Search

For units of the anteroventral cochlear nucleus in the cat—as for primary units—response to a best‐frequency tone f1 can be markedly reduced when a higher‐frequency suppressing tone f2 of equal or greater intensity is presented simultaneously at a frequency separation corresponding, in the present experiments, to an optimal distance on the basilar membrane of about 0.5–0.8 mm. More data are needed, but this distance does not seem to vary systematically with best frequency and may be related to critical bandwidth. In further preliminary experiments, we have compared unit response to the combination tone, 2f2−f3, formed when f1 is omitted and f3 is added above f2, with the unit’s response to a three‐tone stimulus. The three‐primary stimulus consists of tones with equispaced frequencies and controllable phase and amplitude relations. Primary tone f1 and combination tone 2f2−f3 coincide at best frequency, while f2 and f3 lie above the tuning curve. The comparison of response to 2f2−f3 ’’alone’’ and of respons...

Calculations were made of the dimensions of traveling‐wave envelopes in four species. The calculations were based on Bekesy's measurements of the tuning curves of single points on the basilar membrane of each species, and on four empirical functions, one per species, relating frequency to position of maximum amplitude on the basilar membrane. The functions were in approximate agreement with other observations by Bekesy. It was found that the distances over which a traveling wave damps from maximum amplitude to fractions of maximum amplitude—i.e., to one‐half maximum, to two‐tenths maximum, or to “zero” amplitude, respectively—appear to be approximately constant, independent of the position of maximum amplitude on the membrane. In comparing the traveling wave envelopes of the four species, elephant, man, guinea pig, and chicken, it was found that (1) the shape of a traveling‐wave envelope is quite similar among the four species, (2) the distance over which a traveling wave damps from maximum to “zero” amplitude is the greater the longer the basilar membrane, (3) this distance, in the species compared, seems to be an approximately constant fraction of the length of the basilar membrane, independent of the length of the membrane.

In their paper, Rainbolt and Schubert incorrectly attribute at related pilot study to me owing to misrecollection of a letter from me to Schubert in 1963. The personal communication in question and did not represent the study as mine but as a student experiment performed by three graduate students of my acquaintance 3 yr before t heard about it in 1960 and 6 yr prior to my letter to Schubert. It is pointed out here that I did not give permission to cite or publish the data of others. My communication to Schubert indicated that the preliminary experiment, despite its possible importance if confirmable, had certain enumerated limitations, that should now influence the weight attached to it. The 1963 letter also directed attention to the pitch judgments of one subject, which are pertinent to points raised by Rainbolt and Schubert.

The hypothesis was offered some years ago (Greenwood, 1961) that critical bandwidth in man was an exponential function of distance on the basilar membrane. Integration of the function yielded directly a frequency‐position curve fitting Bekesy's data. Thus critical bandwidth in man was apparently proportional, and if distance was scaled in millimeters, numerically equal, to the derivative of the frequency‐position function. It was unlikely, however, that critical bandwidth would correspond to the same distance in species with cochleas of different sizes. Scale relations between frequency‐position functions of a few species, and between displacement envelopes in four species (Greenwood, 1962), suggested that critical bandwidth might correspond to scale‐related distances, implying changes among species in the constant of proportionality to the derivative of the frequency‐position function. Based on a 1‐mm distance in man, some unpublished calculations of scaled distances were made for other species. Subseque...

Analysis of Rhode's data (1971) on basilar membrane response in the squirrel monkey has permitted the phase data to be fitted by empirical functions which give cumulative phase, or travel time, for tones of different frequency to the given position. These functions were transformed, by means of two auxiliary functions, to express the travel time of a given tone to any point on the membrane. Calculations of travel times to positions of maximum amplitude closely fit estimates of travel times derived from VIIIth nerve units by Anderson et al. (1971). With scale changes in constants, the functions can be applied to other species. Calculations quite closely agree with the apical phase data of Bekesy for man. They also agree with Kohlloffel's (1972) relative phase curves from the basal half of the guinea pig cochlea, giving a phase of about 4π at maximum amplitude. Considerable agreement is found with some of the cochlear microphonic phase measurements of Eldredge and others in the guinea pig. The agreement obs...

Earlier findings indicated a decrease of about 4 dB in the ratio of signal level to masker level at a masker level of about 50 dB SL when noisebands were subcritical but not when supercritical [Greenwood, J. Acoust. Soc. Am. 33, 484–502 (1961)]. Schlauch, Weir, and Norton [J. Acoust. Soc. Am. Suppl. 1 71, 573 (1982)] reported a related result. The early data suggested a masking study in which both masker and signal were the minimum width subcritical stimulus, namely pure tones of the same frequency in‐phase, to see if consistent change would occur. Pilot tone‐on‐tone experiments in 1960 [Greenwood, Harvard P. A. L. Status Rpt. 37, 8–9 (1961)] showed a larger drop in ratio since signal and masker were coherent. Further detailed tone‐on‐tone growth of masking curves from over a dozen subjects were obtained in 1967–69 and, with the 1960 results, are reported here. A clear break in slope, of variable abruptness, usually occurs at about 50 dB SL, either rapidly or gradually dropping 5/M ratio by 6 to 8 dB. [c.f. Rabinowitz et al., J. Acoust. Soc. Am. 59, 1506–1509 (1976) for a similar result.] Addition of low‐level wideband and high‐pass noise was found to counteract the change, i.e., to raise the high level section of the growth of masking curve [see also Zwicker, Acustica 6, 365–396 (19561; Viemeister, J. Acoust. Soc. Am. 51, 1265–1269 (1972); Moore and Raab, J. Acoust. Soc. Am. 55, 1049–1054 (1974) for consistent results.] However, use of added narrow bandpass noise at variable frequencies indicates the most effective region in which to add the noise lies, at a frequency ratio of band to masking tone of 1.3 to 1.5, between masking tone and its second harmonic. These results and others [Green, J. Acoust. Soc. Am. 41, 1517–1525 (1967)] appear consistent with the results of increasing bandwidth from subcritical to supercritical widths cited above.

Accurate cochlear frequency‐position functions based on physiological data would assist interpretation of physiological and psychoacoustic data within and across species. In 1961, a family of almost‐exponential functions was developed [D. D. Greenwood, J. Acoust. Sac. Am. 33, 1344–1356 (1961)] by integrating an exponential function fitted to a subset of critical bandwidth estimates. The integral fitted well cochlear data on frequency‐place in human cadavers, and, with changes in constants, similar data on several other species—as well as behavioral‐anatomical data on cats. Since 1961, mechanical data from dead human ears and various data from living animals (cat, guinea pig, chinchilla, monkey, gerbil) have appeared. It is shown that they are fitted by the same basic function. As in 1961, in both cadaver and living ears, the same slope parameter can be used if partition length is normalized to 1; i.e., slope is scaled if distance is expressed in physical units. Slope uniformity suggests that the cadaver data may well describe at least the function's slope in living human cochleas. To set the upper frequency limit requires only the empirical choice of a single other parameter. It is suggested that these cochlear frequency‐position functions be used in plotting not only physiological data but also psychoacoustic data (and in modeling work with speech and other bioacoustic signals), since, if the cochlear data are correct, the functions are consequently independent of, and an appropriate means to examine, psychoacoustic data and assumptions. [Work supported by NSERC, Canada.]

When signals are presented on the immediate high‐frequency side of a pure‐tone masker, it is not the signals, but rather combination tones, that are detected at masked threshold [Greenwood, J. Acoust. Soc. Am. 50, 502–543 (1971)]. At low to moderate masker intensities, the detected combination tone is 2f1 ‐f2, where f1 is the masker frequency and f2 is the signal frequency. At higher masker levels, the combination tone f2 ‐ f1 becomes large enough to play a role. To investigate this role, the combination tones f2 ‐ f1 and 2f1 ‐ f2 were selectively masked in different low‐frequency regions, to identify their effects. As has been known, the notch region on the high side of the pure‐tone masking pattern creates a secondary peak, which shifts upward in frequency and height as masker intensity is raised. At masker levels sufficient to shift this secondary peak upward, beyond 1.5 times f1 (masker), both 2f1 ‐ f2(C1) and 2f2 ‐ f1(D1) are detected with about equal likelihood in the higher‐frequency part of the no...

Unit responses in the AVCN of the cat have been studied using two‐ and three‐tone stimulus combinations. With two‐tone stimulation, f1 was placed at the unit's best frequency and a higher tone (f2) added at an equal or greater intensity. In these preliminary experiments, maximum suppression of the f1 response was observed when the frequency of f2 was located at a position 0.5–0.8 mm from f1 on the basilar membrane. This distance did not appear to vary with the best frequency of neurons and might be related to critical bandwidth. During three‐tone stimulus experiments, f1 was set at the unit's best frequency and f2 and f3 added above the unit's tuning curve. The frequency separations between the three tones were equal; hence f1 and 2f2 − f3 both fell at the unit's best frequency. The phase relations of the three tones could be systematically varied by varying the phase of f1. The neural responses generated by the resultant of f1 and 2f2 −f3 allowed for the determination of their relative amplitudes. These ...

In experiments mapping pure‐tone masked audiograms, in which the signal is either a pure tone or a narrow band of noise, a notch appears on the high‐frequency side of the masker with a low point at a frequency one critical bandwidth from the masker frequency. This notch has been shown to be caused by the detection, on the lower‐frequency side of the masker, of the combination tones that are produced by the addition of masker and signal. The same results are obtained when a very narrow band of noise masks a pure‐tone signal. The implication that the same process that creates combination tones creates combination bands further supports other evidence as to the mechanical nature of combination tones. The immediate further implication bands or tones should therefore produce low‐frequency masking effects in the same way that external stimuli do has been tested and confirmed over a wide range of intensities and frequency separations of the masking primaries. The latter have consisted of pairs of tones, tones plus narrow bands of noise, and pairs of narrow bands of noise. Some of the implications of these results for cochlear physics and critical bandwidth are outlined.

Combination bands generated by a pure tone and a narrow band of noise produce masking patterns in the frequency region where they occur. These patterns take the form of elevations that merge with, or stand in isolation from, the low‐frequency side of the masked audiogram produced by the lower primary, depending on the separation of the bands from the primary. Masked audiograms produced by bands of external origin were also mapped. An “external” band was presented under two conditions: alone and then together with a more intense higher‐frequency tone. Comparisons among conditions lead to these conclusions: (1) Addition of a higher‐frequency more intense tone to a band of noise of external origin does not increase the masking maximum produced by the noise. This masking remains visible as an elevation on the low‐frequency side of the masked audiogram produced by the tone—unless masking by the tone itself in the lower‐frequency region of the noise exceeds the masking effect of the noise. (1) Consequently, it ...