Actual Source Location Research Articles

Binaural Cross-Correlation Predicts the Responses of Neurons in the Owl’s Auditory Space Map under Conditions Simulating Summing Localization

Summing localization describes the perceptions of human listeners to two identical sounds from different locations presented with delays of 0-1 msec. Usually a single source is perceived to be located between the two actual source locations, biased toward the earlier source. We studied neuronal responses within the space map of the barn owl to sounds presented with this same paradigm. The owl's primary cue for localization along the azimuth, interaural time difference (ITD), is based on a cross-correlation-like treatment of the signals arriving at each ear. The output of this cross-correlation is displayed as neural activity across the auditory space map in the external nucleus of the owl's inferior colliculus. Because the ear input signals reflect the physical summing of the signals generated by each speaker, we first recorded the sounds at each ear and computed their cross-correlations at various interstimulus delays. The resulting binaural cross-correlation surface strongly resembles the pattern of activity across the space map inferred from recordings of single space-specific neurons. Four peaks are observed in the cross-correlation surface for any nonzero delay. One peak occurs at the correlation delay equal to the ITD of each speaker. Two additional peaks reflect "phantom sources" occurring at correlation delays that match the signal of the left speaker in one ear with the signal of the right speaker in the other ear. At zero delay, the two phantom peaks coincide. The surface features are complicated further by the interactions of the various correlation peaks.

Investigation of three‐dimensional bathymetric refraction using laboratory scale modeling

Experimental measurements have been carried out to identify the effect of bathymetry on three‐dimensional underwater sound propagation. The measurements have been conducted over a 1/10 000th scale model of the Santa Lucia Escarpment off the coast of California. Experiments measured the bearing error obtained in locating a source with a horizontal line array using a phase‐wave beamformer. The results of this experiment demonstrated significant variations in the measured bearing error. Differences in the detected bearing of the variations beamformer and the actual source location were as large as 18 deg, with the magnitude of these errors being a function of bearing line and source frequency. An analytical solution of the acoustic field in a shear supporting wedge using the method of images has been used to approximate the results of the bearing error experiment. Comparison of experiment and theory shows the same effects with bearing errors of the same magnitude. [Work supported by ONR.]

What happens near a quasi‐linear point source?

The quasi‐linear continuous point source has been central to many studies of multidimensional flow in unsaturated soil. Immediately next to the source, however, the water pressure Ψ is positive and indefinitely large, so the quasi‐linear flow equation (limited to Ψ ≤ 0) fails in principle. We investigate whether this is a serious problem in practice. Exact analysis would be complicated and difficult, since the problem is of mixed saturated‐unsaturated flow with an unknown free boundary between the two regions. When the source strength is not too large, the saturated region is close to spherical. A simple approximate analysis follows with exterior and interior flow velocities matched on the boundary. The correction, taking into account the saturated region, shifts the quasi‐linear source a small distance above the actual source location. In practical applications the correction is usually negligibly small. Physical reasons why the approximation works well and the correction is small are indicated. In addition, these physical reasons suggest that the scope of the quasi‐linear analysis is wider than supposed hitherto. The solution in potential theory for the eccentric point source needed for the study is given in the appendix.

Localization of narrow-band sounds in the medial sagittal plane.

Many experiments have shown that sources in the medial sagittal plane are localized by human listeners according to spectral cues. Therefore, narrow-band sounds in different frequency ranges are perceived as originating from different vertical locations. To further elucidate the cues involved in this effect, listeners were asked to localize two kinds of sound; a sine tone and one-third-octave band of noise. Stimuli were presented, at random, from one of three sources: front, overhead, and rear. The listener’s task was to identify the source. The experimental results supported the spectral theory of localization in that, for both kinds of sound, judgments depended more upon frequency than upon actual source location. However, the results also showed that judgments were considerably more stable and consistent with the one-third-octave bands of noise than with the sine tones. It is suspected, therefore, that the spectral theory of vertical localization is true but that it is incomplete. The results are interpreted to mean either that listeners are capable of using head-related transfer-function cues on a band that is as narrow as one-third octave or else that listeners gain vertical-plane localization information from spectral fluctuations, present in the noise but not in the sine tone. [Work supported by the NIDCD, DC00181.]

MEG versus EEG localization test using implanted sources in the human brain

It is believed that the magnetoencephalogram (MEG) localizes an electrical source in the brain to within several millimeters and is therefore more accurate than electroencephalogram (EEG) localization, reported as 20 mm. To test this belief, the localization accuracy of the MEG and EEG were directly compared. The signal source was a dipole at a known location in the brain; this was made by passing a weak current pulse simulating a neural signal through depth electrodes already implanted in patients for seizure monitoring. First, MEGs and EEGs from this dipole were measured at 16 places on the head. Then, computations were performed on the MEG and EEG data separately to determine the apparent MEG and EEG source locations. Finally, these were compared with the actual source location to determine the MEG and EEG localization errors. Measurements were made of four dipoles in each of three patients. After MEGs with weak signals were discounted, the MEG average error of localization was found to be 8 mm, which was worse than expected. The average EEG error was 10 mm, which was better than expected. These results suggest that the MEG offers no significant advantage over the EEG in localizing a focal source. However, this does not diminish other uses of the MEG.

Magnetic evoked field associated with transcortical currents in turtle cerebellum

The neural basis of magnetic evoked fields of the brain was studied with an isolated turtle cerebellum as a model preparation. The turtle cerebellum is a nearly flat tissue with neural processes arranged along three orthogonal axes of symmetry. According to theoretical results, this geometry should enable us to selectively measure the magnetic field due to a subpopulation of nerve cells whose longitudinal axes are perpendicular to the cerebellar surface, by simply placing the cerebellum vertically in the bath so that these cells are horizontal and by measuring the field along the rostrocaudal axis perpendicular to the longitudinal axis of these nerve cells. To elicit neural activity in these cells the dorsal midline was electrically stimulated with a bipolar electrode. Consistent with our expectations, the one-dimensional profile of the field normal to bath surface (Bz) was antisymmetrical along the rostrocaudal axis, implying that the underlying currents were transcortical. Also, the Bz field at a field extremum varied as a cosine of the orientation of the cerebellum when it was rotated about its rostrocaudal axis with the amplitude being zero when the cerebellum was horizontal. The Bz field was dipolar as judged by statistically excellent fits of the dipolar field to the one-dimensional field profile and to the distance function relating the field magnitude at an extremum to measuring distance. This was the case even for the initial component thought to be due to antidromic action currents invading the soma and dendrites of Purkinje cells. We also showed that the dipolar term of the source could be localized within 1 mm of the actual source location in most cases.

Time domain signal transmission and source location in a waveguide: Matched filter and deconvolution experiments

The experiments were made in an ideal air waveguide having Sheetrock (hard) boundaries. Transmissions of a signal were from a source to a pair of receivers. Matched filter and deconvolution procedures were tested. Matched filters were computed for transmissions from a set of trial source locations to the receivers. The trial matched filters were then applied to the signals. The filtered signals were cross correlated. A map of cross-correlation absolute maxima for trial source locations gave the actual source location. Many smaller maxima indicated false locations. For the second method, the signals were deconvolved with estimates of the impulse transmission from trial locations. At the actual source location, deconvolution operations gave a diverging oscillatory remainder. The oscillations had a higher frequency than the signal. It was possible to low-pass filter the deconvolved signals and recover good replicas of the transmitted signal. The map of cross correlations of the deconvolved and filtered signals gave the source location. The deconvolution–low-pass filter method gave fewer and smaller indications of false locations than the matched filter method. However, the matched filter method was more tolerant of errors in the estimates of the impulse response.

Optimum time domain signal transmission and source location in a waveguide: Experiments in an ideal wedge waveguide

Source location experiments in a wedge waveguide were conducted to test a single-channel time domain technique for a known source. The laboratory measurements were made in the air wedge having rigid sheet rock boundaries. The wedge angle was 6.6°. The source transmitted a short oscillatory signal having a frequency band of 3–13 kHz. The distances of the source and receiver from the wedge apex were 1.85 and 3.13 m. The Biot–Tolstoy wedge solution was used to calculate the source-to-receiver impulse response. For the geometry, all of the arrivals are due to images. The waveguide matched filter was the time reverse of the theoretical impulse response. The output of the matched filter was cross correlated with the known input signal. Maps of possible source locations were made by (i) assuming a trial source location, (ii) computing a theoretical impulse response and the corresponding matched filter, (iii) filtering the signal, and (iv) mapping the peak absolute maximum of the matched filter output. The actual source location was correctly mapped. There were several lower amplitude maxima that gave false source locations. The Appendix includes a tutorial description of the Laplace generating function (z transforms) and their applications.

Diffraction limitations on sound‐imaging experiments for hearing‐aided listeners

It had been proposed by Bauer [J. Acoust. Soc. Am. 33, 1536–1539 (1961)] that two loudspeakers of different source strengths in the horizontal plane could be used to produce a virtual sound image, located at other than the actual source location. It is shown here that, due to the diffraction of sound by the head, this particular method produces simultaneous and interdependent interaural time (ITD) and level differences (ILD). The particular ITDs and ILDs produced by Bauer's method do not generally correspond to a unique spatial location and therefore produce potentially conflicting localization cues. This difficulty is aggravated by unbalancing the hearing aid gains or their phase response. It appears from the study by A. Nabelek, T. Letowski, and D. Mason [J. Speech Hear. Res. 23, 670–687 (1980)] that aided listeners position the sound image on the basis of interaural time differences. The role of interaural pressure level differences on the positioning of sound images could not be established unambiguously from these data.

Binaural horizontal plane localization: Influence of monaural spectral cues

A recent article [Butler and Flannery, Percept. Psychophys. 28, 449–457 (1980)] reported the existence of monaural spatial referent maps (SRMs). These orderly maps related the perceived horizontal plane locus of narrow bands of noise to center frequency, which ranged from 4–14 kHz. Perceived location was governed primarily by spectral composition and not actual source location. Similar SRMs under binaural listening conditions were obtained in the current experiments. Stimuli were presented from a loudspeaker at 270° azimuth. Subjects could report the perceived origin as any of 13 positions, each 15° apart and extending from 360°to 180°. Results were consistent with observations regarding the region of the horizontal plane most sensitive to spectral cues. [Musicant and Butler, J. Acoust. Soc. Am. Suppl. 1 71, S87 (1982)]. Localization tests utilizing specific features of the SRMs decreased binaural localization accuracy. A 4-kHz high-pass noise with a notch at a “map-connecting” region was the stimulus. Stimuli were presented from all 13 loudspeakers in the horizontal plane. Performance decrements were greatest within the 315°–225° arc. The present results strengthen our argument that monaural spectral cues are an important cue to binaural horizontal plane localization.

Effect of training on human underwater sound-localization.

Experimental evidence has suggested that humans have a moderately functional sound-localization capability underwater; it appeared appropriate to determine if this ability could be improved by training. Signals of 1000 Hz and 25 pps were presented to the subjects at five angles (to the head). Four sets of stimuli were employed: the first and last sets constituted the pre- and post-tests, respectively, and the middle sets were a training procedure where subjects were informed of the actual source location after each presentation. The obtained scores demonstrated a significant increase in correct responses for the 1000-Hz signal; improvement for 25 pps was not as great and did not show statistical significance. Subject Classification: 65.62; 30.85.