J Neurophysiol 88: 2322Ð2328, 2002; 10.1152/jn.00119.2002.

Encoding of Localization Cues by an Identified Auditory Interneuron: Effects of Temporal Pattern

ANNIE-HE« LE` NE SAMSON AND GERALD S. POLLACK Department of Biology, McGill University, Montreal, Quebec H3A 1B1, Canada

Received 19 February 2002; accepted in final form 1 August 2002

Samson, Annie-He´le`ne and Gerald S. Pollack. Encoding of sound Two features of the neural response to acoustic stimuli vary localization cues by an identified auditory interneuron: effects of with stimulus intensity: spike count and/or rate increases with stimulus temporal pattern. J Neurophysiol 88: 2322Ð2328, 2002; increasing intensity and response latency decreases (Imaizumi 10.1152/jn.00119.2002. An important cue for is binaural comparison of stimulus intensity. Two features of neuronal and Pollack 2001; Kiang et al. 1965). The binaural difference responses, response strength, i.e., spike count and/or rate, and re- in both of these measures varies systematically with sound sponse latency, vary with stimulus intensity, and binaural comparison location, and either or both might code for sound direction of either or both might underlie localization. Previous studies at the (Boyan 1979; Eggermont 1998; Mo¬rchen 1980). receptor- level showed that these response features are affected Crickets localize sound both to find mates and to evade by the stimulus temporal pattern. When are repeated rapidly, predators. Stridulating male crickets produce songs with rela- as occurs in many natural sounds, response strength decreases and tively low carrier frequency (dominant frequency for Te- latency increases, resulting in altered coding of localization cues. In leogryllus oceanicus: 4.5 kHz), consisting of repeated trains of this study we analyze binaural cues for sound localization at the level of an identified pair of interneurons (the left and right AN2) in the sound pulses at rates varying (for T. oceancius) between 8 and cricket , with emphasis on the effects of stimulus 32 pulses/s (Balakrishnan and Pollack 1996). Echolocating temporal pattern on binaural response differences. AN2 spike count bats produce ultrasound pulses (20 to Ͼ100 kHz), repeated at decreases with rapidly repeated stimulation and latency increases. rates ranging from a few pulses per second to Ͼ100 pulses/s Both effects depend on stimulus intensity. Because of the difference (Griffin et al. 1960). Behavioral experiments have demon- in intensity at the two , binaural differences in spike count and strated that crickets perform positive phonotaxis (movement latency change as stimulation continues. The binaural difference in toward the sound source) in response to cricket-like stimuli and spike count decreases, whereas the difference in latency increases. negative phonotaxis (movement away from the sound source) The proportional changes in response strength and in latency are when presented with bat-like sounds (Moiseff et al. 1978; greater at the interneuron level than at the receptor level, suggesting Nolen and Hoy 1986; Pollack et al. 1984). The importance of that factors in addition to decrement of receptor responses are in- volved. Intracellular recordings reveal that a slowly building, long- the frequency ranges of cricket song and bat sounds is reflected lasting hyperpolarization is established in AN2. At the same time, the by the organization of the cricket’s ; nearly three-quarters of level of depolarization reached during the excitatory postsynaptic the approximately 70 receptor in each ear are tuned to potential (EPSP) resulting from each sound stimulus decreases. Nei- cricket-like frequencies, and more than one-half of the remain- ther these effects on membrane potential nor the changes in spiking der are most sensitive to ultrasonic frequencies (Imaizumi and response are accounted for by contralateral inhibition. Based on Pollack 1999). comparison of our results with earlier behavioral experiments, it is When sensory neurons are stimulated with a long series of unlikely that crickets use the binaural difference in latency of AN2 rapidly repeated stimuli, their response strength (spike count responses as the main cue for determining sound direction, leaving the and/or rate) decreases for successive stimuli, and response difference in response strength, i.e., spike count and/or rate, as the latency increases (Coro et al. 1998; Givois and Pollack 2000; most likely candidate. Pasztor 1983). In cricket auditory receptors, as in mammals (Eggermont and Spoor 1973), the decline in response strength is greater the higher the sound level (Givois and Pollack 2000). INTRODUCTION This implies that neurons ipsilateral to the sound source, where Sound localization in the horizontal plane is based on com- stimulus intensity is higher, will experience a larger response paring sounds at the two ears. Sound arrives earlier at the ear decrement. This may even lead to a reversal in the sign of the closer to the source, and in general stimulus intensity is higher binaural difference in response strength (contralateral re- at that ear. Some animals can detect the small (tens of micro- sponseϾ ipsilateral; Givois and Pollack 2000). The change in seconds) binaural difference in arrival time (for review, see latency of receptor-neuron responses induced by repeated stim- Carr 1993); others, including insects, rely mainly on binaural ulation does not depend on sound level; thus directional infor- intensity difference as the cue for sound location (for review, mation encoded by binaural latency difference is not affected. see Pollack 1998). Phonotaxis responses are not driven directly by receptor

Address for reprint requests: G. S. Pollack, Dept. of Biology, McGill The costs of publication of this article were defrayed in part by the payment University, 1205 Ave. Doctor Penfield, Montreal, Quebec H3A 1B1, Canada of page charges. The article must therefore be hereby marked ‘‘advertisement’’ (E-mail: [email protected]). in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

2322 0022-3077/02 $5.00 Copyright © 2002 The American Physiological Society www.jn.org TEMPORAL PATTERN AND SOUND LOCALIZATION CUES 2323 neurons, but rather by interneurons that receive input from receptors. Negative phonotaxis in response to ultrasound stim- uli is initiated by the identified, bilaterally paired interneuron AN2 (Nolen and Hoy 1984). The effects of the changes in receptor responses described above might be compensated or exaggerated during signal transmission from receptors to in- terneurons, but as yet this issue has not been studied. In this study, we investigate the effects of rapidly repeated stimulation on encoding of sound localization cues by AN2.

METHODS Animals Teleogryllus oceanicus were raised in the laboratory on a diet of Purina Cat Chow and water. Unmated females, ages 12Ð20 days after the final molt, were used in experiments. Crickets were anesthetized by chilling on ice and were mounted on a wax support, ventral side up, after removing their wings and mid- and hind legs. In most experi- ments, the femur of the fore legs was fixed with a bees’ wax- colophonium mixture, parallel to the body axis, and the tibia and tarsus were held flexed against the femur, simulating their position during flight. When recordings were made from receptor neurons, the legs were rotated around the coxa-trochanter-femur joints so as to project perpendicularly from the body axis, exposing the anterior surface of the femur for electrode implantation (see Electrophysiol- ogy).

FIG. 1. Effect of repeated stimulation on spike count (A) and first-spike Electrophysiology latency (B) of AN2 responses. Sound pulses were presented at 16 and 0.3 pps (for clarity, responses to 0.3 pps are shown only for 90 dB). Each point is the For extracellular recordings of AN2, the cervical connectives were response to 1 sound pulse; data are from a single, representative experiment. In exposed and placed on a pair of silver-wire hook electrodes. AN2 A, for time points Ͼ1 s, only every 5th data point is plotted for 16 pps. Solid produces the largest sound-evoked spikes in the cervical connectives, lines (for 16 pps,70 dB in A;70dBinB) are curves fit to an equation of the ϭ ϩ Ϫt/ss ϩ Ϫt/ll and these are easily recognized by visual inspection (Moiseff and Hoy form: r(t) rss ae be , where r(t) is response at time t, rss is the 1983, see inset of Fig. 1A). To record compound action potentials of steady-state response, ss and ll are short and long time constants, and a and b auditory receptor neurons, we inserted a Teflon-coated silver wire are constants. Inset: extracellular recording, from a different experiment, (114 ␮m OD) into the anterio-dorsal surface of the femur, close to the showing AN2 spikes (top; first 8 responses; 90 dB, 16 pps) and stimulus nerve branch carrying the of the receptor neurons from their monitor (bottom). origin in the prothoracic tibia to the prothoracic ganglion. The indif- ferent electrode was placed proximally and ventrally in the femur (see the series. Stimuli were produced, using LabWindows programs (Na- Pollack and Faulkes 1998 for further details). For intracellular record- tional Instruments), by a D/A circuit (ATMIO16F5, National Instru- ings of AN2, the prothoracic ganglion was exposed and supported on ments, Austin; D/A update rate 200 kHz) and were attenuated (PA4, a metal platform. The ganglion was submerged in physiological saline Tucker-Davis, Gainesville, FL), amplified (D150A, Amcron, Elkhart, (Strausfeld et al. 1983) and recordings were made using microelec- IN), and broadcast through loudspeakers (RadioShack, 40Ð1310B) trodes (Ͼ30 M⍀) filled with2MKacetate. Electrophysiological situated on the cricket’s left and right in the horizontal plane, perpen- recordings were digitized (Digidata 1320A, 10 kHz sampling rate, dicular to the longitudinal axis, at a distance of 50 cm. The loud- Instruments) using the program AxoScope 8.0 (Axon Instru- speakers and cricket were housed in a chamber lined with echo- suppressing mineral-wool wedges. Sound pressure level in dB re 2 ϫ ments). Responses were viewed and analyzed using custom-written Ϫ5 Ϫ2 programs (GP) for Scilab (www-rocq.inria.fr/scilab/), and statistical 10 Nm (SPL) was measured at the cricket’s position with a analysis were performed using Statistica for Windows (StatSoft, Bru¬jel and Kjaer (Naerum, Denmark) 4135 microphone and 2610 Tulsa, OK). measuring amplifier.

Stimuli Data analysis Stimuli consisted of trains of trapezoid-shaped sound pulses of The response to each sound pulse was analyzed over a time window 30-ms duration, including 5-ms rise and fall times. Carrier frequency that included the entire response as determined by visual inspection. was 30 kHz, which is within the range of AN2’s greatest sensitivity Time windows for AN2 responses ranged from 38 ms (26 pps) to 50 (Moiseff and Hoy 1983). Pulse trains were 30Ð45 s long and were ms (0.3 pps) in duration and were constant for each pulse rate across preceded by 60 s of silence. Pulses were presented at rates of 0.3, 8, experiments; for receptor responses, windows ranged from 24Ð26 ms, 16, and 26 pulses/s (pps). The lowest pulse rate was presented at the and were constant for all measurements in an individual cricket. beginning, middle, and end of each experiment, and responses were Measuring AN2 spike count and latency was straightforward, because statistically indistinguishable in all cases. This ensured both that each individual spike could clearly be detected (e.g., Fig. 1A, inset). responsiveness remained stable, and that the 60-s pause following For receptor neurons, response strength was quantified by integrating each stimulus was sufficient for return to baseline responsiveness. The the recording over the response time window, following high-pass order of presentation of stimuli varied from experiment to experiment, filtering (100 Hz) and full-wave rectification. Receptor response la- with the two highest pulse rates most often presented in the middle of tency was taken as the time, relative to stimulus onset, of the first

J Neurophysiol ¥ VOL 88 ¥ NOVEMBER 2002 ¥ www.jn.org 2324 A.-H. SAMSON AND G. S. POLLACK negative peak of the compound action potential revealed by signal averaging (see inset, Fig. 4B); this peak represents the nearly syn- chronous firing of many receptors neurons. See Givois and Pollack (2000) and Pollack and Faulkes (1998) for further details.

RESULTS Effects of repeated stimulation on spike count and first-spike latency Responses of AN2 to stimuli presented at low repetition rate are stable, both in spike count and in latency (Fig. 1, 0.3 pps). With more rapid stimulation, responses change systematically through time. An initial very rapid decrement in spike count is followed by a slower asymptotic decline that last for several seconds (Fig. 1A). Short and long time constants, derived from double-exponential curve fits (see Fig. 1 legend), were 303 Ϯ 266 ms and 11.71 Ϯ 8.18 s (mean Ϯ SD; pooled data for 70 and 90 dB, and 8, 16, and 26 pps, averaged from 7 crickets). Neither time constant varied with either sound level or pulse rate (ANOVA, P values range from 0.27 to 0.36). First-spike latencies lengthen with repeated stimulation (Fig. 1B) and also show rapid and slower phases of increase (time constants: 1.39 Ϯ 0.64 ms; 47.87 Ϯ 15.72 s). Note that the variability of latency decreases significantly with intensity (mean variance for 7 crickets, of responses between 8 and 10 s after stimulus onset is 4.37 ms, at 90 dB; 17.32 ms at 70 dB; FIG. 2. Effect of intensity and pulse rate on changes in AN2 responses. ϭ ϭ Mean spike count and latency were calculated for responses to stimuli deliv- F6 6.9, P 0.01; we focus on the interval 8Ð10 s after ered between 8 and 10 s after stimulus onset, by which time the changes in stimulus onset to facilitate comparison of our results with AN2’s response are well established. Points are mean Ϯ SE for 7 crickets. earlier behavioral experiments; Pollack and El-Feghaly 1993; Two-way ANOVA, intensity effect on spike-count: F(4,120) ϭ 61.15, P Ͻ ϭ Ͻ see DISCUSSION). In addition, variability of latency increases as 0.0001; pulse-rate effect on spike-count: F(3,120) 276.78, P 0.0001; interaction: F(12,120) ϭ 8.75, P Ͻ 0.0001; intensity effect on latency: the train of stimuli continues with larger increases at lower F(4,120) ϭ 72.76, P Ͻ 0.0001; pulse-rate effect on latency: F(3,120) ϭ intensities and higher pulse rates (ANOVA on mean variance 116.98, P Ͻ 0.0001; interaction: F(12,120) ϭ 5.73, P Ͻ 0.0001. of responses between 8Ð10 s after stimulus onset: pulse-rate ϭ Ͻ ϭ Ͻ effect, F6 23.9, P 0.001, intensity effect, F6 13.2, P tenbach and Hoy 1999; personal observations). This, together 0.001). Variability of spike count is similar for the range of with the intensity dependence of the response changes de- pulse rates and intensities examined (ANOVA; pulse-rate ef- scribed above, implies that the changes in response resulting fect, P ϭ 0.4; intensity effect, P ϭ 0.5). from repeated stimulation may alter binaural differences. Fig- ure 3 shows binaural differences measured from simultaneous Effects of intensity and pulse rate on changes in spike count recordings of the left and right AN2. Rapidly repeated stimu- and latency lation results in a decrease in the binaural difference in spike Figure 2A summarizes the decremented response as the count, but an increase in the binaural latency difference. The mean of responses to sound pulses delivered between 8 and change in spike-count difference is monotonic with pulse rate, 10 s after stimulus onset. As expected, the number of AN2 whereas the latency difference reaches a maximum at 16 pps. spikes/sound pulse increases with intensity. However, because of the pulse-rateÐdependent response decrement, the increase Response decrement is greater in interneuron AN2 is less pronounced the higher the pulse rate. than in receptor neurons The increase in latency with repeated stimulation (Fig. 1B) also depends on both pulse rate and intensity. The higher the It seems probable that, as for low-frequency receptors pulse rate, and the lower the intensity, the greater the increase (Givois and Pollack 2000), responses of ultrasound receptors in latency (Fig. 2B). decline with repeated stimulation (Coro et al. 1998). If so, then Therefore the effects of repeated stimulation on spike count the effects we describe for AN2 might be accounted for simply and on first-spike latency vary with both intensity and pulse by response decrement of receptor neurons. To determine rate. The effect of pulse rate is similar for both response whether this is the case, we compared the decrement of com- parameters: higher pulse rates produce larger response pound action potentials recorded from the whole auditory changes. The effect of intensity, however, differs: at higher nerve with the decrement of AN2 responses. Ultrasound re- intensities, the decrease in spike count is more pronounced, but ceptors comprise only a small proportion of the entire receptor the change in latency is less pronounced. population (Imaizumi and Pollack 1999), and as a result, clear whole-nerve responses, which reflect synchronous activity of Effects of repeated stimulation on localization cues many receptors (Pollack and Faulkes 1998), can be recorded The stimulus intensity of a 30-kHz sound may be Յ15Ð20 reliably only for high stimulus levels late in the pulse train, by dB higher at the ipsilateral than at the contralateral ear (Wyt- which time responses have decremented. For this reason, we

J Neurophysiol ¥ VOL 88 ¥ NOVEMBER 2002 ¥ www.jn.org TEMPORAL PATTERN AND SOUND LOCALIZATION CUES 2325 In parallel with the build-up of hyperpolarization in AN2, the level of depolarization reached during successive excitatory postsynaptic potentials (EPSPs) decreases [Fig. 5, B (thick arrow) and C). Like the decrement in spike count, the mem- brane potential at the peak of the EPSP, during the period 8Ð10 s after stimulus onset (thick arrow in Fig. 5B), varies with pulse rate and intensity (Fig. 5E). The decrease in EPSP amplitude likely reflects a combination of factors, including decrement of receptor responses (Fig. 4), shunting of excitatory current through channels responsible for the slow hyperpolarization, and perhaps depression of receptor-to-AN2 synaptic transmis- sion as well.

FIG. 3. Binaural differences in spike count and latency vary with pulse rate. DISCUSSION Points are means Ϯ SE (n ϭ 7 crickets) of differences between ipsilateral and contralateral spike counts and latencies, for sound pulses delivered between 8 Our results show that ultrasound localization cues encoded and 10 s after stimulus onset. Two-way ANOVA, pulse-rate effect on spike by AN2 are affected by rapidly repeated stimulation. The count difference: F(3,43) ϭ 35.62, P Ͻ 0.0001; intensity effect on spike-count spiking response decreases with time (Fig. 1), and this change difference: F(1,43) ϭ 0.4, P ϭ 0.53; pulse-rate effect on latency difference: F(3,43) ϭ 5.81, P Ͻ 0.005; intensity effect on latency difference: F(1,43) ϭ 23.12, P Ͻ 0.001. compare responses of AN2 and receptors only for high stim- ulus levels (90Ð100 dB). To compare AN2 response strength (spike-count) to the auditory response (integrated whole-nerve recording, see METHODS), we normalized the decremented re- sponses with respect to the mean response to a low-pulse-rate stimulus (0.3 pps) at the same intensity; as shown in Fig. 1A, stimulation at this repetition rate does not induce response decrement. As shown in Fig. 4, response decrement is more severe for AN2 than for the auditory receptors. In addition, the effects of pulse rate on response decrement and change in latency are stronger for AN2 than for receptors [2-way ANOVA: interaction effect between pulse rate and level of measurement (receptor or AN2) on relative response magni- tude: F(3,62) ϭ 86.27, P Ͻ 0.0001; on latency: F(3,62) ϭ 14.59, P Ͻ 0.0001].

Changes in membrane potential The observation that response decrement of AN2 is more pronounced than that of receptors suggests that factors in addition to decremented receptor responses may also be in- volved. We examined this possibility by recording from AN2 intracellularly. During the course of stimulation, a slowly building, long- lasting hyperpolarization builds up in AN2. This is most clearly evident as a strong hyperpolarization following the termination of stimulation, shown in Fig. 5A. An additional indication is the decrease in membrane potential reached be- tween sound pulses, shown in Fig. 5, B (thin arrow) and C. A known source of inhibition of AN2 is the interneuron ON1 which is excited by input from the contralateral ear (Faulkes and Pollack 2000; Selverston et al. 1985). Following removal of the contralateral ear, both the decrease in membrane poten- tial between successive responses (Fig. 5C) and the inhibition FIG. 4. Response decrement of auditory receptors and AN2. Relative re- sponse is computed as the mean response [spikes-count for AN2 and integrated following the termination of stimulation (data not shown) were responses for leg nerve (see METHODS)] induced by sound pulses delivered still present. Cutting the contralateral leg nerve, which includes 8Ð10 s after stimulus onset, normalized with respect to the mean of 15 the axons of auditory receptors, affected neither the decline of responses to 0.3 pps at the same intensity. Points are mean Ϯ SE of 5 crickets. AN2’s spike count, nor the increase in latency (Fig. 5D). Thus Inset in A: responses to 3 successive pulses (90 dB, 16 pps) at the onset of the pulse train and 10 s later for leg nerve (a and b, respectively) and for AN2 (c neither the hyperpolarization of AN2, nor the decrement in and d). Inset in B: averaged compound action potentials (8Ð10 s) recorded in AN2’s spiking response with repeated stimulation, are due to leg nerve. Arrows indicate onset of sound pulse and first negative peak of the contralateral inhibition. response; time difference between these is latency.

J Neurophysiol ¥ VOL 88 ¥ NOVEMBER 2002 ¥ www.jn.org 2326 A.-H. SAMSON AND G. S. POLLACK

FIG. 5. Intracellular recordings of AN2. A: responses to a 90 dB, 16 pps sound stimulus, (indicated below the trace), as well as the poststimulus period. B: first 3 responses (thick line), superimposed with 3 responses beginning 10 s after stimulus onset (thin line). Thick arrow indicates peak depolarization during the excitatory postsynaptic potential (EPSP); thin arrow indicates minimum membrane potential between EPSPs. C:as in B, after removal of the contralateral ear. Bottom: stimulus monitor (applies to B as well). D: effect of removing contralat- eral inhibition on spike count and latency (means of stimuli delivered 8Ð10 s after stimulus onset). Points are mean Ϯ SE (n ϭ 6 crickets), of responses to 90-dB stimuli, presented at 8 and 26 pps. t-test for paired samples on intact and cut habituated responses; 8 pps: t ϭ 1.31, P ϭ 0.26; 26 pps: t ϭ 1.51, P ϭ 0.23; on intact and cut habituated latency; 8 pps: t ϭ 1.62, P ϭ 0.18; 26 pps: t ϭ 1.63, P ϭ 0.19. E: maximum depolarization (thick arrow in B) during EPSPs between 8 and 10 s after onset of stimulus. Points represent means Ϯ SE for 3 crickets, each stimulated at both 70 and 90 dB SPL. Two-way ANOVA, intensity effect on depolarization: F(1,16) ϭ 13.5, P ϭ 0.002; pulse rate effect on depolarization: F(3,16) ϭ 25.9, P Ͻ 0.001; interaction: F(3,16) ϭ 3.25, P ϭ 0.04.

is more pronounced at higher intensity and pulse rate (Fig. 2). the EPSP increases for successive stimuli, particularly for First-spike latency increases with time and, as for the change in lower stimulus intensities (data not shown). spike count, the increase in latency is greatest for high pulse Our intracellular recordings show that in addition to decreas- rates. However, the effect of intensity is opposite to that on ing excitatory input, a hyperpolarizing current may also play a spike count; the change in latency decreases with increasing role in AN2’s response decrement. Two sources of inhibitory intensity (Fig. 2). synaptic input to AN2 have been described previously: 1) The intensity dependence of these response features is of contralateral inhibition, originating in the contralateral identi- particular importance when considering binaural cues used in fied neuron ON1 (Faulkes and Pollack 2000; Selverston et al. sound localization. The decrement in AN2’s spiking response 1985), and 2) ipsilaterally derived inhibition of unknown origin is greater for the neuron ipsilateral to the sound source, and as (Moiseff and Hoy 1983). The hyperpolarization persists after a result, the binaural difference in spike count decreases. In removing the contralateral ear (Fig. 5C); thus it is either contrast, because the change in latency with repeated stimula- ipsilateral in origin and/or intrinsic to AN2. The ipsilateral tion is smaller at higher intensities, the binaural difference in inhibition described by Moiseff and Hoy (1983) is recruited response latency increases as stimulation continues. most effectively by low-carrier-frequency stimuli. However, The larger latency shift at lower intensities may be due to the some low-frequency-tuned receptor neurons are stimulated by shape of the relationship between latency and stimulus inten- intense ultrasound (Imaizumi and Pollack 1999); thus we can- sity. For AN2, as for most interneurons, latency decreases not rule out this source. Slowly building, long-lasting hyper- asymptotically as stimulus intensity increases (Moiseff and polarization similar in time course to that which we observed Hoy 1983), presumably because the rate of rise of the EPSP in AN2 has been described in other auditory neurons of insects increases to a maximum (perhaps set by membrane biophysics) (Pollack 1988; Ro¬mer and Krusch 2000). In the ON1 neuron of as excitatory input increases. Consequently, intensity changes Acheta domesticus, this hyperpolarization is triggered by ac- have a larger effect on latency at the low end of the intensity tivity-associated Ca2ϩ accumulation (Sobel and Tank 1994). range. Excitatory input to AN2 decreases during repeated The presumed function of negative phonotaxis in crickets is stimulation, in part because of decrementing responses of bat evasion (Hoy 1992). Bat echolocation sounds span a broad receptor neurons (Fig. 4), and possibly also because of synaptic range of temporal patterns, varying both with the species of bat depression. The decrease in excitation should be functionally and with the progression of the bat-insect interaction (Griffinet equivalent to a decrease in stimulus intensity. This decrease, al. 1965). Before a bat has encountered a potential prey, during occurring at an already low stimulus intensity, can be expected the “search” phase of echolocation, echolocation cries are to produce a large shift in latency. Consistent with this inter- emitted over roughly the same range of pulse rates as in our pretation, our intracellular recordings show that the rise time of experiments (6Ð35 pps) (Jones 1999). During the “terminal”

J Neurophysiol ¥ VOL 88 ¥ NOVEMBER 2002 ¥ www.jn.org TEMPORAL PATTERN AND SOUND LOCALIZATION CUES 2327 phase, just before capture, pulse rate may exceed 100/s (Griffin ably, binaural difference in spike count and latency of AN2, et al. 1960). Pulse durations of echolocation sounds are gen- and in particular their changes throughout the course of the erally shorter than the 30-ms sounds we used (which we chose pulse trains, were similar in their experiments to that which we to facilitate comparisons with earlier behavioral work), and it describe. Pollack and El-Feghaly found that the initial phono- seems reasonable to suppose that pulse duration might affect taxis response to stimulation was reliably directed away from the extent of response decrement. However pulse durations of the sound source for all pulse rates tested. However, at high several to tens of milliseconds are typical of high-duty-cycle pulse rates, i.e., under conditions that produce the largest bats (Jones 1999). Northern Australia, where T. oceanicus live changes in response of AN2, these initial, negative, responses (Hill et al. 1972), is home to at least five species of Hipposi- were transient (approximately1sinduration), and by 10 s after daridae (Hoffmann et al. 1982), a subfamily characterized by the onset of stimulation, phonotaxis was on average toward, high-duty-cycle calls (Jones 1999). Thus it is plausible that T. rather than away from, the sound source. Our physiological oceanicus might be exposed, in nature, to bat cries that would results show that the binaural difference in response latency induce considerable decrement of AN2 responses. after 10 s of stimulation is larger than that at stimulus onset, Habituation of ultrasound-induced behavioral responses of suggesting that if the crickets relied mainly on latency differ- T. oceanicus has been studied in the laboratory (Engel and Hoy ence as the cue for sound direction, their behavioral responses 1999; May and Hoy 1991). These studies showed that negative should not have reversed in direction. By contrast, the binaural phonotactic responses declined when crickets were stimulated difference in spike count, which initially strongly favors the with pulses (either 10 or 50 ms in duration) presented at rates ipsilateral AN2, is near zero after 10 s of stimulation. It seems of 1.3 pps and 2/s. We saw no decrement in the response of likely, then, that the reversal in behavioral response direction is AN2 when 30-ms pulses were presented at a rate of 0.3/s, and due in part to the loss of directional information encoded as considerable decrement when the pulse rate was 8/s. Unfortu- binaural spike-count difference. It is unclear, however, why the nately, we have no data on AN2’s response to the pulse rates response reverses in direction, rather than simply disappearing. used in the habituation studies, so we cannot say to what extent Nevertheless, comparison of our physiological results with the decline in AN2 response might contribute to behavioral these earlier behavioral studies suggests strongly that binaural habituation. May and Hoy (1991) suggest that behavioral ha- latency difference is not the chief cue for determining sound bituation is not the result of declining AN2 responses, based on direction. This agrees with recent experiments that indicate that an earlier physiological study of AN2 (Nolen and Hoy 1987) the same may be true for localization of cricket song (Pollack that, they claim, failed to find significant response decrement 2001), but contrasts with a recent model of phonotaxis in even at a pulse rate of 15/s. Our results are clearly at odds with crickets, in which latency difference is the pertinent cue (Webb this. However, the data on which the claim for lack of decre- and Scutt 2000). ment in AN2 is apparently based, Fig. 12 of Nolen and Hoy (1987), show a substantial decline in AN2’s response, from This work was supported by the Natural Science and Engineering Research approximately 12 spikes to the first sound pulse to 7 spikes to Council of Canada. the second. Nevertheless, the phenomenology of AN2’s re- sponse decrement and behavioral habituation do differ, sug- gesting that habituation is not accounted for entirely by AN2’s REFERENCES response decrement. In particular, behavioral habituation is BALAKRISHNAN R AND POLLACK GS. Recognition of courtship song in the field greatest at low stimulus intensities (May and Hoy 1991), cricket, Teleogryllus oceanicus. Anim Behav 52: 353Ð366, 1996. whereas the decline in AN2’s response is most pronounced at BOYAN GS. Directional response to sound in the central nervous system of the high intensities (Fig. 2). cricket Teleogryllus commodus (Orthoptera: Gryllidae). I. Ascending inter- It is unclear to what extent the decrement in AN2’s response neurons. J Comp Physiol 130: 137Ð150, 1979. CARR CE. Processing of temporal information in the brain. Annu Rev Neurosci might affect natural behavior. The initial response to ultra- 16: 223Ð243, 1993. sound, at lower intensities, is a short-latency turn away from CORO F, PEREZ« M, MORA E, BOADA D, CONNER WE, SANDERFORD MV, AND the sound source, and it is possible that this would remove the AVILA H. Receptor cell habituation in the A1 auditory receptor of four cricket from the sound field of the bat’s echolocation cry before noctuoid moths. J Exp Biol 201: 2879Ð2890, 1998. EGGERMONT JJ. Azimuth coding in primary of the cat. II. substantial response decrement could occur (cf. Miller 1983). Relative latency and interspike interval representation. J Neurophysiol 80: When stimulated with higher-intensity ultrasound, crickets re- 2151Ð2161, 1998. act in a nondirectional manner (Nolen and Hoy 1986), so the EGGERMONT JJ AND SPOOR A. Cochlear adaptation in guinea pigs: a quantita- effects of response decrement on sound localization cues might tive description. Audiology 12: 193Ð220, 1973. ENGEL JE AND HOY RR. Experience-dependent modification of ultrasound be irrelevant. Estimating the degree to which stimulus-induced auditory processing in a cricket escape response. J Exp Biol 202: 2797Ð changes in localization cues might play a role in bat-cricket 2806, 1999. interactions must await field observations, which are currently FAULKES Z AND POLLACK GS. Effects of inhibitory timing on contrast enhance- lacking. ment in auditory circuits in crickets (Teleogryllus Oceanicus). J Neuro- No matter what its relevance to natural behavior, the decre- physiol 84: 1247Ð1255, 2000. GIVOIS V AND POLLACK GS. Sensory habituation of auditory receptor neurons: ment we describe, when compared with earlier laboratory implications for sound localization. J Exp Biol 203: 2529Ð2537, 2000. experiments, helps to illuminate how ultrasound location is GRIFFIN DR, FRIEND JH, AND WEBSTER FA. Target discrimination by the represented at the level of the AN2 neurons. Pollack and echolocation of bats. J Exp Zool 158: 155Ð168, 1965. El-Feghaly (1993) described the effects of pulse rate on pho- GRIFFIN DR, WEBSTER FA, AND MICHAEL CR. The echolocation of flying notactic responses of T. oceanicus to ultrasound stimuli. Their insects by bats. Anim Behav 8: 141Ð154, 1960. HOFFMANN RS, GARDNER AL, BROWNELL RL, KOOPMAN KF, MUSSER GG, AND stimuli, like ours, consisted of trains of 30-ms duration sound SCHLITTER DA. Mammal Species of the World. Lawrence, KS: Allen Press, pulses, with pulse rates ranging from 8 to 32 pulses/s. Presum- 1982, p. 136Ð148.

J Neurophysiol ¥ VOL 88 ¥ NOVEMBER 2002 ¥ www.jn.org 2328 A.-H. SAMSON AND G. S. POLLACK

HILL KG, LOFTUS-HILLS JJ, AND GARTSIDE DF. Pre-mating isolation between PASZTOR VM. Graded potentials and spiking in single units of the oval organ, the Australian field crickets Teleogryllus commodus and. T. oceanicus a in the lobster ventilatory system. III. Sensory habituation (Orthoptera: Gryllidae). Aust J Zool 20: 153Ð163, 1972. to repetitive stimulation. J Exp Biol 107: 465Ð472, 1983. HOY RR. The evolution of in insects as an adaptation to predation from POLLACK GS. Selective attention in an insect auditory neuron. J Neurosci 8: bats. In: The Evolutionary Biology of Hearing, edited by Webster DB, Fay 2635Ð2639, 1988. RR, and Popper AN. Berlin: Springer-Verlag, 1992. POLLACK GS. Neural processing of acoustic signals. In: Comparative Hearing: IMAIZUMI K AND POLLACK GS. of sound frequency by cricket Insects, edited by Hoy RR, Popper AN, and Fay RR. New York: Springer auditory receptors. J Neurosci 19: 1508Ð1516, 1999. Verlag, 1998. IMAIZUMI K AND POLLACK GS. Neural representation of sound amplitude by POLLACK GS. Sound localization in crickets: the dominant neural cue is functionally different auditory receptors in crickets. J Acoust Soc Am 109: binaural difference in response strength, not latency difference. Intl Soc 1247Ð1260, 2001. Neuroethol Abstr 98, Bonn, July 29ÐAug. 3, 2001. JONES G. Scaling of echolocation calls in bats. J Exp Biol 202: 3359Ð3367, 1999. POLLACK GS AND EL-FEGHALY E. Calling song recognition in the cricket KIANG NY-S, WATANBE T, THOMAS EC, AND CLARK LF. Discharge Patterns of Teleogryllus oceanicus: comparison of the effects of stimulus intensity and Single Fibers in the Cat’s Auditory Nerve. Cambridge, MA: MIT Press, sound spectrum on selectivity for temporal pattern. J Comp Physiol 173: 1965. 759Ð765, 1993. MAY ML AND HOY RR. Habituation of the ultrasound-induced acoustic startle POLLACK GS AND FAULKES Z. Representation of behaviorally relevant sound response in flying crickets. J Exp Biol 159: 489Ð499, 1991. frequencies by auditory receptors in the cricket. J Exp Biol 201: 155Ð163, MILLER LA. How insects detect and avoid bats. In: and Be- 1998. havioral Physiology, edited by Huber F and Markl H. Berlin: Springer- POLLACK GS, HUBER F, AND WEBER T. Frequency and temporal pattern- Verlag, 1983. dependent phonotaxis of crickets (Teleogryllus oceanicus) during tethered MOISEFF A AND HOY RR. Sensitivity to ultrasound in an identified auditory flight and compensated walking. J Comp Physiol 154: 13Ð26, 1984. interneuron in the cricket: a possible link to phonotactic behavior. J Comp ROMER¬ H AND KRUSCH M. A gain-control mechanism for processing of chorus Physiol 152: 155Ð167, 1983. sounds in the afferent auditory pathway of the bushcricket Tettigonia viri- MOISEFF A, POLLACK GS, AND HOY RR. Steering response of flying crickets to dissima (Orthoptera: Tettigoniidae). J Comp Physiol 186: 181Ð191, 2000. sound and ultrasound: mate attraction and predator avoidance. Proc Natl SELVERSTON A, KLEINDIENST H-U, AND HUBER F. Synaptic connectivity be- Acad Sci USA 75: 4052Ð4056, 1978. tween cricket auditory interneurons as studied by selective photoinactiva- MORCHEN¬ A. Spike count and response latency. Two basic parameters encod- tion. J Neurosci 5: 1283Ð1292, 1985. ϩ ing sound direction in the CNS of insects. Naturwissenschaften 67: 469, SOBEL EC AND TANK DW. In vivo Ca2 dynamics in a cricket auditory neuron: 1980. an example of chemical computation. Science 263: 823Ð826, 1994. NOLEN TG AND HOY RR. Initiation of behavior by single neurons: The role of STRAUSFELD NJ, SEYAN HS, WOHLERS D, AND BACON JP. Lucifer Yellow behavioral context. Science 226: 992Ð994, 1984. histology. In: Functional Neuroanatomy, edited by Strausfeld NJ. Berlin: NOLEN TG AND HOY RR. Phonotaxis in flying crickets. I. Attraction to the Springer Verlag, 1983, p. 132Ð155. calling song and avoidance of bat-like ultrasound are discrete behaviour. WEBB B AND SCUTT T. A simple latency-dependent spiking-neuron model of J Comp Physiol 159: 423Ð439, 1986. cricket phonotaxis. Biol Cybern 82: 247Ð269, 2000. NOLEN TG AND HOY RR. Postsynaptic inhibition mediates high-frequency WYTTENBACH RA AND HOY RR. Laser-vibrometric measurements of the re- selectivity in the cricket Teleogryllus oceanicus: implications for flight sponse of cricket ears to ultrasound. Soc Neurosci Abstr 25(1Ð2), Miami phonotaxis behavior. J Neurosci 7: 2081Ð2096, 1987. Beach, Oct. 23Ð28, 1999.

J Neurophysiol ¥ VOL 88 ¥ NOVEMBER 2002 ¥ www.jn.org