Review Auditory Processing of Interaural Timing Information: New Insights

Transcription

1 Journal of Neuroscience Research 66: (2001) Review Auditory Processing of Interaural Timing Information: New Insights Leslie R. Bernstein * Department of Neuroscience, University of Connecticut Health Center, Farmington, Connecticut Differences in the time-of-arrival of sounds at the two ears, or interaural temporal disparities (ITDs), constitute one of the major binaural cues that underlie our ability to localize sounds in space. In addition, ITDs contribute to our ability to detect and to discriminate sounds, such as speech, in noisy environments. For low-frequency signals, ITDs are conveyed primarily by cycle-by-cycle disparities present in the fine-structure of the waveform. For high-frequency signals, ITDs are conveyed by disparities within the time-varying amplitude, or envelope, of the waveform. The results of laboratory studies conducted over the past few decades indicate that ITDs within the envelopes of high-frequency are less potent than those within the fine-structure of low-frequency stimuli. This is true for both measures of sensitivity to changes in ITD and for measures of the extent of the perceived lateral displacement of sounds containing ITDs. Colburn and Esquissaud (1976) hypothesized that it is differences in the specific aspects of the waveform that are coded neurally within each monaural (single ear) channel that account for the greater potency of ITDs at low frequencies rather than any differences in the more central binaural mechanisms that serve these different frequency regions. In this review, the results of new studies are reported that employed special highfrequency transposed stimuli that were designed to provide the high-frequency channels of the binaural processor with envelope-based information that mimics waveform-based information normally available only in low-frequency channels. The results demonstrate that these high-frequency transposed stimuli (1) yield sensitivity to ITDs that approaches, or is equivalent to, that obtained with conventional low-frequency stimuli and (2) yield large extents of laterality that are similar to those measured with conventional low-frequency stimuli. These findings suggest that by providing the highfrequency channels of the binaural processor with information that mimics that normally available only at low frequencies, the potency of ITDs in the two frequency regions can be made to be similar, if not identical. These outcomes provide strong support for Colburn and Esquissaud s (1976) hypothesis. The use of high-frequency transposed stimuli, in both behavioral and physiological investigations offers the promise of new and important insights into the nature of binaural processing. J. Neurosci. Res. 66: , Wiley-Liss, Inc. Key words: binaural hearing; auditory processing; interaural temporal disparities Having two ears, rather than just one, makes possible or enhances a number of auditory capabilities that may often be taken for granted. These include our ability to localize sounds in space and our ability to detect, and selectively attend to, relevant signals such as speech in noisy environments. The two ears, together with the neural circuitry responsible for combining information from them, are referred to as the binaural system. The binaural system encodes and processes differences in the physical waveforms that arrive at each ear. An understanding of how such differences arise in natural settings may be gained by considering a source of sound that occupies a position in space closer to one side of the head. Because the path lengths differ between the source and each ear, the sound will arrive at the more proximal ear slightly earlier than it arrives at the more distal ear. Such between-ear differences in time-of-arrival of sounds are referred to as interaural temporal disparities (ITDs). In addition, largely as a result of the acoustic shadow cast by the head, the amplitude or intensity at the more proximal ear will be greater. Such between-ear differences in the intensity of sounds are referred to as interaural intensitive disparities (IID). It has been recognized for the last 100 years or so that ITDs and IIDs are the major binaural cues that underlie our ability to localize sounds and which contribute to our ability to detect and discriminate sounds in noisy environments. Contract grant sponsor: National Institute on Deafness and Other Communication Disorders, National Institutes of Health; Contract grant numbers: DC 04147, DC *Correspondence to: Leslie R. Bernstein, MC 3401, University of Connecticut Health Center, 263 Farmington Ave., Farmington, CT Les@neuron.uchc.edu Received 14 September 2001; Accepted 21 September Wiley-Liss, Inc.

2 1036 Bernstein Considerably more effort has been devoted to assessing binaural processing of ITDs than binaural processing of IIDs. This is largely because binaural processing of ITDs, unlike the case for IIDs, is highly dependent upon the nature of the stimuli employed. Thus, the use of ITDs has proven more revealing in terms of understanding various aspects of the binaural system. Indeed, it is the processing of ITDs that forms the foundation of the majority of theoretical and mathematical descriptions of the binaural system. Most of what is known about how human listeners process ITDs has been gleaned from investigations in which sounds have been presented via earphones (for a review, see Durlach and Colburn, 1978). The primary advantage of this method is that it allows for extremely precise control of the stimuli and, most importantly, of the interaural disparities themselves. When stimuli of like frequency content are presented to each ear via earphones, the perception is typically that of a singular fused acoustic image that is located within the head, or intracranially, along a line bounded laterally by each ear. If the sounds presented to each ear are identical, the intracranial image will be perceived close to the center of the head. Depending on the particular stimulus employed, the introduction of an ITD can result in the intracranial position of the sound being perceived as displaced toward the ear at which the physical waveform leads in time. Such an intracranial image is said to be lateralized. In contrast, sounds emitted by external sources such as loudspeakers are typically perceived as being outside the head and are referred to as being localized. Two classes of experimental paradigms have been employed to measure the potency of ITDs. The first involves measurement of listeners sensitivity to, or ability to discriminate, small changes in ITD. The second, involves measures of the extent of perceived lateral displacement produced by ITDs. I. Sensitivity to Changes in ITD In a typical task designed to measure sensitivity to changes in ITD, the listener is presented with a pair of sounds in succession. During one presentation, an ITD leading in one ear, e.g., the left, is imposed; during the other presentation, no ITD is imposed. The listener s task is to report which of the two intervals contained the stimulus with the ITD. Alternatively, one interval may contain a sound in which the ITD leads at one ear while the other interval contains a sound in which the ITD leads at the opposite ear. In that type of task, the listener is instructed to report whether the pair of sounds was perceived as moving from left to right or from right to left. In either type of task, the ITD is adjusted to achieve some criterion level of performance (typically 75% correct responses). The resulting value of ITD is referred to as the threshold or the just noticable difference (jnd) of interaural time. Before proceeding further, it is important to distinguish between two types, or aspects, of interaural time delay. The first is the temporal disparity between the beginning or onset of the sound at each ear. The second is the temporal disparity between the arrival of each of the individual cycles or features of the ongoing waveform. It is the ongoing ITD that has been of interest to investigators because sensitivity to this cue reflects the ability of the nervous system to code or follow temporal aspects of the waveforms delivered to the ears. As such, the use of onset delays by listeners has usually been avoided in the measurements of thresholds either by turning on the waveforms to each ear very slowly and/or by delaying the waveform destined for one ear prior to turning on the stimuli simultaneously at the two ears. Figure 1 displays data obtained by Zwislocki and Feldman (1956) in which threshold ITD is plotted as a function of the frequency of pure tone (sinusoidal) stimulation. As is shown, human listeners can be extremely sensitive to timing differences across the ears, requiring only about 13 to 30 s in order to reach threshold, depending on the frequency of the tone. Interestingly, when the frequency of the tone was increased beyond 1,300 1,500 Hz or so, Zwislocki and Feldman s listeners were unable to discriminate ITDs. The finding that listeners cannot discriminate ITDs within pure tones for frequencies above about 1,500 Hz bolstered a view put forth by Rayleigh in In his classical duplex theory of sound localization, Rayleigh argued that two separate systems subserve listeners abilities to localize sounds. The theory holds that at low frequencies, ITDs mediate localization, whereas at high frequencies IIDs are the relevant cue. Rayleigh reached these conclusions based on observations made using pure-tone stimulation. He noted that for very low frequencies (e.g., 128 Hz), the wavelengths are such that the head does not cast an effective acoustic shadow and so any intensity differences across the ears would be slight. As the frequency is increased, however, effective IIDs would be produced. On the other hand, while interaural timing differences could signal the location of a low-frequency source of sound, the periodic nature of sinusoidal stimulation would cause the determination of which ear led in time to be ambiguous as the frequency of the sinusoidal source was increased. For example, if a source of sound were positioned opposite to one ear, the interaural delay would be equivalent to half the period of a sinusoid of approximately 800 Hz and it could not be determined to which ear the signal was leading in time. As discussed above, the notion of the binaural system as being insensitive or unable to utilize ITDs at high frequencies was based on observations and analyses using pure-tone stimulation. Although listeners cannot discriminate ITDs within high-frequency pure tones, a number of early studies demonstrated that listeners are, indeed, capable of discriminating ITDs within complex waveforms such as sinusoidally amplitude-modulated (SAM) tones, filtered transients, and bands of noise even when the spectral content of those signals is restricted to frequencies above 1,500 Hz (e.g., Klumpp and Eady, 1956; Leakey et al., 1958; David et al., 1959; van Bergeijk, 1959).

3 Auditory Processing of Interaural Timing Information 1037 Fig. 1. Threshold interaural temporal disparities (ITD) plotted as a function of the frequency of pure tone stimulation (from Zwislocki and Feldman, 1956). Klumpp and Eady (1956) demonstrated that sensitivity to changes in ITD within narrow bands of highfrequency noise rivals that observed with low-frequency pure-tones; their listeners required only sorsoto reach 75% correct detection. Leakey et al. (1958) presented listeners with 4,000-Hz tonal carriers that were sinusoidally amplitude-modulated at a rate of 200 Hz. Such a SAM tone is depicted in Figure 2a along with a 200-Hz pure tone for comparison. The heavy dashed line drawn through the positive peaks of the individual cycles, or fine-structure, of each waveform is referred to as the envelope. The envelope corresponds to the amplitude of the waveform. In the case of the SAM tone, the amplitude varies sinusoidally. In the case of the pure tone, the amplitude does not vary and the envelope is flat. When Leakey et al. (1958) applied ITDs to only the envelopes of high-frequency SAM waveforms, as shown in Figure 2b, their listeners perceived changes in the lateral positions of the sounds. Leakey et al. (1958) concluded that this sensitivity was dependent upon the correlated nature of the envelopes of the signals at the two ears and not upon their carriers. Taken together, the results of these early studies called into question the long-standing view based on the duplex theory, that listeners could not utilize ITDs at high frequencies. More recent investigations have explored, in greater detail, the nature of listeners sensitivities to changes in ITDs within high-frequency complex waveforms. These studies employed a variety of high-frequency stimuli including SAM tones, two-tone complexes, and bands of noise (e.g., Henning, 1974a,b; McFadden and Pasanen, 1976; Nuetzel and Hafter, 1976, 1981; Bernstein and Trahiotis, 1982, 1994; Blauert, 1983). In some cases, the ability of listeners to discriminate ITDs conveyed by highfrequency stimuli was comparable to that observed for low frequency stimuli. For the most part, however, these studies reveal that when the spectra of the stimuli are carefully restricted to high frequencies and only ongoing interaural delays are imposed, threshold ITDs are about two to 10 times larger than those typically obtained with lowfrequency stimuli. That is, listeners are less sensitive to changes in ITD within high-frequency, complex stimuli than they are to changes in ITD within low-frequency stimuli. A number of findings from the more recent investigations cited above supported the view that sensitivity to changes in ITD conveyed by high-frequency complex waveforms is mediated by delays within the envelope of the waveform. For example, such sensitivity persists when only the envelope of a high-frequency waveform (and not its fine-structure) is delayed and this sensitivity is essentially equivalent to that measured when the entire waveform (envelope and fine-structure) is delayed. In contrast, as expected from the earlier findings with high-frequency tones, listeners appear to be insensitive to interaural delays within the fine-structure of high-frequency, complex waveforms (e.g., Henning, 1974b; Nuetzel and Hafter, 1976). Essentially the reverse is true for low-frequency

4 1038 Bernstein Fig. 2. a: A sinusoidally amplitude modulated (SAM) tone consisting of a 4000-Hz carrier modulated at the rate of 200 Hz (upper) and a 200-Hz pure tone (lower). The heavy dashed lines drawn through the positive peaks of each waveform represent their envelopes. b: An example of a delay of only the envelope of the SAM tone depicted in a. In this case, the envelope in the left ear leads that in the right ear. stimuli. That is, while listeners are exquisitely sensitive to ITDs in the fine-structure at low frequencies, they are quite insensitive to delays within the envelope of complex, low-frequency stimuli (Henning, 1980, 1983; Henning and Ashton, 1981). Such observations suggest a dichotomy with regard to the processing of ITDs in low- and highfrequency waveforms. In the case of low-frequency waveforms, such processing appears to be dominated by cycleby-cycle disparities present in the fine-structure of the waveform. In the case of high-frequency waveforms, such processing appears to be mediated by envelope-based disparities. II. Extents of Laterality Produced by ITDs A variety of techniques have been utilized in order to assess directly the effect of ITDs on the lateral position of acoustic images. For example, the listener may be asked to simply report whether the sound is perceived to the left or right of the center of the head (Sayers and Cherry, 1957) or may be required to use some type of numerical scale to indicate the position that best coincides with the perceived locus of the acoustic image (e.g., Sayers, 1964; Watson and Mittler, 1965; Yost, 1981; Blauert, 1982). Several investigators have employed what is referred to as an acoustic pointing task in which the listener adjusts the IID (or ITD) of one sound (the pointer) until its intracranial position matches, or coincides with, the intracranial position of a second, experimenter-controlled sound, referred to as the target (e.g., Moushegian and Jeffress, 1959; Domnitz and Colburn, 1977; Bernstein and Trahiotis, 1985a). The interaural disparity inserted in the pointer by the listener serves as a metric of the extent of laterality produced by the ITD inserted in the target by the experimenter. Consistent with the measures of threshold ITDs described earlier, measures of laterality for pure tones indicate that once the frequency increases beyond 1,300 1,500 Hz, ITDs have little, if any, effect on the lateral position of the acoustic image (e.g., Sayers and Cherry, 1957; Schiano et al., 1986). Figure 3 shows the results of a single, representative listener from Schiano et al. (1986). Those investigators employed an acoustic pointing task in which the listener adjusted the IID of a narrow-band, low-frequency pointer to match the intracranial position of a pure tone target. For the data shown in Figure 3, the

5 Auditory Processing of Interaural Timing Information 1039 Fig. 3. The interaural intensitive differences (IID) inserted in the acoustic pointer in order to match the intracranial position of the target as a function of the frequency of tonal stimulation. For each target, the ITD was 150 s (from Schiano et al., 1986). ongoing ITD was held constant at 150 s. The abscissa indicates the frequency of the pure tone; the ordinate indicates the IID necessary to match the position of the target. Note that an IID of 0 db corresponds to a perceived intracranial image located at the center of the head, while an IID of 10 db represents a position that is nearly at the ear. The data indicate that, as the frequency was increased from 300 to 1,000 Hz, the IID required by the listener increased slightly from 6 to 9 db. This indicates that, for these frequencies, the intracranial image was substantially displaced from the midline. Once the frequency was increased beyond 1,200 Hz, the image moved progressively toward the center of the head and was essentially centered once the frequency reached 1,600 Hz. This demonstrates the inability of ITDs to produce any substantial change in lateral position for high-frequency pure tones. When high-frequency complex stimuli, rather than high-frequency pure tones are employed, the outcome is quite different. ITDs do produce changes in laterality when such complex waveforms are employed and it is clear that these changes are based on ITDs conveyed by their time-varying envelopes (e.g., Blauert, 1982; Bernstein and Trahiotis, 1985b; Trahiotis and Bernstein, 1986). It is the case, however, that the extent of laterality produced by ITDs within high-frequency complex waveforms is far less than that produced when the same delays are applied to low-frequency stimuli. For example, Blauert (1982) asked listeners to rate the sidedness of stimuli presented with several values of ITDs. A rating of 0 corresponded to an acoustic image at the midline; a rating of 5 corresponded to an image fully lateralized toward one side of the head. Figure 4 shows the sidedness judgments obtained by Blauert (1982) as a function of ITD for a 250-Hz tone and for a 4-kHz tone sinusoidally amplitudemodulated at 250 Hz. Figure 4 indicates that, for all values of ITD, the extent of laterality produced by the highfrequency SAM tone is smaller than that obtained for the pure tone. Notably, the sidedness judgment for the SAM tone never exceeds a value of about 2 even for an ITD of 500 s, a value close to the largest delay a human listener can encounter in natural settings. III. Differences in Processing of ITDs at Low and High Frequencies Two important outcomes emerge from the investigations described above. First, for low frequencies, it is the fine-structure of the waveform that conveys or carries virtually all of the information regarding ITDs, while for high frequencies, it is the envelope of the waveform. Second, ITDs within the envelopes of high-frequency stimuli appear to be less potent than those within the finestructure of low-frequency stimuli. This is true for both measures of sensitivity to changes in ITD and for measures of the extents of lateral displacement produced by ITDs. In order to begin to understand these outcomes, it is useful to consider the nature of the processing that occurs at the mammalian auditory periphery. Figure 5 is a schematic of the stages of transduction that form the basis of most current models of binaural processing. For clarity,

6 1040 Bernstein Fig. 4. Sidedness ratings as a function of ITD for a 250-Hz pure tone (squares) and a 4-kHz SAM tone with a modulation rate of 250 Hz (circles). A rating of 0 corresponds to an intracranial image at the center of the head; a rating of 5 corresponds to a fully lateralized image (from Blauert, 1982). the three stages of (peripheral) processing that occur prior to binaural interaction are labeled for the left ear only. The three stages are bandpass filtering, half-wave rectification, and low-pass filtering. Bandpass filtering comes about largely as a result of the mechanics of the inner ear and is manifest in the finding that any given auditory receptor or hair cell only responds to, or is tuned, to a limited range of frequencies. Half-wave rectification occurs because hair cells tend to respond to only one direction of displacement of the stimulating waveform. Finally, lowpass filtering, a process which acts to smooth over rapid changes, is included to incorporate the finding that the degree to which the fine-structure of the stimulating waveform is reflected in the responses of hair cells, and ultimately in the synchronized discharge of auditory nerve fibers, diminishes as the frequency of stimulation is increased beyond some value, typically around 1 to 2 khz (e.g., Palmer and Russell, 1986; Weiss and Rose, 1988; for a review, see Ruggero and Santos-Sacchi, 1997). Figure 6 illustrates the waveforms before and after this type of peripheral processing for two different signals: (1) a 250-Hz tone, and (2) a 4,000-Hz sinusoid sinusoidally amplitude-modulated at 250 Hz. Figure 6 shows that, for the low-frequency 250-Hz tone, the effect of such processing is to pass only the positive values of the waveform. That is, the waveform has been half-waverectified. For the 4,000-Hz SAM tone, the effect of the processing is to extract the sinusoidally varying envelope from the waveform. The fine-structure at 4000 Hz is removed because the low-pass filter smooths over oscillations at this frequency. Were the peripheral processing applied to a 4,000-Hz pure tone, all that would remain at the output would be its flat envelope. If such a highfrequency tone were presented with an ITD, the constantvalued envelope could provide no information to the binaural system regarding the delay. This is consistent with the finding, discussed above, that listeners are completely insensitive to ITDs within high-frequency tones. The simple model of peripheral transduction presented here reflects the findings that at low frequencies, neural impulses are synchronized to the half-waverectified waveform, while the same peripheral processing at high frequencies results in neural impulses synchronized only to the envelope of the waveform (e.g., Johnson, 1980; Weiss and Rose, 1988; Joris and Yin, 1992). Colburn and Esquissaud (1976) hypothesized that it is the differences in the specific aspects of the waveform that are coded neurally within each monaural (single ear) channel that account for the greater potency of ITDs at low frequencies rather than any differences in the more central binaural mechanisms that serve these different frequency regions. Specifically, Colburn and Esquissaud (1976) hypothesized that synchronized neural impulses, independent of whether they arise from low-frequency or highfrequency stimulation, serve as inputs to a central binaural mechanism that functions uniformly across frequency. An intuitive understanding of why ITDs may be more potent when conveyed by low-frequency stimula-

7 Auditory Processing of Interaural Timing Information 1041 Fig. 5. A schematic of the stages of transduction that form the basis of most modern models of binaural hearing. tion may be gained by a close reinspection of Figure 6. For the 250-Hz tone, the peripheral processing results in an output such that there are distinct regions between adjacent peaks of the processed waveform where the waveform remains at or close to a value of zero. The transition between each peak and each of these off regions is rather abrupt and is characteristic of a rectified waveform. In contrast, the 250-Hz envelope that is extracted from the SAM tone at 4,000 Hz is essentially a sinusoid and has no such distinct off regions. Blauert (1982) conjectured that the greater potency of ITDs for low frequencies as compared to high frequency may come about because a peripherally processed low-frequency signal yields more distinct time cues... duetoitstransient features. (p. 68) IV. New Experimental Results A recent series of preliminary investigations in our laboratory was designed to test Colburn and Esquissaud s (1976) hypothesis and Blauert s (1982) conjecture directly (Bernstein and Trahiotis, 2001). Recall that Colburn and Esquissaud (1976) hypothesized that the observed differences in the potency of ITDs at low and high frequencies stem from differences in the waveforms, as processed, at the auditory periphery. Given that premise, we reasoned that if the processed waveforms could be made to be similar in the two frequency regions, then the potency of ITDs, evaluated either in terms of sensitivity to ITDs, or in terms of the extents of laterality they produce, would also be found to be similar in the two frequency regions. The goal was to provide the high-frequency channels of the binaural processor with envelope-based information that mimics waveform-based information normally available only in low-frequency channels. A technique for accomplishing this was recently described by van de Par and Kohlrausch (1997) who studied the use of binaural cues at low and at high frequencies in order to detect the presence of signals embedded in noise (binaural masking). The technique we employed was a modification of the one they described. The method, illustrated in Figure 7, entails multiplying a half-wave-rectified, low-pass-filtered, lowfrequency tone by a high-frequency sinusoidal carrier. The resulting waveform is shown at the bottom of the figure. Note that the envelope of this new stimulus is the rectified, low-frequency sine wave with which we began. That is, the rectified tone has been transposed onto the 4,000-Hz carrier as its envelope. Figure 8 is the same as Figure 6 with the addition of a transposed stimulus generated by transposing a rectified, low-pass-filtered 250-Hz tone to the envelope of a 4-kHz carrier. Note that the assumed peripheral processing results in identical outputs for the 250-Hz tone and the highfrequency transposed stimulus. Both have sharper, more distinct peaks than the output corresponding to the highfrequency SAM tone. It seems reasonable to assume then, that for the processed tonal and transposed stimuli, the corresponding patterns of neural discharges would be less dispersed in time (have less variability) than would be the case for the high-frequency SAM tone. A. Sensitivity to changes in ITD. Sensitivity to changes in ongoing ITD was measured for three listeners using the three types of stimuli depicted in Figure 8: (1) sinusoidally amplitude-modulated (SAM) tones centered

8 1042 Bernstein Fig. 6. Left side: A 250-Hz tone (upper) and a 4,000-Hz tone sinusoidally amplitude modulated at 250 Hz (lower). Right side: The same two stimuli after bandpass filtering, rectification, and lowpass filtering. at 4 khz, (2) transposed stimuli produced by transposing to 4 khz, rectified low-frequency tones whose frequencies matched the frequencies of modulation employed for the SAM tones, and (3) low-frequency pure tones. The highfrequency stimuli were presented along with a continuous low-frequency noise in order to preclude the listeners use of any low-frequency distortion products that might be generated within the earphone or within the ear itself. Figure 9 displays the threshold ITDs averaged across the three listeners as a function of the frequency of modulation (SAM and transposed stimuli) or the frequency of the low-frequency tone. The error bars represent 1 standard error of the mean. Although modulation frequencies of 32 and 64 Hz were employed for the SAM tones and transposed stimuli, threshold ITDs were not measured for pure tones of those frequencies. Those two frequencies were omitted in order to avoid difficulties concerning the coupling of the earphones to the head and because high levels of stimulation would have been required in order to overcome the relative insensitivity of human listeners to such low-frequency stimulation. Two important outcomes are revealed by the data in Figure 9. First, the threshold ITDs obtained with the transposed stimuli were uniformly smaller than those obtained with the SAM tones. Second, the thresholds obtained with the high-frequency transposed stimuli and with the low-frequency pure tones were very similar for frequencies of 128 Hz and 256 Hz. That is, the ability to discriminate changes in ITD using high-frequency transposed stimuli matches, or comes very close to, that measured with low-frequency stimulation. These outcomes are in accord with both Colburn and Esquissaud s (1976) hypothesis and Blauert s (1982) conjecture. At 512 Hz, the threshold ITDs obtained with the SAM and with the transposed stimuli are relatively large and substantially greater than those obtained with the 512-Hz pure tone. This outcome was expected for two reasons. The first concerns the separation in frequency between the individual spectral components that define these stimuli. For both the SAM and transposed stimuli, increasing the rate of modulation causes the individual frequency components that surround the center frequency of 4 khz (the sidebands ) to be separated to a greater and greater extent. This causes the sidebands to eventually fall outside the frequency limits of the nominal bandpass filter that defines the 4-kHz channel (see Fig. 5). Nuetzel and Hafter (1981) discussed how such peripheral filtering could result in reductions in depth of modulation (the envelope, as represented internally, becomes more shallow) which lead to degradations in sensitivity to ITD. van de Par and Kohlrausch (1997) have recently discussed similar effects while interpreting binaural masking thresholds obtained with transposed stimuli. The second reason this outcome was expected is that binaural processing appears to be constrained by an inability of the auditory system to follow rates of fluctuations of the envelope as high as 512 Hz. This limitation appears to originate centrally, after peripheral filtering, and per-

9 Auditory Processing of Interaural Timing Information 1043 Fig. 7. Schematic representation of the method used to generate transposed stimuli. Fig. 8. Same as Figure 6 with the addition of a transposed stimulus generated by multiplying a rectified, low-passfiltered 250-Hz tone by a 4-kHz tone.

10 1044 Bernstein Fig. 9. Threshold ITDs averaged across three listeners as a function of the modulation or pure-tone frequency. The parameter of the plot is the type of stimulus employed. The error bars represent 1 standard error of the mean (from Bernstein and Trahiotis, 2001). haps prior to binaural interaction (Bernstein and Trahiotis, 1994) Ḃ. Extents of laterality produced by ITDs. The effectiveness of providing the binaural processor with similar ITD-based information at low and high frequencies was also assessed by measuring extents of laterality. This was accomplished by employing an acoustic pointing task in which listeners adjusted the interaural intensitive disparity (IID) of a 200-Hz-wide Gaussian noise centered at 500 Hz (the pointer), so that its intracranial position matched that of the experimenter-controlled stimulus (the target; see Bernstein and Trahiotis, 1985a,b; Schiano et al., 1986 for details). Three types of stimuli were employed: (1) a SAM tone centered at 4 khz and modulated at 128 Hz, (2) a 128-Hz tone transposed to 4 khz as described above, and (3) a 100-Hz-wide Gaussian noise centered at 500 Hz. A low-frequency noise was utilized rather than a low-frequency tone because noises produce more punctate intracranial images than pure tones (see Bernstein and Trahiotis, 1985a). The data were obtained from a listener who had previously participated in several types of binaural experiments. In Figure 10, the IID (in db) inserted by the listener in the pointer is plotted as a function of the ongoing ITD inserted in the target. In order to make the presentation more straightforward, the results will be described in terms of lateral position rather than in terms of the IID of the pointer. Recall that an IID of 0 indicates the center of the head while an IID of 10 db corresponds to a position of the intracranial image essentially at the ear. For all three stimuli, lateral position increases monotonically with increases in the ITD of the target. Consistent with earlier investigations, the low-frequency noise results in greater extents of laterality than the high-frequency SAM tone for all (non-zero) values of ITD tested. Even an ITD as large as 600 s does not result in the SAM tone being fully lateralized. For the high-frequency transposed tone, however, the extent of laterality is not only substantially greater than for the SAM tone, but actually exceeds that obtained with the low-frequency noise. These results, like the measures of ITD sensitivity described above, lend support to Colburn and Esquissaud s (1976) hypothesis and are consistent with Blauert s conjecture (1982). V. Summary and Future Research The new experimental results presented here demonstrate that high-frequency transposed stimuli: (1) yield sensitivity to ITDs that approaches, or is equivalent to, that obtained with conventional low-frequency stimuli, and (2) yield large extents of laterality that are similar to those measured with conventional low-frequency stimuli. These results suggest that by providing the high-frequency channels of the binaural processor with information that mimics that normally available only at low frequencies, the potency of ITDs in the two frequency regions can be made to be similar, if not identical. These outcomes provide strong support for Colburn and Esquissaud s (1976) hypothesis which holds that the greater potency of ITDs typically observed for low as compared to high frequencies results from differences in the specific aspects of the waveform that are coded peripherally rather than from differences in the more central binaural mechanisms that serve the different frequency regions. Our laboratory is currently engaged in a program of research designed to evaluate listeners use of highfrequency transposed stimuli in a variety of binaural tasks. As a complement to these behavioral measures, it would be valuable to have physiological measures obtained using

11 Auditory Processing of Interaural Timing Information 1045 Fig. 10. The IID inserted into the acoustic pointer in order to match the intracranial position of the target as a function of the target ITD. The parameter of the plot is the type of stimulus employed. these stimuli. For example, it would be of great interest to compare the patterns of neural discharges obtained with low-frequency stimuli to those obtained when those same stimuli are transposed to high frequencies. Such a comparison would allow one to assess directly the degree to which the neural input to the binaural processor resulting from the transduction of high-frequency transposed stimuli really can mimic that obtained with low-frequency stimuli. With regard to the neural mechanisms that underlie binaural processing per se, it has been demonstrated in a number of species that units at the level of the superior olive (principally, the medial superior olive, or MSO) and inferior colliculus (IC) are tuned to both frequency and ITD (for a review, see Kuwada and Yin, 1987). At high frequencies, units in the IC have been shown to be sensitive to ITDs within the envelopes of SAM tones (e.g., Batra et al., 1993). It would be of interest to know how such sensitivity would be affected, and perhaps enhanced, if high-frequency transposed stimuli were employed. In conclusion, the use of high-frequency transposed stimuli in both behavioral and physiological investigations offers the promise of new and important insights into the nature of binaural processing. Such information could prove valuable for the improvement of mathematical models of binaural mechanisms and, more generally, for an understanding of several important auditory capabilities including our ability to localize sounds in space and our ability to understand speech in noisy environments. ACKNOWLEDGMENTS The author thanks Drs. Shig Kuwada and Klaus Hartung for their helpful comments. REFERENCES Batra R, Kuwada S, Stanford T High-frequency neurons in the inferior colliculus that are sensitive to interaural delays of amplitudemodulated tones: evidence for dual binaural influences. J Neurophys 70: Bernstein LR, Trahiotis C. 1985a. Lateralization of low-frequency, complex waveforms: the use of envelope-based temporal disparities. J Acoust Soc Am 77: Bernstein LR, Trahiotis C. 1985b. Lateralization of sinusoidally-amplitudemodulated tones: effects of spectral locus and temporal variation. J Acoust Soc Am 77: Bernstein LR, Trahiotis C Detection of interaural delay in highfrequency SAM tones, two-tone complexes, and bands of noise. J Acoust Soc Am 95: Bernstein LR, Trahiotis C Detection of interaural delay in high frequency noise. J Acoust Soc Am 71: Bernstein LR, Trahiotis C Using transposed stimuli to reveal similar underlying sensitivity to interaural timing information at high and low frequencies: support for the Colburn-Esquissaud hypothesis. In: Breebaart DJ, Houtsma AJM, Kohlrausch A, Prijs VF, Schoonhoven R, editors. Physiological and Psychophysical Bases of Auditory Function: Proceedings of the 12 th International Symposium on Hearing. The Netherlands: Shaker Publishing. p Blauert J Binaural localization: Multiple images and applications in room- and electroacoustics. In: Gatehouse RW, editor. Localization of sound: theory and application. Groton: Amphora Press. p

12 1046 Bernstein Colburn HS, Esquissaud P An auditory-nerve model for interaural time discrimination of high-frequency complex stimuli. J Acoust Soc Am 59:Suppl 1, S23. David EE, Guttman N, van Bergeijk WA Binaural interaction of high-frequency complex stimuli. J Acoust Soc Am 31: Domnitz RH, Colburn HS Lateral position and interaural discrimination. J Acoust Soc Am 61: Durlach NI, Colburn HS Binaural phenomena. In: Carterette EC, Friedman MP, editors. Hearing, vol IV, handbook of perception. New York: Academic Press. p Henning GB. 1974a. Detectability of interaural delay in high-frequency complex waveforms. J Acoust Soc Am 55: Henning GB. 1974b. Lateralization and the binaural masking-level difference. J Acoust Soc Am 55: Henning GB Some observations on the lateralization of complex waveforms. J Acoust Soc Am 68: Henning GB Lateralization of low-frequency transients. Hear Res 9: Henning GB, Ashton J The effect of carrier and modulation frequency on lateralization based on interaural phase and interaural group delay. Hear Res 4: Johnson D The relationship between spike rate and synchrony in responses of auditory-nerve fibers to single tones. J Acoust Soc Am 68: Joris PX, Yin TCT Responses to amplitude-modulated tones in the auditory nerve of the cat. J Acoust Soc Am 91: Klumpp RG, Eady HR Some measurements of interaural time difference thresholds. J Acoust Soc Am 28: Kuwada S, Yin TCT Physiological studies of directional hearing. In: Yost WA, Gourevitch G, editors. Directional hearing. New York: Springer Verlag. p Leakey DM, Sayers BMcA, Cherry C Binaural fusion of low- and high-frequency sounds. J Acoust Soc Am 30:222. McFadden D, Pasanen EG Lateralization at high frequencies based on interaural time differences. J Acoust Soc Am 59: Moushegian G, Jeffress L Role of interaural time and intensity differences in the lateralization of low-frequency tones. J Acoust Soc Am 31: Nuetzel JM, Hafter ER Lateralization of complex waveforms: effects of fine-structure, amplitude, and duration. J Acoust Soc Am 60: Nuetzel JM, Hafter ER Discrimination of interaural delays in complex waveforms: spectral effects. J Acoust Soc Am 69: Palmer AM, Russell IJ Phase-locking in the cochlear nerve of the guinea-pig and its relation to the receptor potential of inner hair-cells. Hear Res 24:1 15. Rayleigh Lord (Strutt JW) On our perception of sound direction. Philos Mag 13: Ruggero MA, Santos-Sacchi J Cochlear mechanics and biophysics. In: Crocker MJ, editor. Encyclopedia of acoustics (vol. III). New York: John Wiley and Sons. p Sayers BMcA Acoustic-image lateralization judgments with binaural tones. J Acoust Soc Am 36: Sayers BMcA, Cherry EC Mechanism of binaural fusion in the hearing of speech. J Acoust Soc Am 29: Schiano JL, Trahiotis C, Bernstein LR Lateralization of lowfrequency tones and narrow bands of noise. J Acoust Soc Am 79: Trahiotis C, Bernstein LR Lateralization of bands of noise and sinusoidally amplitude-modulated tones: Effects of spectral locus and bandwidth. J Acoust Soc Am 79: van de Par S, Kohlrausch A A new approach to comparing binaural masking level differences at low and high frequencies. J Acoust Soc Am 101: Watson CS, Mittler B Time-intensity equivalence in auditory lateralization: a graphical method. Psychonom Sci 2: Weiss TF, Rose C A comparison of synchronization filters in different auditory receptor organs. Hear Res 33: Yost WA Lateral position of sinusoids presented with interaural intensive and temporal differences. J Acoust Soc Am 70: Zwislocki J, Feldman RS Just noticeable differences in dichotic phase. J Acoust Soc Am 28: