Localization of low-pass sounds was tested in relation to aspects of Wallach's(1939Wallach's( , 1940 hypotheses about the role of head movement in frontJback and elevation discrimination. With a 3-sec signal, free movement of the head offered only small advantage over a single rotation through 45°for detecting elevation differences. Veryslight rotation, as observed using a 0.5-sec signal, seemed sufficient to prevent front/back confusion. Cluster analysis showed that, in detecting elevation, some listeners benefited from rotation, some benefited from natural movement, and some from both. Evidence was found indicating that a moving auditory system generates information for the whereabouts of sounds, even when the movement does not result in the listener facing the source. Results offer significant if partial support for Wallach's hypotheses.Classical theory ofauditory localization suggests that, for motionless listening, interaural cues, at least for simple sounds, provide spatial information that is essentially ambiguous. Interaural cues take the form of time differences for low-frequency acoustic energy and level differences for high-frequency energy. A classical model assumes the head to be a perfect sphere, with holes, centered on opposite sides, for ears. In this model, interaural cues specify the angle of horizontal displacement of a sound source from the median vertical plane (MVP), with the vertex of the angle at the center of the listener's head. From here on, this will be referred to as the azimuth angle. Its properties are consistent with the azimuth angle defined in the double-pole coordinate system used by Middlebrooks, Makous, and Green (1989). Figure 1 illustrates this system and shows that a particular azimuth angle encompasses a range of directions, including the direction of the source. An azimuth angle of 60°to the left, for example, specifies a cone-shaped locus covering positions forward, rearward, above, and below the interaural axisa cone ofconfusion (Woodworth & Schlosberg, 1954). It can thus be seen that, considered purely in terms ofgeometry, interaural cues alone do not specify the elevation of the source, nor whether it is forward or rearward.We wish to thank Malcolm A. Perrett, who helped provide mathematical solutions for processing of the head tracker data and whose development of the DSP software and digital signals used in the experiment formed part of his Honours (Electrical Engineering) project. Appreciation is extended to Frank Niebling for construction ofthe supporting framework for the loudspeakers and the spherical screen, and to Dean Davidson for computer programming. We also thank the reviewers of earlier drafts of the paper, whose comments led to significant developments in our thinking and practice.

Current understanding gives predominant weight to stationary cues for auditory localization. Two experiments were conducted to investigate the possible existence of a dynamic cue. The first experiment involved localization of concealed sources in the upper median vertical plane (MVP) and showed, as expected, that elevation was not detectable with motionless listening when high-frequency energy was absent or when normal pinna function was distorted. Elevation under such conditions did become detectable with horizontal head rotations, provided low-frequency energy was present in the signal. This indicates that the basis of the dynamic cue is variation in the rate of transformation of low-frequency interaural time/phase differences. The second experiment involved localization of sources arrayed throughout upper and lower regions of the MVP and in the left lateral vertical plane (LVP); it showed that upper hemisphere sources can be distinguished somewhat from those in the lower hemisphere, even in motionless listening conditions, but more so with rotation. The greatest benefit for localization from rotation of the head appears to be gained for sources positioned in the front MVP.

Listeners had the task of following a target speech signal heard against two competitors either located at the same spatial position as the target or displaced symmetrically to locations flanking it. When speech was the competitor, there was a significantly higher separation effect (maintained intelligibility with reduced target sound level), as compared with either steady-state or fluctuating noises. Increasing the extent of spatial separation slightly increased the effect, and a substantial contribution of interaural time differences was observed. When same-and opposite-sex voices were used, a hypothesis that the similarity between target and competing speech would explain the role for spatial separation was partly supported. High-and low-pass filtering showed that both parts of an acoustically similar competing signal contribute to the phenomenon. We conclude that, in parsing the auditory array, attention to spatial cues is heightened when the components of the array are confusable on other acoustic grounds.

Successful replication of an experiment by Butler and Humanski (1992) showed that listeners are able to proficiently localize sources on a lateral vertical plane on the basis of interaural differences alone, When a lateral horizontal array was included in the test setup, that finding was replicated only for a broadband signal interacting with the pinna, not for ones (lowpass and pure tone) providing only interaural differences. Cross-plane errors conforming to "cones of confusion" were observed for those latter sounds. In a second experiment, response options were made more unconstrained, which clarified the nature of the cross-plane confusions. Lowpass signals from lateral vertical plane sources tend to be heard at or close to the horizon. Measurement of cue values needs to take account of the response options available to listeners, as well as signal properties.Classical theory of sound localization in space has stressed the role of the binaural system. Early research (Stevens & Newman, 1936) showed that the auditory system is sensitive to interaural time differences for lowfrequency sounds and interaural level differences for sounds of high frequency. The low-frequency cue is considered to be largely an effect of time and phase difference detection; the high-frequency cue is the result ofincreasingly sharp acoustic shadowing by the head. In classical theory, the head is assumed to be a perfect sphere, with the ears as holes located at the extremes of a diameter through the sphere (Mills, 1972). Using this model, interaural differences specify the horizontal angle of the sound source to' the left or right of the median vertical plane (MVP). Absence of difference specifies that the source is somewhere on the MVP. Such differences cannot unambiguously specify the vertical angle of the sound source relative to the horizontal plane nor, relatedly, whether it lies forward or rearward of, or above or below, the interaural axis. In normal listening conditions, and with a broadband noise signal, those aspects of spatial whereabouts seemingly not revealed by "classical" interaural differences can nonetheless be distinguished.In a recent report, Butler and Humanski (1992) presented results showing that listeners could accurately discriminate the whereabouts of sound sources displaced at 15 0 intervals in a quadrant above the interaural axisin the lateral vertical plane (LVP). Proficient performance was observed for bursts of 3-kHz highpass filtered noise. Such short wavelength signals interact withThe authors thank Bruce Stevenson for valuable advice about the design of Experiment 2.