Abstract When insectivorous bats such as Eptesicusfuscus emit ultrasonic signals and analyze the returningechoes to hunt insects, duration selectivity of auditoryneurons plays an important role in echo recognition.The success of prey capture indicates that they caneVectively encode progressively shortened echo dura-tion throughout the hunting process. The present studyexamines the echo duration selectivity of neurons inthe central nucleus of the bat inferior colliculus (IC)under stimulation conditions of single pulses andpulse–echo (P–E) pairs. This study also examines therole of gamma-aminobutyric acid (GABA)ergic inhibi-tion in shaping echo duration selectivity of IC neurons.The data obtained show that the echo duration selec-tivity of IC neurons is sharper when determined withP–E pairs than with single pulses. Echo duration selec-tivity also sharpens with shortening of pulse durationand P–E gap. Bicuculline application decreases andGABA application increases echo duration selectivityof IC neurons. The degree of change in echo durationselectivity progressively increases with shortening ofpulse duration and P–E gap during bicuculline applica-tion while the opposite is observed during the GABAapplication. These data indicate that the GABAergicinhibition contributes to sharpening of echo duration
selectivity of IC neurons and facilitates echo recogni-tion by bats throughout diVerent phases of hunting.
Keywords Bat · Bicuculline · Duration selectivity · Echolocation · GABA · Inferior colliculus · Pulse–echo gap
AbbreviationsBD Best durationBF Best frequencyGABA Gamma-aminobutyric acidMT Minimum thresholdnDW Normalized duration width
Introduction
In the central auditory pathway, the central nucleus ofthe inferior colliculus (IC) receives and integratesexcitatory and inhibitory inputs from many lower audi-tory nuclei and from the auditory cortex (Casseday andCovey 1995; Herbert et al. 1991; HuVman and Henson1990; Saldana et al. 1996; Winer et al. 1998). Duringsignal processing, temporal and spectral interaction ofthese two opposing inputs shapes response propertiesof IC neurons. Presumably, inhibition only occurswhen inhibitory inputs arrive prior to the excitatoryinputs at IC neurons within a certain temporal window.Furthermore, inhibitory inputs with stronger intensityand longer duration should be more eVective in inhibi-tion of auditory response of IC neurons than inhibitoryinputs with weaker intensity and shorter duration (Luand Jen 2002, 2003).
One of the major inhibitory inputs in the IC is medi-ated by GABA (Fubara et al. 1996; Roberts and Ribak
C. H. Wu · P. H.-S. Jen (&)Division of Biological Sciences and Interdisciplinary Neuroscience Program, University of Missouri-Columbia, Columbia, MO 65211, USAe-mail: [email protected]
Present address: C. H. WuDepartment of Life Science, National Taiwan Normal University, Taipei, Taiwan, ROC
123
J Comp Physiol A
1987). By means of application of GABA or bicucul-line, which is an antagonist for GABAA receptors(Bormann 1988; Cooper et al. 1982), many studies haveshown that interaction of excitation and GABAergicinhibition contributes to auditory temporal processingand shapes multi-parametric selectivity (e.g., duration,frequency, amplitude, direction, etc.) of IC neuronsusing single repetitive sound pulses or temporally pat-terned trains of sound pulses (Casseday et al. 1994,2000; Faingold et al. 1991; Jen and Feng 1999; Jen andZhang 2000; Jen et al. 2002; Klug et al. 1995; Koch andGrothe 1998; LeBeau et al. 2001; Lu et al. 1997, 1998;Lu and Jen 2001; Park and Pollak 1993; Yang et al.1992; Vater et al. 1992).
Among the multi-parametric selectivity, durationselectivity of auditory neurons plays an important rolefor sound recognition particularly in human speech,animal communication and bat echolocation (Coveyand Casseday 1999; Popper and Fay 1995; Shannonet al. 1995). For example, the big brown bat, Eptesicusfuscus, must eVectively analyze the changing echo fea-tures including echo duration throughout the entirecourse of hunting for successful prey capture. The neu-ral basis underlying the duration selectivity of bats hasbeen supported by many studies which show that bat’sIC neurons behave as band-, short-, long- and all-passWlters to sound duration (Casseday et al. 1994, 2000;Ehrlich et al. 1997; Faure et al. 2003; Fremouw et al.2005; Fuzessery and Hall 1999; Galazyuk and Feng1997; Jen and Feng 1999; Jen and Zhou 1999; Pinheiroet al. 1991; Zhou and Jen 2001).
However, these studies examined duration selectiv-ity of IC neurons using single sound pulses or tempo-rally patterned pulse trains that consist of sound pulseswith equal amplitude and inter-pulse gap. Yet, in thereal world the bat emits intense pulses and listens togreatly attenuated returning echoes in which the gapbetween the pulse and echo (abbreviated as P–E gap)is always shorter than the gap between two succeedingP–E pairs. Therefore, in reality the bat analyzes theecho from a series of P–E pairs of unequal amplitudewith progressively shortening P–E gap during hunting.For this reason, a study of variation of echo durationselectivity of the bat auditory neurons with shorteningof P–E gap is necessary for better understanding ofecho recognition by bats during diVerent phases ofhunting.
The main objective of the present study is to exam-ine the echo duration selectivity of IC neurons of thebig brown bat to show the following. (1) Echo durationselectivity of bat IC neurons is sharper when deter-mined with the echo pulses of P–E pairs than with tem-porally isolated single pulses. (2) Echo duration
selectivity of IC neurons progressively improvesthroughout a target approaching sequence. (3) GAB-Aergic inhibition shapes the echo duration selectivityof IC neurons.
Materials and methods
Ten E. fuscus (6 males, 4 females, 10–27 g, bodyweight, b.w.) were used for this study. As described inprevious studies (Jen et al. 1987), the Xat head of a1.8 cm nail was glued onto the exposed skull of eachNembutal anesthetized bat (45–59 mg/kg b.w.) withacrylic glue and dental cement 1 or 2 days before therecording session. Exposed tissue was treated with anantibiotic (Neosporin) to prevent inXammation.During recording, the bat was administered the neuro-leptanalgesic, Innovar-Vet (Fentanyl 0.08 mg/kg b.w.Droperidol 4 mg/kg b.w.), and placed inside a batholder (made of wire mesh) that was suspended in anelastic sling inside a double-wall sound-proof room(temperature 28–30°C). The ceiling and inside walls ofthe room were covered with 3-in. convoluted polyure-thane foam to reduce echoes. After Wxing the bat’shead with a set screw, small holes were made in theskull above the IC for insertion of 3 M KCl glass pipetteelectrodes (impedance: 5–10 M�). Additional doses ofInnovar-Vet were administered during later phases ofrecording when bats showed signs of discomfort. Alocal anesthetic (Lidocaine) was applied to the openwound area. The recording depth was read from thescale of a microdrive (David Kopf). A common indiVer-ent electrode (silver wire) was placed at the nearbytemporal muscles. Each bat was used in one to Wverecording sessions on separate days and each recordingsession typically last for 2–6 h. The experiments wereconducted according to NIH publication no. 85–23,“Principles of Laboratory Animal Care” and with theapproval of the Institutional Animal Care and UseCommittee of the University of Missouri-Columbia.
Acoustic stimuli (4 ms with 0.5 ms rise–decay times,delivered at 2 pulses/s) were generated with an oscilla-tor (KH model 1200) and a homemade electronicswitch driven by a stimulator (Grass S88). These stim-uli were then ampliWed after passing through a decadeattenuator (HP 350D) before they were fed to a smallcondenser loudspeaker (AKG model CK 50, 1.5 cmdiameter, 1.2 g) that was placed 23 cm away from thebat and 30° contralateral to the recording site. Calibra-tion of the loudspeaker was performed with a ¼-in.microphone (B & K 4135) placed at the position of thebat’s head during recording using a measuring ampli-Wer (B & K 2607). The output of the loudspeaker was
expressed in dB SPL in reference to 20 �Pa root meansquare.
Upon isolation of an IC neuron with 4 ms soundpulses, its best frequency (BF) was determined bychanging the pulse frequency and intensity. The mini-mum threshold (MT) at the BF was deWned as thesound level that elicited 50% response probabilityfrom the neuron. The echo duration selectivity of theIC neuron was studied by plotting its duration tuningcurves using the number of impulses in response toeight durations (1, 1.5, 2, 4, 6, 8, 10 and 20 ms) of singlepulses and the echo pulses of three P–E pairs. Rise–decay times for these diVerent single pulses and echodurations were typically 0.5 ms but they were 0.25 msfor 1-ms pulse duration. The amplitude of single echopulses was set at 10 dB above the MT while the ampli-tude of pulse and echo was, respectively, set at 30 and10 dB above the neuron’s MT as used in previousstudies (Suga et al. 1983; Tanaka et al. 1992; Wonget al. 1992). The left panel of Fig. 1 shows the sketchesof three P–E pairs with duration, gap (time periodbetween the oVset of pulse and onset of echo) andinterval (time period between onset of pulse and echo)used in this study. These three P–E pairs are compara-ble to the P–E pairs occurring during search, approachand terminal phases of hunting by E. fuscus (GriYn1958; Surlykke and Moss 2000). When these three P–Epairs were used to study the echo duration selectivity,the pulse was always Wxed at a constant value (i.e., 1.5,4.0 or 10 ms) while the echo duration was varied ateight diVerent durations. Because these bat speciesuses frequency modulated (FM) pulses during echolo-cation, we also studied echo duration selectivity of ICneurons using downward sweeping FM P–E pulsesgenerated by means of ramp signals. Each FM pulseswept one octave downward across the BF of the ICneuron.
Iontophoretic application of bicuculline and GABAto recorded IC neurons has been described in previousstudies (Lu et al. 1997, 1998). BrieXy, a three-barrel orWve-barrel electrode (tip: 10–15 �m) was piggybackedto a 3 M KCl single-barrel electrode (tip: less than1 �m; impedance: 5–10 M�) whose tip was extendedabout 10 �m from the tip of the three-barrel electrode.The 3 M KCl single-barrel recording electrode wasused to record neural responses. One of the barrels ofthe three-barrel electrode was Wlled with bicucullinemethiodide (10 mM in 0.16 m NaCl, pH 3.0; Sigma) orgamma-aminobutyric acid (GABA, 500 mM in dis-tilled water, pH 3.5; Sigma). However, when a Wve-bar-rel electrode was used, two barrels were Wlled withboth drugs, respectively, such that both drugs could beapplied to the recorded neuron. The bicuculline and
GABA were prepared just prior to each experimentand the electrode Wlled immediately before use. Thedrug channel was connected via silver–silver chloridewire to a microiontophoresis constant current genera-tor (Medical Systems Neurophore BH-2) that was usedto generate and monitor iontophoretic currents. Dur-ing drug application, a 1-s pulse of positive 40 nA at0.5 pps was applied for 1 min before data acquisition.The application current was changed to 10 nA duringdata acquisition. The other two barrels were Wlled with1 M NaCl (pH 7.4), one of which was used as theground and another as the balanced barrel. The bal-ance electrode was connected to a balance module.The retaining current was negative 8–10 nA.
Recorded action potentials were ampliWed, band-pass Wltered (Krohn-Hite 3500), and then fed through awindow discriminator (WPI 121) before being sent toan oscilloscope (Tektronix 5111) and an audio monitor(Grass AM6). They were then sent to a computer(Gateway 2000, 486) for acquisition of peri-stimulus-time (PST) histograms (bin width: 500 �s and samplingperiod: 300 ms) to 32 presentations of stimuli. Theright panel of Fig. 1 shows the PST histograms and thenumber of impulses of a representative IC neuronobtained with the three P–E pairs.
When IC neurons received drug application, theirecho duration tuning curves were plotted before andduring drug application. The tuning properties of aduration tuning curve were expressed with a best dura-tion (BD) and a normalized duration width (nDW)
Fig. 1 Aa, Ba, Ca Sketches showing the envelops of three pulse–echo (P–E) pairs with P duration (PD), P–E gap and P–E interval(P–E I) comparable to that occurring during search (Ca), ap-proach (Ba) and terminal phases (Aa) of hunting by the bigbrown bat, E. fuscus. The pulse and echo were set at 30 and 10 dBabove the minimum threshold (MT) of each investigated inferiorcollicular (IC) neuron. Ab, Bb, Cb Peri-stimulus-time (PST) his-tograms (bin width: 500 �s and sampling period: 50 ms) showingthe discharge patterns of an IC neuron obtained with 32 presen-tations of three P–E pairs. The neuron’s number of impulses in re-sponse to each pulse (P) and echo (E) are shown atop
(see Fig. 2). These two duration tuning properties of ICneurons obtained under diVerent stimulation condi-tions were then quantitatively studied and statisticallycompared using repeated measures one-way or two-way ANOVA followed with a Student–Newman–Keuls multiple comparisons post-test at P < 0.05.
Results
In this study, 98 IC neurons were isolated at depthsbetween 119 and 1,898 �m. Their BFs and MTs ranged20.3–73.9 kHz (36.5 § 7.4 kHz) and 18–55 dB SPL
(42.5 § 7.1 dB SPL). The Wrst-spike latencies werebetween 9 and 22 ms (12.6 § 1.7 ms). Echo durationselectivity was studied for all 98 neurons. However,echo duration selectivity of only 54 neurons was stud-ied during drug application because of loss of neuronsthroughout the course of study. Among them, 22 neu-rons received bicuculline application and 14 neuronsreceived GABA application. The remaining 18received bicuculline application Wrst. After recoveryfrom the drug eVect, they then received GABA appli-cation. Therefore, echo duration selectivity was studiedin 40 neurons during bicuculline application and in 32neurons during GABA application.
In the following, we Wrst describe diVerent types ofecho duration tuning curves and compare the echoduration selectivity of IC neurons determined with BFand FM pulses. We then compare echo duration selec-tivity of IC neurons determined with single pulses andecho pulses of P–E pairs. The description is followedwith a presentation of the eVect of bicuculline andGABA application on echo duration selectivity.Finally, we describe the relationship between echoduration selectivity, BF and recording depth. For con-venience of description and comparison, we use theterm “echo duration selectivity” to describe the dura-tion selectivity of IC neurons determined with both sin-gle pulses and echo pulses of P–E pairs.
Echo duration tuning curves of IC neurons
The discharge patterns of a representative IC neuronobtained with P–E pairs of 4 ms pulse at varied echodurations are shown in Fig. 2A. The neuron’s numberof impulses in response to the 4 ms pulse did not diVerby three pulses while its number of impulses inresponse to diVerent echo durations varied as many as14 impulses. This neuron’s echo duration tuning curveis shown in Fig. 2Ba.
In total, 928 echo duration tuning curves were plot-ted with the number of impulses discharged to singleecho pulses and to the echo pulses of P–E pairs underdiVerent stimulation conditions. These duration tuningcurves can be described as band-, short-, long- and all-pass using the same criterion adopted in previous stud-ies (Jen and Feng 1999; Jen and Zhou 1999; Wu andJen 2006). The band-pass duration tuning curvesshowed a maximal number of impulses to a speciWcduration and the maximal number of impulsesdecreased at least 50% at both limbs (Fig. 2Ba, n = 215,23%). The short-pass duration tuning curves showed amaximal number of impulses to a short duration andthe maximal number of impulses decreased more than25% at a short duration and more than 50% at a long
Fig. 2 Aa–h PST histograms showing the discharge pattern of anIC neuron determined with P–E pairs. The envelope of each P–Epair is shown below the PST histogram. The PD was always 4 mswhile the echo duration (ED) varied from 1 to 20 ms. Ba–d Fourtypes of ED tuning curves plotted with the number of impulsesdischarged to the echo of each P–E pair against ED. These EDtuning curves are described as band pass (Ba), short pass (Bb),long pass (Bc) and all pass (Bd). Each horizontal dashed line indi-cates the 50% maximal response. The sharpness of each EDcurve is expressed with a best duration (BD indicated with anarrowhead) and a normalized duration width (nDW indicatedwith a double arrowhead) of an ED tuning curve at 75% of max-imum. NA a BD is not available. The BF (kHz), latency (ms), MT(dB SPL) and recording depth (�m) of these neurons were 36.9,11.5, 37.0, 438 (Ba); 34.6, 12.0, 36.0, 334 (Bb); 38.6, 12.5, 41.0, 753(Bc); and 42.5, 13.5.0, 44.0, 1,286 (Bd) (see the text for details)
123
J Comp Physiol A
duration (Fig. 2Bb, n = 195, 21%). Conversely, thelong-pass duration tuning curves showed a maximalnumber of impulses to a long duration and the maximalnumber of impulses decreased more than 25% at along duration and more than 50% at a short duration(Fig. 2Bc, n = 104, 11%). The number of impulses ofall-pass echo duration tuning curves often diVered bymore than 25% but never by more than 50% at alldurations tested (Fig. 2Bd, n = 414, 45%).
In this study, the pulse duration that elicited themaximal number of impulses in the band-, short- andlong-pass duration tuning curves is deWned as the BD(indicated by an arrowhead in Fig. 2Ba–c). For conve-nience of description, IC neurons with any of thesethree types of duration tuning curves are called dura-tion selective neurons. Conversely, IC neurons withall-pass duration tuning curves are called durationnon-selective neurons.
Because the maximal number of impulses inresponse to the preferred pulse duration varied greatlyamong individual IC neurons and during drug applica-tion, we used a nDW to express the sharpness of a dura-tion tuning curve. An nDW was obtained by dividingthe maximum by the width of a duration tuning curve at75% of the maximum (Fig. 2Ba–d, DW indicated by adouble arrowhead). By obtaining the nDW, we excludethe possibility that a change in the duration width ofecho duration tuning curves plotted under diVerentstimulation conditions might only reXect a change inresponse size to presented echo pulses rather than atrue change in echo duration selectivity. Thus, a neuronwith a large nDW has a narrow duration tuning curveand sharp duration selectivity. Every band-, short- andlong-pass echo duration tuning curve has an nDW.Although an all-pass echo duration curve is durationnon-selective according to our criterion, an all-passecho duration curve may have an nDW when the maxi-mum decreases by more than 25% at the two limbs(e.g., Fig. 2Bd). As described below, this nDW becomesuseful to compare the echo duration selectivity of ICneurons that changed from duration non-selective intoduration selective during GABA application.
Echo duration selectivity of IC neurons determined with BF and FM pulses
To determine if echo duration selectivity of IC neuronsmight diVer when determined with pure tone pulsesand FM pulses, we compared echo duration selectivityof 38 IC neurons determined with P–E pairs of BFpulses and FM pulses. As shown in Fig. 3, similar dis-charge patterns and number of impulses were obtainedfrom a representative IC neuron with P–E pairs (pulse:
1.5 ms, P–E gap: 2 ms) of BF and FM pulses. As such,the neuron had similar echo duration tuning curvewhen plotted with the number of impulses in responseto BF and FM echo pulses in varied duration(Fig. 3A2a vs. B2a). In the same token, the neuron’sduration tuning curves obtained with the other twoP–E pairs (pulse: 4 or 10 ms, P–E gap: 4 or 8 ms) werealso similar (Fig. 3A2b, c vs. B2b, c). Overall, the neu-ron’s duration selectivity obtained with both BF andFM echo pulses decreased with lengthening of PD andP–E gap as evident by broadened duration tuningcurves, lengthened BD and decreased nDW (Fig. 3A2,B2).
The average BD of these 38 IC neurons obtainedwith both BF and FM pulses signiWcantly increased andthe average nDW decreased with lengthening of pulseduration (PD) and P–E gap (Fig. 3Ca, b, repeated mea-sures two-way ANOVA; P < 0.01 for unWlled and Wlledbars). A Student–Newman–Keuls multiple compari-sons post-test showed signiWcant diVerences betweeneach set of the BD and the nDW (**P < 0.01 and*P < 0.05). However, the BD and nDW obtained fromBF and FM pulses did not diVer signiWcantly (Fig. 3Ca,b, repeated measures two-way ANOVA; P > 0.05between Wlled and unWlled bars).
As shown in Table 1, the percent distribution of allfour types of duration tuning curves of these 38 IC neu-rons obtained from BF and FM pulses is very similar.The number of band- and short-pass echo durationtuning curves decreases while that of long- and all-passecho duration tuning curves increases with lengtheningof PD and P–E gap.
Because echo duration selectivity of IC neuronsdetermined with BF and FM echo pulses did not diVersigniWcantly and generation of FM pulses at variedduration was time consuming such that recorded ICneurons often lost before completion of study, we onlyused BF echo pulses to study duration selectivity of theremaining 60 neurons. The following presentation
Table 1 Echo duration tuning curves of 38 IC neurons plottedwith the number of impulses discharged to echoes of BF and FMpulse–echo (P–E) pairs
BP Band pass, SP short pass, LP long pass, AP all pass
therefore is based on the data collected from all 98 ICneurons studied with BF pulses.
Echo duration selectivity of IC neurons determined with echo pulses of P–E pairs and single echo pulses
IC neurons had sharper echo duration selectivity whendetermined with echo pulses of P–E pairs than with sin-gle echo pulses. As shown in Fig. 4, a representative ICneuron always discharged more impulses to 4 ms pulsethan to 1.5 and 10 ms pulses regardless of the variationin echo duration at each P–E pair. However, the neuronalways discharged fewer impulses to echo pulses of P–Epairs than to single echo pulses (Fig. 4A–C vs. D).When stimulated with three P–E pairs, the neuron’snumber of impulses discharged to each echo durationbecame larger with lengthening of P–E gap (Fig. 4A–C). The neuron discharged maximally to 4 ms BD echopulses and the maximum progressively decreased as theecho duration became shorter or longer than the BD.The decrease from the maximum was larger for shorterthan for longer pulse duration and P–E gap. For exam-ple, when the 4 ms echo pulse increased to 6 ms, the
number of impulses decreased from 8 to 3 (¡62.5%)when tested with 1.5 ms pulse at 2 ms P–E gap;decreased from 12 to 6 (¡50%) when tested with 4 mspulse at a 4 ms P–E gap and decreased from 16 to 12(¡25%) when tested with 10 ms pulse at 8 ms P–E gap(Fig. 4A–C, at echo duration of 4 vs. 6 ms). As such, theneuron’s echo duration selectivity progressivelyincreased with shortening of pulse duration and P–Egap as evident by increasing nDW (Fig. 4E–G). How-ever, the neuron had the least duration selectivity whenobtained with single echo pulses (Fig. 4H). The neuronhad a short-pass duration tuning curve when obtainedwith single echo pulses but had a band-pass durationtuning curve when obtained with all three P–E pairs(Fig. 4E–H). The neuron had a BD of 4 ms whenobtained under all four stimulation conditions.
Figure 5 shows the discharge patterns and echoduration tuning curves of another IC neuron with var-ied BD under four diVerent stimulation conditions.This neuron also discharged more impulses to singleecho pulses than to echo pulses of P–E pairs (Fig. 5A–C vs. D). When stimulated with three P–E pairs, theneuron always discharged more impulses to 10 ms
Fig. 3 A1, B1 The PST histo-grams of an IC neuron in re-sponse to three P–E pairs of best frequency (BF) tones and frequency modulated (FM) pulses. The PD and P–E gap were both 4 ms while the echo duration (ED) varied be-tween 1 and 20 ms. The BF (kHz), latency (ms), MT (dB SPL) and recording depth (�m) of the neuron were 32.5, 10.5, 35.0 and 424. A2, B2 The neuron’s echo duration tuning curves plotted with the num-ber of impulses in response to varied echo durations of these three P–E pairs of BF and FM pulses. C Bar histograms showing the average BD and nDW of echo duration tuning curves determined with three P–E pairs of BF (unWlled bars) and FM pulses (Wlled bars) (see the text for details)
123
J Comp Physiol A
pulses than to 4 and 1.5 ms pulses. The neuron’s num-ber of impulses discharged to each echo pulses becamesmaller with shortening of pulse duration and P–E gap(Fig. 5A–C). The decrease from the maximumobtained at BD was greater for shorter than for longerpulse duration and P–E gap (Fig. 5A vs. B vs. C). Assuch, the neuron’s echo duration selectivity progres-sively increased with shortening of pulse duration and
P–E gap as evident by increasing nDW (Fig. 5E–G).The neuron had an all-pass echo duration tuning curveand was duration non-selective when obtained withsingle echo pulses (Fig. 5H). When determined withthree P–E pairs, the neuron’s short-pass duration tun-ing curve obtained at 8 ms P–E gap changed into band-pass at shorter P–E gap and the BD shortened from 8to 4 ms (Fig. 5E–G).
Fig. 4 A–D PST histograms of an IC neuron obtained with threeP–E pairs and single echo pulses. The PD, ED and P–E gap are,respectively, shown. The BF (kHz), latency (ms), MT (dB SPL)and recording depth (�m) of the neuron were 33.8, 10.5, 37.0 and446. E–H The neuron’s echo duration tuning curves determinedwith the echo pulses of three P–E pairs and with single echo pulses.
The duration tuning properties, the BD and the nDW are shownwithin each plot. Note that the IC neuron always had a BD of4 ms when determined at all four stimulation conditions. Theneuron had the smallest nDW when determined with single echopulses. The nDW progressively increased with shortening of PDand P–E gap when tested with three P–E pairs
Fig. 5 The echo duration selectivity of another IC neuron deter-mined with three P–E pairs and single echo pulses. The BF (kHz),latency (ms), MT (dB SPL) and recording depth (�m) of the neu-ron were 37.6, 12.5, 41.0 and 896. Note that the neuron also had
the smallest nDW when determined with single echo pulses.When determined with three P–E pairs, the nDW progressivelyincreased and the BD decreased from 8 to 4 ms with shorteningof PD and P–E gap (see Fig. 4 for legends)
123
J Comp Physiol A
Among the 98 IC neurons studied, 45 neurons hadBDs between 1.5 and 10 ms covering the duration ofpluses emitted by E. fuscus during three phases ofhunting. The BD of 27 (60%) neurons did not change(e.g., Fig. 4) while the BD of other 18 (40%) neuronsshortened with decreasing pulse duration and P–E gaps(e.g., Fig. 5).
The distribution of four types of echo duration tuningcurves obtained with echo pulses of three P–E pairs andsingle echo pulses is shown in Table 2. It is clear thatmore duration selective neurons were obtained with theecho pulses of P–E pairs than with single echo pulses. Itis also clear that many all-pass duration non-selectiveneurons obtained with single echo pulses changed intoband-pass duration selective neurons when obtainedwith echo pulses of P–E pairs. As shown in Fig. 6, thenumber of duration selective neurons progressivelydecreases while the number of duration non-selectiveneurons increases with lengthening of pulse durationand P–E gap of the P–E pairs. The variation reaches themaximum when obtained with single echo pulses(Fig. 6a, Wlled vs. unWlled circles). As such, the diVer-ence between the number of duration selective and non-selective neurons progressively decreases with thesediVerent stimulation conditions (Fig. 6a, dashed line).
Variation in the sharpness of echo duration tuningcurves under these four diVerent stimulation condi-tions can be seen in the variation of the BD and nDWof these IC neurons. The average BD signiWcantlyincreases and nDW decreases with lengthening ofpulse duration and P–E gap of the P–E pairs and thevariation reaches the maximum when stimulated withsingle echo pulses (Fig. 6b, c, repeated measures one-way ANOVA, P < 0.01). Newman–Keuls multiplecomparisons post-test showed signiWcant diVerencesbetween each set of the BD and the nDW(***P < 0.001, **P < 0.01 and *P < 0.05).
Echo duration selectivity of IC neurons in relation to pulse duration, P–E interval and P–E gap
Because we examined the echo duration selectivitywith the P–E pairs that varied in pulse duration, P–Einterval and P–E gap, we were able to examine theeVect of these varied parameters on echo durationselectivity of 48 IC neurons. Figure 7 shows the nineecho duration tuning curves of a representative IC neu-ron obtained with the P–E pairs under diVerent combi-nations of pulse duration and P–E interval. Theneuron’s discharge patterns to the P–E pairs at tworepresentative P–E intervals with and without P–Eoverlap are shown in Fig. 7B. When stimulated with10 ms pulse duration at 8 and 18 ms P–E intervals, theneuron’s responses to pulse and varied echo durationswere always recognizable although P–E overlapoccurred when stimulated at 18 ms P–E interval(Fig. 7Ba vs. Bb). However, response to each echopulse was greatly reduced when P and E overlappedsuch that the neuron had poor echo duration selectivity
Table 2 Echo duration tuning curves of IC neurons plotted withthe number of impulses discharged to the echoes of P–E pairs andto single pulses
Fig. 6 a Variation in the number of duration selective (bandpass, short pass and long pass, unWlled circles) and non-selective(all pass, Wlled circles) IC neurons determined with three P–Epairs and single echo pulses. The dashed line indicates the diVer-ence in the number between the two groups of IC neurons. b, cBar histograms showing the average BD and nDW of IC neu-rons determined under diVerent stimulation conditions. TheBD and nDW obtained under all four stimulation conditionsdiVer signiWcantly (repeated measures one-way ANOVA,P < 0.01). Newman–Keuls multiple comparisons post-testshowed signiWcant diVerences between each set of the BD andthe nDW (**P < 0.01 and *P < 0.05)
123
J Comp Physiol A
(i.e., small nDW)(Fig. 7Ah vs. Ai, nDW: 0.9 vs. 3.0).Similarly, the neuron’s nDW was 9.1 when tested with1.5 ms pulse at 3.5 ms P–E interval (i.e., no P–E over-lap) but its nDW was only 0.8 when tested with 10 mspulse at 3.5 ms P–E interval (i.e., P–E overlap) (e.g.,Fig. 7Aa vs. g).
When tested with P–E pairs, variation in echo dura-tion selectivity of the IC neuron with the pulse dura-tion and P–E interval is unsystematic. For example,when determined at 3.5 or 8 ms P–E interval, echoduration selectivity (i.e., nDW) varied unsystematicallywith lengthening of pulse duration (Fig. 7Aa vs. d vs. g,b vs. e vs. h, Ca solid and unWlled circles). When deter-mined at 18 ms P–E interval, duration selectivityhardly changed with pulse duration (Fig. 7Ac vs. f vs. I,Ca solid triangles). Similarly, when determined with1.5-, 4- and 10-ms pulse duration at varied P–E inter-val, the nDW changed unsystematically with increasingP–E interval (Fig. 7Aa vs. b vs. c, d vs. e vs. f, g vs. h vs.i, Cb).
We also plotted the echo duration tuning curve ofthe same IC neuron under diVerent combinations ofpulse duration and P–E gap. DiVerent from the aboveobservation; the echo duration selectivity of the sameIC neuron varied systematically with pulse durationand P–E gap. When determined at 2 or 4 ms P–E gap,the nDW progressively decreased with lengthening ofpulse duration (Fig. 8Aa vs. d vs. g; b vs. e vs. h, Basolid and unWlled circles). However, when determinedat 8 ms P–E gap, duration selectivity hardly changedwith pulse duration (Fig. 8Ac vs. f vs. I, Ba solid trian-
gles). Alternatively, when determined with 1.5-, 4- and10-ms pulse duration, the nDW progressively increasedwith shortening of P–E gap (Fig. 8Aa vs. b vs. c, d vs. evs. f, g vs. h vs. i). However, the change in the nDW wasgreater for shorter PD than long PD (Fig. 8Bb).
Figure 9 summarizes the variation in the averageBD and nDW of 42 IC neurons with P–E gap and inter-val. It is clear that when stimulated with 1.5 and 4 mspulses at 2 ms P–E gap, these neurons had the smallestaverage BD which signiWcantly increased with shorten-ing and lengthening of P–E gap and P–E interval(Fig. 9Aa, Ba, one-way ANOVA, P < 0.05). However,the average BD hardly changed when pulse and echooverlapped (Fig. 9Aa, Ba, shaded). The average BDhardly varied with the P–E gap and interval whentested with 10 ms pulses (Fig. 9Ca, one-way ANOVA,P > 0.05).
Conversely, when stimulated with 1.5 and 4 mspulses at 2 ms P–E gap, these neurons had the largestaverage nDW which signiWcantly decreased with short-ening and lengthening of P–E gap and P–E interval(Fig. 9Ab, Bb, Cb, one-way ANOVA, P < 0.01–0.05).Variation of the average nDW with P–E gap and inter-val was minimal when pulse and echo overlapped(Fig. 9Ab, Bb, Cb, shaded).
The eVect of bicuculline application on echo duration selectivity determined with P–E pairs
In agreement with previous studies (Casseday et al.1994, 2000; Jen and Feng 1999), bicuculline application
Fig. 7 Aa–i Nine echo duration tuning curves of an IC neuronobtained with P–E pairs at diVerent combinations of PD andP–E I. B The discharge patterns of the IC neurons obtained withP–E pairs with (Ba) and without (Bb) the overlap between P andE. The plotted echo duration tuning curves are, respectively,
shown in (Ah, Ai). Note that the neuron had poor ED selectivitywhen P and E overlapped [as shown by smaller nDW, 0.9 in (Ah)than 3.0 in (Ai)]. C Variation in nDW with PD (Ba) and P–E I(Bb). Note that the nDW varied unsystematically with increasingPD and P–E I (see the text for details)
123
J Comp Physiol A
broadened the echo duration tuning curves of IC neu-rons. As shown in Fig. 10, a representative IC neuronhad a band-pass echo duration tuning curve whenplotted with the echo pulses of three P–E pairs beforebicuculline application (Fig. 10Aa–c). Bicucullineapplication produced a greater increase in the numberof impulses for shorter and longer non-BD echo pulsesthan for the BD echo pulse (Fig. 10Aa vs. Ba, Ab vs.Bb, Ac vs. Cc, left ordinate). As a result, all band-passecho duration tuning curves were broadened and
changed into all-pass echo duration tuning curves withdecreasing nDW (Fig. 10A vs. B).
Among 40 neurons studied, bicuculline applicationsigniWcantly increased the BD and decreased the nDWof echo duration tuning curves to varying degree whendetermined with the echo pulses of all three P–E pairs(Fig. 10Ca, Da, Wlled vs. unWlled bars). As such, signiW-cant increase in BD and decrease in nDW with length-ening of pulse duration and P–E gap was only observedbefore but not during bicuculline application (repeated
Fig. 8 Aa–i Nine echo duration tuning curves of an IC neuronobtained with P–E pairs at diVerent combinations of PD and P–Egap. The duration tuning properties, the BD and the nDW are
shown within each plot. B Variation in nDW with PD (Ba) and P–E gap (Bb). Note that the nDW mostly decreases with increasingPD and P–E gap (see the text for details)
Fig. 9 Variation in the BD (Aa, Ba, Ca) and nDW (Ab, Bb, Cb) of echo duration tun-ing curves of IC neurons plot-ted with diVerent combinations of the PD (A 1.5 ms, B 4.0 ms and C 10.0 ms) and P–E gap (0, 1, 2, 4, 6, 8 and 10 ms). The top ab-scissa represents the P–E interval corresponding P–E gap shown in the bottom ab-scissa. Shaded areas indicate overlap between P and E (see the text for details)
123
J Comp Physiol A
measures two-way ANOVA; unWlled bars, P < 0.05 forFig. 10Ca, P < 0.01 for Fig. 10Da; Wlled bars, P > 0.05for Fig. 10Ca, Da). A Student–Newman–Keuls multiplecomparisons post-test showed signiWcant diVerencesbetween each set of the BD and the nDW (**P < 0.01and *P < 0.05). The average percent change of the BDand nDW during bicuculline application signiWcantlydecreased with lengthening of pulse duration and P–Egap (repeated measures one-way ANOVA; P < 0.05 forFig. 10Cb, P < 0.01 for Fig. 10Db). A Student–New-man–Keuls multiple comparisons post-test showed sig-niWcant diVerences between each set of the BD and thenDW (**P < 0.01 and *P < 0.05).
The echo duration tuning curves of 48–58% of these40 IC neurons changed into all-pass during bicucullineapplication. As a result, progressive decrease in the num-ber of band- and short-pass echo duration tuning curvesand increase in the number of all-pass echo duration tun-ing curves with lengthening of pulse duration and P–Egap was only observed before but not during bicucullineapplication (Table 3, predrug vs. bicuculline).
The eVect of GABA application on echo duration selectivity determined with P–E pairs
Contrary to the eVect of bicuculline application, GABAapplication sharpened the echo duration tuning curvesof all IC neurons. Figure 11 shows the echo durationtuning curves of an IC neuron that had band-, short-and all-pass echo duration tuning curves when plottedwith the echo pulses of three P–E pairs before GABA
application (Fig. 11Aa–c). GABA application pro-duced a greater decrease in the number of impulses fornon-BD echo pulses than for BD echo pulse (Fig. 11Aavs. Ba, Ab vs. Bb, Ac vs. Cc, left ordinate). As a result,the neuron’s band-pass echo duration tuning curvebecame even sharper while the short- and all-pass echoduration tuning curves changed into band-pass echoduration tuning curves (Fig. 11Aa–c vs. Ba–c). Theincrease in the neuron’s nDW during GABA applica-tion was greater when tested with longer pulse durationand P–E gap (Fig. 11Ba, 6.25%; Bb, 23.1%; Bc, 61%).
GABA application signiWcantly decreased the BDand increased the nDW of echo duration tuning curvesof 32 IC neurons to varying degree when determinedwith all three P–E pairs (Fig. 11Ca, Da, Wlled vs.unWlled bars). As such, signiWcant increase in BD anddecrease in nDW with lengthening of pulse durationand P–E gap was only observed before but not duringGABA application (repeated measures two-wayANOVA; unWlled bars, P < 0.05 for Fig. 11Ca, P < 0.01for Fig. 11Da; Wlled bars, P > 0.05 for Fig. 11Ca, Da). AStudent–Newman–Keuls multiple comparisons post-test showed signiWcant diVerences between each set ofBDs (**P < 0.01 and *P < 0.05). The average percentchange of the BD and nDW during GABA applicationsigniWcantly increased with lengthening of pulse dura-tion and P–E gap (repeated measures one-wayANOVA; P < 0.05 for Fig. 11Cb, P < 0.01 for Fig. 12d).A Student–Newman–Keuls multiple comparisons post-test showed signiWcant diVerences between each set ofthe BD and the nDW (**P < 0.01).
Fig. 10 A, B Echo duration tuning curves of an IC neuron deter-mined with three P–E pairs before (predrug, A) and during (B)bicuculline application. The PD, P–E gap, BD and nDW areshown within each plot. Ca, Da Bar histograms showing the aver-
age BD and nDW determined with three P–E pairs before (pre-drug, unWlled bars) and during (Wlled bars) bicucullineapplication. Cb, Db Percent change in the BD and the nDW dur-ing bicuculline application (see the text for details)
123
J Comp Physiol A
Among 32 IC neurons studied, GABA applicationchanged most all-pass duration tuning neurons intoshort and band-pass duration tuning neurons. As aresult, noticeable decrease in the number of band- andshort-pass echo duration tuning curves and increase inthe number of all-pass echo duration tuning curveswith lengthening of pulse duration and P–E gap wasonly observed before but not during GABA applica-tion (Table 4, predrug vs. GABA).
Echo duration selectivity, BF and recording depth
Because IC neurons diVered in BF and recording depth,we studied the echo duration selectivity in relation to
BF and recording depth. Regardless of stimulus condi-tions, high BF neurons tend to have long BD and smallnDW than low BF neurons had (Figs. 12, 13). Linearregression analyses of scatter plots of BD and nDW inrelation to BF revealed that the BD signiWcantlyincreased and the nDW decreased with BF (P < 0.01).
Consonant with previous studies (Jen and Schlegel1982; Pinheiro et al. 1991; Poon et al. 1990, Wu and Jen1991), the BF of sequentially isolated neuronsincreased with recording depth (Fig. 14Aa, Ba). Fur-thermore, linear regression analyses of the scatter plotsof BD and nDW in relation to recording depthrevealed that the BD signiWcantly increased and thenDW decreased with recording depth (P < 0.01–0.05;
Table 3 The eVect of bicucul-line application on echo dura-tion tuning curves of 40 IC neurons determined with three P–E pairs
See Table 1 for legends.Underlined values indicate nochange in duration tuningproperties
Fig. 11 A, B Echo duration tuning curves of an IC neuron deter-mined with three P–E pairs before (predrug, A) and during (B)gamma-aminobutyric acid (GABA) application. Ca, Da Bar his-tograms showing the average BD and nDW determined with
three P–E pairs of P–E before (predrug, unWlled bars) and during(Wlled bars) GABA application. Cb, Db Percent change in the BDand the nDW during the GABA application (see the text fordetails)
123
J Comp Physiol A
for simplicity, we only show the scatter plots obtainedwith single echo pulses and echo pulses of one P–Epair). Similar observations were also obtained whenecho duration selectivity of IC neurons was studiedwith other two P–E pairs.
Discussion
Echo duration selectivity of IC neurons determined with single echo pulses and echo pulses of P–E pairs
In the present study, we showed that the durationselective IC neurons had band-, short-, or long-pass
duration tuning curves (Figs. 2, 3, 4, 5, 6). These neu-rons responded maximally to a speciWc duration andshowed high sensitivity to change in the pulse duration.We observed that the echo duration selectivity of ICneurons was sharper when determined with echopulses of P–E pairs than with single echo pulses(Figs. 4, 5, 6). These observations suggest that a bat’secho duration selectivity in the real world is sharperthan what is shown by earlier studies using singlepulses (Casseday et al. 1994, 2000; Ehrlich et al. 1997;Fuzessery and Hall 1999; Galazyuk and Feng, 1997; Jenand Feng 1999; Jen and Schlegel 1982; Pinheiro et al.1991; Zhou and Jen 2001).
A previous study examined the interaction of exci-tation and inhibition in IC neurons using a probe(excitatory pulse) and a masker (inhibitory pulse) (Luand Jen 2002). This study showed that masking ofprobe-elicited responses of IC neurons occurs when amasker is presented within a certain inter-pulse inter-vals (the temporal window) in relation to the probe.Within the temporal window, the strength of this for-ward masking increases with shortening of inter-pulseinterval. Similarly, many studies showed that a neu-ron’s response to a single pulse could be suppressedwhen the single pulse is paired with another pulsewithin a temporal window (Brosch and Schreiner1997; Calford and Semple 1995; Faure et al. 2003;Hocherman and Gilat 1981; Litovs.ky and Yin 1998).As such, neurons show larger responses to singlepulses presented in temporal isolation than to thesame pulse presented in temporally patterned pulsetrains (Moriyama et al. 1994). It is therefore possiblethat this forward neural masking may also account forthe sharper echo duration selectivity of IC neuronsobtained with echo pulses of P–E pairs than with sin-gle echo pulses.
Fig. 12 Scatter plots showing the distribution of BD of IC neu-rons in relation to the BF under four diVerent stimulation condi-tions. The linear regression line is shown with a solid line. N is thenumber of neurons, r the correlation coeYcient and P the signiW-cance level
Table 4 The eVect of GABA application on echo duration tuning curves of 32 IC neurons determined with three P–E pairs
In the present study, the P served as the masker andthe E served as the probe. When determined with theP–E pairs, the forward masking and the recovery prop-erty, which determines a neuron’s ability in response toa succeeding pulse, may be the two predominant fac-tors to shape a neuron’s echo duration selectivity. Theshorter the P–E gap is, the stronger the forward mask-ing becomes to shape the echo duration selectivity.However, our earlier studies in E. fuscus show that therecovery property of IC neurons becomes poor withshortening of P–E gap (Lu et al. 1997; Wu and Jen1998; Zhou and Jen 2003). For this reason, the echo
duration selectivity of IC neurons is shaped by thesetwo opposing forces in relation to the P–E gap. Con-ceivably, there is an optimal P–E gap when forwardmasking and a neuron’s recovery ability complementto each other to shape the neuron’s sharpest echo dura-tion selectivity.
For example, when the P–E gap is within the tempo-ral window of forward masking, a neuron’s echo dura-tion selectivity becomes sharper with shortening of P–E gap because of increasing strength of forward mask-ing (Figs. 6c, 8Bb, 9). However, when the P–E gapbecomes smaller than the optimal P–E gap, theincreasing strength of forward masking would be coun-terbalanced by the increasingly poor recovery ability ofthe IC neuron. As a result, the neuron’s echo durationselectivity becomes poor, in particular when P and Eoverlap (Figs. 7, 9). Conceivably, echo duration selec-tivity determined with overlapping P–E pairs is compa-rable to that determined with single echo pulses. Asshown earlier, echo duration selectivity of IC neuronsis sharper when determined with non-overlapping P–Epairs than with single echo pulses (Figs. 4, 5, 6).
Alternatively, previous studies have shown thatmasking eVect is maximal when the masker and probeoverlap (i.e., simultaneous masking)(Faure et al. 2003;Lu and Jen 2002). It is possible that the simultaneouslymasking during P–E overlap may produce similardegree of decrease in the number of impulses to BDand non-BD echo durations. As such, the simultaneousmasking only lowers but does not change the sharpnessof the echo duration tuning curve.
We observed that when the P–E gap is larger thanthe temporal window, forward masking becomes
Fig. 13 Scatter plots showing the distribution of nDW of IC neu-rons in relation to BF under four diVerent stimulation conditions(see Fig. 12 for legends)
Fig. 14 Scatter plots showing the distribution of BF, BD and nDW of IC neurons in relation to recording depth under diVerent stimulation conditions. Note that the scale is diVerent in (Ac, Bc)
123
J Comp Physiol A
ineVective. For example, echo duration selectivity ofIC neurons changed in a small degree with pulse dura-tion when determined at 8 ms P–E gap or at 10-mspulse duration (Fig. 8Ba, solid triangles; Fig. 9C).These observations suggest that the temporal windowfor forward masking is smaller than 10 ms which is theinter-pulse interval for 1.5 ms pulses at 8 ms P–E gap.Since the inter-pulse intervals for 4 and 10 ms pulses at8 ms P–E gap are 12 and 18 ms, forward masking onecho duration selectivity would be also ineVective(Fig. 8Ba, Wlled triangles). These data are comparableto two previous studies which show that the temporalwindow for forward masking of response properties(i.e., number of impulses, latency, frequency tuningcurves and directional selectivity) of IC neurons of thesame bat species is 6.3–7.0 ms (Lu and Jen 2002; Zhouand Jen 2000).
The role of GABAergic inhibition in shaping echo duration selectivity of IC neurons
The role of GABAergic inhibition in shaping durationselectivity of IC neurons has been reported in previousstudies (Casseday et al. 1994, 2000; Ehrlich et al. 1997;Fuzessery and Hall 1999, Jen and Feng 1999, Wu andJen 2006). In the present study, we plotted the echoduration tuning curves of IC neurons with the P–Epairs before and during application of GABA and itsantagonist, bicuculline (Bormann 1988; Cooper et al.1982). The opposite eVect on the echo duration tuningcurves of IC neurons during both drug applicationsallowed us to double conWrm the role of GABAergicinhibition in shaping the echo duration selectivity of ICneurons (Figs. 10, 11). Our data indicate that GAB-Aergic inhibition sharpens echo duration selectivity ofIC neurons by producing a greater decrease in thenumber of impulses for non-BD echo pulses than forBD echo pulse when stimulated with all three P–Epairs. As such, many duration non-selective IC neuronsbecame duration selective during GABA application(Fig. 11).
Our data also indicate that GABAergic inhibitioncontributes to improving echo duration selectivity ofIC neurons with shortening of pulse duration and P–Egap (Figs. 4–6). This is supported as follows. (1) Thedegree of broadening of echo duration tuning curvesprogressively increased with shortening of pulse dura-tion and P–E gap during bicuculline application(Fig. 10Cb, Db). (2) The degree of sharpening of echoduration tuning curves progressively decreased withshortening of pulse duration and P–E gap duringGABA application (Fig. 11Cb, Db). (3) The number ofband- and short-pass echo duration tuning curves pro-
gressively increases and the number of all-pass echoduration tuning curves decreases with shortening ofPD and P–E gap was only observed before but not dur-ing bicuculline and GABA application (Tables 3, 4).
Because shortening of pulse duration and P–E gap isin essence equivalent to increase pulse repetition rate,all these data are in parallel with our most recent studywhich shows that the improvement of duration selectiv-ity is due to progressively increasing GABAergic inhi-bition with pulse repetition rate (Wu and Jen 2006). Inagreement with our present data, this study showed thatbicuculline application produces more pronouncedbroadening of duration tuning curves at high than atlow pulse repetition rate while GABA application pro-duces more pronounced narrowing of duration tuningcurves at low than at high pulse repetition rate.
A previous study reported that forward masking isbased on the combination of IPSP and after-hyperpo-larization to create a shorter recovery cycle for shorterthan for long duration stimuli such that forward mask-ing is greater for shorter than for long duration stimuli(Eggermont 2000). Based on all these studies and ourpresent data, we suggest that increasing strength ofGABAergic inhibition may be the underlying mecha-nism for increasing forward masking with shortening ofpulse duration and P–E gap to sharpen the echo dura-tion selectivity. On the other hand, we have shown thatthe GBAergic inhibition also shapes the recoveryproperty of IC neurons of this bat species (Lu et al.1997; Zhou and Jen 2003). For this reason, increasingstrength of GABAergic inhibition with shortening ofP–E gap would inevitably deteriorate the recoveryproperty of IC neurons. As such, the sharpest echoduration selectivity is only obtained at the optimal P–Egap as described above.
We observed that the BD became shorter in 40% ofneurons (e.g., Fig. 5) but remained unchanged in 60%neurons (e.g., Fig. 4) with shortening of the pulse andP–E gap. This observation suggests that neural mecha-nism underlying the formation of BD of individual ICneurons is dependent on pulse duration and P–E gap.Future works are needed to determine the diVerence inthe neural mechanisms underlying the formation ofBD of these two types of IC neurons.
Echo duration selectivity, stimulus frequency and recording depth
We showed that IC neurons with short BD and largenDW typically had low BF while IC neurons with longBD and small nDW had high BF (Figs. 12, 13). Fur-thermore, the BD of tonotopically organized IC neu-rons signiWcantly increased and the nDW decreased
123
J Comp Physiol A
with recording depth (Fig. 14). These Wndings suggestthat neurons at upper IC have shorter BD and largernDW than neurons at deep IC have. A previous studyshows that neurons with GABAA receptors are mostlydistributed in the dorso-medial region of the IC but aresparsely distributed in the ventro-lateral region(Fubara et al. 1996). For this reason, high BF neuronsat deeper IC would conceivably receive fewer GAB-Aergic inhibitory inputs than low BF neurons at upperIC. As such, drug application would produce greaterchange in nDW of duration tuning curves of low BFneurons at upper IC than high BF neurons at deeperIC. In sum, our data show that echo duration selectivityof IC neurons shaped by GABAergic inhibitionappears to be systematically organized along the dorso-ventral axis of the IC.
Previous studies show that duration selectivity of ICneurons can also be shaped by glycinergic inhibitionand neurons with glycine receptors are mostly distrib-uted at the ventro-lateral region of the IC but aresparsely distributed at dorso-medial region of the IC(Ehrlich et al. 1997; Fubara et al. 1996). How echoduration selectivity of IC neurons shaped by glycinergicinhibition is organized in the IC remains to be studied.
Behavioral relevance
Consonant with our previous study (Jen and Zhou1999), we found that discharge patterns and echo dura-tion curves obtained with both BF and FM pulses didnot diVer signiWcantly (Fig. 3 and Table 1). Since FMpulses of diVerent duration diVer in frequency sweepdirection and rate, duration selectivity of IC neuronsdetermined with FM pulses represents the result ofinteractions between the sweep rate and variation infrequency within the pulse duration. Our data suggestthat variation in pulse duration rather than frequencysweep rate is the predominant factor in determiningthe duration selectivity of IC neurons.
During hunting, E. fuscus progressively increase therepetition rate, shorten the duration, decrease theamplitude and lower the frequency of emitted pulses asthey search, approach and Wnally intercept the insectsor avoid obstacles (GriYn 1958; Jen and Kamada 1982;Surlykke and Moss 2000). In this study, we determinedecho duration selectivity with P–E pairs mimickingthose occurring at search, approach and terminal phaseof echolocation (Fig. 1). We showed that the echo dura-tion selectivity of bat IC neurons became sharper withshortening of pulse duration and P–E gap but becamepoor when pulse and echo overlap (Figs. 4, 5, 6, 7, 8, 9).These data suggest that shortening pulse duration by E.fuscus during hunting to avoid overlap between pulse
and echo is crucial for accurate echo recognition. Thesedata also suggest that as bats progressively increasepulse emission rate and shorten emitted pulses duringof hunting, echo duration selectivity would improveand therefore facilitate echo recognition.
The BD of IC neurons we studied ranged between1.5 and 10 ms covering the duration of pluses emittedby E. fuscus during three phases of hunting. Weshowed that the BD in 40% of neurons became shorter(e.g., Fig. 5) while the BD in 60% of neurons remainedunchanged (e.g., Fig. 4) with shortening of the pulseand P–E gap. Presumably, IC neurons with unchangedBD can be utilized by bats to encode echoes at eachphase of hunting. Conversely, IC neurons with short-ened BD and improved duration selectivity duringhunting can be utilized by bats to encode the progres-sively shortening echo throughout the entire course ofhunting.
We showed that IC neurons are systematically orga-nized along the dorsoventral axis of the IC based onthe BF, BD and nDW (Figs. 12, 13, 14). Conceivably,low BF neurons with shorter BD and sharper durationselectivity would appear suitable for echo recognitionduring the terminal phase of hunting when the highlyrepetitive echoes are low in frequency and short induration. Conversely, high BF neurons with long BDwould be suitable for echo recognition during searchphase of hunting when the returning echoes are high infrequency and long in duration.
Acknowledgments We thank two anonymous reviewers forcommenting on an earlier version of this manuscript. This workwas supported by a grant and fellowship from the GraduateSchool, the Division of Biological Sciences and College and Artsand Sciences of University of Missouri-Columbia.
References
Bormann J (1988) Electrophysiology of GABAA and GABABreceptor subtypes. Trends Neurosci 11:112–116
Brosch M, Schreiner CE (1997) Time course of forward maskingtuning curves in cat primary auditory cortex. J Neurophysiol77:923–943
Casseday JH, Covey E (1995) Mechanisms for analysis of audi-tory temporal patterns in the brainstem of echolocating bats.In: Covey E, Hawkins HL, Port RF (eds) Neural representa-tion of temporal patterns. Plenum, New York pp 25–51
Casseday JH, Ehrlich D, Covey E (2000) Neural measurement ofsound duration: control by excitatory–inhibitory interactionin the inferior colliculus. J Neurophysiol 84:1475–1487
Casseday JH, Ehrlich D, Covey E (1994) Neural tuning for soundduration: role of inhibitory mechanisms in the inferior colli-culus. Science 264:847–850
Cooper JR, Bloom FE, Roth RH (1982) The biomedical basis ofneuropharmacology. Oxford University Press, New York
123
J Comp Physiol A
Covey E, Casseday JH (1999) Timing in the auditory system ofthe bat. Annu Rev Physiol 61:457–476
Ehrlich D, Casseday JH, Covey E (1997) Neural tuning to soundduration in the inferior colliculus of the big brown bat, Eptes-icus fuscus. J Neurophysiol 77:2360–2372
Faingold CL, Boersma-Anderson CA, Caspary DM (1991)Involvement of GABA in acoustically-evoked inhibition ininferior colliculus neurons. Hear Res 52:201–216
Faure PA, Fremouw T, Casseday JH, Covey E (2003) Temporalmasking reveals properties of sound-evoked inhibition induration selective neurons of the inferior colliculus. J Neuro-sci 23:3052–3065
Fremouw T, Faure PA, Casseday JH, Covey E (2005) Durationselectivity of neurons in the inferior colliculus of the bigbrown bat: tolerance to changes in sound level. J Neurophys-iol 94(3):1869–1878
Fubara BM, Casseday JH, Covey E, Schwartz-Bloom RD (1996)Distribution of GABAA, GABAB and glycine receptors inthe central auditory system of the big brown bat, Eptesicusfuscus. J Comp Neurol 369:83–92
Fuzessery ZM, Hall JC (1999) Sound duration selectivity in thepallid bat inferior colliculus. Hear Res 137:137–154
Galazyuk AV, Feng AS (1997) Encoding of sound duration byneurons in the auditory cortex of the little brown bat, Myotislucifugus. J Comp Physiol A 180:301–311
GriYn DR (1958) Listening in the dark. Yale University Press,New Haven, Conn (Reprinted by Comstock Publ Assoc,Cornell University Press 1986)
Herbert H, AschoV A, Ostwald J (1991) Topography of projec-tions from the auditory cortex to the inferior colliculus in therat. J Comp Neurol 304:103–122
Hocherman S, Gilat E (1981) Dependence of auditory evokedunit activity on interstimulus interval in the cat. J Neurophys-iol 45:987–997
HuVman RF, Henson OW (1990) The descending auditory path-ways and acoustico motor systems: connections with the infe-rior colliculus. Brain Res Rev 15:295–323
Jen PHS, Feng RB (1999) Bicuculline application aVects dischargepattern and pulse-duration tuning characteristics of bat infe-rior collicular neurons. J Comp Physiol A 184:185–194
Jen PHS, Kamada T (1982) Analysis of orientation signals emit-ted by the CF-FM bat, Pteronotus parnellii parnellii and theFM bat, Eptesicus fuscus during avoidance of moving andstationary obstacles. J Comp Physiol A 148:389–398
Jen PHS, Schlegel P (1982) Auditory physiological properties ofthe neurons in the inferior colliculus of the big brown bat,Eptesicus fuscus. J Comp Physiol 147:351 364
Jen PHS, Zhou XM (1999) Temporally patterned sound trainsaVect duration tuning characteristics of bat inferior collicularneurons. J Comp Physiol A 185:471–478
Jen PHS, Zhang JP (2000) The role of GABAergic inhibition ondirection-dependent sharpening of frequency tuning in batinferior collicular neurons. Brain Res 862:127–137
Jen PHS, Sun XD, Chen D, Teng HB (1987) Auditory space rep-resentation in the inferior colliculus of the FM bat, Eptesicusfuscus. Brain Res 419:7–18
Jen PHS, Wu CH, Luan RH, Zhou XM (2002) GABAergic inhi-bition contributes to pulse repetition rate-dependent fre-quency selectivity in inferior colliculus of the big brown bat,Eptesicus fuscus. Brain Res 948:159–164
Klug A, Park TJ, Pollak GD (1995) Glycine and GABA inXuencebinaural processing in the inferior colliculus of the mustachebat. J Neurophysiol 74:1701–1713
Koch U, Grothe B (1998) GABAergic and glycinergic inhibitionsharpens tuning for frequency modulations in the inferiorcolliculus of the big brown bat. J Neurophysiol 80:71–82
LeBeau FE, Malmierca MS, Rees A (2001) Iontophoresis in vivodemonstrates a key role for GABAA and glycinergic inhibi-tion in shaping frequency response areas in the inferior colli-culus of guinea pig. J Neurosci 21:7303–7312
Litovsky RY, Yin TC (1998) Physiological studies of the prece-dence eVect in the inferior colliculus of the cat. I. Correlatesof psychophysics. J Neurophysiol 80:1285–1301
Lu Y, Jen P HS (2001) GABAergic and glycinergic neural inhibi-tion in excitatory frequency tuning of bat inferior collicularneurons. Exp Brain Res 141:331–339
Lu Y, Jen PHS (2002) Interaction of excitation and inhibition ininferior collicular neurons of the big brown bat, Eptesicusfuscus. Hearing Res 169:140–150
Lu Y, Jen PHS (2003) Binaural interaction in the inferior colliculusof the big brown bat, Eptesicus fuscus. Hearing Res 177:100–110
Lu Y, Jen PHS, Zheng QY (1997) GABAergic disinhibitionchanges the recovery cycle of bat inferior colliculus neurons.J Comp Physiol A 181:331–341
Lu Y, Jen PHS, Wu M (1998) GABAergic disinhibition aVects re-sponses of bat inferior collicular neurons to temporally pat-terned sound pulses. J Neurophysiol 79:2303–2315
Moriyama T, Hou T, Wu M, Jen PHS (1994) Responses of inferiorcollicular neurons of the FM bat, Eptesicus fuscus, to pulsetrains with varied pulse amplitudes. Hearing Res 79:105–114
Park TJ, Pollak GD (1993) GABA shapes sensitivity to interauralintensity disparities in the mustache bat’s inferior colliculus:implications for encoding sound location. J Neurosci13:2050–2067
Pinheiro AD, Wu M, Jen PHS (1991) Encoding repetition rateand duration in the inferior colliculus of the big brown bat,Eptesicus fuscus. J Comp Physiol A 169:69–85
Poon PW, Sun XD, Kamada T, Jen PHS (1990) Frequency andspace representation in the inferior colliculus of the FM bat,Eptesicus fuscus. Exp Brain Res 79:83–91
Popper AN, Fay RR (eds) (1995) Hearing by bats. Springer, Ber-lin Heidelberg New York
Roberts RC, Ribak CE (1987) An electron microscopic study ofGABAergic neurons and terminals in the central nucleus ofthe inferior colliculus of the cat. J Neurocytol 16:333–345
Saldana E, Feliciano M, Mugnaini E (1996) Distribution ofdescending projections from primary auditory neocortex toinferior colliculus mimics the topography of intracollicularprojections. J Comp Neurol 371:15–40
Shannon RV, Zeng FG, Kamath V, Wygonski J, Ekelid M (1995)Speech recognition with primary temporal cues. Science270:303–304
Suga N, O’Neill WE, Kujirai K, Manabe T (1983) SpeciWcity of“combination sensitive” neurons for processing complexbiosonar signals in the auditory cortex of the mustached bat.J Neurophysiol 58:643–654
Surlykke A, Moss CF (2000) Echolocation behavior of big brownbat, Eptesicus fuscus, in the Weld and the laboratory. J Acous-tic Soc Am 108:2419–2429
Tanaka H, Wong D, Taniguchi I (1992) The inXuence of stimulusduration on the delay tuning of cortical neurons in the FMbat, Myotis lucifugus. J Comp Physiol A 171:29–40
Vater M, Habbicht H, Kossl M, Grothe B (1992) The functionalrole of GABA and glycine in monaural and binaural process-ing in the inferior colliculus of horseshoe bats. J Comp Phys-iol 171:541–553
Winer JA, Larue DT, Diehl JJ, Hefti BJ (1998) Auditory corticalprojections to the cat inferior colliculus. J Comp Neurol400:147–174
123
J Comp Physiol A
Wong D, Maekawa M. Tanaka H, Taniguchi I (1992) The eVectof pulse repetition rate on delay sensitivity of neurons in theauditory cortex of the FM bat, Myotis lucifugus. J CompPhysiol A 170:393–402
Wu CH, Jen PHS (2006) The role of GABAergic inhibition inshaping duration selectivity of bat inferior collicular neuronsdetermined with temporally patterned sound trains. HearingRes (in press)
Wu M, Jen PHS (1991) Encoding of acoustic stimulus intensity byinferior collicular neurons of the big brown bat, Eptesicusfuscus. Chin J Physiol 34:145–155
Wu M, Jen PHS (1998) The recovery properties of neurons in theinferior colliculus, auditory cortex and the pontine nuclei ofthe big brown bat, Eptesicus fuscus. Chin J Physiol 41(1):1–8
Yang L, Pollak G.D, Resler C (1992) GABAergic circuits sharp-en tuning curves and modify response properties in themustache bat inferior colliculus. J Neurophysiol 68:1760–1774
Zhou XM, Jen PHS (2000) Neural inhibition sharpens auditoryspatial sensitivity of bat inferior collicular neurons. J CompPhysiol 186:389–398
Zhou XM, Jen PHS (2001) The eVect of sound intensity on dura-tion-tuning characteristics on bat inferior collicular neurons.J Comp Physiol A 187:63–73
Zhou XM, Jen PHS (2003) The eVect of bicuculline applicationon azimuth-dependent recovery cycle of inferior collicularneurons of the big brown bat, Eptesicus fuscus. Brain Res973:131–141
