Nonconstant quality of auditory filters in the porpoises, Phocoena phocoena and Neophocaena phocaenoides (Cetacea, Phocoenidae) Vladimir V. Popova兲 and Alexander Ya. Supinb兲 Institute of Ecology and Evolution, The Russian Academy of Sciences, 33 Leninsky Prosp., 119071 Moscow, Russian Federation
Ding Wangc兲 and Kexiong Wangd兲 Institute of Hydrobiology, The Chinese Academy of Sciences, Wuhan, Hubei 430072, People’s Republic of China
共Received 10 October 2005; revised 24 January 2006; accepted 14 February 2006兲 Simultaneous tone-tone masking in conjunction with the envelope-following response 共EFR兲 recording was used to obtain tuning curves in porpoises Phocoena phocoena and Neophocaena phocaenoides asiaeorientalis. The EFR was evoked by amplitude-modulated probes with a modulation rate of 1000 Hz and carrier frequencies from 22.5 to 140 kHz. Equivalent rectangular quality QERB of the obtained tuning curves varied from 8.3–8.6 at lower 共22.5–32 kHz兲 probe frequencies to 44.8–47.4 at high 共128–140 kHz兲 frequencies. The QERB dependence on probe frequency could be approximated by regression lines with a slope of 0.83 to 0.86 in log-log scale, which corresponded to almost frequency-proportional quality and almost constant bandwidth of 3– 4 kHz. Thus, the frequency representation in the porpoise auditory system is much closer to a constant-bandwidth rather that to a constant-quality manner. © 2006 Acoustical Society of America. 关DOI: 10.1121/1.2184290兴 PACS number共s兲: 43.80.Lb 关WWA兴
Pages: 3173–3180
I. INTRODUCTION
The auditory system of mammals functions as a bank of frequency-tuned bandbass filters which performs frequency analysis of sound signals. In the majority of mammals, passbands of these filters are roughly frequency proportional 共except lowest frequencies兲; in other words, filter quality 共the ratio of center frequency to the passband width兲 is almost constant throughout a major part of the frequency range of hearing. This feature is known as a constant-Q 共quality兲 representation of frequencies in the auditory system. In humans, this regularity is known from the very first measurements of critical ratios 共Fletcher, 1940兲 and critical bands, which are psychophysical equivalents of auditory filters 共Zwicker, 1961兲. Later this regularity was confirmed in a variety of psychophysical investigations in humans with the use of various paradigms of frequency-tuning measurements: frequency-tuning curves 关see Zwicker 共1974兲 for a review兴, narrow-band noise masking 关critical bands, see Zwicker 共1982兲 for a review兴, notch-noise masking 共Patterson, 1976; Patterson et al., 1982兲, not to mention numerous other studies. To summarize the great number of psychophysical investigations, several analytical expressions have been proposed to describe filter bandwidth variation with central frequency 共Greenwood, 1961; Zwicker and Terhardt, 1980; Moore and Glasberg, 1983; Glasberg and Moore, 1990兲. All of them are
a兲
Electronic mail:
[email protected] Electronic mail:
[email protected] c兲 Electronic mail:
[email protected] d兲 Electronic mail:
[email protected] b兲
J. Acoust. Soc. Am. 119 共5兲, May 2006
rather close to one another. For example, a simple equation given by Glasberg and Moore 共1990兲 may be considered. Its original form suggested by the authors is ERB = 24.7共4.37F + 1兲,
共1兲
where ERB is the filter equivalent rectangular bandwidth in Hz and F is the central frequency in kHz. With expression of both ERB and frequency in equal units, this equation may be presented as ERB = 0.108F + 24.7 Hz,
共2兲
where F is frequency expressed in the same units as ERB. That is, at all frequencies above a few hundred Hz, where the constant of 24.7 Hz can be neglected, the filter bandwidth is of an almost constant proportion 共10.8%兲 of frequency; respectively, the filter quality QERB is constant and equal to 1 / 0.108= 9.26. The very same regularity has been demonstrated by tuning curves obtained in humans and some mammals using evoked potentials such as cochlear action potentials 共AP兲 共Dallos and Cheatham, 1976; Eggermont, 1977; Harris, 1978; Abbas and Gorga, 1981; Gorga and Abbas, 1981; Harrison et al., 1981兲 and auditory brain-stem responses 共ABRs兲 共Mitchell and Fowler, 1980; Salvi et al., 1982; Gorga et al., 1983; Brown and Abbas, 1987兲. The majority of these studies presented the quality of the obtained tuning curves as Q10 共the center frequency divided by the bandwidth at a level 10 dB above the tip of the curve兲 which is about twice lower than QERB, but the principal result was the same: the quality was almost constant throughout a wide frequency range.
0001-4966/2006/119共5兲/3173/8/$22.50
© 2006 Acoustical Society of America
3173
These and many other data led to a generally adopted idea that each relative frequency unit 共e.g., octave兲 occupies a constant space on the basilar membrane 共about 4 mm per octave in humans兲, as well as each critical band occuping a constant space 共Greenwood, 1961兲. The result is a frequencyproportional filter 共or critical band兲 width, which is the same as a constant filter quality. The only well-known significant deviation from this rule is the auditory system of mustache bats. It has expanded representation of a narrow 共a few kHz兲 frequency band around 60 kHz with much higher filter quality than at frequencies both above and below this “acoustic fovea” 共e.g., Suga, 1978; Pollak and Casseday, 1989; Pollak, 1992兲. This exception of a common rule is obviously associated with properties of the biosonar of mustache bats, which employs narrow-band acoustic pulses of around 60-kHz frequency and requires very fine frequency analysis in just this frequency range to detect Doppler shift arising due to movement of the prey. In the study described herein, we encountered another intriguing deviation from the rule of constant-Q frequency representation in the auditory system. We investigated hearing abilities of porpoises 共Phocoenidae, Cetacea兲. The investigations of hearing of these small cetaceans were motivated both by concern for their protection and conservation and for their unique bioacoustics. Similar to many other odontocetes 共toothed whales, dolphins, and porpoises兲, phocoenids have extremely wide frequency range 共more than 150 kHz兲 and high sensitivity of hearing, obviously associated with their ability to echolocate 共rev. Au, 1993; Supin et al., 2001兲. However, contrary to many other odontocetes, phocoenids use for echolocation narrow-band pips instead of wide-band clicks 共Dubrovskii et al., 1971; Møhl and Andersen, 1973; Kamminga et al., 1986, 1996; Kamminga, 1988; Li et al., 2005兲. Therefore, some specific features of frequency tuning of hearing might be expected. Bearing this in mind, measurements of frequency tuning of porpoise’s hearing were carried out. The measurements were done with use of the evokedpotential method in conjunction with the simultaneous tonetone masking paradigm. The tone-tone masking 共both the probe and masker are tones兲 provides tuning curves 共masked thresholds as a function of frequency difference between the probe and masker兲 which reflects the auditory frequencytuned filter form. The use of tuning-curve paradigm in conjunction with evoked-potential recording has already demonstrated its effectiveness for measurements of frequency tuning of hearing in a few species of cetaceans 共Supin et al., 1993; Popov et al., 1995; Supin et al., 2001兲. Therefore, the same method was used to investigate hearing of porpoises. These measurements provided unexpected results indicative of frequency representation more close to constant-B 共bandwidth兲 rather than to constant-Q 共quality兲. These results are presented herein. II. METHODS A. Subjects, facilities, and experimental design
Experimental animals were one harbor porpoise Phocoena phocoena 共an adult female兲 and two Yangtze fin3174
J. Acoust. Soc. Am., Vol. 119, No. 5, May 2006
less porpoises Neophocaena phocaenoides asiaeorientialis, a male and female. Investigation of the harbor porpoise was carried out in facilities of the Utrish Marine Station of the Russian Academy of Sciences, Black Sea coast, Russia. The animal was kept in a pool 共9 ⫻ 4 ⫻ 1.2 m3兲 filled with sea water. During the experiment, the animal was taken from the home pool and placed on a stretcher in a plastic bath 4 ⫻ 0.6⫻ 0.6 m3 filled with sea water in such a manner that the dorsal surface of the head with the blowhole remained above the water surface. The walls and bottom of the bath and the water surface in front of the animal’s head were covered by soundabsorbing material 共rubber with cone-shaped closed air cavities兲 to reduce sound reflections and make the stimulus sound field more uniform. Investigation of the finless porpoises was carried out in facilities of the Institute of Hydrobiology of the Chinese Academy of Sciences, Wuhan, P. R. China. The animals were kept in a pool 共20⫻ 7 ⫻ 3.0 m3兲 filled with fresh water. During the experiment, the animal was taken out of the home pool, placed on a stretcher in a wooden bath 2.25⫻ 0.85 ⫻ 0.6 m3 filled with water, and positioned in the same manner as the harbor porpoise. The walls and bottom of the bath were covered by open-cell neoprene to reduce sound reflection. The care and use of all the animals were performed under the Guidelines of the Russian Ministry of Higher Education on the use of animals in biomedical research adhering to the Ethical Principles of the Acoustical Society of America. B. Evoked-potential recording
The evoked-potential recording technique was similar for both the harbor porpoise and finless porpoises. For noninvasive evoked-potential recording, suction-cup electrodes were used consisting of a 15-mm stainless steel disk mounted within a 60-mm silicon suction cup. The active electrode was fixed at the vertex head surface, 5 cm behind the blowhole, above the water surface. The reference electrode was fixed at the dorsal or pectoral fin. The electrodes were connected by shielded cables to the input of a custommade EEG amplifier that provided 88-dB gain within a frequency range of 200 to 5000 Hz, as defined at −3-dB points of 6-dB/oct slopes. The amplified signal was digitized and collected using an E-6040 data acquisition board 共National Instruments兲 and stored in computer memory. To extract signal from noise, the digitized signal was coherently averaged 共1000 original records per one averaged record兲 using triggering from the stimulus onset. C. Sound signals
Sound signals were digitally synthesized at an update rate of 500 kHz and digital-to-analog converted by the same E-6040 board, amplified, attenuated, and played through a B&K 8104 transducer. The transducer was positioned at a distance of 60 cm in front of the animal’s head, near the front wall of the bath. The playback channel was calibrated both before and after the experiments by positioning a calibrated Popov et al.: Nonconstant quality of porpoises’ auditory filters
receiving hydrophone 共B&K 8103兲 at the same location as the animal’s head. Both probe and masker levels are specified below as their rms values. Two kinds of sound signals were used to obtain frequency-tuning curves: probes and maskers. Probes were 20-ms-long sinusoidally amplitude-modulated tone bursts, modulation rate was 1000 Hz, and modulation depth was 100%. These parameters of modulation were chosen since they are optimal to produce in odontocetes 共dolphins and porpoises兲 a robust rhythmic evoked-potential response 共Supin et al., 2001兲. Carrier frequencies of probes varied from 22.5 to 140 kHz, separated by 41 - or 81 -oct steps. Modulation always started from a zero phase in coherence with evokedresponse collection. The probe bursts were presented at a rate of 10 s−1. Maskers were continuous nonmodulated tones. Their frequencies varied around the probe-carrier frequency with steps of 0.5 to 1.5 kHz. Intensities of probes and maskers were varied by independent attenuators with 5-dB steps; after attenuation, the probe and masker were mixed and emitted through the common transducer. Even though the walls of the baths were covered with sound-attenuating material, some sound reflections from the bath walls and water surface were inevitable, thus resulting in interference patterns in the small enclosed space. To assess the influence of these interference patterns on the stimulus parameters, sounds were monitored by a B&K 8103 hydrophone near the animal’s head. The monitoring showed that despite the sound reflections within the bath, the real modulation depth of the stimuli remained not less than 70%–80%, and local sound levels varied by not more than 5 dB.
FIG. 1. Examples of EFR records 共A兲 and their frequency spectra 共B兲 at different masker-to-probe ratios. The finless porpoise, male. Probe 128 kHz, 90 dB re : 1 Pa; masker 128 kHz, level 共dB re probe兲 indicated near the records, N/M —no masker 共control兲.
III. RESULTS A. Evoked response waveform and manifestation of masking D. Masked threshold evaluation and tuning-curve derivation
For tuning-curve derivation, probe levels 40 dB above absolute ABR thresholds were used. Thus, at thresholds from 50 to 80 dB re : 1 Pa depending on frequency 共Popov et al., 1995, 2005兲, probe levels varied from 90 to 120 dB re : 1 Pa. For masked threshold evaluation, masker level was varied while keeping the probe level constant. A 16-mslong part of the rhythmic evoked-potential response to the sinusoidally modulated probe stimulus, from 6 to 22 ms, was Fourier transformed to obtain its frequency spectrum. The magnitude of 1-kHz peak was plotted as a function of masker intensity and an oblique near-threshold part of the plot not longer than 20 dB was approximated by a straight regression line 关the criteria for selection of a range for straight-line approximation are motivated in more detail by Supin et al. 共2001兲兴. The intersection of this line with the zero-amplitude level was adopted as a masked threshold estimate. This threshold-determination procedure was repeated at 6 to 12 masker frequencies around the probe frequency. The resulting function of masked threshold versus masker frequency was taken as the tuning curve.
Amplitude-modulated probe sounds evoked robust rhythmic responses, which followed the modulation rate, i.e., the envelope-following response 共EFR兲. As Fig. 1共A兲 exemplifies, a probe of a moderate intensity 共40 dB above the response threshold兲 without a masker evoked a response of more than 1 V peak-to-peak amplitude 共no-masker control兲. The burst onset evoked a small transient on-response, which after a few milliseconds transformed to EFR. Both the start and the end of the response appeared with a few ms lag relative to the stimulus. This lag provided a good opportunity to check artifact contamination of records. The response-free initial part of the records showed clearly that the records were not contaminated with electromagnetic artifacts; as well, the response persistence until about 4.5 ms after the stimulus end showed the physiological nature of the response. The same probe in masker background evoked smaller responses. As the masker level increased 共from 0 to 15 dB above the probe level in Fig. 1兲, EFR diminished until it disappeared in the noise. Frequency spectra of the records are shown in Fig. 1共B兲. These spectra were obtained by Fourier transform of a part of the record, from 6 to 22 ms. This 16-ms window contained a
J. Acoust. Soc. Am., Vol. 119, No. 5, May 2006
Popov et al.: Nonconstant quality of porpoises’ auditory filters
3175
FIG. 2. EFR magnitude dependence on masker-to-probe ratio at different masker-to-probe frequency spacings. The finless porpoise, male. Probe 128 kHz, 90 dB re : 1 Pa. Masker frequencies: 128 kHz 共1兲, 131.5 kHz 共2兲, and 135 kHz 共3兲. Masker level 共abscissa兲 in dB re probe level.
major part of the EFR record but did not contain the latency and the initial transient part of the response. At masker levels below 15 dB, all the spectra featured a peak at the probe envelope frequency of 1 kHz. At the masker level of 10 dB, this peak was comparable with the spectrum noise level but still definite, and at the masker level of 15 dB it completely disappeared. Figure 2 demonstrates the masked threshold determination procedure. The magnitude of the 1-kHz peak of the response spectrum was taken as an estimate of the response magnitude and plotted against masker level. The nearthreshold part of the plot could be satisfactorily approximated by a straight regression line crossing the zeroamplitude level in a point accepted as the threshold estimate. As it usually is in the tuning-curve paradigm, masked thresholds depended on frequency difference between the probe and masker. The lowest masked threshold 共i.e., the most effective masking兲 appeared at a coincidence of the probe and masker frequencies. With the difference increase, the masked threshold increased. Figure 2 exemplifies response amplitude versus masker intensity functions at three probe-to-masker frequency differences: 共1兲 zero, 共2兲 3.5 kHz, and 共3兲 7 kHz. The example demonstrates increasing the masked threshold from 15.0 dB 共relative to the probe level兲 at the zero difference to 35.5 dB at a 3.5-kHz difference and further to 48.6 dB at a 7-kHz difference. Using the same procedure, masked thresholds were determined at a variety of masker frequencies around the probe frequency to obtain a tuning curve as a masked threshold versus masker frequency function 共at a certain probe frequency兲. The same measurements were repeated at probe frequencies from 22.5 kHz 共in the harbor porpoise兲 or 32 kHz 共in the finless porpoises兲 to 140 kHz, thus obtaining a family of tuning curves. Lower probe frequencies were not tested since evoked-response amplitude was too low at lower frequencies 关as described by Popov and Supin 共2001兲兴 and did not provide satisfactory precision of measurements. Higher probe frequencies were not tested because of proximity to the upper limit of the frequency range of hearing. The family of tuning curves obtained from the harbor 3176
J. Acoust. Soc. Am., Vol. 119, No. 5, May 2006
FIG. 3. Tuning curve families obtained in three subjects: the harbor porpoise 共A兲, finless porpoise female 共B兲, and finless porpoise male 共C兲. The curves are masked thresholds as functions of masker frequency. Dots represent experimental data, curves—roex approximations.
porpoise is presented in Fig. 3共A兲, and similar families obtained in the finless porpoises are presented in Figs. 3共B兲 and 3共C兲. All the families looked rather similar. In particular, a noticeable feature of all the tuning curves was that they are not of a constant width when plotted on a log frequency scale, thus indicating nonconstant Q across the frequency scale: high-frequency tuning curves were much more acute than low-frequency ones. In order to characterize the curves quantitatively, the experimental points were approximated by the roundedexponential 共roex兲 function, which is widely used to approximate auditory filters. Its original form suggested by Patterson et al. 共1982兲 is
W共g兲 = 共1 + pg兲e−pg ,
共3兲
where W共g兲 is the filter form in terms of transferred power, g is the normalized frequency deviation from the center, and the term p determines the filter tuning. ERB of the roex function is Popov et al.: Nonconstant quality of porpoises’ auditory filters
B=F
冕
⬁
共1 + pg兲e−pg dg = 4F/p,
共4兲
−⬁
where B is ERB and F is the center frequency. Respectively, the quality Q is Q = F/B = p/4.
共5兲
Curves like those presented in Fig. 3 may be approximated by an inverted roex function with W共g兲 presented on a dB scale: T共g兲 = Tmin − 10 log 关W共g兲兴,
共6兲
where T共g兲 is the masked threshold and Tmin is the minimal masked threshold at the curve tip. To approximate experimental data by roex functions, the parameters Tmin and p were adjusted to reach the best fit to experimental data according to the least-mean-square criterion. These approximating functions are presented in Fig. 3 along with the experimental points. The obtained value of p was used to calculate ERB and Q of the curves according to the equations presented above. The result was obvious dependence of Q on the probe frequency: from 8.3–8.6 at rather low frequencies 共22.5–32 kHz兲 to 44.8–47.4 at high frequencies 共128–140 kHz兲. All the obtained values of Q are presented in Fig. 4 as functions of probe frequency. All the functions look rather similar: neglecting some data scatter, there was an almost proportional increase of Q with frequency. Being presented on log-log scales, these functions were satisfactorily approximated by straight lines 共r2 from 0.85 to 0.91, Table I兲.The slope of these regression lines in the three subjects varied from 0.83 to 0.86, as presented in Table I. In other words, the dependence of Q on frequency might be approximated by a function Q = q 0F k ,
共7兲
where F is frequency, q0 is a quality at F = 1, and k determines the degree of Q dependence on F. Constant 共frequency-independent兲 Q corresponds to k = 0; frequencyproportional Q corresponds to k = 1. The found values of k from 0.83 to 0.86 indicated not exact but almost frequencyproportional Q. With such degree of dependence, approximated Q increased from around 10 to around 40 within a frequency range from 30 to 150 kHz. According to Eqs. 共5兲 and 共7兲, ERB dependence on frequency was approximated as B = F/Q = F1−k/q0 .
FIG. 4. Frequency-tuning quality 共Q兲 dependence on frequency in three subjects: the harbor porpoise 共A兲, finless porpoise female 共B兲, and finless porpoise male 共C兲. Solid lines—experimental data, thin straight lines— approximating regression lines. k is indicated for each regression line.
spatial variation of both the sound level and modulation depth. Nevertheless, we suppose that the obtained tuning curves are credible because of the reasons as follows. 共i兲
共8兲
ERB approximations computed in such a way were little dependent on frequency, e.g., from around 3 kHz at 30-kHz frequency to around 4 kHz at 150-kHz frequency.
IV. DISCUSSION
共ii兲
As shown 共Supin and Popov, 1995兲, variation of modulation depth influences the envelope following response amplitude rather than threshold. So, it could not influence the form of tuning curves based on masked thresholds. Variation of sound 共both the probe and masker兲 levels by ±5 dB does directly influence the masked threshold estimates. However, this variation should increase the data scatter, but could hardly systematically influence the width of tuning curves.
The tuning curves presented above were obtained in conditions of nonideal acoustic field. Even though the walls of the experimental baths were covered with soundattenuating material, sound interference resulted in some
At least two features of the porpoise’s tuning curves described above deserve attention: 共i兲 very high quality 共up to 40兲 in the upper part of the hearing frequency range and
J. Acoust. Soc. Am., Vol. 119, No. 5, May 2006
Popov et al.: Nonconstant quality of porpoises’ auditory filters
3177
TABLE I. Q dependence on frequency in different mammal species. r2
Harbor porpoise Finless porpoise female Finless porpoise male Mean
Present data 0.83± 0.10 0.86± 0.09 0.84± 0.12 0.84± 0.06
0.87 0.91 0.85 0.88
References
22–140 32–139 32–139
Human
Summarizing analytical expression 0.06± 0.01 1–16
Glasberg and Moore 共1990兲
Cat Cat Mouse Chinchilla Macaque
Psychophysical critical bands 共F / B兲 −0.13± 0.05 1–8 0.12± 0.09 4–12 −0.03± 0.05 10–50 0.21± 0.07 1–4 −0.13± 0.07 1–16
Pickles 共1975兲 Nienhuys and Clark 共1979兲 Ehret 共1976兲 Seaton and Trahiotis 共1975兲 Gourevitch 共1970兲
Chinchilla Macaque
Psychophysical tuning curves 共Q10 dB兲 0.32± 0.26 1–12 0.00± 0.02 1–16
McGee et al. 共1976兲 Serafin et al. 共1982兲
Evoked-potential tuning curves 共Q10 dB兲 0.40± 0.06 2–8 0.18± 0.07 11–110 0.54± 0.12 32–108
Mitchell and Fowler 共1980兲 Popov et al. 共1995兲 Klishin et al. 共2000兲
Guinea pig Bottlenose dolphin Beluga whale
共ii兲 tuning dependence on frequency which is much closer to a constant-bandwidth rather than to a constant-quality manner. The frequency tuning of QERB ⬇ 40 is several times more acute than that in many mammals and humans 共QERB ⬇ 10, see Sec. I兲. However, this feature of the porpoise’s tuning curves is not very surprising, since auditory filters with quality higher than 30 were already described in a few dolphin species 共Supin et al., 1993; Popov et al., 1995, 1997; Klishin et al., 2000兲. The cause of very acute frequency tuning in dolphins and porpoises is not known yet, but we may hypothesize that it is associated with the wide frequency range of their hearing. Indeed, both frequency tuning and temporal resolution in the auditory system are limited by passbands of peripheral filters. Acute frequency tuning requires narrow passbands whereas high temporal resolution requires wide passbands. Thus, increasing frequency-tuning acuity above a certain limit is only possible at the cost of decreased temporal resolution and vice versa. However, this contradiction becomes less limiting at high frequencies, because at high frequency F, auditory filters can combine high quality with wide passband. Thus, at high frequencies, acute frequency tuning does not prevent high temporal resolution. Odontocetes 共toothed whales, dolphins, and porpoises兲 have frequency ranges of hearing of more than 100 kHz, in particular, up to 150 kHz in porpoises, which is several times wider than in the majority of terrestrial mammals 共except bats兲. This frequency range allows combining acute frequency tuning with high temporal resolution. The dependence of tuning on frequency found in porpoises was unexpected. Contrary to the majority of terrestrial 3178
Range 共kHz兲
k ± SE
J. Acoust. Soc. Am., Vol. 119, No. 5, May 2006
mammals and humans, this dependence is much closer to constant-B rather than to constant-Q representation. In principle, the constant-B frequency representation is not prohibited by any auditory mechanisms. In particular, it is typical of the lowest part of the frequency range of the human’s hearing: as follows from Eqs. 共1兲 and 共2兲, at low frequencies, ERB cannot be infinitely narrow and therefore approximates a constant value of around 25 Hz. But it is only typical of the lowest part of the frequency range. Within the upper part, a constant-Q 共or approaching constant-Q兲 representation is more typical of both humans and a variety of mammals. We analyzed, retrospectively, data obtained in a few mammal species which are commonly used in hearing investigations, in order to calculate Q dependence on frequency within an upper part of the frequency range. For critical-band data 共Gourevitch, 1970; Pickles, 1975; Seaton and Trahiotis, 1975; Ehret, 1976; Nienhuys and Clark, 1979兲, Q was calculated as the frequency-to-band ratio; for tuningcurve data 共McGee et al., 1976; Mitchell and Fowler, 1980; Serafin et al., 1982; Popov et al., 1995; Klishin et al., 2000兲, Q10 dB estimates were directly taken. The Q dependence on frequency was approximated by a log-log regression line, exactly in the same manner as we did for the experimental data presented above. The values obtained for the factor k 关see Eq. 共5兲兴 are presented in Table I along with our experimental data. Many of them were not far different from zero, thus indicating the constant-Q representation; in some cases, k reached values as high as 0.4–0.5, thus indicating a representation type intermediate between constant-Q and constant-B. But only in porpoises did values as high as 0.85 indicate a representation type much closer to constant-B rather than to constant-Q. Popov et al.: Nonconstant quality of porpoises’ auditory filters
Abbas, P. J., and Gorga, M. P. 共1981兲. “AP responses in forward-masking paradigms and their relationship to responses of auditory-nerve fibers,” J. Acoust. Soc. Am. 69, 492–499. Au, W. W. L. 共1993兲. The Sonar of Dolphins 共Springer-Verlag, New York兲. Brown, C. J., and Abbas, P. J. 共1987兲. “A comparison of AP and ABR tuning curves in guinea pig,” Hear. Res. 25, 193–204. Dallos, P., and Cheatham, M. A. 共1976兲. “Compound action potential 共AP兲 tuning curves,” J. Acoust. Soc. Am. 59, 591–597. Dubrovskii, N. A., Krasnov, P. S., and Titov, A. A. 共1971兲. “On the emission of echolocation signals by the Azov sea harbor porpoise,” Sov. Phys. Acoust. 16, 444–448. Eggermont, J. J. 共1977兲. “Compound action potential tuning curves in normal and pathological human ears,” J. Acoust. Soc. Am. 62, 1247–1251. Ehret, G. 共1976兲. “Critical bands and filter characteristics of the ear of the housemouse 共Mus musculus兲,” Biol. Cybern. 24, 35–42. Fletcher, H. 共1940兲. “Auditory patterns,” Rev. Mod. Phys. 12, 47–65. Glasberg, B. R., and Moore, B. C. J. 共1990兲. “Derivation of auditory filter shapes from notched-noise data,” Hear. Res. 47, 103–138. Gorga, M. P., and Abbas, P. J. 共1981兲. “Forward-masked AP tuning curves in
normal and acoustically traumatized ears,” J. Acoust. Soc. Am. 70, 1322– 1330. Gorga, M. P., McGee, J., Walsh, E. J., Javel, E., and Farley, G. R. 共1983兲. “ABR measurement in the cat using a forward-masking paradigm,” J. Acoust. Soc. Am. 73, 256–261. Gourevitch, G. 共1970兲. “Detectability of tones in quiet and in noise by rats and monkeys,” in Animal Psychophysics: The Design and Conduct of Sensory Experiments, edited by W. C. Stebbins 共Appleton, New York兲, pp. 67–97. Greenwood, D. D. 共1961兲. “Critical bandwidth and the frequency coordinates of the basilar membrane,” J. Acoust. Soc. Am. 33, 1344–1356. Harris, D. M. 共1978兲. “Action potential suppression, tuning curves and thresholds: Comparison with single fiber data,” Hear. Res. 1, 133–154. Harrison, R. V., Aran, J.-M., and Erre, J.-P. 共1981兲. “AP tuning curves from normal and pathological human and guinea pig cochleas,” J. Acoust. Soc. Am. 69, 1374–1385. Kamminga, C. 共1988兲. “Echolocation signal types of odontocetes,” in Animal Sonar, Processes and Performance, edited by P. E. Nachtigall, and P. W. B. Moore 共Plenum, New York兲, pp. 9–22. Kamminga, C., Cohen, S. A. B., and Silber, G. K. 共1996兲. “Investigation of cetacean sonar XI: Intrinsic comparison of the wave shapes of some members of the Phocoenidae family,” Aquat. Mamm. 22, 45–55. Kamminga, C., Kataoka, T., and Engelsma, F. J. 共1986兲. “Investigation of cetacean sonar VII: Underwater sounds of Neophocaena phocaenoides of the Japanese coastal population,” Aquat. Mamm. 12, 52–60. Klishin, V. O., Popov, V. V., and Supin, A. Ya. 共2000兲. “Hearing capabilities of a beluga whale, Delphinapterus leucas,” Aquat. Mamm. 26共3兲, 212– 228. Li, S., Wang, K., Wang, D., and Akamatsu, T. 共2005兲. “Echolocation signals of the free-ranging Yangtze finless porpoise 共Neophocaena phocaenoides asiaeorientalis兲,” J. Acoust. Soc. Am. 117, 3288–3296. McGee, T., Ryan, A., and Dallos, P. 共1976兲. “Psychophysical tuning curves of chinchillas,” J. Acoust. Soc. Am. 60, 1146–1150. Mitchell, C., and Fowler, C. 共1980兲. “Tuning curves of cochlear and brainstem responses in the guinea pig,” J. Acoust. Soc. Am. 68, 896–900. Møhl, B., and Andersen, S. 共1973兲. “Echolocation: High frequency component in the click of the harbour porpoise 共Phocoena phocoena L.兲,” J. Acoust. Soc. Am. 54, 1368–1373. Moore, B. C. J., and Glasberg, B. R. 共1983兲. “Suggested formulae for calculating auditory filter bandwidths and excitation patterns,” J. Acoust. Soc. Am. 74, 750–753. Nienhuys, Y. W., and Clark, G. M. 共1979兲. “Critical bands following the selective destruction of cochlear inner and outer hair cells,” Acta OtoLaryngol. 88, 350–358. Patterson, R. D. 共1976兲. “Auditory filter shapes derived with noise stimuli,” J. Acoust. Soc. Am. 59, 640–654. Patterson, R. D., Nimmo-Smith, I., Weber, D. L., and Milory, R. 共1982兲. “The deterioration of hearing with age: Frequency selectivity, the critical ratio, the audiogram, and speech threshold,” J. Acoust. Soc. Am. 72, 1788–1803. Pickles, J. O. 共1975兲. “Normal critical bands in the cat,” Acta Oto-Laryngol. 80, 245–254. Pollak, G. D. 共1992兲. “Adaptation of basic structures and mechanisms in the cochlea and central auditory pathway of the mustache bat,” in The Evolutionary Biology of Hearing, edited by D. B. Webster, R. R. Fay, and A. N. Popper 共Springer, New York兲, pp. 751–778. Pollak, G. D., and Casseday, J. H. 共1989兲. The Neural Basis of Echolocation in Bats 共Springer, Heidelberg兲. Popov, V. V., and Supin, A. Ya. 共2001兲. “Contribution of various frequency bands to ABR in dolphins,” Hear. Res. 151, 250–260. Popov, V. V., Supin, A. Ya., and Klishin, V. O. 共1995兲. “Frequency tuning curves of the dolphin’s hearing: Envelope-following response study,” J. Comp. Physiol., A 178, 571–578. Popov, V. V., Supin, A. Ya., and Klishin, V. O. 共1997兲. “Frequency tuning of the dolphin’s hearing as revealed by auditory brain-stem response with notch-noise masking,” J. Acoust. Soc. Am. 102, 3759–3801. Popov, V. V., Supin, A. Ya., Wang, D., Wang, K., Xiao, J., and Li, S. 共2005兲. “Evoked-potential audiogram of the Yangtze finless porpoise Neophocaena phocaenoides asiaeorientalis 共L兲,” J. Acoust. Soc. Am. 117, 2728– 2731. Salvi, R. J., Ahroon, W. A., Perry, J. W., Gunnarson, A. D., and Henderson, D. 共1982兲. “Comparison of psychophysical and evoked potential tuning curves in the chinchilla,” Am. J. Otolaryngol. 3, 408–416. Seaton, W. H., and Trahiotis, C. 共1975兲. “Comparison of critical ratios and
J. Acoust. Soc. Am., Vol. 119, No. 5, May 2006
Popov et al.: Nonconstant quality of porpoises’ auditory filters
It is remarkable that frequency-tuning measurements in a few dolphin species carried out with the use of the very same technique 共tuning curves obtained with sinusoidally amplitude-modulated tones as a probe, continuous tones for simultaneous masking, and EFR as an indicative response兲 resulted in a representation type much closer to constant-Q 共Popov et al., 1995; Klishin et al., 2000兲. So, the unusual manner of frequency-tuning dependence on frequency in porpoises cannot be explained by some peculiarities of the methods used for hearing investigation in cetaceans. It may be considered as a real feature of the porpoise’s auditory system. It should be noted that the results presented above do not cover the entire frequency range of hearing but only the upper part of this region. We failed to perform measurements precise enough at lower frequencies, so the type of frequency representation in the low-frequency range in porpoises remains uncertain. But the investigated high-frequency region is just one where constant-Q representation appears normally in other mammals. So the representation manner in porpoises really differs from that in other mammals and humans. Again, we know nothing about the cause of such type of frequency representation. One of the advantages of such an organization is that a constant bandwidth of peripheral auditory filters provides a constant limit for temporal resolution throughout all the frequency range. A filter of a bandwidth B is capable of transferring modulation rates up to B / 2; thus, the porpoise’s auditory filters of a bandwidth of 3–4 kHz are capable of transferring amplitude modulations of a signal as fast as 1.5–2 kHz. This value is very close to estimates of temporal resolution of the dolphin’s and porpoise’s hearing obtained by both psychophysical and evoked-potential methods 关see reviews Au 共1993兲 and Supin et al. 共2001兲兴. ACKNOWLEDGMENTS
Grants provided by National Foundation of Natural Sciences of China 共30170142兲, the Chinese Academy of Sciences 共KSCX2-SW-118兲, the Institute of Hydrobiology, the Chinese Academy of Sciences 共220103兲, Russian Foundation for Basic Research 共02-04-39017, 03-04-48117兲, and Russian Ministry of Science and Education 共NSh-2152.2003.4兲 are greatly appreciated.
3179
critical bands in the monaural chinchilla,” J. Acoust. Soc. Am. 57, 193– 199. Serafin, S. V., Moody, D. B., and Stebbins, W. C. 共1982兲. “Frequency selectivity of the monkey’s auditory system: Psychophysical tuning curves,” J. Acoust. Soc. Am. 71, 1513–1518. Suga, N. 共1978兲. “Specialization of the auditory system for reception and processing of species-specific sounds,” Fed. Proc. 37, 2342–2354. Supin, A. Ya., and Popov, V. V. 共1995兲. “Envelope-following response and modulation transfer function in the dolphin’s auditory system,” Hear. Res. 92, 38–46. Supin, A. Ya., Popov, V. V., and Klishin, V. O. 共1993兲. “ABR frequency tuning curves in dolphins,” J. Comp. Physiol., A 173, 649–656.
3180
J. Acoust. Soc. Am., Vol. 119, No. 5, May 2006
Supin, A. Y., Popov, V. V., and Mass, A. M. 共2001兲. The Sensory Physiology of Aquatic Mammals 共Kluwer, New York兲. Zwicker, E. 共1961兲. “Subdivision of the audible frequency range into critical bands 共Frequenzgruppen兲,” J. Acoust. Soc. Am. 33, 248. Zwicker, E. 共1974兲. “On a psychoacoustical equivalent of tuning curves,” in Facts and Models in Hearing, edited by E. Zwicker and E. Terhardt 共Springer, Berlin兲, pp. 132–141. Zwicker, E. 共1982兲. Psychoacustic 共Springer, Berlin兲. Zwicker, E., and Terhardt, E. 共1980兲. “Analytical expression for criticalband rate and critical bandwidth as a function of frequency,” J. Acoust. Soc. Am. 68, 1523–1525.
Popov et al.: Nonconstant quality of porpoises’ auditory filters
