Page 47 - 2017Winter
P. 47

   Figure 2. Spectrograms for a series of FM echolocation sounds emit- ted by a big brown bat during an aerial interception. Time is in sec- onds; frequency is in kilohertz. Most of the information illustrated by these spectrograms would have been available to anyone thinking about echolocation by the late 1950s (see Griffin, 1958, Plate 8). The sounds shorten in duration from 20 ms down to 0.3 ms as the ma- neuver progresses and the bat flies closer to its target. The bat likely detects the insect from the echo of the broadcast that occurs at about 0.27-0.28 s because subsequent broadcasts have more gradual on- sets and increase progressively in the slope of the FM sweep at the very beginning of the sound. Each sound contains several harmonics (curved sloping ridges), with the strongest shown as the darkest. The 1st harmonic changes from shallow FM sweeps in the 25- to 30-kHz region to wide FM sweeps from 60 to 25 kHz.
extends far enough to encompass the arrival of echoes from objects at different distances (dt).
A major concern for Griffin and Galambos was the potential- ly deleterious effect of intense outgoing broadcasts on sensi- tivity to weak echoes arriving immediately thereafter. Radar systems “blank” the receiver at the moment of transmission to avoid blinding the system to weak reflections from targets (Skolnik, 1980). The mammalian auditory system does have a mechanical means, the middle ear muscles, to attenuate sounds traveling from the external ears to the highly sensitive
Figure 3. Diagram of information in a bat-like downsweeping FM echolocation broadcast and an echo relevant to determining target range. FM sweep is from 90 kHz to 10 kHz. The wide bandwidth of bat sounds favors accurate determination of distance from delay. The broadcast is emitted (at t = 0) to propagate outward, impinge on an object, and return as an echo (at dt). Two-way travel time is 6 ms/m of target range. The bat has to clock the sound’s travel time (one operation) and then register target range on a perceptual scale from near to far (a different operation). For large objects such as background vegetation, operating ranges are up to 20-30 m or delays of 120-180 ms (Stilz and Schnitzler, 2012). For small objects such as flying insects, operating ranges are only about 3-5 m or delays of 18- 30 ms. Broadcasts and echoes (black), the envelopes (orange), or the spectrograms (green) supply the information used by bats. Central to these theories is the nature of the time-measurement process for target ranging and the time origin (t = 0). For a receiver that utilizes the envelopes, the onsets or the offsets are the prominent markers in both the broadcast and the echo. Other than delay, most information about the target’s effect on echo waveforms is lost.
receptors of the inner ears, but can they contract to suppress the broadcast while relaxing quickly enough to allow weak echoes to pass through? Eventually, both physiological and behavioral tests clarified the role of the middle ear muscles for attenuating broadcasts and also for regulating sensitivity to echoes (Pollak and Casseday, 1989; Neuweiler, 2000).
However, although masking of weak echoes by the preced- ing strong broadcast is a serious worry, the masking effect of one echo on another has to be considered as well. If several echoes arrive close together in time, can the bat resolve them as separate sounds well enough to perceive the correspond- ing objects? The newly available sonograms of the bat’s FM sounds suggested a solution to the echo-on-echo masking problem. Each of the bat’s FM sounds is milliseconds long (20 ms down to 2 ms during the course of most of the inter- ception in Figure 2). In contrast, the spectrogram of each sound consists of sloping ridges that trace the frequency sweep from high to low over time. Importantly, the width of this ridge along the time axis of the spectrogram is roughly half a millisecond, substantially shorter than the duration of the sound itself. If the bat receives only one echo following each broadcast, the echo in isolation would stand out well enough to represent a discrete object that could be avoided or tracked and intercepted.
 Winter 2017 | Acoustics Today | 45

   45   46   47   48   49