885095936

3:10-3:20

Break

Contributed Papers

3:20

7AB5. Physical constraints of shaliow water on acoustic communication by aquatic insects. T. G. Forrest, G. L. Miller, J. R. Zagar, and K. E. Gilbert (Natl. Ctr. for Physical Acoust. and Dept. of Biol., Univ. of Mississippi, Univcrsity, MS 38677)

Frequency responses of shaliow, freshwater ponds in northem Mississippi wcre measured. The response has a highpass characteristic with a sharp cutoff frequcncy due to the modal properties of the system. The cutoff frequency of the system is inversely related to the dcpth of water at the shallower transducer (projector or receiver). Frequencies below the first modę are nonpropagating. and the overall eflect of this envi-ronment on propagation is about 50 dB. Several species of aquatic in-sects communicate in thcse shallow-water ponds using acoustic signals, and they must contend with the physical constraints imposed by the system. Data on the calling song (long rangę) of a common aquatie beetle (Tropisternus collańs, Hydrophilidae) are presented and dis-cussed in relation to the propagation characteristics of their shaliow pond habitats. (Work supported by USDA.J

3:35

7AB6. Freguency response of the swimbladders of fish. Thomas N. Lewis, Peter H. Rogcrs, David B. Rogers (School of Mech. Eng., Georgia Insi. of Techno?., Atlanta, GA 30332), and Steven N. Flanagan (Georgia Insi. of Technol., Atlanta, GA 30332)

The gas-filled swimbladder of a fish resonates in the ambient noise field, scattering significant amounts of acoustic energy. This characteristic scattered field is thought to assist a fish’s own hearing and also may allow for the detection and Identification of other scatterers by the receiver (P. H. Rogers et ai, J. Acoust. Soc. Am. Suppl. I 85, S35 (1989)]. The frequency response of the swimbladder can be measured in uioo by a noninvasive vibration amplitudę measurement system (M. Cox and P. H. Rogers. J. Vib. Acoust. Str. Rei. Des. 109. 55-59 (1987)]. The response for a variety of fish of the species Carassius auratus (common goldfish) and Aslronotus ocellatus (oscar) was measured and cor-related with rcspect to fish size. One aspect of the results further exam-ined was the appearance of twin peaks in the response of sonie goldfish. X rays of the subjects indicated that differences in size of the anterior and posterior chambers of the swimbladder may be responsible. (Work supported by ONR.]

3:50

7AB7. Neural network simulatlon for ranging mechanism of FM-FM cortical neurons in FM bats. Zhen-Biao Lin, S. K. Chittajallu, S. Kayalar, D. Wong, and H. O. Yurtseven (Indiana University-Purdue University at Indianapolis, 1201 E. 38th St., Indianapolis. IN 46205)

A multilayer neural network is designed for modeling FM-FM neurons in the auditory cortcx of FM bai.v Thcsc neurons play a critical role as a “delay-dependent multtplier" to perform cross correlation. Data from FM-FM neurons in Myotis lucifugus were used to train differcnt artificial neural networks. Results show that one hidden layer is adequate for modeling biological FM-FM neurons. Morę hidden lay-ers give little improvement in precision, but the cost of computation

1967 J. Acoust. Soc. Am.. Vol. 89, No. 4, Pt. 2. April 1991

grows iinearly with the number of synaptic connections. Very large networks and very high-training cycles lead to the memorization of the training patterns but result in a poor prediction for testing patterns. A neural network model with one hidden layer (15 nodes) is proposed. The results from threc trained models showed that for testing patterns not included in the training set, the prediction error was smali (MSE = 0.007). The dynamie behavior of the delay-dependent multi-plicr (described by the contour of 70% of the maximum value) is reasonable from a hunting behavior standpoint. In the search phase (low-rcpctition rates), a high-output sensitivity and a very wide delay window is necessary for dctecting a target at an unknown distance. After detection, the repetition ratę inereases and the delay width drops dramatically corrcsponding to focusing to a specific target distance.

4:05

7AB8. Echolocation signals of the smallest odontocetes. Whitlow W. L. Au (Naval Ocean Systems Ctr., P.O. Box 997, Kailua, HI 96734)

Echolocation signals of the smallest odontocetes, Cephalorhynchus commersonii. Cephalorhynchus hectori (genas Cephalorhynchus), Phoc-oenoides dalii. Phocoena phocoena. and Neophocoena phocoena (family phocoenidae) are compared with those of larger dolphins. Available data seem to indicate that two distinct classes of signals exist. Echolocation signals of Tursiops truncatus. Pseudorca crassidens, and Delphi-napterus leucas studied at the Naval Ocean Systems Center have peak frequencies between 100-120 kHz, with high amplitudes (210-225 dB re: 1 p?a), short durations (50-70 ps), and wide bandwidths (30-40 kHz). Tlie smallest odontocetes tend to cmit signals having peak fre-quencies between 120 and 140 kHz, with Iow amplitudes ( < 170 dB ne:l /iPa), long durations (170-430 /is) and narrow bandwidths (7-11 kHz). Double pulses are also emitted rcgularly by the smaller odontocetes and very infrequently by the larger dolphins. The signals used by the smaller animals may reflecl an adaptation resulting from constraints associated with their smali size and may also rcflect differences in the generation mechanism, Sonie of the properties of high-frequency, narrow-bandwidth biosonar signals used by the smaller odontocetes will be discussed, along with comparative advantages and disadvantages of these signals.

4:20

7AB9. Acoustic response times (RTs) for Tursiops truncatus. S. H. Ridgway, D. A. Carder, P. L. Kamolnick, D. J. Skaar (Biosciences Div., Naval Ocean Systems Ctr., San Diego, CA 92152), and W. A. Root (Dept. of Math. C-012, Univ. of California at San Diego, La Jolla, CA 92093)

Seven dolphins (4 males and 3 females age 5-30 -f ) were trained to make underwater acoustic responses (ARs = whistles or pulsc trains) to lonal or click train (10-ms interclick interva!) stimuli (St). After first training, St delivery and AR and RT recording was Computer con-trolled. St duration varied from 60-450 ms. St were 120 dB (re: I p Pascal peak to peak at 1 m) ±24 dB in 6-dB steps. With the dolphin at 1-m depth and 1 m from St hydrophone, the trainer started a randomly variablc 3- to 20-St błock. The Computer selected St from a file in random sequence and interval (1.1-2.1 s in 0.1-$ steps) and offered St

121 st Meeting: Acoustical Society of America 1967