FM Sweep Research Articles

Results of tonotopic and echo delay‐tuning mapping of the auditory cortex of the Big Brown Bat

Characterization of the response properties of neurons in the auditory cortex of Eptesicus fuscus is an initial step toward defining the neural mechanisms that mediate echolocation in FM emitting bats. Multiple‐ and single‐unit recordings from the cortex were obtained during multiple presentation of pure tones, FM sweeps, and FM sweeps in pulse‐echo pairs decreasing in delay from 40 to 0 ms. The existence of a distribution of high to low frequencies from the anterior to posterior poles of auditory cortex suggests a tonotopic map. The frequencies comprising the first harmonic of the echolocation emission appear to be magnified in their cortical representation. Frequency tuning and onset latencies in response to FM sweeps with a 20‐kHz bandwidth are similar to responses to pure tones. While the majority of cells sampled are not echo delay tuned, delay‐tuned cells have been found with best delays ranging from approximately 3–30 ms. These neurons cluster into two distinct groups suggesting regions of function...

Design of an experiment to measure low-frequency acoustic monostatic backscatter in the open ocean

An experiment has been designed to measure acoustic backscatter from the ocean wave surface simultaneously with surface wave measurements of the SWADE experiment. The objective is to measure scattering cross section Doppler broadening for grazing angles of 15 to 2 deg, at frequencies of 100, 200, 400, and 800 Hz for the purpose of comparison with perturbation theory. Constraints are that the system be operated under battery power at sea, unattended for a period of 2 months or more, with measurements at 1-h intervals. Design parameters of the source and receiver array are driven by signal-to-noise considerations, whereas the signaling format is driven by predicted Doppler broadening and data storage. Depth of the source and receiver is optimized to give time-gated arrivals for low grazing angle before the fathometer return and long enough in transmissions in time to resolve Doppler broadening. Ambiguity functions for cw pulses, pseudorandom codes, and FM sweeps are considered and compared. A solution appears possible, but just within the constraints of experimental geometry and power and storage capabilities.

The response of tympanate moths to the echolocation calls of a substrate gleaning bat, Myotis evotis

1. Most studies examining interactions between insectivorous bats and tympanate prey use the echolocation calls of aerially-feeding bats in their analyses. We examined the auditory responses of noctuid (Eurois astricta) and notodontid (Pheosia rimosa) moth to the echolocation call characteristics of a gleaning insectivorous bat, Myotis evotis. 2. While gleaning, M. Evotis used short duration (mean ± SD = 0.66 ± 0.28 ms, Table 2), high frequency, FM calls (FM sweep = 80 − 37 kHz) of relatively low intensity (77.3 + 2.9, −4.2 dB SPL). Call peak frequency was 52.2 kHz with most of the energy above 50 kHz (Fig. 1). 3. Echolocation was not required for prey detection or capture as calls were emitted during only 50% of hovers and 59% of attacks. When echolocation was used, bats ceased calling 324.7 (±200.4) ms before attacking (Fig. 2), probably using prey-generated sounds to locate fluttering moths. Mean call repetition rate during gleaning attacks was 21.7 (±15.5) calls/s and feeding buzzes were never recorded. 4. Eurois astricta and P. rimosa are typical of most tympanate moths having ears with BFs between 20 and 40 kHz (Fig. 3); apparently tuned to the echolocation calls of aerially-feeding bats. The ears of both species respond poorly to the high frequency, short duration, faint stimuli representing the echolocation calls of gleaning M. evotis (Figs. 4–6). 5. Our results demonstrate that tympanate moths, and potentially other nocturnal insects, are unable to detect the echolocation calls typical of gleaning bats and thus are particularly susceptible to predation.

A gating mechanism for sound pattern recognition is correlated with the temporal structure of echolocation sounds in the rufous horseshoe bat

The rufous horseshoe bat, Rhinolophus rouxi, was trained to discriminate differences in target distance. Loud free running artificial pulses, simulating the bat's natural long-CF/FM echolocation sounds, interfered with the ability of the bat to discriminate target distance. Interference occurred when the duration of the CF component of the CF/FM artificial pulse was between 2 and 70 ms. A brief (2.0 ms) CF signal 2–68 ms before an isolated FM signal was as effective as a continuous CF component of the same duration. When coupled with the bat's own emissions, a 2 ms FM sweep alone was effective in interfering when it came 42 to 69 ms after the onset of the bat's pulse. The coupled FM artificial pulses did not interfere when they began during the bat's own emissions.

Foraging behavior and echolocation of wild horseshoe bats Rhinolophus ferrumequinum and R. hipposideros (Chiroptera, Rhinolophidae)

1. Echolocation and foraging behavior of the horseshoe bats Rhinolophus ferrumequinum and R. hipposideros feeding under natural conditions are described. 2. The calls of both species consisted predominantly of a long CF segment, often initiated and terminated by brief FM sweeps of substantial bandwidth. 3. R. hipposideros typically flew close to vegetation, and fed by aerial hawking, gleaning and by pouncing on prey close to the ground. R. hipposideros called with a CF segment close to 112 kHz which is the second harmonic of the vocalization; its calls included low intensity primary harmonics, and had prominent initial and terminal FM sweeps of considerable bandwidth. When searching for prey on the wing it had longer interpulse intervals than R. ferrumequinum, but emitted shorter pulses at a higher repetition rate; overall it had a similar duty cycle to R. ferrumequinum. 4. R. ferrumequinum, calling with a CF segment close to 83 kHz, also used harmonics other than the dominant secondary in its calls, and modified its echolocation according to ecological conditions. This species showed certain parallels with R. rouxi of Asia. It was observed feeding by aerial hawking and by flycatching. When scanning for prey from a perch (perch hunting), calls were of shorter duration, and interpulse intervals were on average longer, than when bats were flying. Mean duty cycle was longer in flight, and the bandwidths and frequency sweep rates of the FM segments in the calls increased in comparison with perched bats. 5. FM information may facilitate determination of target range and the location and nature of obstacles; it may also be involved in the interpretation of echoes and the detection of moving targets among clutter. The rising FM sweep initiating the call in both species when flying (and to a lesser extent perch hunting) in the wild must have a significant adaptive role, and should be considered an essential component of the call; rhinolophids should be termed ‘FM/CF/FM’ bats.

Low-frequency measurements of bottom reflectivity at a thin sediment site in the North Pacific

Bottom reflectivity measurements were made in the North Pacific, during Pacific Echo I, using a low-frequency, towed source. A MK VI source transmitted 50-s FM sweep signals (5 to 15 Hz) to a vertical array of 16 hydrophones. The FM ramps were deconvolved and beamformed to separate direct and bottom arrivals in time and angle. In addition, ray theory predictions of the arrival structure, scaled by spreading loss, were beamformed. A comparison between the time integrated energy in the measured and theoretical bottom arrivals, led to a measure of the bottom reflectivity. FM ramps were processed along the source ship track to vary the bottom grazing angle from 10 to 75 deg. Reflectivity as a function of bottom grazing angle, smoothed over angle to reduce variance, will be presented. Theoretical computations of reflectivity from a simulated bottom that includes a sediment layer with sound-speed gradient and an elastic basement, with shear and compressional speed gradients and attenuations, will be presented. Theory and measurements are compared to infer the sediment thickness and the wave speed profiles in the upper 200 m of the oceanic crust.

The echolocation and hunting behavior of the bat, Pipistrellus kuhli.

The echolocation and hunting behavior of Pipistrellus kuhli was studied in the field using multi-exposure photography synchronized with high-speed tape recordings. During the search phase, the bats used 8-12 ms signals with sweeps (sweep width 3-6 kHz) and pulse intervals near 100 ms or less often near 200 ms. The bats seemed to have individual terminal frequencies that could lie between 35 and 40 kHz. The duty cycle of searching signals was about 8%. The flight speed of hunting bats was between 4.0 and 4.5 m/s. The bats reacted to insect prey at distances of about 70 to 120 cm. Given the flight speed, the detection distance was estimated to about 110 to 160 cm. Following detection the bat went into the approach phase where the FM sweep steepened (to about 60 kHz bandwidth) and the repetition rate increased (to about 30 Hz). The terminal phase or 'buzz', which indicates prey capture (or attempted capture), was composed of two sections. The first section contained signals similar to those in the approach phase except that the pulse duration decreased and the repetition rate increased. The second section was characterized by a sharp drop in the terminal frequency (to about 20 kHz) and by very short pulses (0.3 ms) at rates of up to 200 Hz. Near the beginning of the buzz the bat prepared for capturing the prey by extending the wings and forming a tail pouch. A pause of about 100 ms in sound emission after the buzz indicated a successful capture (Fig. 4).(ABSTRACT TRUNCATED AT 250 WORDS)

High‐resolution time of arrival estimation via linear prediction

A new method is presented for estimating the arrival times of overlapping signal pulses that are separated by less than the duration of the signal autocorrelation function. The method is based on the observation that if the signal has a flat band‐limited spectrum, then the maximum‐likelihood estimator for the time of arrivals can be approximately transformed into an equivalent high‐resolution exponential parameter estimation problem. Then, an improved linear prediction algorithm [D. W. Tufts and R. Kumaresan, Proc. IEEE 70, 975–989 (1982)] for high resolution exponential parameter estimation is used to estimate the time of arrivals. The proposed method is computer simulated for two linear FM sweep signal pulses at various separations and signal‐to‐noise ratios. The simulation results are compared to the Cramer‐Rao lower bound (CRLB). These results indicate performance close to the CRLB over a reasonable range of pulse separation and signal‐to‐noise ratios.

Response properties of FM-FM combination-sensitive neurons in the auditory cortex of the mustached bat.

For echolocation, the mustached bat, Pteronotus parnellii rubiginosus, emits orientation sounds (pulses) and listens to echoes. Each pulse is made up of 8 components, of which 4 are constant frequencies (Cf 1.4) and 4 are frequency-modulated (FM 1-4). Target-range information, conveyed by the time delay of the echo FM from the pulse FM, is processed in this species by specialized neurons in a part of the auditory cortex known as the FM-FM area. These cortical neurons are responsive to pulse-echo pairs at specific echo delays. The essential components in the sound pair include the pulse FM1 followed by an echo FMn (n = 2, 3 or 4). Downward sweeping FM1-FMn sounds that are similar to those the animal naturally hears during echolocation are the most effective in evoking facilitative responses. Most FM-FM neurons, however, still exhibit facilitative responses to stimulus pairs consisting of upward sweeping FM sounds and/or pure tones at frequencies found in FM sweeps. The magnitude of facilitation is altered by changes in echo rather than pulse amplitude. Neurons characterized by shorter best delays (or echoes from closer targets) do not require larger best echo amplitudes for facilitation.

Echolocation sound features processed to provide distance information in the CF/FM bat,Noctilio albiventris: evidence for a gated time window utilizing both CF and FM components

Bats of the speciesNoctilio albiventris, trained to discriminate differences in target distance, emitted pairs of pulses at a rate of 7–10/s, the first a constant frequency (CF) pulse of about 8 ms duration and 75 kHz frequency, followed after about 28 ms by a CF/FM pulse having a 6 ms, 75 kHz CF component that terminates in a 2 ms FM sweep to about 57 kHz.

Auditory sensitivity in the fish-catching bat,Noctilio leporinus

Behavioral and auditory brainstem response (ABR) audiograms are described for the fish-catching bat,Noctilio leporinus, which uses short constant frequency/frequency modulated (short-CF/FM) sonar pulses with a CF component of 56–59 kHz, followed by an FM sweep down to 28–32 kHz. Social communication signals contain the frequencies found in sonar pulses, but may extend to lower frequencies (16 kHz). 2. Behavioral thresholds, obtained by operant conditioning in three bats, display maximum sensitivity in the region of the bats' CF sonar component, but show good sensitivity (thresholds less than 10 dB SPL) between 32 and 57 kHz. Below 24 to 32 kHz, sensitivity declined at a rate of 35 to 45 dB/octave. No response was obtained below 1 kHz. 3. The most prominent feature of the behavioral audiogram is an abrupt increase in threshold above 56 to 58 kHz, with an initial roll-off as high as 550 dB/octave. A threshold ‘plateau’ exists between 64 and 96 kHz, but sensitivity declines rapidly above 96 kHz. No response was obtained above 120 kHz. 4. ABR audiograms were obtained in two animals in which behavioral thresholds had been measured previously. These display a broadly tuned peak of maximum sensitivity at 24 kHz and a more sharply tuned sensitivity peak in the region between 56 and 59 kHz. No responses were obtained below 1 kHz or above 96 kHz. 5. Differences in the shape of the behavioral and ABR threshold curves are discussed. It is suggested that the two sensitivity peaks in the ABR may result from a disproportionately large representation of frequencies corresponding to social communication signals (24 kHz) and the bat's CF sonar pulse (56 to 59 kHz). 6. The behavioral audiogram ofN. leporinus is compared to those of other bats, and functional implications are discussed.

Auditory response properties and spatial response areas of superior collicular neurons of the FM bat,Eptesicus fuscus

1. Electrophysiological properties of 279 single units in the superior colliculus (SC) of the big brown bat,Eptesicus fuscus, were studied by recording their responses to pure tone pulses and frequency-modulated stimuli. 2. The SC units generally fired only a few impulses to acoustic stimulus. They are not tonotopically organized along the dorsoventral axis of the SC (Fig. 1). 3. The response latency of 276 units was between 3.6 and 20 ms, but most (253 units) were below 12.5 ms (Fig. 2). 4. Tuning curves of 164 units are either narrow, intermediate, or broad, according to the bandwidth of the tuning curves (Fig. 3). Q10 dB values range between 1.21 (best frequency (BF) =33.52 kHz) and 68.9 (BF=69.94 kHz), but the majority are below 20. The minimum thresholds (MTs) of these units were between 15 and 104 dB SPL, but most were above 40 dB SPL. 5. MTs to pure tone pulses and one octave upward and downward sweeping FM stimuli were compared (Fig. 4). SC units generally had their lowest MTs to downward sweeping FM stimuli and their highest MTs to pure tone pulses (Table 1). 6. Intensity-rate functions of 16 SC units were measured with both pure tone pulses and upward and downward sweeping FM stimuli. There was no correlation between the type of intensity-rate function obtained and the type of acoustic stimulus used (Fig. 5). 7. Auditory spatial response areas were measured for 47 units. The size of the spatial response area increases with stimulus intensity (Fig. 6). The auditory space is not orderly represented in the SC of a bat (Fig. 7).

Responses of cerebellar units in FM and CF-FM bats to acoustic stimuli

With pure tone pulses (35- and 0.5-ms rise-decay times), a total of 499 units from FM bats (Myotis lucifugus and Eptescius fuscus) and 401 units from the CF-FM bats (Pteronotus parnellii parnellii and Pteronotus parnellii rubiginosus) were isolated from a large area of the, cerebellar vermis and hemispheres. Response latencies ranged between 1.2 and 68 ms but most were below 10 ms. Phasic units, phasic bursters, and tonic units were observed. Threshold curves of these units were either narrow, broad, or irregular. In the CF-FM bats, units with BFs between 60 and 63 kHz corresponding to the predominant CF frequency were sharply tuned and showed off responses upon cessation of the stimulus. In both bats, variation of BFs of isolated units from penetrations within the same lobe was smaller than those from different lobes. A study of frequency tuning and representation suggests that a bat's cerebellum can effectively process the predominant components of biologically significant signals. Responses to a 4 ms upward and downward sweeping FM stimuli and pure tone pulses were studied in 70 units of the Eptesicus fuscus. Most units were more sensitive to a downward sweeping FM stimulus than to a pure tone pulse. An upward sweeping FM stimulus was as effective as a 35-ms pure tone pulse. A study of directional sensitivity of 47 units showed that 33 units were most sensitive to a sound delivered from the front. The other 14 units were most sensitive to a sound delivered from 20°–30° lateral. [Work supported by NSF 80 07348 and USPH 1-K04-NS-00433 to P. Jen.]

Tonotopic Representation and Space Map in the Non-Primary Auditory Cortex of the Mustached Bat

As auditory system has no sensory epithelium into which auditory space are projected, we studied the physiological map of the auditory space in the non-primary auditory cortex of the mustached bat by using the echo of their orientation sound. Ten bats were used as experimental subjects. Tungsten wire electrodes were inserted obliquely in the dorsomedial (DM) and ventroposterior (VP) areas of the non-primary auditory cortex. When single neuron was isolated, best frequency (BF), best azymuth (BAZ) and best elevation (BEL) were measured and were plotted on a schematic figure. To mimic its biosonar, one loudspeaker, delivering synthesized orientation sounds, was placed in front of the animal, and another loudspeaker delivering synthesized echo was mounted on a movable hoop. Tonotopic representation was observed but complicated in both areas, and those areas could be divided into several subdivisions consisting of the neuron groups characterized by three frequency bands. The neurons were thought to be related to the processing of biosonar informations from the facts that their BFs agreed with the scope of the FM sweep of each echo harmonics. The magnitude of the response showed rapid increase at their BAZ or BEL, so that the neurons seemed to tune to a certain direction in the auditory space. Especially in the DM area, neurons assumed a systematic arrangement of their BAZs on the cerebral surface and showed some tendency of a systematic arrangement of their BELs. The DM area was thought to have a kind of neural map of the auditory space.

Responses of primary auditory fibers to transient sounds

Responses of primary auditory fibers to clicks, short tone bursts, FM sweeps and consonant‐vowel syllables were obtained in the anesthetized cat. Tone‐burst responses depend on the fiber's characteristic frequency (CF). Fibers of high CF (above 10 kHz) responded to 2‐ms bursts with practically steady‐state frequency selectivity. Fibers with low CF (below 1 kHz) needed 10‐ms bursts to approach steady‐state rate selectivity. However, temporal encoding of the stimulus frequency was established in low‐CF fibers within 5 ms. Rapid establishment of temporal encoding is also characteristic of responses to the consonant‐vowel sounds: response patterns in successive 8‐ms epochs accurately reflected formant frequencies in those epochs. Responses to FM sweeps essentially encoded instantaneous stimulus frequency. The basic characteristics of click responses for any fiber were generally predictable by a minimum‐phase model derived from its threshold‐response curve. This model also generated realistic tone‐burst responses. Our evidence indicates that each primary fiber is a short‐term spectral analyzer whose transient behavior approximates that of a filter having its threshold‐response curve. [Work supported by NIH.]

Neural basis of amplitude-spectrum representation in auditory cortex of the mustached bat.

Neural basis of amplitude-spectrum representation in auditory cortex of the mustached bat.

Subglottic pressure and the control of phonation by the echolocating bat,Eptesicus

1. The respiratory dynamics of phonation, particularly the relationship between subglottic pressure and the sound pressure level (SPL), frequency and duration of vocalizations by nine bats (Eptesicus fuscus) is described. 2. Subglottic pressure rises to about 30 or 40 cm H2O immediately prior to emission of a single FM pulse or group of pulses, respectively, and drops 5 to 20 cm H2O during the course of each pulse (Fig. 2). 3. The maximum SPL of the echolocative pulse is positively correlated with the magnitude of the subglottic pressure at the onset of phonation in eight bats. The unusually high subglottic pressure in bats is an adaptation for the production of high intensity orientation pulses, thus increasing the range of echolocation. 4. The maximum SPL was proportional to the subglottic pressure at the pulse onset raised to a power which ranged from a mean of 0.4 to 1.36 for individualEptesicus (Fig. 3 and Table 1). The approximately linear relationship between sound pressure level and onset subglottic pressure in several bats suggests the vocal tract has a minimal effect on pulse intensity. 5. Within a given vocalization, upward sweeping FM is usually associated with either constant or increasing subglottic pressure (Figs. 4 and 5), but the absolute value of the subglottic pressure at pulse onset is correlated with the initial frequency of vocalization in only four of these bats (Table 2). 6. Subglottic pressure drops most rapidly during short pulses, but during long vocalizations (exceeding 20 ms) it attains a minimum rate of decline of about 0.2 cm H2O/ms (Fig. 6 and Table 2), which may be determined by the maximum glottal resistance at which phonation can occur. 7. The relatively non-compliant lung ofEptesicus (mean static compliance = 0.021 ±0.002 ml/cm H2O) may be an adaptation for the production of the unusually high subglottic pressures needed to produce high intensity echolocative pulses.

Perception of transmission-echo periodicity in bat sonar

Echolocating bats (Eptesicus fuscus) detect displacements in the arrival times of sonar echoes as small as 500 ns. They can thus detect differences in target range or fluttering motions of prey resulting in displacements in range as small as 100–200 μm. The range-axis acoustic image of a planar target corresponds to the half-wave rectified crosscorrelation function between the bat's broadband, FM sonar transmissions and echoes. The bat perceives an intersignal periodicity histogram for the sounds. The FM sweep is rapid enough that the signal frequency is within the tuning curve of any given primary auditory neuron for a few periods at most. Thus, to the auditory system of Eptesicus, the sonar sound is a set of impulses with center frequencies distributed across the echolocation bandwidth. Only the first or “on” discharge of each primary neuron needs to be locked to individual stimulus waves for the neural representation of the signal to preserve transmission-echo phase or period. These bats model sonar targets as acoustic “glints” in the nervous system, and they discriminate among target features with a time-domain sonar receiver.

Acoustic reflexes of middle-ear and laryngeal muscles in the FM bat Myotis lucifugus

In Myotis lucifugus both the middle-ear muscles (MEM's) and the laryngeal muscles (LM's) are highly developed for echolocation. These muscles respond to acoustic stimuli with a shortest latency of 3 and 6 msec, respectively. The tuning curves of the MEM fibers are broader than those of the LM fibers. The lowest threshold of the acoustic reflex of the MEM's is 20 dB lower than that of the LM's. The majority of the MEM and LM fibers are tuned to sounds between 35 and 45 kHz. On the average, the LM fibers are 5 dB more sensitive to 4-msec FM signals sweeping downward 10 kHz across their BF's than to pure tones, but this is not true for equivalent upward sweeping FM signals. On the other hand, the MEM fibers show poorer responses to FM signals than pure tones, regardless of sweep direction, Electrical stimulation of the superior laryngeal nerve evokes action potentials in the MEM's with a shortest latency of 5 msec. The LM's and MEM's are apparently coupled through sensory fibers in the superior laryngeal nerve. [Work supported by University of Missouri summer research fellowship, Studienstiftung des deutschen Volkes, and NSF Grant 75-17077.]

Long-range echo sounding with a chirp source

An experiment has been performed to determine the applicability of coherent signal processing techniques to a long-range low-frequency side-looking bathymetric survey system. A linear FM sweep of 280-Hz center frequency, 10-Hz bandwidth, and 20-s duration was produced by a source at depths between 120 and 150 m towed parallel to the Blake escarpment at a range of about 250 km. Reflected signals were received by a line array towed at depths ranging from 765 to 825 m. The received signals were passed through a digital pulse compressor and averaged. Groups of the received signals were easily identified as reflections from the Blake escarpment. The experiment has shown that coherent signal processing techniques are applicable to low-frequency side-looking bathymetric survey systems.