www.elsevier.com/locate/brainres

Brain Research 1019 (2004) 281–288

Short communication

Azimuth-dependent recovery cycle affects directional selectivity

of bat inferior collicular neurons determined with

sound pulses within a pulse train

Xiaoming Zhou, Philip H.-S. Jen*

Division of Biological Sciences and Interdisciplinary Neuroscience Program, University of Missouri-Columbia, 208 Lefevre Hall,

Columbia, MO 65211-6190, USA

Accepted 4 June 2004

Available online 17 July 2004

Abstract

In our previous study, we have shown that the recovery cycle of most neurons in the inferior colliculus (IC) of the big brown bat,

Eptesicus fuscus, is typically longer at ipsilateral azimuth than at contralateral azimuth under free-field stimulation conditions. The present

study is to test the hypothesis that this azimuth-dependent recovery cycle may contribute to the variation of directional selectivity of IC

neurons with sequential presented sound pulses within a pulse train. A 300-ms pulse train containing nine sound pulses of 4-ms with an inter-

pulse interval of 33.3 ms was delivered at several selected azimuthal angles between F 80j lateral in the frontal auditory space of a bat. A

family of nine directional selectivity curves was plotted with a neuron’s number of impulses in response to each individual pulse against the

azimuthal angles. The type and sharpness of these directional selectivity curves were then compared in relation to pulse position within the

pulse train. All 675 directional selectivity curves obtained from 75 IC neurons could be described as directionally selective (423, 63%),

hemifield (220, 32%), or non-directional (32, 5%). The directional selectivity curves of 45 (60%) neurons did not vary with pulse position.

However, those of the remaining neurons (30, 40%) changed from one type to another such that the number of neurons with directionally

selective curves progressively increased and the number of neurons with hemifield and non-directional selectivity curves decreased with

increasing pulse position within the pulse train. Among 68 IC neurons whose directional selectivity curves were compared quantitatively,

directional selectivity determined with sequentially presented sound pulses significantly increased in 38 (56%) neurons; decreased in 18

(26%) neurons but did not change in 12 (18%) neurons. This change of directional selectivity was due to the variation in recovery cycle of

these IC neurons with azimuthal angle as we hypothesized.

D 2004 Elsevier B.V. All rights reserved.

Theme: Sensory systems

Topic: Auditory systems: central physiology

Keywords: Bat; Directional selectivity; Inferior colliculus; Inter-pulse interval; Recovery cycle

Previous studies suggested that the directional selectiv- that most neurons in the inferior colliculus (IC) of bats

ity of auditory neurons is one of the essential response

properties underlying detection and localization of prey by

insectivorous bats during hunting. These studies showed

0006-8993/$ - see front matter D 2004 Elsevier B.V. All rights reserved.

doi:10.1016/j.brainres.2004.06.004

Abbreviations: AR, angular range; BA, best angle; BF, best frequency;

IC, inferior colliculus; MT, minimum threshold; nAR, normalized angular

range; PRR, pulse repetition rate; PST, peri-stimulus time; SPL, sound

pressure level

* Corresponding author. Tel.: +1-573-882-7479; fax: +1-573-884-

5020.

E-mail address: jenp@missouri.edu (P.H.-S. Jen).

discharged maximally or displayed lowest threshold to

sounds delivered from an angle or a range of angles and

their response magnitude changed significantly with a 1–

2j shift in sound direction along the midline of the

frontal auditory space [2,5–8,11,13,19,20,24–26]. These

studies also showed that the directional selectivity of IC

neurons could be influenced by frequency, intensity, and

type of sound stimulus. For example, the sharpness of

directional selectivity curves of most IC neurons increased

with pulse frequency [2,11,13,20]. The shape of direc-

tional selectivity curve and the angle of maximal sensi-

X. Zhou, P.H.-S. Jen / Brain Research 1019 (2004) 281–288282

tivity of IC neurons also varied with pulse intensity

[7,20].

We reported previously that the directional selectivity of

IC neurons in terms of spatial receptive field significantly

increased when a sound was paired with another one that

was delivered within a temporal window and with frequen-

cy–intensity combinations within the neuron’s inhibitory

area [25]. Furthermore, the directional selectivity of IC

neurons improved with increasing pulse repetition rate

(PRR) of temporally patterned pulse trains [24,26]. This

improvement of directional selectivity was mostly due to a

selective decline of response at the two sides of the best

angle (BA) at high PRRs [26]. These studies indicate that

the temporal sequence of sound pulses may influence the

directional selectivity of IC neurons. Thus, a neuron’s

directional selectivity determined with single pulses in

temporal isolation might not predict well its response

property in the real world where sound pulses occur in

more complex temporal conditions.

Our recent study showed that the recovery cycle of most

IC neurons of the big brown bat, Eptesicus fuscus, was

longer when determined with paired sound pulses delivered

at ipsilateral than at contralateral azimuthal angles. Presum-

ably, this azimuth-dependent recovery cycle was mainly due

to the variation in binaural interaction with sound azimuth,

although the effects of head shadowing were not disen-

tangled due to the free-field stimulus conditions [27].

Because a neuron’s recovery cycle determines its ability to

respond to closely repetitive sound pulses, we hypothesize

that this azimuth-dependent recovery cycle may contribute

to variation of directional selectivity of IC neurons with

sequentially presented sound pulses within a pulse train.

To test this hypothesis, we plotted a family of directional

selectivity curves with a neuron’s number of impulses in

response to each of nine sequentially presented sound pulses

against the azimuthal angles. We also determined the

variation of recovery cycle of these neurons with azimuthal

angle. We then compared and correlated the type and

sharpness of these directional selectivity curves in relation

to pulse position within the pulse train as well as their

azimuth-dependent recovery cycle.

Ten E. fuscus (18–22 g, body weight, b.w.) were used for

this study. Surgical procedures and the experimental setup

were basically the same as in previous studies [13,26,27].

Briefly, 1 or 2 days before the recording session, a 1.8-cm

nail was glued onto the exposed skull of Nembutal-anes-

thetized (45–50 mg/kg b.w.) bat with acrylic glue and

dental cement. During recording session, the bat was ad-

ministered the neuroleptanalgesic Innovar-Vet (Fentanyl

0.08 mg/kg b.w., Droperidol 4 mg/kg b.w.). The bat was

then placed inside a bat holder (made of wire mesh) that was

suspended in an elastic sling inside a double-wall sound-

proof room (Industrial Acoustics Company, temperature

28–30 jC). The ceiling and inside walls of the room were

covered with 3-in. convoluted polyurethane foam to reduce

echoes. After immobilizing the bat’s head by fixing the

shank of the nail into a mental rod with a set screw, its eye–

snout line was then pointed visually to 0j in azimuth and 0jin elevation of the frontal auditory space by adjusting the

mental rod. Small holes were bored in the skull above the IC

for insertion of 3 M KCl glass pipette electrodes (imped-

ance: 5–10 MV) to record auditory responses of IC neu-

rons. Additional doses of Innovar-Vet were administered

during later phases of recording when the bat became

restless. Each bat was used in one to five recording sessions

on separate days and each recording session typically lasts

for 4–6 h. These procedures were conducted in compliance

with NIH publication No. 85-23, ‘‘Principles of Laboratory

Animal Care’’ and with the approval of the Institutional

Animal Care and Use Committee (#1438) of the University

of Missouri-Columbia.

Acoustic stimuli (4 ms with 0.5 ms rise–decay times)

were generated with an oscillator (KH model 1200) and a

homemade electronic switch driven by a stimulator (Grass

S88). These stimuli were then amplified after passing

through a decade attenuator (HP 350D) before they were

fed to a small condenser loudspeaker (AKG model CK 50,

1.5 cm diameter, 1.2 g) that was placed 23 cm away from

the bat. The loudspeaker could be positioned at 0j elevation

and in any specific azimuth within F 90j in the frontal

auditory space through a remote control system driven by

two small electric motors. To avoid potential influence of

noises generated by the motors on response of a recorded

neuron, the motors were always turned off after positioning

the loudspeaker at each selected azimuthal angle. Calibra-

tion of the loudspeaker was performed with a 1/4 in.

microphone (B&K 4135) placed at the position of the bat’s

head during recording using a measuring amplifier (B&K

2607). The output of the loudspeaker was expressed in

decibels of sound pressure level (SPL) in reference to 20

APa root mean square.

Upon isolation of an IC neuron with 4 ms sound pulses

delivered at a rate of two per second from contralateral 40jrelative to the recording site, its best frequency (BF) and

minimum threshold (MT) were determined by changing the

frequency and intensity of sound pulses. At the MT, the

probability of the neuron’s response to BF sound pulses was

about 50%. The directional selectivity of an IC neuron was

studied with 32 presentations of a 300-ms pulse train

containing 4-ms BF sound pulses delivered at two trains

per second. Thus there was always a 200-ms silent period

between pulse trains. The onset-to-onset inter-pulse interval

within the pulse train was set at 33.3 ms (i.e., the number of

pulses was nine within the pulse train) and the intensity of

pulses was set at 20 dB above the neuron’s MT. The number

of impulses of a neuron discharged to pulse train at nine

selected azimuthal angles (contralateral 20j, 40j, 60j, 80j;ipsilateral 20j, 40j, 60j, 80j relative to the recording site,

abbreviated as c20j, c40j, c60j, c80j; or i20j, i40j, i60j,i80j, and the front, e.g., 0j) was recorded. A family of

directional selectivity curves of each neuron was then

plotted using the number of impulses in response to each

X. Zhou, P.H.-S. Jen / Brain Research 1019 (2004) 281–288 283

of nine sequentially presented sound pulses against these

azimuthal angles. The pulse train was presented randomly

from different azimuthal angles to avoid the potential effect

of sequential presentation of pulse train at selected azimuth-

al angles on a neuron’s response. A neuron’s BA was

defined as the angle to which the number of impulses was

maximal and at least 50% greater than that obtained at other

azimuthal angles. The sharpness of a directional selectivity

curve was expressed by half of the angular range (abbrevi-

ated as AR, see Fig. 1Ba,b) at 50% below the maximum of

each directional selectivity curve. A normalized AR (abbre-

viated as nAR) was then obtained by dividing the maximum

by the AR. The ARs and nARs of directional selectivity

curves of each IC neuron were compared statistically across

each of the nine pulses within pulse trains using repeated

measures one-way ANOVA at p < 0.01.

Fig. 1. A: Peri-stimulus time (PST) histograms of three representative IC

neurons determined with the first pulse within a pulse train at selected

azimuthal angles (shown at the top). The number of impulses discharged to

each sound pulse is shown within each PST. B: The directional selectivity

curves of these three neurons that are directionally selective (a), hemifield

(b), or non-directional (c). The sharpness of the directionally selective and

hemifield curves is expressed by half of the angular range at 50% below

maximal response (shown by AR in a, b). A normalized angular range

(nAR) is then calculated by dividing the number of impulses obtained at the

best angle (BA, indicated by unfilled arrow in a) by AR. The non-

directional curve does not have an AR because the number of impulses does

not vary more than 50% for all azimuthal angles tested. The BF (kHz) and

MT (dB SPL) of these three neurons were 23.6, 63 (a); 35.9, 48 (b); 36.4,

63 (c). C: Histogram distribution of three types of directional selectivity

curves determined for each of nine sequentially presented sound pulses.

The ordinate and abscissa represent number of neurons and pulse position

within the pulse train. Note that the number of neurons with directionally

selective curves (unfilled) progressively increased and the number of

neurons with hemifield (shaded) and non-directional (filled) selectivity

curves decreased with increasing pulse position. N, number of neurons (see

text for details).

The recovery cycle of a neuron was studied by a pair of

identical BF sound pulses with different inter-pulse intervals

delivered at three azimuthal angles: the BA, contralateral

40j, and ipsilateral 40j from the BA relative to the neuron’s

recording site (conveniently abbreviated as c40j re BA and

i40j re BA). The intensity of BF sound pulses was set at 20

dB above each neuron’s MT. The percentage recovery was

obtained by dividing the neuron’s response to the second

pulse by the response to the first pulse. The percentage of

recovery at each inter-pulse interval was used to plot the

neuron’s recovery cycle. The 50% recovery times of all

neurons obtained at three azimuthal angles were statistically

compared.

Recorded action potentials were amplified, band-pass

filtered (Krohn-Hite 3500), and fed through a window

discriminator (WPI 121) before being sent to an oscillo-

scope (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 As, sampling period: 300 ms) to 32 train

presentations. During data analysis, the sampling window

was adjusted to include only the part of the PST histogram

that was generated by one pulse to obtain a neuron’s number

of impulses discharged to each pulse. This adjustment of

sampling window allowed us to easily exclude the sponta-

neous activity. The number of impulses discharged to each

pulse was then used to plot a neuron’s directional selectivity

curve or recovery cycle.

A total of 89 IC neurons was isolated in this study. Their

BFs ranged from 17.2 to 68.2 kHz encompassing the

frequency range of the FM sweep emitted by E. fuscus

during hunting [21]. Their MTs were between 23 and 75 dB

SPL with a modal value of 49 dB SPL. These neurons were

tonotopically organized along the dorsoventral axis of the

IC such that BF progressively increased with recording

depth. When stimulated with a pulse train, most (n = 75,

84%) neurons discharged impulses to each individual pulse.

The remaining (n= 14, 16%) neurons discharged impulses

to the initial few pulses or the first pulse only. These

observations are in agreement with previous studies

[1,10,14,16,17,26]. For consistent comparison, we only

studied the variation of directional selectivity with sequen-

tial presented sound pulses within a pulse train in those 75

neurons that responded to each individual pulse.

As described earlier, a directional selectivity curves was

plotted with a neuron’s number of impulses in response to

each individual pulse against the azimuthal angles (Fig. 1A).

All 675 directional selectivity curves plotted with the

number of impulses elicited by sequentially presented sound

pulses at selected azimuthal angles can be described as

directionally selective, hemifield, or non-directional. A

directionally selective curve has a clear peak at a certain

angle which is larger than the minimum by at least 50%

(Fig. 1Ba). A hemifield selectivity curve rises from an

ipsilateral angle by more than 50% and either reaches a

plateau or declines by less than 50% over a wide range of

X. Zhou, P.H.-S. Jen / Brain Research 1019 (2004) 281–288284

contralateral angles (Fig. 1Bb). A non-directional selectivity

curve does not show any clear maximum at a particular

angle and the number of impulses obtained from all angles

tested never differs by more than 50% (Fig. 1Bc).

Fig. 1C shows the distribution of these three types of

directional selectivity curves of 75 neurons obtained with

sequentially presented sound pulses. Directional selectivity

curves of 45 (60%) neurons did not change with pulse

position, including 25 directionally selective, 17 hemifield,

and 3 non-directional curves. Directional selectivity curves

of the remaining neurons (30, 40%) changed from one type

to another such that the number of neurons with direction-

ally selective curves progressively increased and the number

of neurons with hemifield and non-directional selectivity

curves decreased with increasing pulse position.

The top panels of Fig. 2 shows the PST histograms of

three representative IC neurons (neurons A, B and C)

Fig. 2. Aa,b,c,Ba,b,c,Ca,b,c: PST histograms of three representative IC neurons d

angles (shown at the right). The number of impulses discharged to each sound puls

curves for each neuron plotted with the number of impulses in response to sequenti

the AR (left ordinate) and nAR (right ordinate) of each family of directional select

definition of AR and nAR). The BF (kHz) and MT (dB SPL) of these three neur

obtained with sequentially presented sound pulses at se-

lected azimuthal angles. All three neurons discharged

impulses to each pulse delivered at c20j. The number of

impulses discharged to each pulse hardly varied in neurons

A and C but progressively decreased in neuron B (Fig.

2Ab,Cb vs Bb). When sound pulses were delivered at i20jand c60j, neuron C still discharged impulses to each pulse

(Fig. 2Ca,c). Neuron A also discharged impulses to each

pulse at c60j but only to the initial three pulses at i20j(Fig. 2Aa,c). Neuron B only discharged impulses to the

initial three pulses at c60j but only to the first pulse at

i20j (Fig. 2Ba,c).

Because of different variations in response to sequen-

tially presented sound pulses with azimuthal angle, varia-

tion in each family of nine directional selectivity curves

with pulse position was different among these three

neurons (Fig. 2Ad,Bd,Cd). The directional selectivity

etermined with 9 sequentially presented sound pulses at selected azimuthal

e is shown within each PST. Ad,Bd,Cd: The family of directional selectivity

ally presented sound pulses against azimuthal angles. Ae,Be,Ce: Variation in

ivity curves with sequentially presented sound pulses (see Fig. 1 legend for

ons were 35.9, 48 (A); 23.6, 63 (B); 35.0, 33 (C).

X. Zhou, P.H.-S. Jen / Brain Research 1019 (2004) 281–288 285

curve of neuron A changed from hemifield to the first

pulse to directionally selective to the remaining pulses

(Fig. 2Ad). The AR of directional selectivity curves

decreased drastically for the first four sound pulses and

then hardly changed for the following presented pulses

(Fig. 2Ae, filled circles). The decrease of AR was mainly

due to a selective drop in response at two sides of the BA.

Neuron B always had a directionally selective curve to

each pulse and the AR also progressively decreased with

increasing pulse position (Fig. 2Bd,Be, filled circles).

However, the decrease of AR was due to a decline in

response over the entire range of azimuthal angles includ-

ing the BA. Unlike neurons A and B, the shape and AR of

directional selectivity curves of neuron C hardly changed

with increasing pulse position (Fig. 2Cd,Ce, filled circles).

This little change in AR was due to the minor change in

response magnitude to sequentially presented sound pulses

at each azimuthal angle (Fig. 2Ca,b,c).

Because the decrease of AR with pulse position in

individual neurons (such as neurons A and B) may be due

to different ways of decline in response to sound pulses

presented at each azimuthal angle, we divided a neuron’s

maximal number of impulses to each pulse by the AR to

obtain an nAR. This normalized nAR allows us to determine

if variation in the sharpness of directional selectivity curve

of a neuron with pulse position within a pulse train is mainly

due to a change in directional selectivity or simply due to a

decrease in response over a wide range of azimuthal angles

(e.g., Fig. 2A d vs. Bd).

As shown in the bottom panel of Fig. 2, the nAR of

neuron A progressively increased with sequentially pre-

sented sound pulses indicating an increase in directional

selectivity (Fig. 2Ae, unfilled circles). Conversely, the nAR

of neuron B progressively decreased with increasing pulse

position indicating a decrease in directional selectivity as a

result of overall decline in response magnitude (Fig. 2Be,

Fig. 3. The variation of average AR and nAR of three groups of IC neurons determ

half of a standard deviation. n, number of neurons.

unfilled circles). The nAR of neuron C hardly varied with

increasing pulse position indicating little change in direc-

tional selectivity with sound position (Fig. 2Ce, unfilled

circles).

In total, the AR and nAR were obtained for 68 neurons

because the other 7 had non-directional selectivity curves.

As shown in Fig. 3, these neurons could be divided into

three groups according to the variation of AR and nAR with

sequentially presented sound pulses. The average AR and

nAR of group I (38, 56%) neurons (like neuron A of Fig. 2)

varied significantly in opposite ways with increasing pulse

position (Fig. 3Aa,b; one-way ANOVA, p < 0.0001). This

significant variation in the AR or nAR was due to the large

differences between the average values of AR or nAR

determined with the initial two or three pulses and with

the remaining ones. Both the AR and nAR of group II (18,

26%) neurons (like neuron B of Fig. 2) significantly

decreased with increasing pulse position (Fig. 3Ba,b; one-

way ANOVA, p < 0.0001). The AR and nAR of group III

(12, 18%) neurons (like neuron C of Fig. 2) hardly varied

with sequentially presented sound pulses (Fig. 3Ca,b; one-

way ANOVA, p>0.1).

To determine whether the variation of AR and nAR with

sequentially presented sound pulses might be related to the

BF or MT, we compared the average BF and MT of these

three groups of IC neurons. As shown in Table 1, these three

groups of IC neurons did not differ significantly in BF and

MT (Table 1; one-way ANOVA, p>0.1).

The recovery cycles of three neurons of Fig. 2 are shown

in the top panel of Fig. 4. Neuron A had a very fast recovery

cycle when determined at the BA (Fig. 4Aa, filled circles).

Its recovery cycle became longer when determined at two

lateral azimuthal angles, in particular at i40j re BA (Fig.

4Aa, unfilled circles and filled squares). Neuron B also had

a shorter recovery cycle at the BA than at two lateral

azimuthal angles (Fig. 4Ba). However, neuron B had a

ined with sequentially presented sound pulses. Each vertical bar represents

Table 1

The range and average BF and MT of three groups of IC neurons

Group BF (kHz) MT (dB SPL)

I (n = 38) Range 18.2 – 68.2 23 – 68

MeanF S.D. 36.8F 11.5 47F 11

II (n = 18) Range 23.6 – 58.1 23 – 75

MeanF S.D. 34.9F 8.9 52F 13

III (n = 12) Range 25.9 – 58.6 29 – 68

MeanF S.D. 39.3F 10.6 48F 13

p (one–way ANOVA) >0.1 >0.1

n, number of neurons.

X. Zhou, P.H.-S. Jen / Brain Research 1019 (2004) 281–288286

longer recovery cycle at the BA than neuron A had (Fig.

4Ba vs. Aa, filled circles). Different from neurons A and B,

neuron C had a fast recovery cycle at all three azimuthal

angles (Fig. 4Ca).

As shown in the inset, the 50% recovery time was short

at the BA and long at two lateral azimuthal angles for

neurons A and B but hardly changed at three azimuthal

angles for neuron C (Fig. 4Aa,Ba,Ca, inset). However,

neuron A had a far shorter 50% recovery time at the BA

than neuron B had. Fig. 4Ab, Bb and Cb show the average

50% recovery time of three groups of neurons (as repre-

sented by neurons A, B and C of Fig. 2) determined at

different azimuthal angles. The average 50% recovery time

of group I neurons increased significantly from 15F 5.5 ms

at BA to 54.4F 10.6 and 40.5F 9.3 ms at i40j and c40j re

BA (Fig. 4Ab; one-way ANOVA, p < 0.0001). The average

50% recovery time of group II neurons also increased from

27.3F 4.4 ms at BA to 49.0F 8.3 ms and 37.6F 7.7 ms at

i40j and c40j re BA (Fig. 4Bb; one-way ANOVA,

Fig. 4. Aa,Ba,Ca: Recovery cycles of the three neurons of Fig. 2 determined at thre

indicated by a horizontal dashed line. Each inset shows the each neuron’s 50% rec

average 50% recovery time of three groups of IC neurons determined at three azim

interval of sequentially presented sound pulses (see Fig. 3 for legends).

p < 0.0001). The average 50% recovery time of group III

neurons did not differ significantly at three azimuthal angles

(10.9F 5.2 ms at BA, 13.7F 4.3 and 12.4F 4.9 ms at i40jand c40j re BA) (Fig. 4Cb; one-way ANOVA, p>0.01).

In this study, we showed that the directional selectivity

of most IC neurons varied with sequentially presented

sound pulses within a pulse train. Specifically, the number

of neurons with directionally selective curves progressive-

ly increased while the number of neurons with hemifield

and non-directional selectivity curves decreased with se-

quential sound pulses within the pulse train (Fig. 1C).

Also, the AR and nAR of most (82%) neurons varied with

sequential sound pulses (Fig. 3A,B). This study together

with the previous ones [3,12,14,15,23–26] indicate that a

neuron’s response in multi-parametric domains to sound

stimuli in isolation cannot always predict well how it

would respond to naturally occurring sound signals in

rapid succession.

We showed that the nAR determined with sequentially

presented sound pulses significantly increased in 56% of

neurons, decreased in 26% of neurons but hardly changed in

18% of neurons. This different variation of nAR with

increasing pulse position is due to different change in

response magnitude with sequentially presented sound

pulses. As described earlier, a selective drop in response

at two sides of the BA to sequentially presented sound

pulses would result in an increase of nAR while an overall

decline in response over a wide range of azimuthal angles

would result in a decrease of nAR with increasing pulse

position (Fig. 2A,B). A little change in response to all

e azimuthal angles (BA, c40j re BA, and i40j re BA). The 50% recovery is

overy time (50% RT) determined at three azimuthal angles. Ab,Bb,Cb: The

uthal angles. Each horizontal dashed line indicates the 33.3 ms inter–pulse

X. Zhou, P.H.-S. Jen / Brain Research 1019 (2004) 281–288 287

presented sound pulses would hardly change the nAR value

(Fig. 2C). This different variation of nAR in three groups of

IC neurons is due to their different recovery cycles.

For example, the average 50% recovery time of the first

group of neurons was 15F 5.5 ms at BA but significantly

increased more than 33.3 ms at two lateral azimuthal angles

(Fig. 4Ab). For this reason, the response magnitude of this

group of neurons to sequentially presented sound pulses

would recover fully at the BA but not at two lateral sides of

the BA (Fig. 2Aa,b,c). Therefore, response magnitude

would only selectively drop at the two sides of the BA

but not at the BA (Fig. 2Ad). As a result, the AR and nAR

of this group of IC neurons changed oppositely with

increasing pulse position (Figs. 2Ae and 3Aa,b). The

average 50% recovery time of the second group of neurons

was 27.3F 4.4 ms at BA but was larger than 33.3 ms at two

lateral azimuthal angles (Fig. 4Bb). Thus the response

magnitude of this group of neurons would not fully recover

not only at the BA but also even less at two lateral azimuthal

angles (Fig. 2Ba,b,c). For this reason, the response magni-

tude of this group of neurons would progressively drop at all

azimuthal angles (Fig. 2Bd). As a result, the AR and nAR of

this group of IC neurons both significantly decreased with

increasing pulse position (Figs. 2Be and 3Ba,b). The aver-

age 50% recovery time of the third group of neurons was

always shorter than 14 ms at all three azimuthal angles (Fig.

4Cb). Therefore, the response magnitude of this group of

neurons would recover fully at all three azimuthal angles

such that they would discharge almost equally well to each

presented sound pulse at each azimuthal angle (Fig.

2Ca,b,c). As a result, the AR and nAR of this group of IC

neurons varied very little with sequentially presented sound

pulses (Figs. 2Ce and 3Ca,b).

Previous studies have shown that the head shadowing

effect is greater on responses of high BF than low BF

neurons because of frequency-dependent pinna transfer

functions [9,18]. In the present study, we showed that three

groups of IC neurons whose recovery cycles varied differ-

ently with sound azimuth did not differ significantly in BF

(Table 1). This observation suggests that the azimuth-de-

pendent recovery cycle of IC neurons observed in the

present study is most likely due to the binaural inhibition.

While the contribution of head shadowing effect cannot be

excluded under the present free-field stimulation conditions,

we have recently shown that the azimuth-dependent recov-

ery cycle of IC neurons in the same bat species is abolished

or greatly reduced during iontophoretic application of bicu-

culline [27]. This observation suggests that the azimuth-

dependent recovery cycle of IC neurons is mainly, if not

only, attributed to GABAergic binaural inhibition. Clearly,

future work with the combined ear-phone system and free-

field study would be necessary to determine the contribution

of head shadowing effect on azimuth-dependent recovery

cycle of IC neurons.

We showed that directional selectivity of 56% of IC

neurons increased with sequentially presented sound pulses

at inter-pulse interval of 33.3 ms. The 33.3 ms inter-pulse

interval is equivalent to that occurring during the approach

phase of hunting [4,22]. Because the effect of sound

position within a pulse train on the nAR of IC neurons is

dependent upon the change in recovery cycle with sound

azimuth, directional selectivity of IC neurons with extreme-

ly short recovery cycle would increase with sequentially

presented sound pulses even at shorter inter-pulse interval.

For example, while directional selectivity of those 18% IC

neurons with short recovery cycle was hardly affected by

pulse position at 33.3 ms inter-pulse interval, their direc-

tional selectivity would conceivably increase with sequen-

tially presented sound pulses at shorter inter-pulse intervals.

We have previously shown that response magnitude of those

IC neurons that could discharge impulses to each presented

sound pulse dropped even greater at lateral azimuthal angles

at inter-pulse intervals of 11.1 ms as that occurring during

the terminal phase of hunting [26]. It is therefore conceiv-

able that IC neurons with shorter recovery cycle would be

systematically recruited to improve directional selectivity to

increasing number of echoes throughout the course of

hunting to facilitate interception of prey.

On the other hand, natural sound signals such as vocal

communication signals of many animal species including

speech by humans, typically occur in rapid succession rather

in isolation. It is conceivable that the present findings might

contribute to the understanding of central auditory process-

ing in complex acoustic environments. For example, the

increase of responsiveness of IC neurons to sequentially

presented sound pulses might serve as a neural basis

underlying facilitation of speech processing in humans.

Acknowledgements

We thank three anonymous reviewers for reading an

earlier version of this manuscript. This work is supported by

the Research Council of the University of Missouri-

Columbia and by a Sorenson Fund.

References

[1] J.H. Casseday, E. Covey, Frequency tuning properties of neurons in

the inferior colliculus of an FM bat, J. Comp. Neurol. 319 (1992)

34–50.

[2] Z.M. Fuzessery, G.D. Pollak, Determinants of sound location selec-

tivity in bat inferior colliculus: a combined dichotic and free-field

stimulation study, J. Neurophysiol. 54 (1985) 757–781.

[3] A.V. Galazyuk, D. Llano, A.S. Feng, Temporal dynamics of acoustic

stimuli enhance amplitude tuning of inferior colliculus neurons,

J. Neurophysiol. 83 (2000) 128–138.

[4] D.R. Griffin, Listening in the Dark, Yale Univ. Press, New Haven,

1958, Reprinted by Comstock, Ithaca, 1986.

[5] A.D. Grinnell, The neurophysiology of audition in bats: directional

localization and binaural, J. Physiol. (Lond.) 167 (1963) 97–113.

[6] A.D. Grinnell, V.S. Griennell, Neural correlates of vertical localiza-

tion by echolocating bats, J. Physiol. (Lond.) 181 (1965) 830–851.

[7] B. Grothe, E. Covey, J.H. Casseday, Spatial tuning of neurons in the

X. Zhou, P.H.-S. Jen / Brain Research 1019 (2004) 281–288288

inferior colliculus of the big brown bat: effects of sound level, stim-

ulus type and multiple sound sources, J. Comp. Physiol. 179 (1996)

89–102.

[8] G. Harnischfeger, G. Neuweiler, P. Schlegel, Interaural time and in-

tensity coding in superior olivary complex and inferior colliculus of

the bat, Molossus ater, J. Neurophysiol. 53 (1985) 89–109.

[9] P.H.S. Jen, D.M. Chen, Directionality of sound pressure transforma-

tion at the pinna of echolocating bats, Hear. Res. 34 (1988) 101–117.

[10] P.H.S. Jen, P.A. Schlegel, Auditory physiological properties of the

neurons in the inferior colliculus of the big brown bat, Eptesicus

fuscus, J. Comp. Physiol. 147 (1982) 351–363.

[11] P.H.S. Jen, M. Wu, Directional sensitivity of inferior collicular neu-

rons of the big brown bat, Eptesicus fuscus, to sounds delivered

from selected horizontal and vertical angles, Chin. J. Physiol. 36

(1993) 7–18.

[12] P.H.S. Jen, X.M. Zhou, Temporally patterned pulse trains affect du-

ration tuning characteristics of bat inferior collicular neurons,

J. Comp. Physiol. 185 (1999) 471–478.

[13] P.H.S. Jen, X.D. Sun, D.M. Chen, H.B. Teng, Auditory space repre-

sentation in the inferior colliculus of the FM bat, Eptesicus fuscus,

Brain Res. 419 (1987) 7–18.

[14] P.H.S. Jen, X.M. Zhou, C.H. Wu, Temporally patterned sound pulse

trains affect intensity and frequency sensitivity of inferior collicular

neurons of the big brown bat, Eptesicus fuscus, J. Comp. Physiol. 187

(2001) 605–616.

[15] P.H.S. Jen, C.H. Wu, R.H. Luan, X.M. Zhou, GABAergic inhibition

contributes to pulse repetition rate-dependent frequency selectivity in

the inferior colliculus of the big brown bat, Eptesicus fuscus, Brain

Res. 948 (2002) 159–164.

[16] Y. Lu, P.H.S. Jen, M. Wu, GABAergic disinhibition affects responses

of bat inferior collicular neurons to temporally patterned sound pulses,

J. Neurophysiol. 79 (1998) 2303–2315.

[17] D.A. Pinheiro, M. Wu, P.H.S. Jen, Encoding repetition rate and du-

ration in the inferior colliculus of the big brown bat, Eptesicus fuscus,

J. Comp. Physiol. 169 (1991) 69–85.

[18] P.W. Poon, X.D. Sun, T. Kamada, P.H.S. Jen, Frequency and space

representation in the inferior colliculus of the FM bat, Eptesicus fus-

cus, Exp. Brain Res. 79 (1990) 83–91.

[19] P.A. Schlegel, Directional coding by binaural brainstem units of the

CF-FM bat Rhinolophus ferrumequinum, J. Comp. Physiol. 118

(1977) 327–352.

[20] P.A. Schlegel, P.H.S. Jen, S. Singh, Auditory spatial sensitivity of

inferior collicular neurons of echolocating bats, Brain Res. 456

(1988) 127–138.

[21] J.A. Simmons, A view of the world through the bat’s ear: the

formation of acoustic images in echolocation, Cognition 33 (1989)

155–199.

[22] J.A. Simmons, M.B. Fenton, M.J. O’Farrell, Echolocation and pursuit

of prey by bats, Science 203 (1979) 16–21.

[23] J.M. Smalling, A.V. Galazyuk, A.S. Feng, Stimulation rate influen-

ces frequency tuning characteristics of inferior colliculus neurons in

the little brown bat, Myotis lucifugus, NeuroReport 12 (2001)

3539–3542.

[24] M. Wu, P.H.S. Jen, Temporally patterned pulse trains affect direction-

al sensitivity of inferior collicular neurons of the big brown bat,

Eptesicus fuscus, J. Comp. Physiol. 179 (1996) 385–393.

[25] X.M. Zhou, P.H.S. Jen, Neural inhibition sharpens auditory spatial

selectivity of bat inferior collicular neurons, J. Comp. Physiol. 186

(2000) 389–398.

[26] X.M. Zhou, P.H.S. Jen, The role of GABAergic inhibition in shaping

directional selectivity of bat inferior collicular neurons determined

with temporally patterned pulse trains, J. Comp. Physiol. 188

(2002) 815–826.

[27] X.M. Zhou, P.H.S. Jen, The effect of bicuculline application on azi-

muth-dependent recovery cycle of inferior collicular neurons of the

big brown bat, Eptesicus fuscus, Brain Res. 973 (2003) 131–141.