Journal of Neurophysiology

Control of Cricket Stridulation by a Command Neuron: Efficacy Depends on the Behavioral State

Berthold Hedwig


Crickets use different song patterns for acoustic communication. The stridulatory pattern-generating networks are housed within the thoracic ganglia but are controlled by the brain. This descending control of stridulation was identified by intracellular recordings and stainings of brain neurons. Its impact on the generation of calling song was analyzed both in resting and stridulating crickets and during cercal wind stimulation, which impaired the stridulatory movements and caused transient silencing reactions. A descending interneuron in the brain serves as a command neuron for calling-song stridulation. The neuron has a dorsal soma position, anterior dendritic processes, and an axon that descends in the contralateral connective. The neuron is present in each side of the CNS. It is not activated in resting crickets. Intracellular depolarization of the interneuron so that its spike frequency is increased to 60–80 spikes/s reliably elicits calling-song stridulation. The spike frequency is modulated slightly in the chirp cycle with the maximum activity in phase with each chirp. There is a high positive correlation between the chirp repetition rate and the interneuron's spike frequency. Only a very weak correlation, however, exists between the syllable repetition rate and the interneuron activity. The effectiveness of the command neuron depends on the activity state of the cricket. In resting crickets, experimentally evoked short bursts of action potentials elicit only incomplete calling-song chirps. In crickets that previously had stridulated during the experiment, short elicitation of interneuron activity can trigger sustained calling songs during which the interneuron exhibits a spike frequency of ∼30 spikes/s. During sustained calling songs, the command neuron activity is necessary to maintain the stridulatory behavior. Inhibition of the interneuron stops stridulation. A transient increase in the spike frequency of the interneuron speeds up the chirp rate and thereby resets the timing of the chirp pattern generator. The interneuron also is excited by cercal wind stimulation. Cercal wind stimulation can impair the pattern of chirp and syllable generation, but these changes are not reflected in the discharge pattern of the command neuron. During wind-evoked silencing reactions, the activity of the calling-song command neuron remains unchanged, but under these conditions, its activity is no longer sufficient to maintain stridulation. Therefore stridulation can be suppressed by cercal inputs from the terminal ganglia without directly inhibiting the descending command activity.


The coordinated release and control of episodic or rhythmic motor patterns by central commands or sensory information involves neuronal processes of general importance for a neurobiological analysis of behavior. Neuronal design principles underlying the release and coordination of different motor activities are being increasingly understood (Dickinson 1995; Edwards et al. 1999; Getting 1989; Kristan and Shaw 1997; Morton and Chiel 1994; Pearson 1993; Stein 1978). The central release and activation of rhythmic and/or stereotyped motor activity seems to be controlled mainly by labeled lines; command neurons that activate dedicated networks, or pattern generators for specific motor programs (Edwards et al. 1999; Kupferman and Weiss 1978; Larimer 1988; Wiersma and Ikeda 1964). More plastic motor activity generated by distributed networks seems to be activated by populations of command-like interneurons (Kien 1983; Larimer 1988). Understanding the design principles for the control of motor patterns leads to the question of how the control mechanisms interact when incompatible motor patterns are evoked simultaneously. Experimental paradigms have been used in which different motor patterns are generated under changing experimental conditions. These demonstrate that distributed circuits may be reconfigured to produce a new output, whereas dedicated networks for particular motor patterns may interact by various degrees of inhibition (Dickinson 1995;Kristan and Shaw 1997; Morton and Chiel 1994).

The stridulatory behavior of crickets offers a favorable opportunity to study the intersegmental control of rhythmic motor patterns. Three motor programs drive the wing movements that produce the acoustic signals calling, rivalry, and courtship songs. The motor programs are generated within the thoracic ganglia and are only released under appropriate behavioral situations (Huber et al. 1989). Evidence from extracellular electrical stimulation of the brain or descending axons (Bentley 1977; Huber 1960,1964; Otto 1971) indicates that the brain controls the occurrence of the stridulatory motor patterns via descending interneurons. Huber (1960, 1964) assumed that the brain drives the thoracic stridulatory chirp pattern by descending rhythmic activity, but the stimulation experiments of Otto and Bentley indicated that a temporal pattern of descending activity is not required. The first aim of this study was therefore to identify the descending brain neurons that control stridulatory behavior and to analyze how the thoracic stridulatory network is set into action by central commands [some of these results have been reported in a short communication (Hedwig 1996)]. The second aim was to analyze principle control mechanisms of stridulatory behavior by eliciting another incompatible reaction. In resting crickets cercal wind stimulation evokes escape running (Gras and Hörner 1992; Stabel et al. 1985). Cercal wind stimulation of stridulating crickets can alter and impair the song pattern (Dambach et al. 1983) or evoke a silencing reaction, during which stridulation transiently stops (Dambach and Rausche 1985). By stimulating stridulating crickets with cercal wind stimuli I tested how any changes in the performance of stridulation are reflected in the activity patterns of the brain interneurons controlling the behavior. These experiments were designed to analyze whether the central commands are altered during sensory conditions that modify or suppress stridulatory behavior.


Animals and dissection

Mature male Gryllus bimaculatus (de Geer) were taken from cricket colonies at the Departments of Zoology in Göttingen, Germany and Cambridge, United Kingdom. The animals were kept on sawdust in plastic containers at 25–28°C with a 12-h light:dark cycle. They were fed a protein-rich diet and water.

Before dissection the animals were cooled to 4°C and immobilized. They then were positioned upright on a block of Plasticine, and all legs were restrained by small metal clamps. Care was taken not to damage the auditory organs in the front legs. The head was rigidly waxed into a U-shaped holder. The whole preparation then was positioned so that the head faced upward. The anterior head capsule was opened, the antennae and the median ocellar nerve were cut, and the brain was exposed. A platform attached to a micromanipulator was positioned below the brain, and a second ring-like platform applied gentle pressure to the anterior surface of the brain to stabilize it for intracellular recordings. One of the platforms also was used as a reference electrode. The brain was rinsed constantly with saline (Fielden 1960). Experiments were performed on 85 crickets.

Intracellular recordings

Microcapillaries were pulled from capillary glass (Clark, Hilgenberg, 1 mm OD) with a David Kopf puller (type 700C). The tip of the microcapillaries was filled with 5% Lucifer yellow Ch (Sigma) in double-distilled water and the shaft with 0.1 M LiCl. Microelectrodes were positioned with a micromanipulator (Leica). A digital gauge (Mitutoyo, Digimatic Indicator 543; resolution, 1 μm) attached to the manipulator measured the vertical progress of the electrode. Intracellular recordings were made in the bridge mode (npi SEC10 amplifier) from the medial anterior region of the protocerebrum. Recordings of the command neurons were obtained at a depth of ∼200 μm. Intracellular current injection was used to modulate the spike rate of the neurons or to inject dye iontophoretically into the cells. Because of vigorous ventilatory movements, some of the recordings were made in AC-coupled mode.

Sensory stimulation and recordings of behavior

Air puffs were delivered to the cerci and posterior area of the abdomen from a small tube with 1.5 mm ID positioned at a distance of 20 mm behind the animal. The strength and duration of the air puffs were controlled by a pressure regulator and an electronic valve. For acoustic stimulation, a piezoelectric speaker was positioned 31 cm above the dorsal midline of the animal. Sound pulses (4.5 kHz, 80 dB SPL amplitude, 25-ms duration; rising and falling ramp, 2 ms) were delivered by a custom made stimulator.

To monitor the stridulatory movements, a 1.5-mm-diam round piece of reflecting foil (3 M, Scotchlite type 7610) was stuck to the lateral part of the right forewing. The animal was illuminated by a DC light source, and the light reflected by the foil was picked up by a linear position sensitive diode (Laser Components, Type 1L30-UV) built into the film plane of a reflex camera. The up-down movements of the wing, which were recorded by the measuring system, mirror the opening-closing movements during sound production because the wings perform an inward-outward rotation during stridulation (Kutsch 1969). In some experiments, sound was recorded simultaneously with the movements by a microphone (Audio-Technica, AT853Rx) positioned lateral to the animal. Experiments were done at an ambient temperature of 20–23°C.

Data analysis

All data were sampled on-line to the hard disk of a PC with Turbolab 4.3 software (Stemmer Technology, Gröbenzell, Germany) and a Data Translation A/D board (DT2821 G8DI). The sampling rate was 10 kHz per channel. Data evaluation was carried out with NEUROLAB software (Hedwig and Knepper 1992; Knepper and Hedwig 1997). Data for correlation and regression analysis were calculated with NEUROLAB and then exported to a spreadsheet program (MS Excel 97).


After recording, the brain was dissected for histological analysis. The tissue was fixed in 4% paraformaldehyde, dehydrated in ethanol, and cleared in methylsalicylate. Drawings and photographs of the stained neurons were made with an epifluorescent microscope (Leica Dialux 20 or Zeiss Axiophot). After embedding in polyesterwax, sagittal sections (20 μm) of the tissue were cut with a microtome (Reichert Jung Model 1130/Biocut).


Intracellular recordings and stainings were obtained in the anterior protocerebrum in an area where electrical stimulation has been shown to elicit singing behavior (Huber 1960). Several hundred neurons were tested for their behavioral effects by intracellular stimulation. Some neurons evoked episodic or rhythmic movements of the abdomen, legs or wings. One type of descending interneuron that elicited stridulatory behavior was identified from a total of 15 recordings and nine stainings.

Morphology of the calling-song command neuron

Stainings of this interneuron always revealed a typical structure (Fig. 1, A and B). The soma lies at the dorsal surface of the protocerebrum ∼100–200 μm lateral to the midline. The primary neurite loops ventrally in the protocerebrum where it has a diameter of ∼7 μm. In the ventral protocerebrum, the neurite divides into major dendritic branches that project bilaterally toward the pedunculi and especially arborize in the neuropil between the α-lobe and pedunculus (Fig. 1 B, left and middle). The descending axon runs anterior to the central body (Fig. 1 B, right). Collaterals arising from the axon project parallel to the axon and arborize on both sides of the anterior protocerebrum. The axon crosses the midline and enters the connective in a dorsal medial position. In the connectives, the axon has a diameter of 3–4 μm, and in the subesophageal ganglion, it runs in a very medial dorsal position. Stained axons could be traced as far as the prothoracic ganglion, but the terminal arborizations within the thoracic ganglia have not been identified. The neuron is bilaterally paired and so far the stainings provide no evidence for sibling command neurons as in the grasshopperO. viridulus (Hedwig 1994).

Fig. 1.

A: arborization pattern of the descending calling-song command neuron in the brain of Gryllus bimaculatus in dorsal view. Soma position is dorsal, the primary neurite loops to ventral areas of the protocerebrum. Axon and dendrites project anterior to the central body. Lateral dendrites extend toward the pedunculus of the mushroom body. Axon crosses the midline and enters the contralateral connective in a median dorsal position. B: 3 compositions of parasagittal sections indicate the arborization pattern of the interneuron in the median protocerebrum. MB, mushroom body; Ped, pedunculus; CB, central body; α-L, α-lobe; PB, protocerebral bridge; DC, deutocerebrum.

Stimulation of calling-song stridulation in resting crickets

Crickets stridulate by rhythmical movement of their forewings. The wings are lifted into singing position and sonorous syllables are produced by opening-closing movements of the wings. During calling songs, three to five syllables are grouped within a chirp. Chirp cycles are ∼350–400 ms in duration, and the syllables last 16–20 ms each (Fig. 2 C) (Huber 1960).

Fig. 2.

Release of calling-song stridulation in G. bimaculatusby intracellular current injection into a single descending brain neuron. A: after an initial activity peak, the current pulse (top) elicits a spike rate of 60–80 spikes/s (2nd and 5th traces). Cricket raises its wings into singing position (3rd trace) and starts sonorous stridulation (4th trace). Singing stops after the end of stimulation. During stridulation, the neuron activity is modulated rhythmically in the chirp rhythm. B: gradual repositioning of the wing after the sequence of calling-song stridulation takes 50 s. Slow movements of the wing are due to ventilatory contractions of the abdomen. C: chirp and syllable pattern during calling song. The calculation of phase values for syllables and spikes was done corresponding to phase =t/chirp cycle. D: average wing movement during the chirp cycles. Prior to averaging, all chirp cycles were normalized to a unit length of 1.0 (top). Phase histogram of the syllables, the maximum of the opening movement, is taken as indicator of a syllable (middle).Bottom: phase histogram of the neuron activity together with the average instantaneous spike frequency (left ordinate for peristimulus time (PST) histogram and right ordinate for average spike frequency). All averages are calculated corresponding to the phase of the chirp cycle. In phase with the chirps, the neuron activity reaches 81 spikes/s where it is 61 spikes/s in the chirp interval. The number of evaluated chirps, and the number of syllables and spikes that occur during these chirps, is indicated by n in each histogram. Number of bins, 100; binwidth corresponds to 3.65 ms. All chirps from a single animal.

In some experiments, crickets started stridulating when the microelectrode probed the protocerebrum. This suggested the close proximity of the crucial descending interneuron. If the interneuron was impaled and a transiently enhanced spike frequency was caused, the cricket immediately started to stridulate with a calling song. With a stable recording, calling-song stridulation could be released by depolarizing current injection that elicited spike frequencies of 60–80 spikes/s (Fig. 2 A). About 500 ms after current injection, the cricket began to raise its forewings into the singing position, and stridulation started with low movement amplitude and soft incomplete chirps sometimes consisting of only one syllable. After a few seconds, the wings had reached the singing position, and fully developed chirps with three to four syllables were generated. Stridulation continued as long as the enhanced spike frequency was maintained by current injection. After the end of the applied depolarization, a few further chirps occurred; however, the chirp interval distinctly increased and stridulation came to a stop within the next few seconds. When the stridulatory movements had ceased, the wings remained in the elevated stridulatory posture for many seconds or even minutes before being lowered gradually into the resting position (Fig. 2 B).

During stridulation, the activity of the interneuron showed a rhythmical modulation in phase with the chirp cycle (Fig. 2,A and D). The phase relationship between the wing movements (Fig. 2 D, top) and the neuron spikes shows a broad distribution of the neuron activity within the chirp cycle (Fig.2 D, bottom). The maximum mean spike frequency of 81spikes/s occurs at phase ϕ = 0.1 and coincides with the first and the second syllable (see Fig. 2 D, middle). However, there is no obvious modulation of the spike activity in the syllable rhythm. The mean spike frequency gradually decreases to 61 spikes/s toward the end of the chirp cycle at ϕ = 0.9. It increases again in phase with the opening movement of the first syllable at ϕ = 0.9–0.1. Thus there is a modulation of the spike rate within the chirp cycle with the major spike activity occurring in phase with the generation of the chirps. The phase histograms, however, demonstrate that the maximum spike frequency does not precede the generation of chirps, indicating that the timing of the chirps is not directly triggered by the rhythmic command activity. Even after current stimulation, the interneuron remained rhythmically active in the chirp rhythm when stridulation gradually waned (Fig. 2 A, see also following text).

Repeated depolarizations of the interneuron, each for several seconds in duration, reliably elicited repeated bouts of calling-song stridulation (Fig. 3 A). During the first current application, the wings were raised into the singing position and stridulation started. If the wings were held in the singing position until the next stimulus, stridulation started with a larger movement amplitude (e.g., Fig. 3 A, beginning of 2nd current pulse). On the basis of the criteria used for the characterization of command neurons (Kupferman and Weiss 1978), this descending brain neuron clearly is sufficient to release calling-song behavior in the cricket G. bimaculatus.

Fig. 3.

A: reliable stimulation of calling-song stridulation in a resting cricket. Interneuron was depolarized by 3 consecutive current pulses of 7- or 8-s duration that elicited a spike frequency of 120–130 spikes/s. During the first current pulse, the wings were raised into singing position and stridulation started with wing movements of low amplitude. Wings remained raised, and at the beginning of the following current pulse, stridulatory movements started with greater amplitude. B: repeated stimulation with 1-s depolarizing current pulses increases the intensity of calling-song stridulation. Each current pulse elicits a rate of 140–150 spikes/s in the interneuron. During the 1st stimulus, the wings were raised and several incomplete chirps were generated. During the following current pulses, the wings were more raised and more complete chirps with 4 syllables were produced.

The effectiveness of interneuron stimulation in eliciting stridulation depended on the history of any preceding stridulation. A series of 1-s-long current pulses at intervals of 1.5–2 s, each of which always elicited a spike frequency of 140–150 spikes/s, were used in resting crickets, which—besides a short test to identify the interneuron—had not stridulated before in the experiment (Fig. 3 B). The first of the pulses elicited only a slight elevation of the forewings and some rapid wing movements vaguely resembling chirps. During the second and third current pulse, the wings were elevated further, and incomplete chirps were generated at the end of the third pulse. With the next four current pulses, wing position and stridulation improved, and finally complete chirps with four syllables were generated by the end of the stimulation series. This suggests that the sensitivity of the stridulatory pattern generator to command input is enhanced by maintained stridulation. The effectiveness of the interneuron activity therefore depends on the preceding activity of the thoracic pattern-generating network. This becomes particularly clear in crickets that performed long-lasting stridulatory sequences.

Release and inhibition of calling-song stridulation in “aroused” crickets

Electrical stimulation experiments within the brain and of the cervical connectives previously have indicated that the threshold for the release of stridulation decreases once a cricket has generated a few song sequences. In crickets that had performed continuous stridulatory activity, for example, a short sequence of electrical pulses is sufficient to trigger prolonged stridulation (Huber 1960, 1964; Otto 1971). This change in threshold could be demonstrated in three crickets at the level of the descending brain neurons.

The cricket in Fig. 4, A andB, had performed sequences of prolonged stridulation elicited by maintained intracellular injection of current into the interneuron. A subsequent 1,500-ms pulse of current caused a burst of action potentials at 130 spikes/s followed 500 ms later by incomplete calling-song chirps consisting of single syllables. At the end of the depolarization, stridulation was not abolished. Moreover the interneuron activity triggered a sequence of calling-song chirps with three to four syllables at an average chirp interval of 440 ms. During this “self-maintained” stridulation, the spikes of the interneuron were modulated in the chirp rhythm (Fig. 4 C) even though their frequency was now much lower than during stridulation maintained by current injection (see Fig. 2). The highest average spike rate of 27 spikes/s occurred in phase with the syllables at ϕ = 0.2, and this frequency gradually decreased to an average of 18 spikes/s at ϕ = 0.85 close to the end of the chirp cycle. Thus during this self-maintained stridulation, a much lower discharge rate of the interneuron was sufficient to maintain stridulatory activity. Before stridulation (Fig. 4 A, left) the spike frequency of the interneuron was as high as the maximum discharge rate during self-maintained stridulation but was not sufficient to start stridulation.

Fig. 4.

A: brief increase in interneuron firing from 30 to 130 spikes/s caused by injection of depolarizing current triggered a long-lasting sequence of stridulation in an aroused cricket. Interneuron activity was modulated rhythmically during stridulation.B: suppression of the interneuron‘s spike activity by hyperpolarizing current injection stopped the calling song.C: average of the wing movements during chirps (top), phase histograms of the syllables (middle), phase histogram of the action potentials(bottom left), and average of the instantaneous spike frequency in the chirp cycle (bottom right). Maximum spike rate occurred in phase with the generation of the chirps.

Interneuron activity was necessary for self-maintained stridulation to continue. When the interneuron spikes were abolished by hyperpolarizing current injection (Fig. 4 B), stridulation stopped immediately, and no additional chirps were produced. Thus under the circumstances of self-maintained stridulation the interneuron also fulfilled the necessity criterion for the characterization of command neurons (Kupferman and Weiss 1978).

Interneuron impact on chirp rate and syllable rate

Crickets use two different pattern generators to produce the motor patterns for the chirp rhythm and the syllable rhythm. The timing of both generators is set independently (Bentley 1969;Kutsch 1969). To characterize more precisely the relationship between the interneuron activity and the chirp and syllable generators, the correlations between the interneuron activity and the chirp and syllable rates were analyzed quantitatively. The instantaneous spike frequency was calculated at the onset of chirps and at the time of maximum syllable rate of each chirp (Fig.5, A and B, insets). Calling-song sequences either were evoked by intracellular current injection or were self-maintained so that a wide range of interneuron spike frequencies could be pooled for the analysis. The chirp repetition rate and the maximum syllable repetition rate within each chirp were determined for stable parts of stridulatory sequences only. Chirps immediately after the onset or offset of current injection were not considered.

Fig. 5.

Statistical analysis of the correlation between the spike frequency and the chirp and syllable repetition rate, respectively. A: there is a high correlation (r = 0.83) between the interneuron spike frequency and the chirp rate. B: there is no correlation (r = 0.1) between the maximum syllable rate and the interneuron spike rate. C: intracellular current injection into the interneuron during self-maintained stridulation evoked a transient increase of the spike rate that caused a transient increase in the chirp rate and reset the chirp pattern. Dots indicate the expected continuation of the chirp pattern if no current would have been applied. There was no change in the maximum syllable rate.

The correlation analysis between the interneuron's spike frequency and the chirp repetition rate (Fig. 5 A) shows a high positive correlation coefficient (r = 0.83, significant at 0.001% niveau) and a linear regression function with y = 0.0237x + 1.69. This demonstrates a tight coupling between the interneuron activity and the chirp repetition rate. The chirp rate increases linearly from 2.4 chirps/s at 30 spikes/s to 4.5 chirps/s at 120 spikes/s. Only a very weak correlation was revealed between the interneuron's activity and the maximum syllable rate of the song (r = 0.1, significant at 0.01% niveau) with the linear regression function y = 0.0042x + 26.44 (Fig.5 B). Independent of the neuron activity, the maximum syllable rate stays between 26.6 syllables/s at 30 spikes/s and 26.9 syllables/s at 120 spikes/s. The interneuron activity, therefore drives the thoracic chirp pattern generator but has no impact on the repetition rate of the syllables.

To test the effect of experimentally increased neuron activity on the performance of the ongoing behavior (Fig. 5 C), intracellular current injection was used to modulate the interneuron spike frequency during self-maintained stridulation. During such sequences, the discharge rate was rather low, ∼30 spikes/s, and the chirp repetition rate was 3.6 chirps/s. When the interneuron discharge rate was transiently enhanced to 146 spikes/s, the chirp rate also transiently increased to 5.6 chirps/s. Thus increasing the spike rate drives the chirp generator and shifts the chirp repetition rate corresponding to the regression function (Fig. 5 A). This effect of course also led to a reset of the chirp generator. The chirps after the onset of stimulation occurred before moment based on the previous chirp rate. Note that the time between the onset of the intracellular stimulus and the consecutively following chirp is only 150 ms and well beyond the duration of a chirp cycle. Therefore a sudden short increase in the interneuron activity will be sufficient to reset the thoracic chirp pattern generator.

There was no change of the maximum syllable repetition rate during the intracellular interneuron stimulation; it was 26.2 syllables/s before and at 26.4 syllables/s during the current injection. There was, however, a reduction in the number of syllables generated during each chirp; this also is described by Otto (1971).

Sensory feedback to the calling-song command neuron

To identify possible sources of the rhythmical modulation of the interneuron spike frequency during stridulation, the effect of different sensory modalities was tested. Changes in light intensity and auditory stimulation (80 dB SPL, 4.5-kHz pulses with 25-ms duration; Fig. 6 A) did not evoke any synaptic input in the interneuron in resting or stridulating crickets. A sensory stimulus so-far identified that evoked activity of the neuron was puffs of air blown across the cerci (Fig. 6 B). The air puffs increased the spike activity of the interneuron and also elicited abdominal contractions similar to ventilatory contractions, which slightly lifted the wings (see Fig. 2 B). The enhanced interneuron activity therefore also might have been caused by any proprioceptive feedback from the abdomen and wings. Because abdominal movements alone did not elicit any enhanced interneuron activity, it is suggested that cercal wind stimulation is a source for sensory feedback to the descending command neurons.

Fig. 6.

A: interneuron did not respond to acoustic stimulation at 4 kHz, 80 dB SPL. Interneuron was hyperpolarized with 3 nA throughout the recording. B: stimulation of the abdomen with air puffs was followed by excitation of the interneuron but also caused abdominal contractions which passively elevated the wings.

Command neuron activity and cercal wind stimulation during stridulation

In naturally singing crickets (G. campestris), stimulation of cercal hairs with air puffs can elicit two different behavioral effects; single weak air puffs change the chirp rhythm (Dambach et al. 1983). Stronger air puffs elicit a silencing response during which crickets lower their wings and stop stridulation (Dambach and Rausche 1985). Here puffs of air disturbed the chirp rhythm and when delivered within a chirp, also disrupted the performance of the syllables within that chirp (Fig.7 A). Silencing reactions could be evoked by increasing the amplitude of the air puffs.

Fig. 7.

A: air puffs of 50-ms duration delivered to the cerci disturbed ongoing calling-song activity. Air puffs impair the chirp rhythm and terminated ongoing chirps. B: air puff prematurely terminates a chirp, but the change in the stridulatory motor activity is not accompanied by any change in the firing rate of the interneuron.

The behavioral effects of air puffs applied to the cerci can be demonstrated when self-maintained stridulation occurs during intracellular recordings (Fig. 7 B). In these experiments, calling-song stridulation occurred after intracellular current injection of the descending command neuron. Air puffs of different strength were delivered to the cerci. When presented during ongoing chirps, air puffs led to a slight increase in the interneuron spike rate, for example, from 20 to 29 spikes/s for the duration of 160 ms. An increase in the interneuron activity also should increase the singing activity, but the chirps ended prematurely. The evoked change in the syllable pattern was not reflected in any obvious change in the command neuron activity. This may correspond to the finding that there also was no correlation between the syllable rate and the interneurons spike frequency.

Higher strength of air puffs elicited a transient increase in the interneuron activity and a silencing of the motor pattern (Fig.8 A). The cricket remained silent for several seconds and then began to stridulate again. During the pause in stridulatory behavior, however, the command neuron continued to spike with the same average spike frequency as during stridulation. Data from five experiments in three animals show the average wing position and spike histogram and interneuron spike rate during the silencing reaction (Fig. 8 B). The wing position indicates a rapid lowering of the wings starting ∼110 ms after the beginning of the air puff. In the interneuron, the average discharge rate increased during the air puff to ∼58 spikes/s. Before the air stimulus, while the cricket was still stridulating and after the stimulus during the silencing reaction, the average spike frequency was constant at ∼33 spikes/s. The silencing reaction changed the performance of the stridulatory motor activity without an obvious modification of the descending command neuron activity. Thus during the silencing reaction, the command neuron activity was not effective in maintaining calling-song stridulation. In this experimental situation, it failed to meet the sufficiency criterion.

Fig. 8.

A: silencing reaction caused by an air puff delivered to the cerci of a singing cricket. Cricket lowered its wings and stopped stridulation for 4 s. Interneuron responded to the air puff with a slight transient increase in the spike rate but then continued to fire with the same rate as before the silencing reaction. B: average response of the wing (2nd trace) and the interneuron (2 bottom traces) in 5 different silencing reactions obtained in 3 singing crickets. Wings are lowered after 110 ms. There is an increase in the spike activity after ∼70 ms. During stridulation and during the silencing reaction, the interneuron activity remains at ∼33 spikes/s.


Within the brain of G. bimaculatus, a descending interneuron was identified as a command neuron for calling-song stridulation. The function of this interneuron seems to be rather complex because its discharge is modulated rhythmically during stridulation, it can modulate and reset the stridulatory pattern generator, it may receive sensory feedback, and its effectiveness changes depending on the behavioral state of the animal. Together these features are beyond a simple control of a rhythmic motor pattern by a command interneuron and indicate a high degree of plasticity within the stridulatory motor system.

Stimulation sites and descending brain neurons in the cricket

In the cricket, electrical stimulation in the anterior protocerebrum (Huber 1960, 1964) and extracellular electrical stimulation of a dorsal medial axon within the connectives (Bentley 1977) were most effective in eliciting calling song. The arborization pattern of the calling-song command interneuron corresponds to these stimulation sites. I therefore suggest that the effects of earlier stimulation experiments were mediated by the direct or indirect activation of the calling-song command neuron identified in the present paper.

A survey of interneurons and soma cluster in the cricket brain reveals ∼200 descending brain neurons projecting into each circumoesophageal connective (Staudacher 1998). Individual descending interneurons are involved in the processing of visual stimuli (Richard et al. 1985), acoustic stimuli (Boyan and Williams 1981; Brodfuehrer and Hoy 1990), or exhibit multimodal response properties (Hörner and Gras 1985). Some descending interneurons alter their sensory responses during walking and may even function as command neurons for walking (Böhm and Schildberger 1992; Staudacher and Schildberger 1998). All these descending interneurons exhibit great differences in arborization patterns and none of them matches the structure of the descending calling-song command neuron.

Descending interneuron as a command neuron for calling song

Wiersma and Ikeda (1964) introduced the concept of command neurons, postulating that the brain of invertebrates controls coordinated movements by tonic activity of specific sets of interneurons. More rigid criteria for the characterization of command neurons were established by Kupferman and Weiss (1978): command neurons should be sufficient and necessary for the performance of a particular motor behavior. Stimulation of the interneuron should evoke the behavior in a resting animal and inhibition of the interneuron should prevent occurrence of the behavior. Although only a small number of command neurons fulfill both of these criteria (e.g.,Edwards et al. 1999; Fredman and Jahan-Pawar 1983; Frost and Katz 1996; Hedwig 1994; Nolen and Hoy 1984) many higher order “command-like” interneurons have been shown to be sufficient and not necessary to release a rhythmic motor pattern (e.g., Boyan et al. 1986; Brodfuehrer et al. 1995;Granzow and Kater 1977; Hedwig 1992;McCrohan 1984; Pearson et al. 1985;Weeks and Kristan 1978). This may be due to a redundant organization of these command systems. The described interneuron in the brain of G. bimaculatus must be regarded as a command neuron for calling-song stridulation. In resting crickets, it is sufficient to elicit stridulation, and stridulation stops whenever the neuron is prevented from firing during self-maintained calling song. The stainings so far do not indicate more than one of these neurons in each side of the brain. If there would be multiple copies of the neuron, the necessity criterion might not always be fulfilled as in redundantly organized command systems (see Hedwig 1994 for discussion of this point).

Control of calling-song chirp rate

Singing activity in crickets can be released by electrical stimulation of the neck connectives (Otto 1971) or of axon bundles of the cervical connectives (Bentley 1977). At stimulus repetition rates of 55 and 75 stimuli/s, chirp rates of 2.9 and 3.4, respectively, were released (Bentley 1977) and at 75 stimuli/s, Otto (1971) obtained 3.3 chirps/s. A linear extrapolation based on intracellular stimulation gives chirp rates of 3.0 chirps/s for 55 spikes/s and of 3.5 chirps/s at 75 spikes/s. This suggests that the same descending interneurons were investigated in the different studies.

There is a linear relationship between the spike frequency of the interneurons and the chirp repetition rate. Linear relationships between the interneuron activity and the expression of a motor pattern also have been found in other command systems, e.g., swimmeret movement of the crayfish (Davis and Kennedy 1972). Such a relationship indicates that the driving cell has direct control over the motor pattern. It also implies that any sufficient increase in the tonic command neuron activity can accelerate and even reset the motor rhythm as demonstrated in Fig. 5 C. A reset of the motor pattern also has been obtained with command-like interneurons for swimming in the leech (Weeks and Kristan 1978, Fig. 8), for feeding in Lymnaea (McCrohan 1984, Fig.2), and with command neurons for stridulation in a grasshopper (Hedwig 1994, Fig. 7). Generally, a neuron that can reset a motor pattern is thought to be an element of that pattern-generating network (e.g., Ramirez 1998). The present data and data on leech neuron 204 (Weeks 1981), however, demonstrate that a pattern generator can be reset and speeded up by a neuron that is not an element of the pattern-generating network but that has only a suitable driving input. This may force us to analyze critically whether or not a reset was obtained as an epiphenomenon by modulating the driving input to a pattern generator or by modulating an element of the pattern-generating network itself.

Activity of command neurons and production other song types

During extracellular brain and connective stimulation experiments, there are often transitions between the calling song and rivalry song or between the calling song and courtship song (Huber 1960; Otto 1971). Transitions between song patterns did not occur during any intracellular stimulation experiments. Activation of the interneuron only elicited the calling song even at experimentally induced spike rates of 160 spikes/s. It is therefore unlikely that different song types in G. bimaculatus are controlled by the activity of a single interneuron as proposed by Otto (1971) or Bentley (1977). Further, preliminary data indicate that a different type of descending interneuron is involved in the control of courtship song (B. Hedwig, unpublished data). The control of the different stridulatory motor patterns in the cricket rather seems to be organized with labeled lines as in the acridid grasshopper O. viridulus where three different types of descending brain neurons specifically control three different types of stridulatory motor patterns (Hedwig 1994; Hedwig and Heinrich 1997). A multifunctional command-like interneuron contributing to six different types of behaviors involving head movements has been described in Aplysia (Xin et al. 1996). So far, however, there is no experimental evidence that the calling song command neuron is multifunctional and is involved in the control of other behaviors as well. It may have to be considered that under the pressure of sexual selection that drives the evolution of acoustic communication systems, the emergence of a hardwired neuronal control system for the generation of acoustic signals may have been favored.

Rhythmic modulation

Command neurons are thought not to carry specific timing information for the production of a motor program (Pearson 1993; Stein 1978). Tonic activation of a command neuron generally is sufficient to elicit a corresponding behavior. However, command-like neurons in several systems exhibit a pronounced rhythmic modulation of their activity during the performance of the behavior, e.g., CV1 interneuron initiating feeding motor output inLymnea stagnalis (McCrohan 1984), DRI during swimming in Tritonia (Frost and Katz 1996); PCN neurons for feeding in Pleurobranchaea, (Gillette et al. 1978, 1982). In Pleurobranchaea and also inLymnea, the modulation is mediated by an inhibitory feedback from the motor pattern-generating network and contributes to rhythm generation. Huber (1960, 1964) postulated that the brain of crickets generates a patterned descending activity controlling the chirp rhythm. The calling-song command neuron exhibits a slight and gradual modulation of its spike frequency by ∼25–30% in the chirp cycle. The phase diagrams (Figs. 2 D and 4 C) indicate that the discharge rate increases simultaneously with the generation of the chirp. Therefore it rather seems that the modulation of the descending command neuron activity is not properly timed to initiate the single chirps and its functional role is not yet clear.

A central feedback from the thoracic pattern-generating network could account for the rhythmic modulation of the command activity. Another source could be a sensory feedback to the cricket calling song interneuron by the wind-sensitive system of the cerci. The ascending giant fiber system of crickets responds to low-frequency air currents that occur during stridulation (Kämper 1984;Kämper and Dambach 1981), and the chirp rhythm can be modified by cercal wind stimulation (Dambach et al. 1983). Cercal wind stimulation elicits responses in a number of local (Schildberger 1984) and descending brain interneurons (Hörner and Gras 1985). Because wind stimulation also elicits responses in the calling-song command neuron, it may indeed contribute to the rhythmic modulation of the command activity.

Command neuron activity and behavioral state

Extracellular stimulation experiments have demonstrated previously a high degree of facilitation and threshold reduction of the thoracic stridulatory network. After consecutive sequences of stimulation, the stridulatory sequences became longer until the stridulatory activity could go on without any additional stimuli (Huber 1960,1964; Otto 1971). Because these effects occur even when the connectives or single command neurons are stimulated, the underlying changes must occur in the thoracic pattern-generating system rather than in the brain. As a consequence, the impact of the calling-song command neuron clearly depended on the activity state of the thoracic system. Stimulation elicited premature calling-song chirps in resting crickets or triggered continuous calling-song sequences after considerable preceding stridulation had occurred. Data on the thoracic stridulatory pattern-generating mechanism in crickets are still very limited (Bentley 1969; Henning 1989,1990). We may understand the changes in the effectiveness of command neuron activity once the thoracic pattern-generating interneurons postsynaptic to the command neurons are identified and their physiological properties are characterized.

Stein (1978) distinguished between trigger and gating command neurons. Trigger command neurons evoke behavioral sequences that last longer than the command neuron activity, whereas the typical (gating) command neurons are activated continuously to maintain motor activity. In the leech, both types of command neurons, the trigger neuron TR1 and the gating neurons 204, 61, and 21 (seeBrodfuehrer et al. 1995 for review) are involved in the swim-activating system. Properties of both functional types can be assigned to the calling song command neuron depending on the activity state of the thoracic system. Although it is not yet clear whether or not both properties are functional in the control of stridulation, on balance it seems that the interneuron is acting more like a gating than like a trigger neuron. This observation, however, demonstrates the difficulties in categorizing command neurons. Of course stimulation experiments of single interneurons do not necessarily tell something definite about the function of that particular interneuron within a network. They moreover demonstrate how the postsynaptic network responds to the input from that interneuron. Thus the effectiveness of the interneuron is a function of the state of the postsynaptic network.

Decisions about behavior: stridulation and silencing reaction

Stridulating crickets respond to a variety of sensory stimuli in their surroundings with a silencing rection (Dambach and Rausche 1985). One sensory pathway mediating this reaction is the cercal wind-sensitive system. The reaction stops self-generated sensory feedback and prepares the animal for escape, which also can be evoked in resting crickets by cercal wind stimulation (Gras and Hörner 1992; Stabel et al. 1985). Stridulation and silencing are incompatible behaviors, so understanding their control helps clarify how decisions are made within the CNS about the performance of behavior. Depending on the network organization, quite different control principles can underlie the choices between different motor patterns. One possibility is that a motor network may be reconfigured for different behaviors to produce the appropriate motor response (Morton and Chiel 1994; Wu et al. 1994). Recordings during stridulation and flight in the cricket (Henning 1990) revealed separate interneuronal pools that are dedicated to each behavior and interact by inhibitory connections. In the locust, Ramirez and Pearson (1988)report interneurons that are specific for walking or flight. A reconfiguration of the thoracic motor system of the cricket for stridulation and silencing seems to be unlikely.

If two mutually exclusive behaviors are driven by two dedicated circuits (Dickinson 1995), there are three different modes of interaction between the circuits to explain the precedence of one behavior. First, circuit A might suppress the commands that are driving circuit B. This economic possibility emphasizes the role of command neurons as decision points for motor patterns. It is realized when the activity of the lateral giant–mediated escape behavior is suppressed during feeding in the crayfish (Krasne and Lee 1988). Such a mechanism ensures the precedence of feeding over withdrawal (Kovac and Davis 1980) and the precedence of escape swimming over feeding inPleurobranchaea (Jing and Gillette 1995). It further is the basis for motor pattern selection in the stomatogastric ganglion of Cancer borealis, where the modulatory proctolin neuron suppresses the gastric mill rhythm by inhibiting projecting interneurons, that activate the gastric mill (Blitz and Nusbaum 1997). Second, circuit A might inhibit both the commands and the pattern-generating elements of circuit B. This seems to be the safest but also most costly solution making sure pattern B will not be expressed when circuit A is activated. The system coming closest to this mechanisms may beClione limacina networks, which decide between wing retraction and swimming (Huang and Satterlie 1990) and feeding, withdrawal, and swimming (Norekian and Satterlie 1996). As a third option, circuit A might inhibit circuit B without suppressing the command to B. This solution is realized in the leech, where descending cephalic swim-initiating interneurons remain active while segmental interneurons of the swim pattern generator are inhibited during shortening (Kristan and Shaw 1997; Shaw and Kristan 1997). In the cricket, the stridulatory command neuron remains active during silencing. In both systems, however, a presynaptic inhibition of the command entering the pattern-generating network cannot be excluded. In the cricket, the organization of the control mechanism may be the result of the particular spatial arrangements of the neurons releasing stridulation and silencing. Ascending information about wind stimulation of the cerci is carried by the cercal giant and nongiant fiber system (Edwards and Palka 1974;Kohstall-Schnell and Grass 1994; Mendenhall and Murphey 1974) and first will reach the thoracic ganglia before entering the brain. A motor response to cercal wind stimulation can be mediated fastest when the ascending pathway directly activates the thoracic motor networks, avoiding any loop via the brain. The neuronal decision, allowing the silencing reaction to dominate over stridulation, seems to be located at the level of the thoracic ganglia. Because the stridulatory command is not switched off, it allows stridulation to recommence when the silencing reaction wanes. The synaptic mechanisms within the thoracic ganglia that underlie the precedence of silencing over singing now may be analyzed in future experiments.


I thank my Cambridge colleagues M. Burrows, T. Matheson, M. Wildman and J. Poulet for critical and constructive comments on the manuscript and M. Schink (Göttingen) for support with the histology.

One part of this study was conducted at the Department of Zoology, Göttingen, and was supported by the Heisenberg Programm of the Deutsche Forschungsgemeinschaft (He 2018/1-3). The other part was done at the Department of Zoology, Cambridge, and was supported by a Biotechnology and Biological Sciences Research Council Grant to M. Burrows and B. Hedwig.


  • Address for reprint requests: Dept. of Zoology, Downing Street, Cambridge CB2 3EJ, UK.

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract