is
HEARES 01X79
The influence of temporal pattern of stimulation on delay tuning of neurons in the auditory cortex of the FM bat, Myu~~~Zz~G~~~g~~ Hidekazu Tanaka Departmen
I
’
and Donaid Wong
of Anatomy. Indiana Unic~ersify&hoof of Medicine, Indianapolis, Indiana. USA
(Received 1June 1902: Revision received 7 October 1992; Accepted 12 October 1992)
In echolocating bats, delay-sensitive neurons show facilitative responses to simulated pulse-echo pairs at particular echo delays, Three experiments examined how the temporal pattern of stimulation affected the delay tuning of neurons in the auditory cortex of the awake FM hat. Myofis lucifilgus. First, delay tuning was compared using a series of pulse-echo pairs fixed in echo delay (‘standard’ stimuli). and a series of pulse-echo pairs in which successive sound pairs decreased by a fixed echo-delay step (‘approach’ stimulif. Similar best delays were measured with both stimulation patterns presented at repetition rates in which the neuron was delay-sensitive. At the higher delay-sensitive pulse repetition rates, approach stimuli evoked larger delay-dependent responses. Second, approach stimuli were fixed at different intertrial intervals. The best delay was unaffected by intertrial interval. although some neurons showed larger responses for longer intertriat intervals (0.5, I.0 sf, erpeciaiiy at the higher delay-sensitive pulse repetition rates. Third, approach stimuli were fixed at different echo-delay steps to simulate target velocity. The majority of neurons showed some sensitivity to echo-delay step, with clear preference for target velocity mainly between 1.S7.0 m/s. This suggests that delay-sensitive neurons compute target velocity by rate of change of echo delay over successive echoes. Thus, response properties of cortical neurons are influenced by dynamic acoustic conditions found in target-directed flight. Auditory cortex: Delay sensitivity: Temporal sequence: Velocity sensitivity; Echolocation: FM bat
Introduction Echoiocating bats perceive targets as acoustic images by analyzing the returning echoes of their emitted pulses. In target-directed flight, these bats actively CORtrol the temporal pattern of their sonar emission (Griffin 1958; Griffin et al., 1960; Schnitzler and Henson. 1980). For example, the pulse repetition rate increases and pulse duration decreases as the echo delay (target distance) decreases. To simulate the acoustic pattern of stimulation confronting the echolocating bat, a number of neurophysiological studies have employed sound sequences of different complexity to study response properties of neurons in the bat’s brain. Delay-sensitive neurons have been the focus of many cortical studies since these neurons are thought to provide the primary cue for target-distance determination. Delay-sensitive neurons show strong facilitative responses to simulated pulse-echo pairs at particular echo delays (Feng et al., 1978; Suga et al., 1978; O’Neill
Correspondence to: Donald Wong, Department of Anatomy, Indiana University School of Medicine, Medical Science Building, Room 205, Indianapolis, IN 46202-5120, USA. Fax: (317) 278-2040. ’ Present address: Department of Otoia~ngolo~, Tokyo Medical and Dental University, l-5-45 Yushima, ~unkyo-ku~ Tokyo 113, Japan.
and Suga, 1982; Sullivan, 1982; Wong et al., 1992). In the auditory cortex of the mustached bat (Pter~r~u~~u~~ ~~~ell~i~~ delay sensitivity was originally characterized using a series of pulse-echo pairs fixed in echo detay (O’NeiIl and Suga, 19821, termed ‘standard’ stimufi (Suga and Horikawa, 1986). However, most temporal stimulation patterns used in delay-tuning studies simulate target approach by decreasing the echo delay within a series of pulse-echo pairs. Using approach stimuli fixed at different pulse repetition rates, we have recently found that the delay-tuning characteristics of cortical neurons in the FM bat, My&b hcifugus, are influenced by pulse repetition rate and pulse duration (Wang et al., 1992; Tanaka et al., 1992). To explore further how the temporal pattern of stimulation affects the delay tuning of neurons in Myotis cortex, three separate studies are reported in this paper. First, the delay tuning is compared using the approach and standard stimuli. No previous study has actually compared the delay-tuning properties using these different stimulation patterns, especially at different pulse repetition rates. Such comparison reveals whether the temporal pattern naturally found during echolocation may contribute to shaping the cortical response properties. Second, approach stimuli containing different intertriat intervals are used to detcrminc the effect of silent periods between sound sequences on delay tuning. Third, the sensitivity to the echo-delay
securing the skull-mounted rod onto a post with a set-screw. All equipment suspending the bat and holding micromanipulators was fixed onto an experimental table, which in turn, floated on a vibration-isolation table (TMC Micro-g). The auditory cortex was visualized through the thin skull by its overlying, characteristic cerebral-vessel pattern. With the aid of an operating microscope, miniature craniotomies ( - 50 pm) for electrode insertion were made with the sharp tip of a scalpel blade. Animals showed no excessive movements associated with pain during this procedure. During the cxperiment, the animal was frequently given drinking water and fed minced mealworms. To prevent local infection, all exposed tissue was cleaned after every recording session.
step used in successive pulse-echo pairs of the approach stimuli is examined. Delay-sensitive neurons that are also sensitive to rate of change of echo delay can provide target-velocity information.
Methods
Animal preparation Six brown bats (Myotis fucifugus), with a body weight of 5-10 g were used. All experimental protocols and animal care for this study have been approved by the Laboratory Animal Resource Center (LARC) of Indiana University School of Medicine (approval number MD No. 621). After injecting the animal with Innovarvet (0.04 mg/kg fentanyl and 2 mg/kg droperidol mixture, i.m.), the bat’s skull was surgically exposed, and a rod (1 .O cm long) was mounted onto the dorsum of the skull with dental cement (Wang, 1984; Wong et al., 1992). During surgery, a local anesthetic (Lidocaine) was topically applied onto exposed tissue to minimize discomfort or pain. All surgical wounds were treated with antibiotic ointment (Furacin). Animals were given a post-surgical recovery period of 2-7 days before neural recordings. Experiments on an individual bat were conducted on alternate days, with each recording session lasting up to 8 h. The bat was situated in the center of an echo-attenuated, soundproof room that was temperature-controlled at 30-32°C. The bat’s body was restrained in a plexiglass holder that was suspended with an elastic band. To obtain stable neural recordings from the awake animal, its head was immobilized by
Acoustic stimulation Ultrasonic free-field sounds were presented from a condenser loudspeaker located 75 cm in front of the bat. FM sweeps were generated to sweep 60 kHz linearly downward in 4 ms with rise-fail times of 0.25 ms. FM sweeps were presented either singly or in pairs as loud emitted biosonar sounds (pulse) and soft echoes (echo) to simulate the natural sonar sounds emitted by Myotis luci’ugus (Griffin 1958; Sales and Pye. 1074). Two basic temporal patterns of paired stimulation were used: the ‘standard’ and the ‘approach’ stimuli (Fig. 1). The standard stimuli consisted of a series of SO pulseecho pairs in which successive sound pairs were fixed at a single echo delay and pulse repetition rate (Fig. 1, bottom). At each pulse repetition rate in which the neuron was delay-sensitive, termed ‘delay-sensitive
Approach stimuli
Standard stimuli
P:
pulse
ID:
initial
n:
number
PRR:
pulse
E:
echo delay of
P-E
repetition
pairs rate
echo
t: echo-delay
step
ITI:
Intertrial
interval
ED:
echo
delay
Fig. I. Two temporal patterns of stimulation: approach (top) and standard (bottom) stimuli. Approach stimuli consist of a series of pulse-echo pairs in which successive pulse-echo pairs decrease in echo delay by a fixed time interval (echo-delay step). An unpaired pulse and unpaired echo are also presented at the end of each trial of the approach stimuli. The pulse repetition rate (PRR) and intertrial interval (ITI) are fixed. Standard stimuli consist of a series of pulse-echo pairs in which successive pulse-echo pairs are fixed at a single echo delay and at a single pulse repetition rate. Both the approach and standard stimuli were presented for 50 trials. Thick lines labeled P and E show the onset of pulse and echo respectively.
pulse repetition rate’, the standard stimuli were presented at different fixed echo delays, The approach stimuli consisted of a series of pulse-echo pairs in which successive sound pairs decreased in echo delay by a particular time interval, termed ‘echo-delay step’. Au unpaired pulse and echo were also presented at the end of each trial to permit subsequent comparison of single-sound and sound-pair evoked responses. The series of pulse-echo pairs and the single sounds in each trial of the approach stimuli were presented at a fixed pulse repetition rate and a fixed intertrial interval (Fig. 1, top). An intertrial interval of 0 ms has no silent period between trials, i.e., the time interval between the onset of the last sound of one trial (echo alone) and the onset of the first sound of the next trial (pulse of first sound pair) is equal to l/PRR (Fig. 1, top), where PRR is the pulse repetition rate. The approach stimuli were presented for 50 trials. In the first experiment, delay tuning was examined using standard stimuli and approach stimuli fixed at an echo-delay step of 1 ms and an intertrial interval of 0 s (i.e., no silent period). In the second experiment, approach stimuli with a 1-ms echo-delay step were presented using three different intertrial intervals (0, 0.5, 1.0 s). In the third experiment, approach stimuli, fixed at both a pulse repetition rate of 10/s and an intertrial interval of 0 s, were presented using four different echo-delay steps (0.5, 1.0, 2.0, 4.0 ms). Stimulation with the smaller echo-delay steps was initially used to estab.lish the neuron’s delay-tuning characteristics. The echo-delay range was then chosen for stimulation with the larger echo-delay steps by pre-setting the initial echo delay and number of pulse-echo pairs to include the neuron’s best delay. At each delay-sensitive pulse repetition rate, the best delay is the echo delay of pulse-echo pairs that evoked maximal delay-dependent responses. In the third experiment, an echo-delay step was chosen to simulate a particular target velocity as follows: velocity (m/s)
= (c/2) ( t//PI)
where c (speed of sound) is y 350 m/s at 30-32”C, t is echo-delay step in ms, and ZPZ is interpulse interval in ms (or l/FRR in s -‘I. In comparing the delay tuning using different echo-delay steps, a pulse repetition rate of 10/s was chosen for two reasons. First, the largest percentage of cortical neurons in M. lucifugus were found to be delay-sensitive at pulse repetition rates between 10 and 20/s (Wang et al., 1992). Second, a relatively low pulse repetition rate permitted a larger number of delay steps to be chosen that correlate to behaviorally-relevant target velocities. For example, at a pulse repetition rate of 10/s, echo-delay steps of 0.5, 1.0, 2.0, and 4.0 ms correspond to target velocities of
approximately 0.88, 1.8, 3.5, and 7.0 m/‘s rcspcctively. ;I range of veIocities encompassing the 1-5 m/s previously reported in behavioral studies of echolocating bat (Suthers, 1965; Habersetzcr and Voglcr, 1983; Wenstrup and Suthers, 1984; Schnitzler et al.. 1987; Simmons and Grinnell, 1988).
Extracellular single-unit recordings were obtained from the awake bat with lacquer-coated tungsten electrodes (- 6 pm tip diameter). Single neurons were isolated by their spike amplitude using an upper and lower threshold setting of a window discriminator (Bak Electronics, model DIS-1). In a single penetration, the recording electrode was advanced at least 100 pm between units. An indifferent electrode (- 10 pm tip diameter) was placed on the surface of a non-auditor cortical location. The search stimuli used to identify a delay-sensitive neuron were approach stimuli at a pulse repetition rate of 10/s, echo-delay step of 1.0 or OS-ms, intertrial interval of 0 s, and echo delays scanning a lo-to-20 ms range. The pulse and echo were initially set to sweep from 100 to 40 kHz, and at amplitudes of 60-80 and 40-60 dB SPL, respectively. Once a delay-sensitive neuron was isolated and characterized in terms of its best frequencies and amplitudes (Suga et al., 1983; Wong et al., 1992), post-stimulus-time (PST) histograms were generated on-line using an IBM-compatible, 386SX personal computer (Gateway 2000) to represent the sound-evoked responses for the different temporal patterns of stimulation. PST histograms were generated using a temporal window of 50 ms to sample neural activity for pulse repetition rates of 20/s or less; the temporal window for pulse repetition rates higher than 20/s was set equal to the interpulse interval. PST histograms were constructed using a binwidth of 0.5 ms for SO trials. The neural spikes collected for each temporal window were then used to plot delay-tuning curves in the form of spike-count versus echo-delay functions (Sullivan, 1982; Suga and Horikawa, 1986; Wong et al., 1992). Sound-pair responses were deemed delay-dependent when the spike counts evoked by 50 pulse-echo pairs at a particular echo delay were greater than the spike counts evoked by 50 pulse alone plus SO echo alone. Results
This study is based on 71 delay-sensitive recorded throughout the auditory cortex. Delay tuning measured stimuli
with approach
neurons
and standard
The delay-tuning curves obtained with the approach and standard stimuli varied with sound repetition rate.
2A: 2 ms, Fig. 2B: 3 f 1 ms). Of 26 neurons, no significant difference was found in the best delay measured by the two stimulation patterns. The two stimulation patterns also generated similar delay-tuning curves at relatively low pulse repetition rates (Fig. 2A: 10/s, Fig. 2B: 5/s and 20/s). However, at relatively high pulse repetition rates (2 30/s), the approach stimuli evoked noticeably larger delay-dependent responses than the standard stimuli. For the neuron shown in Fig. 2A, larger delay-dependent responses were evoked with approach stimuli presented at pulse repetition rates of 30 and SO/s. At a pulse repetition rate of 30/s, this difference in the response magnitude at best delay was at least two-fold, whereas at 50/s clear delay sensitivity was found with the approach stimuli but no delay sensitivity was found with the standard stimuli. Similarly, in Fig. 2B, larger delay-dependent responses were also evoked by the approach stimuli, mainly at a pulse rcpctition rate of 30/s. In 19 neurons examined with pulse repetition rates at or lower than 20/s, 84% (N = 16) exhibited delay-dependent responses of similar strength for both stimulation patterns, and the remaining 16% (N = 3) exhibited stronger delay-dependent responses to the standard than to the approach stimuli. In 20 neurons examined at pulse repetition rates higher than 30/s. all exhibited larger delay-sensitive responses to the approach than to the standard stimuli, and some exhibited delay-dependent responses to only the approach stimuli.
PKR
PRR
Fig. 2. Comparison of delay tuning measured with approach and standard stimuli. Series of delay-tuning curves are shown for two neurons (A. B) at their delay-sensitive pulse repetition rates. The approach stimuli were presented with a I-ms echo-delay step and an intertrial interval of 0 b.
Fig. 2 shows the delay-tuning curves of two neurons at pulse repetition rates evoking delay-dependent responses. In each unit, identical best delays were measured with the two stimulation patterns presented at the unit’s delay-sensitive pulse repetition rates (Fig.
B
A
IT,
PRR
Echo
delay
PRR
(ms)
Fig. 3. Comparison of delay tuning measured with approach stimuli fixed at different intertrial intervals (ITIs). Series of delay-tuning curves are shown for two neurons (A, B) at their delay-sensitive pulse repetition rates (PRRs). The approach stimuli were presented with a I-ms echo-delay step.
in Fig. 3A, the delay-dependent rcsponscs at an echo delay of 8 ms were only evoked using intcrtrial intcrvals of 0.5 and 1.0 s, and these responses were comparable to the best-delay response at 41 ms. For the neuron in Fig. 3B, intertrial intervals of 0.5 and 1.0 s resulted in delay-dependent responses at echo delays of 12, 8, 5, and 5 ms for pulse repetition rates of 50. 70. 100, 150/s respectively; these delay-dependent responses were comparable to the best-delay responses at the corresponding pulse repetition rates. The best-delay responses often became noticeably smaller at relatively high pulse repetition rates with no silent period. In Fig. 3A, stimulation at a pulse repetition rate of 50/s evoked best-delay responses that
Approach stimuli fked at different intertrial inten*als Delay tuning was examined using the approach stimuli fixed at intertrial intervals of 0, 0.5, and 1.0 s. Fig. 3 shows two delay-sensitive neurons with best delays that were not significantly affected by different intertrial intervals at each delay-sensitive pulse repetition rate. In all 24 neurons examined, the best delay was not altered significantly (i.e., + 1 ms> by intertrial interval. However, at particular delay-sensitive pulse repetition rates, the initial echo delay used in the approach stimuli (first pulse-echo pair of each trial) evoked larger responses for the longer intertrial intervals, and these delay-dependent responses were often as large as the best-delay response. At a pulse repetition rate of 70/s
Echo
30
A i
A
delay
Velocity
step 0.5 Ins
0.88 m/s
o-------o .
.
1.0 al8
1.8
m/e
b
4
2.0 Ins
3.5
m/a
(r
-3
4.0 ms
7.0
m/s
67.4d6
SPL
FM:96.15-36.15kHz. E: 49.4dB
SPL.
P: MOZlZSZ-7
0
100
FM:
122.17-62.17kHr.
E: 53.4dE
SPL.
P:
69.4dE
SPL
75.4dB
SPL
M021291-17
0
50
FM: 100.30-40.30kHz. P: E: 59.4dB SPL. M021292-16
0
5
Echo
10
delay
15
20
(ms)
Fig. 4. Sensitivity to echo-delay step. Series of delay-tuning curves are measured for three neurons (A, B, C) using approach stimuli fixed at a pulse repetition rate of 10/s and an intertrial interval of 0 s. (A) neuron shows a preference for a I-ms echo-delay step that corresponds to a velocity of 1.8 m/s. (B) neuron shows a preference for relatively large delay steps that correspond to velocities of about 3.5 and 7.0 m/s. (C) neuron shows no preference for echo-delay step.
were two to three times larger with intertrial intervals of 0.5 and 1.0 s than of 0 s. In Fig. 3B, the best-delay responses were also larger with intertrial intervals of 0.5 and 1.0 s in stimulation for most of the range of delay-sensitive pulse repetition rates. Stimulation with different intertrial intervals typically did not affect the range of pulse repetition rates at which the neuron was delay-sensitive. In I6 neurons examined, 75% (N = 12) showed similar ranges of delay-sensitive pulse repetition rates for the three intertrial intervals, and 4 neurons changed their range of delay-sensitive pulse repetition rates with intertrial interval. The neuron in Fig. 3B showed delay sensitivity with stimulation at pulse repetition rates of 10/s to 100/s and for intertrial intervals of 0 and 0.5 ms. The range of delay-sensitive pulse repetition rates extended to 150/s with stimulation containing an intertrial interval of 1.0 s. Semiticity to echo-delay step
I-2
Delay tuning was measured using approach stimuli in which successive pulse-echo pairs were fixed at four different echo-delay steps. Fig. 4A shows a neuron whose delay-tuning curve was strongly influenced by the echo-delay step used. Stimulation with a 1-ms echo-delay step resulted in the sharpest delay tuning (as determined at half-maximal response), and evoked the largest delay-dependent responses. For example, at the neuron’s best delay (* 5 ms>, the response evoked with a 1-ms echo-delay step (best echo-delay step) was about three times larger than the response evoked with a 4-ms echo-delay step (worse echo-delay step). Fig. 4B shows another neuron whose delay tuning was also influenced by specific echo-delay steps. This neuron showed a clear preference for larger echo-delay steps (2.0, 4.0 ms), and responded maximaily at its best delay with a 4-ms echo-delay step. Fig. 4C shows a neuron whose delay tuning was not significantly affected by stimulation at the different echo-delay steps. For this neuron, most of the variation in the response magnitude for the different echo-delay steps were at or above the unit’s best delay.
3-4
Best
5-6
deiay
7-8
.9
(msf
Fig. 6. Best-delay distribution of delay-sensitive neurons shown separately according to their best echo-delay step. No apparent relationship is found between a neuron’s best echo-delay step and its best echo delay.
Of the 37 delay-sensitive neurons examined at the four echo-delay steps, 40% (N = 15) showed clear preference for echo-delay step. These neurons showed delay-dependent responses that were at least twice as large with the best echo-delay step as those with the worst echo-delay step (e.g., Fig. 4A and 4Bl. In another 38% (N = 141, some sensitivity to echo-delay step was evident. In the remaining 22% (N = 81, no sensitivity to echo-delay step was observed (e.g., Fig. 4CI. Fig. 5 shows the distribution of neurons according to their best echo-delay step. Neurons with best echo-delay step of 1, 2 and 4 ms were about equally distributed. Only one neuron had a best echo-delay step of 0.5 ms. Fig. 6 shows best-delay distributions, each for neurons with a particular best echo-delay step. No apparent correlation was evident between the best echo-delay step and the neuron’s best delay.
Discussion hfluence of temporal pattern of sound sequences on delay t~~i~~
1
05
Best Fig. 5. Distribution
echo
of delay-sensitive echo-delay
2
delay neurons step.
4
step
(ms)
according
to their best
Comparison of the delay tuning of cortical neurons using approach and standard stimuli revealed both similarities and differences. First, the two stimulation patterns resulted in almost identical delay-tuning cumes at the lower delay-sensitive pulse repetition rates, whereas differences in their profiles often became apparent at the higher delay-sensitive pulse repetition
rates. Second, no significant difference was found in the best delays measured with the two temporal patterns of stimulation at each delay-sensitive pulse repetition rate. This result confirms a previous finding in the mustached bat in which similar best delays were also measured in delay-sensitive neurons using standard stimuli and temporal-pattern-stimulating (TPS) stimuli, a more complex stimulation pattern simulating target approach (Suga and Horikawa, 1986). Third, the delay-dependent responses evoked at the best delay were greatly diminished with the standard stimuli at the higher repetition rates. Thus, at the higher repetition rates, delay sensitivity was still observed with the approach stimuli, but disappeared with the standard stimuli. The delay-dependent responses evoked by the two stimulation patterns were differentially affected by habituation. In a previous study of ~y~~~~ cortex, the delay-dependent responses of some neurons habituated rapidly to stimulation with the echo delay fixed at or near the unit’s best delay, whereas the delay-dependent responses were consistently evoked by stimulation with the echo delay sweeping through the unit’s best delay (Sullivan, 1982). However, habituation to approach stimuli became evident with a progressive decline in delay-dependent response as repetition rate increased (Wang et al., 1992). In the present study, the delay-dependent responses of some neurons habituated even more dramatically to the standard than to the approach stimuli as the repetition rate was increased. The temporal pattern of stimulation may also influence the delay sensitivity at higher repetition rates. Neurophysiological studies of the mammalian auditory cortex have shown that a unit’s responsiveness to a particular stimulus element in a sound sequence can profoundly influence its responsiveness to succeeding elements (Abeles and Goldstein, 1972; Hocherman and Giiat, 1981; Phillips, 1985; Shamma and Symmes, 1985). In the present study, it is possible that stimulation at different echo delays in the approach stimuli enhanced the cortical responsiveness to subsequent stimulation at or near the best delay. Although the mechanism underlying delay-dependent facilitation is currently not well understood, it has been proposed that delay-dependent responses involve a temporal interaction between the pulse- and echo-evoked excitatory and inhibition events (Berkowitz and Suga, 1989; Suga et al., 1990; Olsen and Suga, 1991). Thus, at higher repetition rates the exact temporal interactions occurring over shorter interstimulus intervals would more critically influence the cortical responsiveness to successive stimuli. The present study, however, cannot distinguish the temporal interactions that may underlie the different responses evoked by the two stimulation patterns. Delay sensitivity may be more effectively preserved
by approach stimuli at wider range of repetition rams. since the approach rather than the standard stimuli more accurately mimics the temporal pattern of stimulation naturally found in target-directed flight. This would suggest that the dynamic acoustic conditions confronting the bat during echolocation may play a role in shaping the delay tuning, and hence the cchoranging function of corticai neurons. Recocery between sound sequences
The present study demonstrated that the addition of an intertrial interval as large as 0.5-1.0 s (Fig. 3) did not significantly affect the best delay and the range of delay-sensitive pulse repetition rates. With a siient period interposed between sound sequences, some neurons showed noticeably larger delay-dependent response at relatively high pulse repetition rates (Fig. 3A, B). This suggested that a long silent period permitted neurons sufficient recovery time from each sound sequence. Moreover, in some neurons relatively large delay-dependent responses were evoked by the longest echo delay (first sound-pair of trial) in sound sequences with silent periods between trials, This is reasonable since the neuron has its longest recovery time between stimulation (i.e., the silent period) at the beginning of each new trial. ~though FM bats emit pulses as high as l~-Z~/s during the terminal phase of echolocation (Griffin, 1958; Griffin et al., 1960; Sales and Pye, 1974), most cortical neurons were found to lose their delay sensitivity at pulse repetition rates above 100/s (for discussion, see Wong et al., 1992; Tanaka et al., 1992). Since no silent period was previously used in the approach stimuli, it was suggested that delay sensitivity could be sustained at high pulse repetition rates with the addition of a silent period. However, only 25% (4/l@ of the neurons in the present study changed their range of delay-sensitive pulse repetition rates in stimulation with sufficient silent periods (e.g., neuron in Fig. 3B). Thus, approach stimuli with no silent period probably cannot fully account for the loss of delay sensitivity by 100/s in most neurons. It remains to be determined whether the addition of harmonics, as found in the emitted pulses during the terminal phase, is critical for delay sensitivity at high putse repetition rates.
A majority of delay-sensitive neurons showed some sensitivity to echo-delay steps in approach stimuli. Many of these neurons showed clear preference for a particular echo-delay step, mainly of 1 ms or greater (Fig. 5). For neurons in which stimulation at relatively small echo-delay steps evoked small delay-dependent responses (e.g., neuron in Fig. 4B), delay sensitivity, especially at or near the best delay in each trial, may be more profoundIy influenced by preceding stimula-
OS
tion at other echo delays earlier in the trial. However, this factor cannot account for the delay-tuning characteristics of either the neurons with clear preference for echo-delay step (Fig. 4A) or those insensitive to echodelay step (Fig. 40. The relative velocity between a bat and target can be determined either by measuring the rate at which target range (echo delay) changes in successive echoes (range-rate information), or by evaluating the Doppler shift in target-reflected echoes (Simmons et al., 1975; Schnitzler and Henson, 1980; Schnitzler, 1984). The use of range-rate information from the FM component to measure target velocity has only been demonstrated behaviorally in Noctilio leporinus, a bat that echolocates with short constant-frequency (CF) and FM components (Wenstrup and Suthers, 1984). A similar strategy for measuring target velocity has also been suspected in Myotis and Eptesicus, FM bats that do not exhibit Doppler-shift compensating behavior. The present study provides the first neurophysiological evidence of delay-sensitive neurons using the rate of change of echo delay over successive echoes to compute target velocity. It has been suggested that long CF-FM bats, such as Pteronotus and Rhinolophus, measure relative target velocity by evaluating the amount of Doppler shift in the echo CF component (Schnitzler, 1968, 1970; Schuller et al., 1974; Simmons, 1974; Henson et al., 1982). Based on this mechanism for velocity measurement, Suga and his colleagues (1983) postulated that the CF/CF area of the mustached bat auditory cortex plays a primary role in processing velocity information. Since these CF,/CF, (n = 2, 3) neurons are tuned to the frequency deviation of CF, from its exact harmonic relationship with CF,, they could theoretically represent the magnitude of Doppler shift in the echo, and thus compute target velocity in the frequency domain. Relative target velocities ranging from about -2 to 9 m/s were mapped in the CF/CF area, with velocities between 0 and 5 m/s overrepresented (Suga et al., 1983; Suga 1990). In the present study, Myotis showed sensitivity to velocities of up to about 7 m/s, values similar to the l-5 m/s reported in behavioral studies of other FM bats (Schnitzler et al., 1987; Habersetzer and Vogler, 1983; Simmons and Grinnell, 1988). The fact that long CF-FM bats exhibit Dopplershift-compensating behavior does not necessarily mean that these bats are actually measuring velocity by frequency analysis. First, deriving velocity information by analyzing the time delay between the FM components of the pulse and echo is also available to long CF-FM bats. In fact, Schuller and Pollak (1979) proposed that the CF component may be more suitable for encoding wing-beat information in prey recognition rather than for precise velocity measurement. These authors argue that wing beats of ‘fluttering’ insects can produce sig-
nificant frequency modulations in the echo CF that will result in ambiguity of _t 1 m/s in velocity measurements. This velocity resolution is worse than the relative velocities of about 0.1 m/s that Rhinolophus should be able to detect by evaluating the Doppler-shift produced only by target-flight movement (see Schnitzler, 1984). Second, in the long CF-FM bat, Rhinolophus rouxi, the frequency tuning of CF/CF cortical neurons does not correspond to stimulus conditions appropriate for Doppler-shift compensation (Schuller et al.. 1991). Most of the CF/CF neurons appear to code for negative Doppler shifts, since these neurons arc tuned to a CF, frequency at a negative deviation from the exact harmonic relationship with the CF, frequency. Ncgative relative target velocities will mainly be represented if Doppler shift is used by CF/CF neurons for velocity determination. In contrast, most neurons in the CF/CF cortical area of the mustached bat represent positive Doppler shifts. Thus, in the Doppler-compensating Rhinofophus bat, the actual role of the CF/CF cortical area remains unclear. It is presently not known whether long CF-FM bats share with FM bats a similar strategy for velocity measurement. It would thus be interesting to use a neurophysiological approach similar to the present study to examine the sensitivity of FM-FM cortical neurons to echo-delay steps in both Pteronotus and R. rouxi. Such a study would directly establish whether the FM-FM area has the capacity for processing vclocity information. Moreover, the FM-FM and CF/CF areas are completely segregated in the Pteronotus cortex. If FM-FM neurons are velocity-sensitive in this long CF-FM bat, separate cortical areas could process velocity information by frequency and/or temporal analysis. Such a functional organization in Pteronotus cortex would then permit different cortical processing strategies to be used in maximizing velocity resolution according to the perceptual demands at the different phases of echolocation. Cortical organization for integrating target information The presence of cortical neurons that are sensitive to both target distance and velocity lends support to the hypothesis that single neurons in Myotis cortex have the capacity for processing different target information (see Wong et al., 1992). These cortical neurons are distributed throughout the auditory cortex, which contains delay-sensitive and tonotopic regions that overlap extensively (Wang and Shannon. 1988). This cortical organization may reflect a processing strategy for target perception that is well-suited for FM bats.
Acknowledgements We thank Douglas Fitzpatrick, George Pollak, and David Suzuki
Mathew Palakal, for valuable com-
ments on this manuscript. William G. Paschal assisted with some of the graphics. Brent Lebo designed software programs for data acquisition. The authors are also grateful to Scott Johnson of the Indiana Department of Natural Resources for supplying us with bats for this study. This research was supported by NIH grant DC00600 and a project Development Program grant from Indiana University to Donald Wong.
References Abeles, M and Goldstein, M.H., Jr. (1972) Responses of single units in the primary auditory cortex of cat to tones and to tone pairs. Brain Res. 42, 337-352. Berkowitz, A. and Suga, N (1989) Neural mechanisms of ranging arc different in two species of bats. Hear. Res. 41, 255-264. Feng, AS., Simmons, J.A., and Kick, S.A. (1978) Echo detection and target ranging neurons in the auditory system of the bat Eptesicus fuscus. Science 202, 645-648. Griffin, D.R. (19581 Listening in the Dark. Cornell University Press (reprinted), Ithaca, New York. Griffin, D.R., Webster, F.A., and Michael, C.R. (1960) The echolocation of flying insects by bats. Animal Behav. 8, 141-154. Habersetzer, J. and Vogler, B. (1983) Discrimination of surfacestructured targets by the echolocating bat Myotis myotis during flight. J. Comp. Physiol. 152, 275-282. Henson, O.W., Pollak, G.D., Kobler, J.B., Henson, M.M.. and Goldman, L.J. (1982) Cochlear microphonic potentials elicited by biosonar signals in flying bats, Pteronotm p. parnellii. Hear. Res. 7. l27- 147. Hocherman, S. and Gilat, E. (1981) Dependence of auditory cortex evoked unit activity on interstimulus interval in the cat. J. Neurophysiol. 45, 987-997. Olsen, J. and Suga, N. (1991) Combination-sensitive neurons in the medial geniculate body of the mustached bat: encoding of target range information. J. Neurophysiol. 65, 1275-1296. O’Neill, W.E and Suga, N. (1982) Encoding of target range information and its representation in the auditory cortex of the mustached bat. J. Neurosci. 2, 17-31. Phillips. D.P. (1985) Temporal response features of cat auditory cortex neurons contributing to sensitivity to tones delivered in the presence of continuous noise. Hear. Res. 19, 253-268. Sales. G. and Pye, D. (1974) Ultrasonic communication by animals. Chapman and Hall, London. Schnitzler, H.-U. (1968) Die Ultraschall-Ortungslaute der Hufeisen Fledermause (Chiroptera-Rhinolophidae) in verschiedenen Orientierungs-situationen. Z. Vergl. Physiol. 57, 376-408. Schnitzler, H.-U. (1970) Echoortung bei der Fledennaus Chilonycteris rubiginosa. 2. Vergl. Physiol. 68, 25-39. Schnitzler, H.-U. (1984) The performance of bat sonar systems. In: Varji, H.U. Schnitzler (Eds.), Localization and Orientation in Biology and Engineering, Springer-Verlag, Berlin, pp. 21 l-224. Schnitzler, H.-U. and Henson, O.W. (1980) Performance of airborne animal sonar systems: I. Microchiroptera. In: R.G. Busnel and J.F. Fish (Eds.), Animal Sonar Systems, Plenum, New York, pp. 1099181. Schnitzler, H.-U., Kalko, E., Miller, L. and Surlykke, A. (1987) The echolocation and hunting behavior of the bat, Pipistrellus kuhli. J. Comp. Physiol. A, 161, 267-274.
Schuller, G.. Beuter K., and Schnitzler. H.-1~;. I 1074)KesponreII) frequency shifted artificial echoes in the hat Rhirrolophu.\ _~cJ-rumequinutn. J. Comp. Physiol. 89. 275 280. Schuller, G., O’Neill, W.E., and Radtke-Schuller, S. (lY9ll Facilitation and delay sensitivity of auditory cortex neurons in CF-Flvl bats, Rhinolophus rouxl rrml Ptrronotus p. pnrnellii. Eur. .I. Ncurosci. 3. llhS-1181. Schuller, G. and Pollak G.D. (1979) Disproportionatc frequency representation in the inferior colliculus of Doppler-compensating greater horseshoe bats: evidence for acoustic fovea. J. Camp. Physiol. 132, 37-54. Shamma, S.A. and Symmes, D. (1985) Patterns of inhibition in auditory cortical cells in awake squirrel monkeys. Hear. Res. 19. l-13. Simmons, J.A. (1974) Response of the Doppler echolocation system in the bat Rhinolophus ferrumequinum. J. &oust. Sot. Am. 56. 672-682. Simmons, J.A. and Grinnell. A.D. (1988) The performance of echolocation: Acoustic images perceived by echolocating bats. In: P.E. Nachtigall and P.W.B. Moore (Eds.), Animal Sonar: Processes and Performance, Plenum, New York, pp. 3533385. Simmons, J.A.. Howell, D.J. and Suga, N. (1975) The information content of bat sonar echoes. Am. Sci. 63. 204-215. Suga, N. (1990) Cortical computational maps for auditory imaging. Neural Networks 3, 3-21. Suga, N. and Horikawa, J. (1986) Multiple time axes for representation of echo delays in the auditory cortex of the mustached bat. J. Neurophysiol. S5, 776-805. Suga, N., Olsen, J. and Butman, J.A. (1990) Specialized subsystems for processing biologically important complex sounds: cross-correlation analysis for ranging in the bat’s brain. Cold Spring Harbor Symp. Quant. Biol. 55. 585-597. Suga, N.. O’Neill, W.E., Kujirai. K. and Manabc, T. (1983) Specificity of combination-sensitive neurons for processing of complex biosonar signals in auditory cortex of the mustached bat. J. Neurophysiol. 49, 1573-1626. Suga, N., O’Neill, WE. and Manabe, T. (lY78l Cortical neurons sensitive to particular combinations of information bearing elements of bio-sonar signals in the mustache bat. Science 200, 778-781. Sullivan, W.E. (1982) Neural representation of target distance in auditory cortex of the echolocating bat Myotis hkfirgus. J. Neurophysiol. 48, 1011-1032. Suthers, R.A. (1965) Acoustic orientation by fish-catching bats. J. Exp. Zool. 158. 319-348. Tanaka, H., Wong, D. and Taniguchi, I. (1992) The influence of stimulus duration on the delay tuning of cortical neurons in the FM bat, Myotis /uci&gus. J. Comp. Physiol. A, 171. 29-40. Wenstrup, J.J. and Suthers, R.A. (1984) Echolocation of moving targets by the fish-catching bat, Noctilio lepotinus. J. Comp. Physiol. A 155, 75-89. Wong, D. (1984) Spatial tuning of auditory neurons in the superior colliculus of the echolocating bat, Myotis lucifugus. Hear. Res. 16. 261-270. Wong, D., Maekawa, M. and Tanaka, H. (1992) The effect of pulse repetition rate on the delay sensitivity of neurons in the auditory cortex of the FM bat, Myotis lucifugus. J. Comp. Physiol. A 170, 393-402. Wong, D. and Shannon, S.L. (1988) Functional zones in the auditory cortex of the echolocating bat, Myotis lucifugus. Brain Res. 453. 349-352.