Diplacusis binauralis (IPD)


Diplacusis binauralis (binaural diplacusis; interaural pitch difference IPD) was first observed in cases of hearing loss, in particular, unilateral hearing loss (Stumpf 1883a, 1890a). In such a case one can notice that the pitch of a pure tone presented to both ears is different at the two ears. The term diplacusis refers to the concept that the pitch of a tone is dependent on the place of maximal excitation of the cochlear partition. Indeed, there is a high correlation of the occurrence of an IPD with local damage of the inner ear (e.g., Shambaugh 1940a, Gaeth & Norris 1965a).

When the magnitude of the IPD is less than about 1 semitone, it is not normally noticed in real life, i.e., when both ears are stimulated simultaneously and with about the same intensity. Yet an IPD may exist. It escapes notice because the pitches of the two ears merge in conscious perception such that only one pitch is heard whose height essentially is the mean of the two monaural pitches (Liebermann & Revesz 1914a, van den Brink 1974a). To measure the ITD in the laboratory, one alternately presents tones to the ears such that at any instant the listener is exposed to one tone only, i.e., either on the right or the left ear. By adjustment of one of the tone frequencies such that the two pitches are equal, an appropriate measure of the IPD is obtained. A reasonable formal definition for the IPD is

dL = (fR-fL)/fL = fR/fL - 1

where fL, fR denote the frequencies at the left and right ear, respectively, that are obtained by adjustment of either of them for a pitch match. For a particular reference frequency fL, dL indeed is a direct measure of the relative difference of the pitches (as opposed to frequencies) between the left and the right ear, i.e., of the IPD. For instance, dL > 0 indicates that, to achieve a pitch match, fR had to be made greater than fL. This means that one and the same tone with the frequency fL, when presented to both ears, will on the left ear evoke a higher pitch than on the right ear.

The correlation between occurrence of an IPD and hearing loss holds not only for a permanent hearing loss (cf. Burns & Turner 1986a, [9]), but also for temporal hearing loss. For instance, Ward (1963a) demonstrated considerable ITDs of 6 normal-hearing listeners who were monaurally exposed to high-intensity pulses, with subsequent measurement of IPD.

The most important conclusion which can be drawn from the above observations is that the pitch of sine tones is basically created by each ear individually. (Here, the term "ear" must not be taken too literally. One should, perhaps, rather say "ear channel", meaning the ascending auditory pathway pertinent to a particular ear.) The conclusion appears inevitable that the sine-tone pitch (i.e., spectral pitch) is established laterally and peripherally. Most remarkably, in cases of permant hearing loss the ITD proves to be permanent as well. This implies that there is no interaural adaptation of pitch. Frequency-to-pitch conversion in one ear is independent of that in the other ear.

Occurrence of IPDs is not just a pathological phenomenon, however (i.e., dependent on a hearing loss). Rather, IPDs are found in any normal hearing system. The main phenomenological difference between pathological and non-pathological cases is that in the former a pronounced ITD typically exists in a fairly broad frequency region, while in the normal hearing system the ITD rarely exceeds plus/minus 2%, and is in a quasi-random manner variable with frequency such that the mean across frequencies essentially is zero. The pattern of the dL(fL)-function is for each individual and for each ear different, but consistent and reproducible (Stevens & Egan 1941a, Jeffress 1944a, van den Brink 1970a, 1975b).

Most remarkably, also in the normal ear there appears to exist a relationship between ITD and local structural conditions of the inner ear. Van den Brink (1970a) has demonstrated that the pattern of the dL(fL)-function, i.e., the IPD as a function of frequency, for an individual listener is stable through many years, and that, most remarkably, it is highly correlated with the same listener's fine-structure pattern of the absolute thresholds of hearing of the two ears. The conclusion that the latter type of pattern indeed is dependent on physiological structural irregularities of the cochlear partition in turn is supported by the finding that the distribution of spontaneous otoacoustic emissions on the frequency axis also is correlated with the fine structure of the hearing threshold (cf. Schloth 1983a, Zwicker 1990b, Zweig & Shera 1995a). This indeed is strong evidence that ITDs in normal ears are not only dependent on, but even originating from, structural irregularities of the cochlear partition.

These observations, together with what was said above about the pathological type of IPD, provide strong evidence for the assumption that the "place theory" of spectral pitch basically is correct. Some pathological cases described in the literature provide additional support to this conclusion. Listeners suffering from some cochlear damage have reported both extreme pitch deviations of musical tones and "pitch splitting" of monaurally presented sine tones such that one tone frequency yielded two or more pitches at a time that were in a constant, yet unsystematic relationship (Liebermann & Revesz 1908, Ward 1955a, Corliss et al. 1968a, Schelleng 1975a, Turner et al. 1983a).

Again, the above findings imply that the formation of spectral pitch is brought about by a rigid, lateral, and fairly peripheral auditory mechanism. It is thus justified to conceptualize a "frequency-to-pitch" function (van den Brink 1970a, 1975b) that exists for each ear of a listener individually, and which is only grossly one and the same for all ears and all listeners. In detail it shows variations as a function of frequency which on a long-term time scale are consistent for each ear, originate from structural irregularities of the cochlear partition, and in that sense can be regarded as the origin of IPD.

From these considerations it is apparent that corresponding "irregularities" of the frequency-to-pitch function of an individual ear should also exist in monaural sine-tone matches to certain well defined pitch-intervals. The pitch interval which is most suitable for that purpose is the octave interval. Indeed, in monaural octave matchings ear-specific structural variations were found (Ward 1954a) that resemble those of IPD (cf. van den Brink 1977a, [100] , [104] p. 341).

Finally it should be noted that the pitch of a sine tone can be shifted, i.e., by a change of intensity or by superimposing another sound. As spectral pitch obviously is produced in each ear independently of the other, one should be able to enhance, or compensate for, an existing IPD at a particular frequency by changing the tone intensity at one ear in an appropriate direction, or by superimposing another sound at one ear. When, for instance, dL is positive at a fairly low frequency, say, 200 Hz (meaning that a binaurally presented 200-Hz tone evokes on the left ear a slightly higher pitch than on the right ear), this IPD can be reduced to zero or even to a negative value by sufficiently increasing the tone intensity at the left ear - since the pitch of a 200-Hz sine tone drops with increasing intensity. Superposition effects of this kind were indeed observed, cf. van den Brink (1965a, 1979a), Burns (1982a), Burns & Turner (1986a), [11].

All the above notions about IPD pertain to sine tones as stimuli - that is, spectral pitch as the corresponding auditory sensation. With complex tones, a number of experiments were described by van den Brink (1965a, 1970b, 1974a, 1975a, 1975b, 1977a). For theoretical reasons, in these experiments complex tones were used as stimuli that comprised only a small number of adjacent part tones ("residue" tones). So as yet there is little experimental data on the ITD of broad-band complex tones such as those of music instruments, the human voice, etc. However, as the pitch of the latter type of signal essentially is of the virtual-pitch type, and since it is well established that virtual pitch emerges from the cooperation of multiple part tones, one can safely presume that for the harmonic complex tones of real life IPD cannot be as significant and pronounced as it is for sine tones. This can be deduced from the notion that, in normal ears, the IPD of sine tones as a function of frequency varies in a quasi-random way such that a positive IPD of one part tone with high probability is accompanied by a negative IPD of one of its neighbours.

The experimental results obtained by van den Brink are of high significance for the understanding of pitch perception, in particular, of virtual pitch. By and large, those results show that the IPD of a harmonic complex tone does not correlate with the IPD of the corresponding sine tone (i.e., with the same oscillation frequency), but that it highly correlates with the IPDs of the part tones that are actually included in the Fourier spectrum. This is another strong support for my own view, namely, that virtual pitch is on a more central stage of the auditory system "deduced" from the spectral pitches of part tones [10], [18], [22], [55], [56].


Author: Ernst Terhardt terhardt@ei.tum.de - Mar 10 2000


main page

File translated from TEX by TTH, version 2.61