rate. The instantaneous spike rate of ON1 was calculated as the inverse of the duration of each interspike interval, e.g., if the interval was 5 ms, the instantaneous spike rate was 200 Hz. The value of 200 Hz was then assigned to that interspike interval (Hedwig and Knepper 1992; Janiszewski and Otto 1989). A Gaussian smoothing was not used to preserve the full dynamic of the instantaneous spike rate. Averages of spike rates across many trials used the beginning of chirps as the temporal reference point. Data of single experiments were exported from NEUROLAB to Excel for further pooled analysis.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
We first focused on the response pattern of the 60 primary auditory afferents. The synchronous activation of auditory afferents is reflected in the amplitude of extracellular recordings of the auditory nerve. When a chirp pattern similar to the species-specific calling song (SP 42 ms, SD 21 ms) was presented, the auditory afferents responded with bursts of spikes (Fig. 1A). These summed up to a typical extracellular multiunit recording with a phasic-tonic response. The dynamics of the response became obvious when we filtered and then averaged the afferent activity (see METHODS). The averaged response demonstrated a pronounced synchronous activity of the primary afferents triggered by the beginning of the syllables. The initial afferent response lasted for 5–10 ms and subsequently dropped to a tonic level of 50% of the initial peak. We systematically varied SP from 10 to 98 ms and quantitatively analyzed the afferent response pattern (Fig. 1B). At SP10 and SP18, the primary afferents clearly responded with phasic synchronized activity to all the individual syllables of a chirp. At SP98, the phasic-tonic response was similar to the response at SP42 with an initial peak activity triggered by the beginning of each syllable and a decline in activity to a tonic level after 5–10 ms. For nine of the crickets, we analyzed the amplitude of the peak response to the syllables of each sound pattern and pooled the results (Fig. 1C). The peak response showed a slight increase in peak amplitude by 10% from SP10 to SP98, although this summed activity of the auditory afferents was not tuned to any particular sound pattern. The population of afferents marked the onset of syllables, and this onset response was independent of syllable duration. This effect has a significant consequence for the coding of sound patterns. The overall duration of sound presented during the paradigms was similar for all SPs, but the number of afferent peak responses was not and corresponded to the number of syllables presented in each of the chirp patterns.
Analysis of interneuron spike activity
We studied the response properties of a prothoracic local auditory interneuron, the Omega neuron (ON1) with intracellular recordings. The ON1 neurons form mutual inhibitory connections (Selverston et al. 1985). They are involved in signal-to-noise filtering (Pollack 1988) and directional information processing (Wiese and Eilts 1985). We stimulated the ON1 neurons with the same stimulus paradigms as the auditory afferents.
When a chirp pattern with SP42 ms was presented, each syllable elicited a burst of spikes in the interneuron after a latency of 22 ms (Fig. 2A).
|
The instantaneous spike rate (see METHODS for details of calculation) gave a stair-like function (see Fig. 6C) that could be averaged over many trials to reveal the dynamics of the neuronal response (Fig. 2A, bottom). The instantaneous spike rate of ON1 changed distinctly in response to each syllable, and peak rates of 250–300 Hz were reached at the beginning of each response. The spike rate peak then produced a shoulder-like effect (Fig. 2A, 
) and rapidly declined to a background value that was determined by the last spike of a burst and the consecutive first spike of the next burst (Fig. 2A, *). The instantaneous spike rate of ON1 represented each syllable of the sound pattern as a distinct and rapid change in spike activity and matched the afferent input into the auditory pathway (see Fig. 1A).
|
|
300 Hz, which rapidly declined to 140 Hz. Evaluation of spikes/chirp versus instantaneous spike rate
In previous studies on thoracic auditory interneurons, the number of action potentials per chirp had been evaluated to analyze the coding properties of auditory interneurons but not the instantaneous spike rate (Huber 1983, Fig. 16C; Wohlers and Huber 1982, Fig. 10). From the ON1 data sets of 12 crickets, we calculated the number of spikes elicited by each of the chirp patterns as well as the maximum of the instantaneous spike rate averaged over many trials evoked by the syllables of each pattern. Both parameters were plotted against the SP (Fig. 3). The number of spikes per chirp varied between 17 action potentials (APs)/chirp at SP10 to 23 APs/chirp at SP58. This variation can easily be explained by the stimulus design in which an integer number of syllables is fitted into 250-ms chirps. As a consequence, the overall duration of sound presented in each of the 12 chirp patterns was not identical. For example, there was 120 ms of sound for SP50 but 145 ms for SP58 although both patterns contain the same number (5) of syllables. These differences in the overall duration of the sound presented were reflected in the small variations of the number of APs/chirp (Fig. 3A). In spite of this, there was still no apparent correlation between the stimulus patterns and this parameter of the neuronal response.
|
Plotting the maximum instantaneous spike rate against the SP patterns revealed a continuous increase in the maximum spike rate from 65 APs/s at SP10 to 240 APs/s at SP50 (Fig. 3B). The instantaneous spike rate of ON1 reached its maximum when the SP attained a value similar to the species-specific value of 42 ms. The maximum spike rate then stayed almost constant at 240 APs/s between SP50 and SP98. Such a response pattern with low discharge rate responses to short SPs represents a low-pass filter for syllable patterns.
Calculating neuronal tuning from instantaneous spike rate peaks
From our results we concluded that the syllables of the chirp pattern were reflected in the peaks of the instantaneous spike rate. As a consequence, the coding of the syllables in the spike rate of auditory interneurons should be sufficient to explain the tuning of cricket phonotaxis.
The relevance of instantaneous spike rate peaks has to be considered in the context of phonotactic behavior. The threshold for phonotaxis decreases with sexual maturation of the females (Sergejeva and Popov 1994) and phonotactic behavior occurs more frequently when females are male-deprived (Cade 1979). Within 1 h after copulation, the opposite happens and females become unresponsive, but phonotaxis can be restored when the abdominal connectives are cut (Loher et al. 1993). These results raise the question of whether changes in behavior correspond to changes in the activity of auditory neurons or whether they are related to threshold changes at the sensory-motor interface that drives phonotactic walking. There are contradictory reports on activity changes in auditory interneurons during sexual maturation (Loher et al. 1992; Sergejeva and Popov 1994; Stout et al. 1991). The restoration of phonotaxis after cutting the abdominal connectives may point toward a direct mechanosensory control of the motor networks by ascending abdominal activity. In any case, the auditory activity has to overcome a threshold for the release of phonotactic behavior. At present, the interaction between the activity of auditory neurons and the threshold for phonotaxis is not known, and we cannot discriminate between changes in threshold or changes in auditory activity. The authors are fully aware that therefore an analysis of what effect different thresholds have on the tuning of phonotactic behavior can only be obtained empirically. To calculate the tuning of the neuronal discharge activity, we consequently assumed several different instantaneous spike rate thresholds for the release of phonotaxis. As thresholds we used 180, 200, 212, and 225 Hz, which cover the range from 72 to 90% of the maximum ON1 response.
The calculation of the mean number of instantaneous spike rate peaks above a given threshold in a chirp can be performed directly by counting those peaks in different trials. We use an alternative method simply multiplying the probability that each syllable elicits spike rate peaks that reach a certain threshold by the number of syllables in a chirp. The probability to reach a spike rate peak above a threshold in a syllable is obtained using the variances in the instantaneous spike rate-tuning curves of ON1 (Fig. 3B) and approximating the distributions by Gaussians. The average number of spike rate peaks above a threshold of 200 Hz is given in Fig. 4, B and C. There are practically no syllables triggering an instantaneous spike rate of 200 Hz at SP10 and SP18 (Fig. 4C). There was then a steep increase in the number of syllables eliciting an instantaneous spike rate of 200 Hz, and the maximum was reached at SP34 with 6.2 discharge peaks and SP42 with 5.7 discharge peaks. The number of spike rate peaks then gradually dropped following the number of syllables constituting a chirp to about three at SP98. This result is indicative of an active band-pass filter with the maximum number of spike rate peaks reaching 200 Hz in accord with the species-specific sound pattern. In this calculation, the different syllable repetition rates were not taken into account. For instance, SP50 and SP58 both contained five syllables; however, the syllable repetition rate for SP50 was 20 Hz and for SP58 it was 17.2 Hz. The intervals between the spike rate peaks, however, might have considerable effects on temporal summation in the auditory pathway. If the number of discharge peaks reaching 200 Hz additionally was multiplied with the peak frequency, the result produced a band-pass function with a sharp rise to an optimum at SP34 and a slow gradual decrease in amplitude toward SP98 (Fig. 4D). Thus the tuning of this band-pass results from two principle variables: the increase in the instantaneous spike rate when SP increases from 10 to 42 ms and the decrease in the number of discharge peaks and the peak frequency elicited by each chirp pattern when SP increases from 42 to 98 ms.
|
The optimum curve obtained depended on the initially chosen threshold to be reached by the spike rate. In the next step we used different thresholds of 180, 200, 210, and 225 Hz and calculated the corresponding tuning curves (Fig. 5A). If the calculation was done for a low threshold (180 Hz), the resulting maximum was higher and shifted toward SP34. With increasing threshold the amplitude of the maximum decreased, the tuning curve became more broad and shifted to longer syllable periods. With a threshold of 225 Hz, the optimum of the tuning curve was at a SP of 50 ms. Thus our empirical choice of different thresholds resulted in a range of different neuronal tuning curves.
We then examined to what degree the calculated tuning curves fitted the observed phonotactic behavior. Data of the phonotactic performance of G. bimaculatus walking at 22°C (Doherty 1985a; Schildberger 1985) were used to compare the calculated tuning curves with the behavior. In these behavioral tests, syllable periods of 20–80 ms with corresponding syllable numbers had been tested. If a threshold level of 212 Hz was assumed (Fig. 5B), there was congruence between the calculated tuning and the performance of phonotactic walking. Both curves had an almost identical onset; both peaked at SP42 and then declined gradually toward longer SP. The close match between the calculated tuning curve based on neural instantaneous spike rates and the observed phonotactic behavior is consistent with three conclusions. First, phonotactic behavior obviously scales with the product of peak frequency and the number of spike rate peaks that reach a certain threshold. This means, e.g., that five syllables presented at 20 Hz are more effective than at 17.2 Hz. Second, changes in the magnitude of the phonotactic response will occur when one of these parameter changes. Third, the data indicate that phonotactic walking may be driven directly by the discharge peaks coding the syllable pattern.
Responses to tone bursts of chirp duration
Female G. bimaculatus not only tracked the species-specific syllable pattern, but 40% also oriented toward tone bursts of chirp duration (Doherty 1985b; Tschuch 1977). Because these sound pulses do not contain any syllable pattern, the orientation toward them is not explained by the properties of the proposed pattern recognition system in the brain (Schildberger 1984). We therefore analyzed the instantaneous spike rate of the neuronal response to such stimuli.
ON1 responded to a 250-ms constant tone burst with a phasic-tonic depolarization (Fig. 6A). The instantaneous spike rate averaged over many trials showed an initial peak of 310 Hz, which then rapidly declined to a tonic level of 155 Hz with no obvious modulation (Fig. 6B). If, however, an individual ON1 response was considered, the spiking of the neuron was not as regular as indicated by the average, and the variability in the neuron's instantaneous spike rate became evident (Fig. 6C). After the initial onset response, pairs and triplets of spikes occurred that caused transient peaks in the instantaneous spike rate that exceeded 250 Hz (Fig. 6C 
). We measured the time intervals between those consecutive spike rate peaks that reached 212 Hz (Fig. 6D). The frequency distribution of the peak intervals had its maximum at 10 ms and then tailed off to about 45 ms. Longer intervals occurred only with a low probability. The mean value for the spacing of the discharge peaks was 24.4 ms (SD 15.3 ms), which corresponds to a syllable period of 48.8 ms. The mean value was different if different thresholds for the discharge rates were considered. As a consequence, however, even during tone bursts of chirp duration, there was variation in the neuronal discharge activity that led to a statistical pattern of high discharge peaks that was very close to the maximum of G. bimaculatus phonotactic tuning curve (compare with Fig. 5B) and may therefore be sufficient to release and drive phonotactic behavior.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Afferent activity
The synchronous activity encountered in the summed recordings of the auditory afferents demonstrates that high and low syllable repetition rates are equally coded. In agreement with Esch et al. (1980), there seems to be no afferent tuning to syllable period. However, final conclusions about temporal filtering at the afferent level will have to be based on an analysis of the instantaneous spike rate of single afferents. In contrast to previous data describing tonic afferent activity (Esch et al. 1980), the summed afferent activity showed a strong phasic-tonic onset-response and marked the beginning of a syllable with a peak of synchronous activity. We conclude that the auditory afferents act as syllable-onset detectors. Interestingly, the duration of the syllables is not relevant for phonotaxis. Cricket phonotaxis can be released by syllables of only 2-ms duration and by duty cycles of 90% if presented at an optimum syllable rate (Thorson et al. 1982). Taken together, the behavioral experiments and the afferent data strongly indicate that the crucial and important coding of syllables occurs at their onset and that any further extension of sound pulses is not relevant for phonotaxis.
Instantaneous spike rates versus time-averaged spike rates
Our analysis of the number of spikes per chirp does not reveal a preferential response of the auditory ON1 neuron to any syllable pattern. Based on similar results, Wohlers and Huber (1982) concluded that thoracic neurons do not act as temporal filters for pattern recognition but merely relay any sound pattern up to the brain (Huber 1983). However, an analysis of neuronal activity by the number of spikes per chirp neglects the temporal dynamics of the neuronal response. Temporal filtering in the auditory pathway, however, must emerge from the continuous flow of afferent information in time. For the evaluation of this information by postsynaptic neurons, a temporal reference like the stimulus onset is not available. Surprisingly, we have noted that coding by instantaneous spike rate has been almost completely neglected in studies that analyze the activity patterns of vertebrate and invertebrate central auditory interneurons, although it is fundamental to temporal summation of postsynaptic potentials. Generally PST histograms with bins smaller than the interspike interval were derived from the overall neuronal response [e.g., Eggermont 2001; Langner and Schreiner 1988 (cat); Kuwada and Batra 1999 (rabbit); Grothe et al. 1997, 2001 (bat); Condon et al. 1991; Penna et al. 1997; Rose and Capranica 1985 (frog); Rheinlaender et al. 1976 (cricket); Surlykke et al. 1988 (moth)]. The importance of instantaneous spike rate coding, however, was demonstrated for acoustic startle responses in moths (Roeder 1964) as well as in Teleogryllus (Nolen and Hoy 1984), and its importance becomes evident in other sensory systems where information is encoded with high instantaneous spike rates (Reich et al. 2000; Wessel et al. 1996; see also Koch 1999).
When a quantitative analysis of instantaneous spike rates is applied to the activity patterns of ON1, filter properties become evident that are not detected by an analysis of time-averaged spike activity. These thoracic interneurons do not code high syllable repetition rates in their activity but reach a maximum spike rate at the species-specific syllable rate. Their spike rate demonstrates that a low-pass filter process has occurred at the very first stage of auditory information processing. The nature of this filter process is not yet clear, but it may relate to the gradual inhibition of ON1 that is triggered by its own suprathreshold activity (Pollack 1988; Sobel and Tank 1994).
Calculating phonotactic tuning from instantaneous spike rate peaks
The onset of syllables caused maximum activity in the auditory afferents and a peak response in the instantaneous spike rate of the ON1 interneuron. Because only the syllable onset is crucial for phonotactic behavior, these spike rate peaks must represent the syllable pattern at the neuronal level. We therefore propose that this onset activity is the neuronal representation of sound patterns that is crucial for phonotaxis.
Present data do not allow for discrimination between changes in the performance of phonotactic behavior due to changes in the threshold for phonotaxis and those due to changes of the activity level in the auditory pathway. Therefore empirical evaluations were made based on an analysis of arbitrary thresholds for the release of phonotaxis in relation to behavioral phonotactic data obtained from published literature. Considering the tuning curve of the instantaneous spike rate (Fig. 3B) and testing different thresholds for the processing of sound pulses allowed us to calculate how many syllables of a sound pattern are actually represented as spike rate peaks in the neural activity (Fig. 4C). The resulting curves are similar to phonotactic tuning but represent a rather broad band-pass filter. The increasing rising phase of the tuning curves depends on the increase in the instantaneous spike rate and the descending phase is simply a result of the decrease in the overall number of syllables presented in each chirp pattern. However, if additionally temporal summation at the postsynaptic site is considered and the frequency of the peaks is also taken into account, the number of spike rate peaks multiplied by discharge peak repetition rate reveals a tuning curve for the neuronal response that very closely matches the tuning of phonotactic behavior (Fig. 5B). The close correlation between the calculated tuning curves and phonotaxis indicates that phonotactic walking depends on both the number of syllables per chirp and the syllable repetition rate. This algorithm for temporal filtering takes into account the instantaneous spike rate and postsynaptic temporal summation and consequently can explain behavioral results that are not accounted for by previous assumptions or models of cricket temporal filtering. The algorithm uses instantaneous spike rates averaged over many trials. However, in the cricket the phonotactic decision process must be organized by a continuous processing of afferent information. Therefore experiments to test the algorithm with real-time spike rate data are currently being performed.
Shifts in the tuning of phonotactic walking
There is very little information about the intensity dependence of the tuning of the pattern recognition process. Thorson et al. (1982) tested G. campestris at 80 dB SPL and assumed that the temporal tuning may not be different at lower sound intensities. Doolan and Pollack (1985) speculate that tuning may shift with sound intensity. At low intensity, the tuning may be broad, and it may get more specific with higher sound intensity. Our data may provide an experimental basis for this assumption. If a relative low threshold for the release of phonotaxis is considered, the calculated tuning of phonotaxis peaks at short SP (Fig. 5). However, the number of discharge peaks that drive phonotaxis is high and covers a wide range of SPs. With higher thresholds, the tuning shifts to longer SPs, but the overall number of discharge peaks is low and the tuning curves become rather flat. Based on these calculations we expect a threshold dependent shift in phonotactic tuning. Interestingly, G. bimaculatus phonotaxis shifts with temperature from preferences of long syllable periods (about SP50–SP60) at 15°C to shorter syllable periods (about SP30–SP40) at 30°C (Doherty 1985a).
Comparison with 30-Hz hypothesis and trade-off phenomena
Thorson et al. (1982) claimed a syllable modulation at 30 Hz as the sufficient and necessary parameter of a song to trigger phonotaxis in G. campestris. In our calculation of the neural tuning curves, the syllable rate is an important factor but is not the exclusive one. The 30-Hz hypothesis cannot explain the trade-off experiments by Doherty (1985b) in G. bimaculatus and the findings on Acheta domesticus (Stout and McGhee 1988; Stout et al. 1983) that indicate that the animals also evaluate chirp rate, syllable rate, and the number of syllables. The attractiveness of a song is shifted when any of these parameters is varied. However, because the tuning of phonotactic behavior as described by Doherty (1985a) matches the product of number of spike rate peaks times frequency of spike rate peaks, phonotaxis should depend on both the number of syllables presented and the syllable repetition rate. A change in one of these parameters should lead to a change in phonotactic performance. Therefore the assumption of a multifactor analysis by the cricket (Doherty 1985b; Stout et al. 1983) and the simplified 30-Hz hypothesis (Thorson et al. 1982) both do not fully describe the mechanism of phonotactic tuning. Different combinations of syllable number and syllable rate provide sound patterns of different attractiveness. However, changes in one parameter may be compensated by the other parameter.
Phonotactic response to tone burst of chirp duration
Instantaneous spike rate coding also explains the attractiveness to tone burst of chirp duration (Doherty 1985b; Popov and Shuvalov 1977; Tschuch 1977) for phonotactic behavior in G. bimaculatus that cannot be explained by previous hypotheses or by the recognition process proposed to occur in the brain (Schildberger 1984). An analysis of individual ON1 responses demonstrated a considerable variability in its instantaneous spike rate during these constant pulses. Although a regular pattern of spike rate peaks is missing, the intervals of the peaks cover the natural range of effective syllables and may therefore be effective to drive phonotaxis. These patterns may not be ideal to release phonotactic walking, but it should be emphasized that in walking females the neural representation of the sound pattern is considerably distorted by central inputs and mechanical noise (Schildberger et al. 1988). Nonetheless these distorted neuronal activity patterns are sufficient to drive phonotactic behavior.
Temporal pattern recognition in the cricket auditory pathway
A processing of auditory information in the brain of G. bimaculatus by low-, high-, and band-pass neurons has been proposed as a basis for the recognition process underlying phonotaxis (Schildberger 1984). However, from that data, it cannot be concluded that the filter process is actually due to the integrative properties of these band-pass brain neurons. The filter process may occur at the thoracic level in the auditory pathway and may just be reflected in the activity patterns of the brain neurons. Any postsynaptic neuron in the auditory pathway that is excited by a discharge activity like the one encountered in the Omega neuron will automatically exhibit band-pass properties when stimulated with the range of syllable patterns.
So what kind of recognition process may underlie the cricket's behavior? Our data argue that temporal pattern recognition in the cricket could be based on two stages. First, the generation of instantaneous spike rate peaks as described here for the Omega neuron. Second, an instance that postsynaptically counts this spike rate peaks and activates phonotactic behavior accordingly.
The location of the second stage process is an open question. A simple implementation of this two-stage process could be realized at the thoracic sensory-motor level. Low-level auditory information processing has been used to design a robot performing auditory steering similar to crickets (Webb and Scutt 2000). The robot uses a direct excitatory connection of the artificial auditory neurons to the motor network. The tuning of robot phonotaxis closely resembled the phonotactic tuning of crickets and emerged without a high level pattern recognition process. ON1 is a central neuron of the cricket auditory pathway. Although it may not be necessary for pattern recognition as demonstrated by hyperpolarization of its spike activity in phonotactically walking G. bimaculatus (Schildberger and Hörner 1988) and by cell killing experiments in A. domesticus (Atkins et al. 1984), its spike activity indicates the kind of neural information that is available for temporal processing in the cricket auditory pathway (see also discussion by Pollack 1986). Although we cannot claim that the Omega neuron is part of the pattern recognition network, our data on ON1 demonstrate that neuronal processing based on instantaneous spike rate coding is sufficient to explain the tuning of phonotactic behavior in G. bimaculatus. This coding principle may also crucially contribute to temporal processing in other invertebrate and vertebrate auditory systems.
|
DISCLOSURES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
FOOTNOTES |
|---|
Address for reprint requests and other correspondence: B. Hedwig, Dept. of Zoology, Downing Street, Cambridge CB2 3EJ, UK (E-mail: bh202{at}cam.ac.uk).
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Atkins G and Pollack GS. Response properties of prothoracic interganglionic, sound-activated interneurons in the cricket Teleogryllus oceanicus. J Comp Physiol [A] 161: 681–693, 1987.
Ball EE, Oldfield BP, and Michel-Rudolph K. Auditory organ structure, development, and function. In: Cricket Behaviour and Neurobiology, edited by Huber F, Moore TE, and Loher W. Ithaca, NY: Cornell Univ. Press, 1989, p. 391–422.
Cade WH. Effect of male deprivation on female phonotaxis in field crickets (Orthoptera: Gryllidae; Gryllus). Can Ent 111: 741–744, 1979.
Casaday GB and Hoy RR. Auditory interneurons in the cricket Teleogryllus oceanicus: physiological and anatomical properties. J Comp Physiol [A] 121: 1–13, 1977.
Condon CJ, Chang S-H, and Feng AS. Processing of behaviourally relevant temporal parameters of acoustic stimuli by single neurons in the superior olivary nucleus of the leopard frog. J Comp Physiol [A] 168: 709–725, 1991.[Medline]
Doherty JA. Temperature coupling and "trade-off" phenomena in the acoustic communication system of the cricket, Gryllus bimaculatus deGeer (Gryllidae). J Exp Biol 114: 17–35, 1985a.
Doherty JA. Trade-off phenomenoa in calling song recognition and phonotaxis in the cricket Gryllus bimaculatus (Orthoptera, Gryllidae). J Comp Physiol [A] 156: 787–801, 1985b.
Doolan JM and Pollack GS. Phonotactic specificity of the cricket Teleogryllus oceanicus: intensity-dependent selectivity for temporal parameters of the stimulus. J Comp Physiol [A] 157: 223–233, 1985.
Eggermont JJ. Temporal modulation transfer function in cat primary auditory cortex: separating stimulus effects from neural mechanisms. J Neurophysiol 87: 305–321, 2001.
Esch H, Huber F, and Wohlers DW. Primary auditory neurons in crickets: physiology and central projections. J Comp Physiol [A] 137: 27–38, 1980.
Grothe B, Covey E, and Casseday JH. Medial superior olive of the big brown bat: neuronal responses to pure tones, amplitude modulations and pulse trains. J Neurophysiol 86: 2219–2230, 2001.
Grothe B, Park TJ, and Schuller G. Medial superior olive in the free tailed bat: response to pure tones and amplitude–modulated tones. J Neurophysiol 71: 1553–1565, 1997.
Hedwig B and Knepper M. NEUROLAB, a comprehensive program for the analysis of neurophysiological and behavioral data. J Neurosci Methods 45: 135–148, 1992.[ISI][Medline]
Huber F. Neural correlates of orthopteran and cicada phonotaxis. In: Neuroethology and Behavioural Physiology, edited by Huber F and Markl H. Heidelberg, Germany: Springer, 1983, p. 108–135.
Janiszewski J and Otto D. Response and song pattern copying of Omega-type I-neurons in the cricket, Gryllus bimaculatus, at different prothoracic temperatures. J Comp Physiol [A] 164: 443–450, 1989.
Koch C. Biophysics of Computation. New York: Oxford Univ. Press, 1999.
Kohne R, Atkin S, Stout J, and Atkins G. Enhanced calling song syllable period discrimination during one-eared phonotaxis by the female cricket (Acheta domesticus) J Comp Physiol [A] 170: 357–362, 1992.
Kuwada S and Batra K. Coding of sound envelopes by inhibitory rebound in neurons of the superior olivary complex in the unaesthetized rabbit. J Neurosci 19: 2273–2287, 1999.
Langner G and Schreiner CE. Periodicity coding in the inferior colliculus of the cat. I. Neuronal mechanisms. J Neurophysiol 60: 1799–1822, 1988.
Loher W, Weber T, and Huber F. The effect of mating on phonotactic behaviour in Gryllus bimaculatus (DeGeer). Physiol Entomol 18: 57–66, 1993.
Loher W, Weber T, Rembold H, and Huber F. Persistence of phonotaxis in females of four species of crickets following allatectomy. J Comp Physiol [A] 171: 325–341, 1992.
Machens CK, Schütze H, Franz A, Kolesnikova O, Stemmler MB, Ronacher B, and Herz AVM. Single auditory neurons rapidly discriminate conspecific communication signals. Nature Neuroscience 6: 341–342, 2003.[ISI][Medline]
Meyer J and Hedwig B. The influence of tracheal pressure changes on the responses of the tympanal membrane and auditory receptors in the locust Locusta migratoria L. J Exp Biol 198: 1327–1339, 1995.[Abstract]
Michelsen A. The tuned cricket. New Physiol Sci 13: 32–38, 1998.
Nocke H. Physiological aspects of sound communication in crickets (Gryllus campestris L.) J Comp Physiol [A] 80: 141–162, 1972.
Nolen TG and Hoy RR. Initiation of behavior by single neurons: the role of behavioural context. Science 226: 992–994, 1984.
Penna M, Lin W-Y, and Feng AS. Temporal sensitivity for complex signals by single neurons in the torus semicircularis of Pleurodema thaul (Amphibia: Leptodactylidae). J Comp Physiol [A] 180: 313–328, 1997.[Medline]
Pollack GS. Discrimination of calling song models by the cricket, Teleogryllus oceanicus: the influence of sound direction on neural encoding of the stimulus temporal pattern and on phonotactic behaviour. J Comp Physiol [A] 158: 549–561, 1986.
Pollack GS. Selective attention in an insect auditory neuron. J Neurosci 8: 2635–2639, 1988.[Abstract]
Pollack GS. Neural processing of acoustic signals. In: Comparative Hearing: Insects, edited by Hoy RR, Popper AN, and Fay RR. New York; Springer, 1998, p. 139–196.
Pollack GS. Who, what, where? Recognition and localization of acoustic signals by insects. Curr Opin Neurobiol 10: 763–767, 2000.[ISI][Medline]
Pollack GS. Analysis of temporal patterns of communication signals. Curr Opin Neurobiol 11: 734–738, 2001.[ISI][Medline]
Pollack GS and Hoy RR. Temporal pattern as a cue for species-specific calling song recognition in crickets. Science 204: 429–432, 1979.
Popov AV and Markovich AM. Auditory interneurons in the prothoracic ganglion of the cricket, Gryllus bimaculatus. II. A high-frequency ascending neuron (HF1AN). J Comp Physiol [A] 146: 351–359, 1982.
Popov AV, Markovich AM, and Andjan AS. Auditory interneurons in the prothoracic ganglion of the cricket, Gryllus bimaculatus deGeer. I. The large segmental auditory neuron (LSAN). J Comp Physiol [A] 126: 183–192, 1978.
Popov AV and Shuvalov VF. Phonotactic behaviour of crickets. J Comp Physiol [A] 119: 111–126, 1977.
Reich DS, Mechler F, Purpura KP, and Victor JD. Interspike intervals, receptive fields, and information encoding in primary visual cortex. J Neurosci 20: 1964–1974, 2000.
Rheinlaender J, Kalmring K, Popov AV, and Rehbein H. Brain projections and information processing of biologically significant sounds by two large ventral-cord neurons of Gryllus bimaculatus DeGeer (Orthoptera, Gryllidae). J Comp Physiol [A] 110: 251–269, 1976.
Roeder KD. Aspects of the noctuid tympanic nerve response having significance in ther avoidance of bats. J Insect Physiol 10: 529–546, 1964.[ISI]
Rose GJ and Capranica RR. Sensitivity to amplitude modulated sounds in the anuran auditory nervous system. J Neurophysiol 53: 446–465, 1985.
Schildberger K. Temporal selectivity of identified auditory neurons in the cricket brain. J Comp Physiol [A] 155: 171–185, 1984.
Schildberger K. Recognition of temporal patterns by identified auditory neurons in the brain. In: Acoustic and Vibrational Communication in Insects, edited by Kalmring K and Elsner N. Berlin, Germany: Parey, 1985, p. 41–50.
Schildberger K and Hörner M. The function of auditory interneurons in cricket phonotaxis. I. Influence of hyperpolarisation of identified neurons on sound localization. J Comp Physiol [A] 163: 621–631, 1988.
Schildberger K, Huber F, and Wohlers DW. Central auditory pathway: neuronal correlates of phonotactic behavior. In: Cricket Behaviour and Neurobiology, edited by Huber F, Moore TE, and Loher W. Ithaca, NY: Cornell Univ. Press, 1989, p. 423–458.
Schildberger K, Milde JJ, and Hörner M. The function of auditory interneurons in cricket phonotaxis. II. Modulation of auditory responses during locomotion. J Comp Physiol [A] 163: 633–640, 1988.
Selverston A, Kleindienst HU, and Huber F. Synaptic connectivity between cricket auditory interneurons as studies by selective photoablation. J Neurosci 5: 1283–1292, 1985.[Abstract]
Sergejeva MV and Popov AV. Ontogeny of positive phonotaxis in female crickets, Gryllus bimaculatus de Geer: dynamics of sensitivity, frequency-intensity domain, and selectivity to temporal pattern of the male calling song. J Comp Physiol]A] 174: 381–389, 1994.
Sobel EC and Tank DW. In vivo Ca2+ dynamics in a cricket auditory neurone: an example of chemical computation. Science 263: 823–826, 1994.
Stabel J, Wendler G, and Scharstein H. Cricket phonotaxis: localization depends on recognition of the calling song pattern. J Comp Physiol [A] 165: 165–177, 1989.
Stout J, Atkins G, and Zacharias D. Regulation of cricket phonotaxis through hormonal control of th threshold of an identified auditory neuron. J Comp Physiol [A] 169: 765–772, 1991.[Medline]
Stout JF, DeHaan CH, and McGhee RW. Attractiveness of the male Acheta domesticus calling song of females. I. Dependence on each of the calling song features. J Comp Physiol [A] 153: 509–521, 1983.
Stout J and McGhee R. Attractiveness of the male Acheta domestica calling song to females. II. The relative importance of syllable period, intensity, and chirp rate. J Comp Physiol [A] 164: 277–287, 1988.
Surlykke A, Larsen OL, and Michelsen A. Temporal coding in the auditory receptor of the moth ear. J Comp Physiol [A] 162: 367–374, 1988.
Thorson J, Weber T, and Huber F. Auditory behaviour of the cricket. II. Simplicity of the calling-song recognition in Gryllus, and anomalous phonotaxis at abnormal carrier frequencies. J Comp Physiol [A] 146: 361–378, 1982.
Tschuch G. The influence of synthetic songs on female Gryllus bimaculatus DeGeer (Part2). Zool Jb Physiol 81: 360–372, 1977.
Webb B and Scutt T. A simple latency-dependent spiking-neuron model of cricket phonotaxis. Biol Cybern 82: 247–269, 2000.[ISI][Medline]
Weber T, Thorson J, and Huber F. Auditory behaviour of the cricket. I. Dynamics of compensated walking and discrimination paradigms on the Kramer treadmill. J Comp Physiol [A] 141: 215–232, 1981.
Wessel R, Koch K, and Gabbiani F. Coding of time-varying electric field amplitude modulation in a wave-type electric fish. J Neurophysiol 75: 2280–2293, 1996.
Wiese K and Eilts K. Evidence for matched frequency dependence of bilateral inhibition in the auditory pathway of Gryllus bimaculatus. Zool Jb Physiol 89: 181–201, 1985.
Wohlers D and Huber F. Intracellular recording and staining of cricket auditory interneuyrons (Gryllus campestris L, Gryllus bimaculatus DeGeer). J Comp Physiol [A] 127: 11–28, 1978.
Wohlers D and Huber F. Processing of sound signals by six types of neurons in the prothoracic ganglion of the cricket, Gryllus campestris L. J Comp Physiol [A] 146: 161–173, 1982.
This article has been cited by other articles:
|
|
 |
|
  H. M. ter Hofstede and J. H. Fullard The neuroethology of song cessation in response to gleaning bat calls in two species of katydids, Neoconocephalus ensiger and Amblycorypha oblongifolia J. Exp. Biol., August 1, 2008; 211(15): 2431 - 2441. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. S. Jacobs, J. M. Ratcliffe, and J. H. Fullard Beware of bats, beware of birds: the auditory responses of eared moths to bat and bird predation Behav. Ecol., July 22, 2008; (2008) arn071v1. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Hedwig and J. F. A. Poulet Mechanisms underlying phonotactic steering in the cricket Gryllus bimaculatus revealed with a fast trackball system J. Exp. Biol., March 1, 2005; 208(5): 915 - 927. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Marsat and G. S. Pollack Differential Temporal Coding of Rhythmically Diverse Acoustic Signals by a Single Interneuron |
