Neural Mechanism of Hearing in Insects

Since the work of Wever & Bray (1933) many electrophysiological studies have been made on sound reception in insects and it has been established that their receptive organs can readily detect the direction of sound and can discriminate its intensity, but not its frequency (Pumphrey, 1940; Autrum, 1955). More recently Haskell (1956, 1957), Haskell & Belton (1956), and Roeder & Treat (1957) have studied this problem further, and the ultrasonic reception in noctuid moths has also been confirmed by electrophysiological methods.

From the view point of the comparative auditory physiology the present authors have tried to clarify the neural mechanism of hearing in insects by means of the same technique as they have used in recent studies of mammals. This report will be concerned with the results obtained from several kinds of insect which are very common in Japan.

As material, insects belonging to four families: Cicadidae (Tanna japonensis, Platypleura kaempferi and Graptopsaltria nigrofuscata) ; Acridiidae (Oxya japonica and Locusta migratoria danica);, Tettigoniidae (Mecopoda elongata, Gampsocleis buergeri and Hexacentrus japonicus japonicus) ; and Gryllidae (Xenogryllus marmoratus and Homoeogrylhis japonicus) were used. Experiments were performed during summer and autumn when those insects were obtainable; the experiments were made on Cicadidae early in summer, then on Tettigoniidae and Gryllidae, and late in autumn on Acridiidae.

These families have their tympanal organs in different parts of their bodies; Tettigoniidae and Gryllidae have them at the proximal end of the tibiae of the first forelegs, and Acridiidae and Cicadidae at the abdomen. In order to expose the tympanal nerve the body of the animal was fixed with pins on a cork board upside down and the exoskeleton was removed from the region of the thoracic ganglion with which the tympanal nerve connects, that is, from the prothoracic ganglion in Tettigoniidae and Gryllidae, the mesothoracic ganglion in Cicadidae and the metathoracic ganglion in Acridiidae. In studies on the cereal nerve which also shows response to sounds the animals were pinned in the normal position and the sternites were cut away from the posterior half of abdominal segments at the dorsal side; the intestine, rectum and trachea were removed to expose the last abdominal ganglion.

After the tympanal nerve or the cereal nerve had been exposed, the nerve was cut as close as possible to the ganglion and then the cut end was raised into the air on a silver wire electrode of 200 μ in diameter, by which the responses of the nerve to sound stimuli were recorded.

In order to obtain the response of a single neuron to the sound stimuli from the tympanal nerve bundle, 3 M-KCI capillary micro-electrodes with ohmic resistances between 30 and 50 Mil were used. The electron-microscope revealed the tip diameters to be less than o·2μ.. The electrodes were slowly introduced into the root of the tympanal nerve by a micromanipulator through the small hole which was made at the ventral surface of the thoracic ganglion. The cathode-follower preamplifier with a very low grid current (10^-12A.) was assembled with Z-729 by triode connexion. As main amplifiers an r.c. and, if necessary, high-gain d.c. amplifier were employed simultaneously. An indifferent electrode was placed on an exposed abdomen.

Recordings were made photographically with a cathode-ray oscilloscope, the beam of which was divided into three channels by means of electronic switches. In this way responses of neurons, time signals and sound waves could be displayed simultaneously. Most records were obtained on running film by the use of a long-recording camera, the stimulus sounds being presented in succession.

Four loud-speakers, one of which was for ultrasonic waves, were used as sound sources and delivered automatically in succession through an attenuator 44 tone bursts of different frequencies fixed between 30 and 100,000 c./s. The frequency characteristics of those four combined speakers was examined and found to be flat within ± 5 db. The duration of tone bursts was of the order of scores milliseconds, were varied in will, if necessary. The sound stimuli were delivered to the auditory organ of the animal in a free field. During the presentation of sounds their intensity was controlled by the attenuator.

Our reference level (zero db.) corresponded to a sound intensity of approximately 100 db. above the lowest average human threshold at 2000−3000 cycles.

Operated animals were isolated in a sound-proofed room and directed towards the loud-speakers at a distance of about 50 cm. The temperature in the room was kept by air conditioning at about 27°C.

The frequency range in which the tympanal or cereal nerve was activated by tone bursts of a certain intensity could be determined on the records which had been photographed on a running film. Such serial records obtained by tone bursts in successive different intensities gave the thresholds of the nerve responses for respective frequencies of sounds. Thus the response ranges of these nerves, in other words those of the end-organs innervated by them could be represented by plotting the thresholds, the frequency of sound being shown on the abscissa and the intensity in decibel units on the ordinate (Fig. 1).

Fig. 1 shows the average response ranges which are obtained from Cicadidae, Acridiidae, Tettigoniidae and Gryllidae.

(A) In Cicadidae the responses to sound stimuli were obtained from the tympanal nerve bundle which connects with the mesothoracic ganglion. Fig. 1A shows the response ranges of the tympanal nerves of three species of Cicadidae. Those response ranges are the average threshold curves of the tympanal nerves of male specimens in each species. The sexual difference in the response range was examined in Tanna japonensis and Platypleura kaempferi, and no particular difference was found except for the slightly higher threshold in the female. The response ranges in which the tympanal nerves were activated by the sounds of o db. for respective species were as follows: in Tanna japonensis, from 0·2 to 20 kc./s.; in Platypleura kaempferi, from 0.4 to 15 kc./s.; and in Graptopsaltria nigrofuscata, from 0·45 to beyond 20 kc./s. The most effective frequencies were 13,5 and 5 kc./s., respectively.

(B) The insects of the family Gryllidae are commonly provided with well-developed cerci. Fig. 1B shows the average response ranges of the tympanal and the cereal nerves of Homoeogryllus japonicus and Xenogryllus marmoratus.

The most effective frequencies for the tympanum and for the cereal hair sensilla were almost the same (H.j. and H.j. (c.), X.m. and X.m. (c.)). However, the response ranges to sound stimuli of the tympanal organ and the cereal hair sensilla in Homoeogryllus japonicus and also Xenogryllus marmoratus were from 0·2 to 8 kc./s. (H.j.), from 0·03 to 3 kc./s. (H.j. (c.)), from 0·08 to 13 kc./s. (X.rn.) and from 0.03 to 2 kc./s. (X.m. (c.)), respectively. In Homoeogryllus japonicus it was recognized that the cereal nerve can respond to a tone burst of up to 300 c./s. with synchronous discharge and can respond to gross air movements with bursts of spikes. Similar responses were also recognized in Haskell’s studies (1956) with certain species of Acridiidae.

(C) Fig. 1C shows the average response ranges obtained from the tympanal nerves of three species of Tettigoniidae. In these insects the tympanal organ is situated as described above at the tibia of the first foreleg and the tympanal nerve is involved in the nerve going out to the first foreleg from the prothoracic ganglion.

The response ranges of each species were as follows: in Gampsocleis buergeri, from 0·6 to 75 kc./s.; in Mecopoda elongata, from 0·14 to 85 kc./s.; and in Hexa-centrus japonicus japonicus, from o-8 to 100 kc./s.; the most effective frequency for each species being 10, 10 and 17 kc./s., respectively.

(D) The responses were obtained for two species of Acridiidae from the proximal part of the tympanal nerve which connected with a metathoracic ganglion, or from its distal part which connected with the chordotonal organ attached to the tympanal membrane. No remarkable differences were found between the two records apart from the tendency to flatness on the response curve in the region of the most effective frequency in the former case.

In Fig. 1D the curve L.m. represents the response range of the tympanal organ of Locusta migratoria danica. The response range is from 0·6 to 45 kc./s., the most effective frequency range being 4−9 kc./s. The response range and the most effective frequency of the tympanal organ of Oxya japonica are from 0·6 to 30 kc./s. and from 4 to 10 kc./s., respectively, while those two of the cereal hair sensilla were from 0.04 to 1·7 kc./s. and from 0.4 to 0·5 kc./s. (curve O.j. (c.)), respectively.

It is of great interest that the tympanal organs of Acridiidae and Tettigoniidae can respond to unusually high-frequency ultrasonic waves and that the most effective frequency of the latter is very high, even higher than 10 kc./s. The response ranges of the other two families, Cicadidae and Gryllidae, are found to be almost within that of man. As seen in Gryllidae and Acridiidae, the tympanal organ takes charge of the high-frequency sound, while the cereal hair sensilla take charge of the low-frequency sound. These insects thus can receive sounds over a wide frequency range with their two separate organs.

The responses of the tympanal nerve to natural stridulations of groups of insects were recorded from the tympanal nerves of Meimuna opalifera (♀), Gampsocleis buergeri (♂), and Mecopoda elongata (♂) simultaneously with the stridulatory sound. Fig. 2 is a case of Gampsocleis buergeri. In each figure the upper, middle and lower beams represent the nerve response, the sound of natural stridulation, and the time mark of 10 msec., respectively. When the tympanal organ is stimulated by the natural stridulation of a group the tympanal nerve sends the volleys of impulses which are synchronous with the pulsatory sound. It has been noted since the original work of Pumphrey (1940) that the tympanal organ receives effectively the pulsatory sound in cicadas (Pringle, 1953, 1954), in moths (Roeder & Treat, 1957) and in the grasshopper (Haskell, 1956, 1957).

It may now be asked what component frequencies were involved in the stridu-latory sound. The stridulatory sound recorded with a tape recorder which was specially designed for both sonic and ultrasonic waves at the Technical Laboratory of Japan Broadcasting Corporation, was analyzed by means of the Sona-Graph. The stridulatory sound is mere ‘noise ‘except for that of Homoeogryllus japonicus, so that the results of sound analysis appear as a continuous sound spectrum. The case of Gampsocleis buergeri is shown in Fig. 3. In the figure, C shows the wave-form of the stridulatory sound which was recorded with a precision microphone. B is the sonagram, in which the intensities of component sounds involved in the stridulatory sound are shown as the grade of darkness. In the sonagram the temporal change of both the frequency and the intensity of sound are well shown. A is the sectioner, which represents with the height of bar the intensities of component sounds involved at the moment indicated by an arrow. The stridulatory sound of G. buergeri contained a high-frequency sound of 40 kc./s., and the sounds of about 8 and 12 kc./s. were especially dominant in it. Those frequencies are called in this article the upper frequency limit and the dominant frequency range, respectively. It was also found by sonagrams and sectioners that the stridulatory sound of Mecopoda elongata had an upper frequency limit of 40 kc./s. and a dominant frequency range of 10 kc./s., and that of Homoeogryllus japonicus an upper frequency limit of 20 kc./s. and a dominant frequencies of 0·7 and 7 kc./s.

On the other hand, the most effective frequencies in the tympanal organs of these insects were 10, 10, and 0·7 kc./s., respectively (C, C and B of Fig. 1). It is a matter of surprise that the dominant frequency range involved in the sound produced by the animal itself shows good agreement with the most effective frequency range in the tympanal organ of that animal. Though the author did not actually analyze the stridulatory sound of Acridiidae, Haskell’s studies (1957) show that the dominant frequency of the wing-beat sound of desert locust is about 4 kc./s.

The most effective frequency of the tympanal organ of Acridiidae obtained by the present authors is about 4-9 kc./s.: Locusta migratoria danica 4−9 kc./s., and Oxya japonica 4−10 kc./s. (D of Fig. 1). Such a good agreement between sound production and its reception must show that stridulation has played an important role in communication among insects. The sound reception of Tettigoniidae and Gryllidae was studied only on the male because of the difficulty in getting females. It is inconceivable that in those insects there are sexual differences in sound reception between male and female. Indeed, the present authors confirmed in other species, for instance, Cicadidae and Acridiidae, that there were no sexual differences in sound reception.

Haskell’s study (1956) also reported the same results in Acridiidae. This too may be quite reasonable from the view point of mutual communication. However, in the case of Cicadidae there was disagreement between our results obtained from the tympanal nerve of these insects and the sound analyses made by other authors. Quite recently Hagiwara of our laboratory analysed the natural sounds of several kinds of Cicadidae by the same methods as we used and the results were also considerably different from the analyses of sound reception. The reason for these discrepancies may be the variable tension of the tympanal membrane resulting from the creasing action of the detensor muscle of the cicada (Pringle, 1953).

It is conceivable that the ultrasonic wave produced by stridulation may play a critical role in the orientation of the insect. The production as well as the reception of the ultrasonic waves was confirmed electrophysiologically on bats (Griffin & Galambos, 1941) and noctuid moths (Roeder & Treat, 1957). The present examination was performed on the tympanal organ of Locusta migratoria danica to find the difference of sensitivity of the tympanal organ to sonic as well as to ultrasonic waves coming from various directions. The tympanum makes an angle of about 45° backward with the body axis and is partly protected by a shield of exoskeleton. The material was mounted with a manipulator and a cathode-follower pre-amplifier on a rotatable round table and about 1·25 m. distant from the loud-speakers which were placed at the same height. The thresholds of responses of the tympanal nerve were measured for sonic and ultrasonic waves coming from different directions as shown in Fig. 4. Concentric circles represent the intensity of sound, 0, –10, –20, –30, –40 and –50 db., respectively. Eccentric curves represent the thresholds of nerve responses to the incident waves from different directions of 6, 10 and 30 kc./s., respectively. The solid lines show the threshold of the tympanal organ which faced the loud-speakers and the dotted lines that of the organ on the opposite side. The spaces between the solid and dotted lines show the differences of sensitivity between right and left tympanal organs. The higher the frequency, the more distinct was the difference of threshold. The threshold was highest for the incident wave coming from the direction of the head and lowest for the wave coming perpendicularly to the body axis. The tympanal organ has sharp directional sensitivity to sounds, especially to ultrasonic waves, in spite of its small size.

Pumphrey (1940) and later Autrum (1955) suggested that insects might be unable to discriminate the frequency of sounds. No definite evidence for it has, nevertheless, been found. It thus seemed likely that the recordings of responses of single auditory neurons to sonic and ultrasonic stimuli might provide a conclusive answer. Attempts were therefore made to record the response of a single neuron from the tympanal nerve of a tettigoniid by the use of a superfine microelectrode.

As already described the tympanal nerves of these animals connect with the prothoracic ganglion. The ganglion and also the tympanal nerve are covered with a hard sheath, through which the insertion of a micro-electrode was found to be almost impossible. An electrode was therefore inserted into the nerve bundle through a small hole which was made at the ventral surface of the prothoracic ganglion by cutting the sheath. The recordings of the responses of single units were generally successful in the very superficial layer but not in the deep layer. By histological studies it was found that in the ganglia of the nerve cord the nerve cells were in general grouped most densely at the ventral side. Thus it is highly probable that many of the responses obtained might have come from the nerve cells The recordings were mostly made extracellularly and very rarely intracellularly, and even when they were made from the ganglion itself the responses recorded from the superficial layers were clearly distinguished in their pattern from those of large nerve fibres involved in the connectives.

The response ranges of single neurons were determined by the same method as described above. Those ranges obtained from more than ten neurons were referable to two types. One type had a response range of 3−60 kc./s. and the most effective frequency at about 10 kc./s. (Fig. 5), while the other had a response range of 0·6−30 kc./s. and the most effective frequency at 6−7 kc./s. The neuron activated by sounds of higher frequencies was relatively easily obtained, while the neuron activated by those of lower frequencies was rather difficult to find. The latter was obtained only when the isolation of the unit was incomplete. This fact suggests that the neurons responding to the sounds of lower frequencies are of small size, namely, thin fibres and small cell bodies. The neurons activated only by sounds of higher frequencies were more sensitive to stimuli than the others activated by sounds of lower frequencies. The tympanal nerve of noctuid moths is said to consist of two fibres (Eggers, 1919), which can be distinguished only by the difference in threshold of their respective end-organs for stimulating sounds (Roeder & Treat, 1957). More details of the nature of these two fibres in moths are needed. In relation to these results, it is of great interest that a large number of fibres which compose the tympanal nerve of Gampsocleis buergeri can also be referred to only two types.

On the neuron which responded to the higher frequency range of sound the relation between the number of spikes per second and the intensity of stimulus in decibels was explored for different stimulus frequencies. The curves were sigmoid and almost parallel to one another for sounds with different frequencies. It is quite obvious that a neuron responds with the most frequent spikes to that sound to which the neuron is the most sensitive, among sounds with different frequencies but of the constant intensity, and it is also quite certain that there are various sounds with different frequencies and intensities which produce the same responses from the same neuron. Therefore a single auditory neuron cannot discriminate the frequency of sound, but can discriminate the change of its intensity.

Based upon the experimental results described above, the problems of frequency discrimination and recognition of the group in insects will be discussed. Frequency discrimination cannot of course be performed by a single auditory neuron as described above. If the insects have the ability to discriminate the frequency of sound, they should have many nerve fibres, with different response ranges. However, only two types of nerve fibres having different response ranges have been found in the tympanal nerve of Gampsocleis buergeri. Therefore it is concluded that the tympanal organ of insects has almost no ability to discriminate frequencies. However, it cannot be concluded from this fact alone that insects have no ability at all to discriminate the frequencies of sounds, because they have in addition many hair sensilla on various parts of the body. It can be shown as a notable example that there is the distinct difference in the response range between the tympanal organ and the cereal hair sensilla in Homoeogryllus japonicus and in Oxya japonica (Fig. 1). The tympanal organ which is exclusively adapted to sound stimuli responds to relatively high-frequency sounds, while the hair sensilla responds to relatively low-frequency sounds. It is indeed true that each of these receptors itself is almost unable to discriminate the frequency of sound. But generally speaking, it may not be so, because records obtained from the connective between the brain and the suboesophageal ganglion, which will be reported elsewhere, show that the impulses in response to sound stimuli are transmitted to the brain from the various parts of the body through different fibres which show different response ranges. Insects must discriminate stimulus frequency from the temporal and spatial pattern of impulses of many fibres which are sent to the upper brain, but of course such discrimination may not be sharp.

From the present experimental results alone the authors cannot discuss what role stridulation plays in mutual communication, but it may be said at least that stridulation does play some important role in natural communication from the fact that the most effective frequency range of the tympanal organ shows good agreement with the most dominant frequency range involved in the stridulatory sound produced by the insect itself. Such agreement also means that the stridulatory sound of the insect can most effectively stimulate the tympanal organ of the group. The stridulatory sound consists of pulsatory component sounds. If the tympanal organ of the insect has no ability at all to discriminate frequencies, the insect may still recognize its group solely by the rhythm of the pulsatory sound. Moreover, it has been found that the insect may discriminate the complex sound by impulses coming not only from the tympanal organ, but also from various sound receptive organs which are widely distributed on the whole body. Thus the insects may recognize their group by discriminating sounds by means of the temporal and also spatial patterns of impulses which are sent up to the upper brain through many different nerve fibres in the cord. The functional analysis of the central nerve cord element is needed.

It is well known that the ultrasonic wave is very useful for the detection of the direction of sound. In this sense the fact that some insects belonging to Acridiidae and Tettigoniidae can produce ultrasonic waves and hear them is a matter of interest. It is indeed surprising that many insects have the ability to produce and to respond to ultrasonic waves. Such a function in insects must also be very important from the ecological point of view. The present studies have confirmed that the stridulatory sounds produced by certain insects involve high-frequency and even ultrasonic waves and the most effective frequency of sound to the insect ear is found to be the dominant frequency involved in the stridulatory sound. For natural communication and also for localization of sound, the higher the dominant frequency the more useful the sound should be to the insect. Insects of various species seem to live in a world of different sounds from ours.

We are indebted to the Ministry of Education of Japan for the financial support of this work and to the Technical Laboratory of Japan Broadcasting Corporation for the analysis of sound by the sonagraph.