An Acridid Auditory Interneurone: I. Functional Connexions and Response to Single Sounds

An abdominal tympanum and associated array of more than 80 sensilla (Grey, 1960) is present in many of the Acridoidea, and many, though not all, of the species with such an ear also stridulate (see, for example, Dirsch, 1966). The complexity of the ear, and its occurrence in groups which do not stridulate, suggests that it is a general purpose hearing organ which can be expected to provide a variety of environmental information to the C.N.S. : in this it resembles the other orthopteran ears and contrasts with the lepidopteran ears of the noctuid type (Roeder & Treat, 1957) which seems to be a simple mechanism serving only very specialized functions.

Grey (1960), Popov (1965), Michelsen (1966) and Murray (1968) have recently investigated the anatomy and physiology of the acridid ear and have demonstrated heterogeneity among the sensilla. There have been a number of accounts of recordings from the tympanic nerves and the central nervous connectives, but relatively few individual auditory neurones have been individually characterized (see Table 4; Discussion). The tympanic nerve enters the metathoracic ganglion and at least some primary sensory axons run to the mesothoracic ganglion (Yanagisawa, Hashimoto & Katsuki, 1967). A majority of authors describe the first undoubted auditory interneurones as arising in the mesothoracic ganglion, though there are exceptions (see Table 4); whether or not these mesothoracic neurones are supplied by primary receptor axons, or by small interneurones synapsing with the receptor axons in the metathoracic ganglion, or by both these inputs, is not yet clear. There are a considerable number of auditory interneurones in the pro-mesothoracic connectives (Horridge, 1961), differing in size of recorded impulse and in response characteristics. At least some of them receive input from both ipsilateral and contralateral tympanic nerves.

In this paper we give a more detailed description of one of these ascending auditory interneurones. This particular unit was selected because it was easily activated by white-noise clicks made with the fingernails, and because its action potentials record as large and unambiguous signals from the intact connectives. We will refer to this unit as the alpha neurone (alpha, and not a, to avoid confusion with the noctuid ear terminology). This first paper deals with the connexions and response of this unit to the parameters of a single stimulus; we also present some data on another auditory interneurone (the beta neurone), which has properties which contrast interestingly with those of the alpha neurone. A second paper (Rowell & McKay, 1969) concerns the response of the alpha unit to repeated stimuli, including its habituation characteristics and its long-term variation in response level, and also concerns the central factors influencing its response.

The insect used was the locally common Gastrimargus africanus Saussure, a grasshopper closely related and very similar to the more familiar Locusta migratoria. Adult males and females were obtained from laboratory culture. The legs were removed and the nervous system was exposed ventrally. Silver wire hook electrodes were used, and, as indifferent electrode, a stainless steel pin in the head. The sound stimuli used were white-noise clicks and pure sine-wave tones generated by an oscillator via an envelope-shaping circuit and an ionic high-frequency loudspeaker. The upper and lower limits of this circuitry were about 2 and 45 kcyc./sec. The shaping circuit ensured a tapering start and finish to sine-wave signals, devoid of transients (Figs. 1, 5, 7). When sound was directed at alternatively the contralateral and ipsilateral ear relative to the recording electrode, the animal was rotated relative to the loudspeaker, and not vice versa. This maintains the same pattern of echoes from the environment, a pattern which is often important with high-frequency sound. Stimuli were monitored by a calibrated Bruel and Kjaer $12$ in. microphone placed in the position of the insect ear, connected either to an oscilloscope or to a Bruel and Kjaer sound-level meter Type 2203 with octave filter set Type 1613. The amplified nervous activity, microphone output and a sweep-triggering signal were recorded on a multi-channel AM/FM magnetic tape recorder and subsequently analysed by photography or long-persistence oscilloscope display, or with a stimulusgated electronic counter. Sound intensities are given in text in dB relative to a reference intensity of 0·0002 dynes/cm².

The results here presented are based on an analysis of experiments on 67 different animals.

No differences between the sexes of the grasshoppers were noticed in any of our experiments.

An appropriate auditory stimulus produces activity in the tympanic nerve which is recorded with an external electrode as a compound potential (Fig. 1A). Activity is also recorded c. 5 msec, later in the meta-mesothoracic connective (Fig. 1 B). This may be the same compound potential seen in the tympanic nerve modified by the passive electrical properties of the connective, or it may include small spikes of interneuronal origin, as implied in the accounts of Horridge (1961), Suga (1963) and Popov (1965); the matter was not investigated further. No element of this signal has a fixed time-relationship with the alpha neurone. The alpha unit is first recorded in the meso-prothoracic connective, and is delayed some 10 msec, relative to the activity recorded from the meta-mesothoracic connective. It transmits 1:1 through the prothoracic ganglion with a delay of less than 2 msec. (Fig. 1D) and similarly through the suboesophageal ganglion, entering the brain. There is one of these units on each side of the animal, and except in so far as they share inputs (see below) they appear to be independent of each other. In contrast to Horridge (1961), we found no activity in response to sound in the metathoracic-abdominal connective (Fig. 1C). Presumably the units he described in Locusta are not activated by the same stimuli as the alpha neurone.

The contribution of each tympanic nerve to one alpha unit can be ascertained by the experiment summarized in Fig. 2, in which recordings are made of the activity of both units to both ipsilateral and contralateral sound, first with both ears connected, then with only one, and lastly with none. The results are summarized in Table 1, and partly illustrated in Figs. 9, 10. Cutting the tympanic nerve of the ear ipsilateral to the sound reduces on the average to 14% the response in the alpha neurone on the same side as the lesion, and reduces less markedly but considerably (to 23 %) the response in the alpha neurone on the opposite side. Cutting the contralateral tympanic nerve approximately halves (43 %) the response of the alpha neurone ipsilateral to the sound, and reduces still more (to 20%) the contralateral response. These results show that each unit receives excitatory inputs from both ears, the ipsilateral ear contributing more, and that there is no inhibitory connexion between the ipsilateral and contralateral systems. In this it contrasts with the ‘T’ fibre of the tettigoniids Gampsocleis (Suga & Katsuki, 1961) and Homorocoryphus (McKay, 1968), and with the beta neurone of Gastrimargus (§C, below). Table 1 also shows that the combined excitatory input of the two tympanic nerves to the alpha neurone is greater than would be expected on the basis of the effect of each summed, suggesting that spatial summation acts in a nonlinear fashion at the synapse.

These findings on the connexions of the alpha neurone are summarized in Fig. 3.

The unit is very sensitive to white-noise clicks, giving a good response to a fingernail click more than 20 ft. away. It gives no response at all to low sine-wave frequencies, and at a sound intensity of + 90 dB it first responds to 40 msec, pulses of sine wave of increasing frequency at between 5 and 8 kcyc./sec. The response gets larger with increasing frequency, and is largest at the highest test frequency used, 40 kcyc./sec. (Table 2). The upper limit was not ascertained, nor the frequency response in terms of amplitude of stimulus required for a constant response, because of the limitations of the apparatus.

Absolute threshold intensity could not be measured reliably, as the auditory system was more sensitive than the monitoring apparatus. Threshold for 40 msec. pulses at 30 kcyc./sec. is between +50 and +6odB and the number of nerve impulses elicited by such a pulse increases with increasing intensity up to approximately +90 dB. Above this value the system saturates and there is no regular increase at least up to +110 dB (Fig. 4).

A louder stimulus produces on average not only more action potentials per response, but also a shorter latency of the first action potential. The negative correlation between number of action potentials and latency is low (r = − 0·6 to −0·2) though significant; the correlation is lower when the response level is low. The constants of regression equations describing response and latency vary greatly even between replicates of the same experiment on the same animal; the source of the variation appears to be the latency. This variation probably invalidates any attempt to correlate response level and latency over a range of experimental conditions or between different animals, and the most that can be said is that stimulus conditions or lesions to the c.N.s. which tend to increase responsiveness also shorten the average latency.

The response to a sustained tone is more or less tonic, slowly adapting, but there is great individual variation (Fig. 5). Some animals cease to respond after 500 msec, of stimulation, others continue for many seconds to the same stimulus. It is possible that this variation is due to an as yet uninvestigated central factor influencing responsiveness. Adaptation is more apparent in the response to short pulses of sound of increasing duration, in which mean impulse frequency declines exponentially with increasing pulse length (Fig. 6).

Acridid ears, like those of noctuid moths, are known to have considerable directional sensitivity, especially at high frequencies (Katsuki & Suga, 1960; Autrum, Schwartzkopf & Swoboda, 1961; Payne, Roeder & Wallman, 1966). The characteristics of the alpha neurone are adequate to preserve at least some of this directional information. Differences in amplitude at the two ears could be represented by differences in either or both frequency of impulses or latency in the response of the two alpha units.

The directional properties of the system were investigated by recording simultaneously from both connectives with a sound source placed laterally, normal to the animal’s long axis, and then rotating the animal through 180° so that the sound impinged on the opposite ear. This 180° transposition is the extreme directional shift, and any directional capability should show most clearly under these conditions.

The response to a standard series of 20 pulses of 30 kcyc./sec. sound, each 40 msec, long and of amplitude + 82 dB, pulse repetition rate i/sec., was recorded in both alpha neurones; 5 animals were tested with both sound directions, receiving a total of 213 sound pulses. A specimen response is shown in Fig. 7. The responses were analysed with respect to (a) the number of impulses in ipsilateral and contralateral units; (b) the latency of the first impulse in ipsilateral and contralateral units. As both these measures were shown to vary with intensity, they are expected to be related.

Table 3 shows that on criterion (a) alone the animals on average discriminated direction correctly in 57% of all trials, discriminated wrongly in 13% and achieved no discrimination in 30%. On criterion (b) alone the corresponding figures were 54, 21 and 25%. When both criteria are considered simultaneously, on the basis that compatible or single indications of direction are accepted, but contrary indications or no indications rejected as indiscriminable, the performance improves perceptibly;,68 % of trials are correctly discriminated, only 10% incorrectly discriminated and 22 % are not discriminated. Thus the animal gains in performance by the lack of complete correlation between latency and impulse number.

In some though not all preparations a second auditory interneurone is seen when recording from the thoracic connectives, which responds with a high-frequency burst to at least some of the same stimuli as the alpha unit. Its action potentials record considerably smaller than those of the alpha unit, and presumably they are usually lost when recording from the intact connective. Its properties can best be illustrated by the following experiment, originally designed to test directionality and input connexions of the alpha unit, and which was shown in Fig. 2. The abdominal and neck connectives were cut in these animals, and in each experiment the animal was subjected to a train of sine-wave pulses, 30 kcyc./sec., p.r.r. 1/sec.

When both tympanic nerves are intact, only the ipsilateral beta unit fires. After section of one tympanic nerve, sound presented on the same side as the lesion elicits no response from the ipsilateral beta unit, but a large response from the contralateral one, which was previously silent. When the sound is presented to the side with the one remaining intact nerve, it elicits a very large response in the ipsilateral unit, and nothing in the contralateral one (Figs. 8, 9).

These results indicate that there is mutual inhibition between the left-hand and right-hand beta-unit systems. The sound stimulus activates the receptors of both ipsilateral (Expts. 1, 2 and 4) and contralateral (Expt. 3) ears, but in the intact animal inhibition of the contralateral system by the ipsilateral is sufficient totally to suppress its response, while the inhibition of the ipsilateral system by the contralateral is only adequate to reduce the evoked response. This difference presumably reflects the directional sensitivity of the two ears (§B(iv)) probably caused by the acoustic shielding of the body. When the ipsilateral tympanic nerve is cut, the contralateral system is disinhibited and gives a response comparable to the ipsilateral unit of an intact animal. When the contralateral tympanic nerve is cut, the disinhibited ipsilateral system gives a very large response.

No attempt has been made to locate the site of this mutual inhibition; it could take place between the primary tympanic inputs, or between the two interneurones, or between the interneurone and the contralateral tympanic nerve; and in any of these it could be direct or mediated by an intermediate interneurone.

The functional significance of this arrangement is to increase greatly the directionality of the signal. Whereas the alpha unit correctly discriminated only 68%, at best, of all presentations, the beta unit discriminates under the same stimulus conditions with 100% accuracy—the contralateral unit never fires.

If the alpha and beta neurones were both fed by the same sensilla, as might be implied by the at least partial overlap in their frequency range, then one would expect the feedback applied to the beta system to result in a relatively lower sensitivity than that of the alpha unit. Experiments to test this proposition were performed on relatively few animals, but showed that, on the contrary, in at least some individuals the beta unit fires at sound intensities less than required to activate the alpha unit. This suggests that the two units receive inputs from populations of sensilla which differ in either their number or their sensitivity or their frequency response.

What conclusions about the function of the two auditory interneurones can be drawn from the measurements of their performance?

The alpha unit is clearly specialized for the reception of high-frequency sound. This has obvious uses. The genus stridulates by rubbing the hind femora on the elytra, but lacks the well-developed tooth or peg structure, as for example in the truxaline grasshoppers, which results in a peak of emission in the audible range. The sound produced is a high-frequency hiss, and oscilloscope examination confirms that most of the emission is ultrasonic. The animal lives in grass, and grass moved by the wind or other animals emits much high-frequency sound. Insectivorous predators such as shrews and bats emit ultrasound at high intensity; large-scale nocturnal flights are characteristic of many species of solitary grasshopper, and we have caught Gastri-margus at light at night. Haskell (1957) describes a take-off response by Schistocerca gregaria to the noise generated by a flying locust, and S. vaga responds similarly to the noise made by a vacuum cleaner. Haskell’s recordings did not extend much beyond 10 kcyc./sec., but it seems probable that high frequencies are important in both cases.

Popov (1965) performed ablations on the three morphological groups of sensilla in the ear of Locusta, and found that only those with a predominantly low-frequency (less than 10 kcyc./sec.) response gave rise to patterned summated potential, mirroring the characteristic amplitude modulation of the species song, in the recording from the tympanic nerve. The Group I sensilla, responding preferentially to higher frequencies (greater than 10 kcyc./sec.), did not; the author accepts that amplitude modulation is the most important parameter of song, and therefore suggests that the high-frequency sensilla are not important in song reception. The alpha neurone, which is driven mainly by high-frequency receptors, might by this argument be adapted for some function other than song reception. However, this argument contains too many assumptions about both song reception and the as yet unrecorded ear sensilla to be particularly attractive.

The input arrangements to the alpha neurone seem to favour high sensitivity at the price of directionality. The unit is certainly very sensitive to at least some elements of a white-noise click, and we suspect that its most sensitive frequency is higher than our highest test frequency of 40 kcyc./sec. Amplitude changes are reasonably well encoded over some 30 dB of the response range. The unit’s rather tonic response to medium length pulses would make it very suitable for signalling the length of such sounds. In a second paper (Rowell & McKay, 1969), we show that there is marked habituation to repetitive stimuli, but that at practical repetition frequencies the response never fails completely.

We have much less information about the beta unit, but it is of great interest to see that it is clearly specialized for accurate directional discrimination, and that the feed-back mechanism by which this is achieved is functionally identical to that used by the tettigoniid grasshoppers Gampsocleis and Homorocoryphus (Suga & Katsuki, 1961 ; McKay, 1968). In other ways the tettigoniid interneurone differs considerably from the beta unit; for example, it is highly phasic in response, shows little habituation in the intact animal, gives a good low-frequency response and is very large in size (McKay, 1968); but directionality is achieved in the same way.

Neither the alpha nor beta units seem to have been described before individually, though there is no doubt that they must be included in the sample described by Horridge (1961). The small size of the beta unit action potentials in most recordings and the absence of response of the alpha unit to most of the human audible range suffice to explain this. Table 4 summarizes those auditory interneurones which have been more or less characterized in the acridid system. Even allowing for spurious diversity created by different experimental techniques or interspecific differences, it is clear that the auditory system is most complex. So far no recordings have been made from the split connective, a technique which is almost certain to reveal more auditory units (see McKay, (1968) with the tettigoniid Homorocoryphus). A corresponding complexity of receptors is also being demonstrated. Recordings of the summated potential of the tympanic nerve indicate a response from 0·4−45 kcyc./sec. and possibly beyond; although several different classes of receptor have already been demonstrated (see Introduction), they do not yet cover this frequency range, and it may be presumed that further types await description.

The auditory system is clearly capable of most sophisticated handling of acoustic information, and is quite unlike the picture commonly presented until recently, sensitive only to amplitude modulation and possibly consisting of only a single type of primary receptor and one ascending interneurone.

This work was supported in part by the United States Government under Contract 61052 67 C0016 and by Makerere College Research Fund, Grant no. 242. We are most grateful to Dr M. C. Pike for advice on statistical treatments, and to Dr R. G. Whitehead for computer facilities. One of us (J.M.M.) was supported by an S.R.C. Studentship.