Springer-Verlag 1998 Exp Brain Res (1998) 118:541±550 RESEARCH ARTICLE Randolph D. Easton ´ Anthony J. Greene Paul DiZio ´ James R. Lackner Auditory cues for orientation and postural control in sighted and congenitally blind people Received: 11 November 1996 / Accepted: 29 July 1997 Abstract This study assessed whether stationary auditory information could affect body and head sway (as does visual and haptic information) in sighted and congenitally blind people. Two speakers, one placed adjacent to each ear, significantly stabilized center-of-foot-pressure sway in a tandem Romberg stance, while neither a single speaker in front of subjects nor a head-mounted sonar device reduced center-of-pressure sway. Center-of-pressure sway was reduced to the same level in the two-speaker condition for sighted and blind subjects. Both groups also evidenced reduced head sway in the two-speaker condition, although blind subjects head sway was significantly larger than that of sighted subjects. The advantage of the twospeaker condition was probably attributable to the nature of distance compared with directional auditory information. The results rule out a deficit model of spatial hearing in blind people and are consistent with one version of a compensation model. Analysis of maximum cross-correlations between center-of-pressure and head sway, and associated time lags suggest that blind and sighted people may use different sensorimotor strategies to achieve stability. Key words Balance ´ Auditory information ´ Vision ´ Blindness ´ Human Introduction It is well known that vision exerts a strong influence on the postural stability of humans. During quiet, unperturbed stance, for example, normal subjects can maintain balance when vision is denied, but body sway typically increases by a factor of 2 (Dichgans and Brandt 1978). ) R.D. Easton ( ) ´ A.J. Greene Boston College, Department of Psychology, Chestnut Hill, MA 02167, USA Fax: +1-617 552-0523, e-mail: Randolph.Easton@bc.edu P. DiZio ´ J.R. Lackner Brandeis University, Waltham, MA 02154, USA The effect is generally attributed to visions high resolution of spatial position, including body position, relative to vestibular or proprioceptive inputs. Recently, it has also been demonstrated that haptic cues at the fingertip can enhance postural stability in the absence of vision, even though the forces between a stable surface and the finger are far below those constituting mechanical support (Holden et al. 1994; Jeka and Lackner 1994, 1995). The haptic sensory cues, arising from cutaneous, kinesthetic, and proprioceptive inputs, presumably also provide high spatial resolution in terms of specifying body position and change of body position. It has also been demonstrated that haptic cues facilitate postural stability in labyrinthine-defective individuals (DiZio et al. 1997) as well as congenitally blind people (Jeka et al. 1996a), indicating that the haptic information can substitute for the absence of vestibular information or a visual spatial reference system, respectively. Together these findings regarding the effects and interactions of different sensory cues on postural stability suggest an additional question: Can the auditory system, like vision and haptics, provide sufficiently precise spatial information regarding ongoing body position to effect postural stability? There is reason to believe that auditory information is only minimally suited to influence postural stability. Of relevance here are findings from the sensory dominance or bias paradigm, where, in an attempt to discern a sensory control hierarchy, a discrepancy is created between spatial information processed in two modalities (see Welch and Warren 1980 for a review). The response of the system typically reveals a biasing effect, that is, one modality dominates the other in terms of perceptual experience. This is interesting, for the case of perceived object position vision is found to strongly dominate audition and proprioception, while audition and proprioception exert only small biasing effects on vision. When a discrepancy is created between audition and proprioception (with the use of a pseudophone), audition is substantially dominated by proprioception and exerts little influence on proprioception. In the context specifically of perceived self-position, it is additionally known that a rotating 542 sound field (external or internal) can induce an illusory experience of self-rotation, just as a rotating visual field does, but the effect is eliminated if a stable visual environment is present (Lackner 1977). As perceived self-position is complementarily related to perceived object position (Lee 1978, 1990), the auditory system may exert only a limited influence on perceived body position and resultant postural stability. The suitability of auditory information for control of postural stability may be particularly important in the context of its effects in congenitally blind people, who typically tend to rely on auditory information to garner spatial knowledge that would otherwise be gathered by vision. Good examples in the current context of spatial orientation are blind peoples use of traffic sounds for parallel or perpendicular alignment in street crossing (Hill and Ponder 1976) or the approach of a motor vehicle (Schiff and Oldak 1990). Two theoretical perspectives on the role of congenital blindness in the development of hearing can be identified: deficit and compensation models (Ashmead et al. 1996). The deficit model argues that nonvisual information is customarily encoded within a visual spatial frame of reference, and individuals lacking such a reference system will evidence impaired spatial hearing (Pick 1974; Warren 1984, 1970; Warren and Pick 1970). The compensation model argues that, in the absence of typical visual inputs, nonvisual areas of perception may become more highly developed than in sighted people (for a review, see Rauschecker 1995). In their strongest versions, these models lead to opposite predictions about the differences in spatial hearing in sighted compared with blind people and, by extension, the potential effects of auditory information on the control and regulation of postural stability. That is, given the lack of a visual-spatial frame of reference, the deficit model would predict little or no effect of auditory information on the postural control of blind people, while the compensation model would predict that, especially in congenitally blind people, auditory information could prove capable of influencing postural stability. In the present study the postural stability of sighted and congenitally blind individuals was assessed using a tandem Romberg stance (heel-to-toe). Auditory information was provided through either a single directional speaker positioned directly in front of the subjects head, at the intersection of the median and transverse planes, or through two speakers, one placed adjacent to each of the subjects ears (see Fig. 1). Given that lateral plane (right-to-left) body sway is enhanced in the tandem Romberg stance, the single-speaker condition results in the traditional information regarding object-to-self direction ± interaural intensity, time of arrival, and phase differences. In contrast, the two-speaker condition results in amplitude changes at the two ears regarding object-to-self distance. Moreover, intensity change at the two ears obeys the inverse square law, that is, intensity varies reciprocally with the square of the distance. Thus the information available is particularly pronounced at short distances (Ashmead et al. 1995, p. 241), especially those used in the present study. In addition to these auditory conditions, the effect Fig. 1 Three auditory information conditions: (a) a single speaker placed in front of the subject in the median plane, (b) two speakers, one placed adjacent to each ear, or (c) a head-mounted sonar device. All conditions were presented in a normal ambient environment with sound-reflecting surfaces. A 2.54-cm-diameter (1-inch) vertical pole was placed 30 cm in front of the subject in the sonar condition to provide self and object cues of a head-mounted sonar device known as the Trisensor (L. Kay, 1985 and unpublished work) was assessed. The Trisensor was designed for use as a mobility and orientation aid for blind people. It is purported to be an ªenvironmental sensorº (Brabyn 1985) in the sense that returning echoes from a transmitted signal inform the user of the direction, distance, and texture of objects within the range of the device, as well as providing acoustic flow during locomotion. In addition, it has been demonstrated that the Trisensor can improve the ability of blind people (who received over 12 h of distributed practice with the device in spatial localization) to perform a challenging balance task on a moving (rocking side-to-side) stabilometer platform placed before three cylindrical, 2.54-cmdiameter poles (Easton 1992). In contrast, stationary directional speakers emitting highly localizable sounds mounted on the poles did not increase stability on the stabilometer platform, a result attributed to the superior precision of the head-mounted sonar in specifying head and body position relative to the environment during 543 Table 1 Characteristics of blind subjects a Retinopathy of prematurity is a vascular abnormality of the retina characterized by neovasularization and resultant sequelae, occurring almost exclusively in premature infants b Leber©s congenital amaurosis is an inherited disorder characterized by retinal pigmentary degeneration c Usher©s syndrome is an inherited disorder, primarily affecting males, which is characterized by degeneration of the retinal pigment epithelium, cataracts, and hearing loss Participant Age (years) Sex Travel aid Cause of blindness B.M. 44 M Long cane and dog guide Retinopathy of prematuritya M.S. 52 F Long cane Retinopathy of prematurity T.W. 28 M Long cane Retinopathy of prematurity P.M 44 F Dog guide Retinopathy of prematurity G.C. 36 F Long cane and dog guide Retinopathy of prematurity J.B. 24 F Dog guide Agricultural pesticide during first of 3 months© gestation M.P. 19 M Long cane Leber©s congenital amaurosisb M.B. 24 M Dog guide Probable Usher©s syndromec large, side-to-side head or body rocking movements (and compensations) which typically occur for the stabilometer task.1 For the present experiment, subjects were tested in the dark with and without auditory information in a tandem Romberg (heel-to-toe) stance, while center-of-foot-pressure as well as head displacement were monitored. In our previous work on postural stability of blind people in this situation, it was found that haptic cues reduced center-of-pressure sway to the same level and extent for both blind and sighted people, but blind people evidenced significantly more head sway than their sighted counterparts (Jeka et al. 1996). One interpretation of the finding is that head movement control in sighted humans and animals may be driven primarily by movements directed toward orienting and stabilizing the eyes in space (Goldberg and Peterson 1986; Outerbridge and Melvill Jones 1971), which may not develop normally in the congenitally blind. Thus, the larger levels of head displacement in blind subjects may not be subserving spatial orientation and postural control, but may result from the inability to coordinate head movements in terms of eye-head synergies and gaze control. An additional question addressed therefore in the present research was whether auditory information, which is processed by receptors located on the head (indeed, in the case of the sonar, receptors as well as a transmitter are located on the head), would attenuate head displacement in blind people, substituting as a kind of gaze stabilization mechanism. 1 The choice of these forms of auditory information was not motivated theoretically. A 500-Hz square wave was chosen over a sine wave because of the presence of transients and resultant localizability. The sonar was chosen because of its stabilizing effect on the stabilometer task (Easton 1992). However, other auditory information could prove particularly effective at stabilizing posture, especially highly localizable, broad-band acoustic wave forms. Additional information Mild hearing loss. Active participation in outdoor activities such as skiing and climbing Materials and methods Subjects Eighteen individuals participated in the experiment, ten sighted and eight congenitally blind, ranging in age from 25 to 50 years (see Table 1). They were healthy and physically active, with no known musculoskeletal or neurological disorders that might have affected their ability to maintain balance. Human subject use was approved by the appropriate ethics committee and all participants gave informed consent prior to participating. Apparatus and measurement Figure 2 depicts our test situation. The participants stood in the tandem Romberg position (heel-to-toe) on a force platform, with their arms at their sides. The force platform measured the reaction forces generated by the feet, indicating displacement of the center of pressure. Medial-lateral (CPx) and anterior-posterior (CPy) coordinates of foot pressure were computed from the medial-lateral (Fx), anterior-posterior (Fy), and vertical (Fz) force components registered by piezo-electric crystals in the corners of the force platform (Kistler model 9261A) and the distances of the crystals from the center of the platform. Medial-lateral (Hx) and anterior-posterior (Hy) head movements were measured with an ISCAN video system. An LED was attached to a headband and the ISCAN camera tracked the movements of the LED to measure Hx and Hy directions of head sway. The ISCAN system measures two-dimensional (2-D) movement in a field of view 512 pixels (Hx)256 pixels (Hy). Because of differing subject heights, we normalized the field of view across participants by measuring the distance between the camera and LED and computing a calibration factor for each participant. The mean resolution across subjects was 0.48 mm (Hx) and 0.96 mm (Hy). All signals were sampled at 60 Hz and collected in real time on a personal computer instrumented with a Data Translation analog-digital board. Figure 1 depicts the configuration of the different auditory information conditions. Auditory information was presented in a normal laboratory room in the following manner: For a single-speaker condition, a speaker was mounted 30 cm in front of participants and level with their ear canals. For a two-speaker condition, speakers were 544 Fig. 2 Participant in the tandem Romberg position on the force platform. Primary sway is in the lateral, side-to-side direction mounted approximately 5 cm to the left and right of the participants pinnas, level with their ear canals. The speakers were driven by a 500-Hz square wave and delivered at 73 dB. In addition to these auditory conditions, participants wore a head-mounted sonar, the Trisensor (Kay 1985 and unpublished work), which is an air-borne sonar patterned after its predecessor, the Binaural Sensory Aid or BSA (Kay 1974). The basic principle on which this aid operates is that a small, head-mounted transmitter irradiates the field of view (essentially a 60 cone) with very high frequency acoustic waves (50±100 kHz), which encounter objects and surfaces in the environment. Reflected waves, which return to the observer as echoes, are detected by head-mounted transducers and these are transformed into an audible acoustic display (essentially, returning echoes are multiplied with the transmitted signal) that potentially informs the user about objects distances, directions, and surface textures. Distance of objects is coded by the pitch of the audible input. The display also receives binaural inputs from angled receivers that yield interaural differences in amplitude level, creating a sense of spatial direction and movement. A unique feature of the Trisensor sonar is that it adds monaural signals derived from a narrow beam (10) in front of the observer. This latter feature provides the user with higher acuity information regarding objects directly ahead. For the present situation, a 2.54-cm-diameter, 182.88-cm-tall pole was placed 30 cm from the participant in the median plane, matching the placement of the one-speaker condition. Given lateral sway in the Romberg-tandem stance, the sonar primarily created an interaural amplitude cue to object or self direction. Procedure Participants stood with their right foot placed directly behind their left foot along the center of the anterior-posterior axis of the force platform. Adhesive tape was used to mark the positions of the feet on the platform, so that the same configuration could be precisely repeated during each trial. The experimental trials included five conditions: eyes open (sighted participants only), eyes closed and no sound, one speaker, two speakers, and sonar feedback sound. Participants began each trial by assuming the tandem stance and positioning their arms at their sides with their heads level, facing forward, with their eyes open or closed, depending on the condition. The visual environment provided a rich, complex scene with many vertical and horizontal cues. Participants were told to take as much time as desired to assume a comfortable stance. Their instructions were to maintain the tandem stance as comfortably as possible while Fig. 3 Raw data time-series wave forms of center-of-pressure sway in the lateral direction (CPx) and head sway in the lateral direction (Hx) for a blind and a sighted subject for the no-sound control and the two-speaker auditory input conditions. The two-speaker condition significantly attenuated both center-of-pressure and head sway in sighted and blind subjects swaying as little as possible. Once they felt ready, subjects said ªgoº and the experimenter initiated data acquisition. Practice trials were given for each condition before the experiment began. The experimental trials were run in four blocks of 5 (sighted) or 4 (blind) trials (one of each condition per block) for a total of 20 or 16 trials. Conditions were randomized within a block. Trial duration was 24 s. If a participant was unsuccessful in a particular trial (e.g., lost balance), then the trial was repeated until performed correctly. After each trial, the participant stepped off the platform and sat comfortably for at least 1 min. The experiment lasted approximately 1 h. Analysis Raw data consisted of time-series wave forms of CPx, CPy, Hx, and Hy. Figure 3 presents examples from a typical sighted and blind subject for the no-sound control condition and the most stabilizing auditory condition ± two speakers. The four time series were reduced to four summary values as follows. First, 4 s were cut from the beginning and end of each series to minimize possible contamination from anticipation of the start and finish of a trial. The absolute deviation of each data point from the mean of the entire 16-s raw signal was found, and these were averaged to yield an index of mean sway amplitude (MSA). Cross-correlations between CP and H sway were calculated to determine the temporal relation of body displacement and head sway. Jeka et al. (1996) found greater head sway in blind than sighted subjects, suggesting the possibility that the two groups use different coordinative strategies to achieve body sway stability. Correlations were performed at each of 100 steps (15.624 ms/step) in both the forward and backward directions to determine whether correlations were strongest at times other than t=0 (i.e., in phase). Finally, power spectral density analyses were performed to determine the component frequencies of CPx and Hx displacement, with a frequency resolution of 0.0586 Hz. Mean power spectra were calculated by collapsing across subjects and trials for each condition. To determine the effects of the different auditory input conditions as well as to directly compare sighted and blind participants, we conducted a 24 mixed ANOVA (subject groups, between; sen- 545 sory input condition, within). We used the omnibus MSe for withinsubjects planned comparisons and the pairwise MSe for betweensubjects planned comparisons. Correlations between CP and H sway were transformed to the Fischers exact z-test for statistical analysis, because raw correlations are not normally distributed (Senders 1958). Results CPx mean sway amplitude Mean lateral center-of-foot-pressure sway is presented in Fig. 4 (top) for sighted and blind subjects under the different sensory input conditions. (As a point of reference, mean CPx displacement for sighted subjects with eyes open was 0.25 cm (SE 0.02). An ANOVA revealed a significant effect of sensory input (F3,48=4.29, MSe 0.004, P<0.01) and no effect associated with the groups factor or the interaction between groups and sensory input. Planned contrasts between the auditory input conditions and the control condition, collapsed across groups, revealed that only the two-speaker condition resulted in significantly reduced lateral body sway (F1,48=8.77, P<0.01). As an index of individual differences or the consistency of the two-speaker effect, it was found that eight of ten sighted and seven of eight blind subjects showed reduced CPx sway in the two-speaker condition compared with the no-sound condition. Hx mean sway amplitude Mean lateral head sway is presented in Fig. 4 (bottom) for subject groups under sensory input conditions. (Mean Hx sway for sighted subjects with eyes open was 0.33 cm (SE 0.03). An ANOVA revealed a significant effect for the group factor (F1,16=8.9, MSe 0.25, P<0.01), with the sighted evidencing smaller head sway than their blind counterparts. Planned contrasts between the auditory input conditions and the control condition, collapsed across groups, indicated that both the two-speaker and the sonar conditions resulted in significantly reduced, mean lateral head sway (F1,48=3.1, P<0.08 and F1,48=2.9, P<0.09, respectively). As an index of individual differences and the consistency of the two-speaker effect, it was found that seven of ten sighted subjects and six of eight blind subjects showed a reduction in Hx sway in the two-speaker condition compared with the no-sound condition. For the sonar condition, five of ten sighted subjects and seven of eight blind subjects showed reduced Hx sway compared with the no-sound condition. CPy mean sway amplitude Mean anterior-posterior center-of-foot-pressure displacement is presented in Fig. 5 (top) for the subject groups across sensory input conditions. (Mean CPy for sighted subjects with eyes open was 0.25 cm (SE 0.03). An AN- Fig. 4 Top Means and standard errors for center-of-pressure sway in the lateral direction (CPx). Bottom Means and standard errors for head sway in the lateral direction (Hx) OVA revealed a significant effect for sensory input conditions (F3,48=3,58, MSe 0.004, P<0.02) and no effect associated with the groups factor or the interaction between groups and sensory input. Planned contrasts between the auditory input conditions and the control condition, collapsed across groups, revealed that only the two-speaker condition resulted in significantly reduced anterior-posterior body sway (F1,48=9.14, P<0.01). Hy mean sway amplitude Mean anterior-posterior head sway is presented in Fig. 5 (bottom) for subject groups across sensory input conditions. (Mean Hy for sighted subjects with eyes open was 0.39 cm (SE 0.03). An ANOVA revealed no significant main or interaction effects. Planned contrasts between the auditory input conditions and the control condition, collapsed across groups, nevertheless revealed that both 546 Fig. 5 Top Means and standard errors for center-of-pressure sway in the anterior-posterior direction (CPy). Bottom Means and standard errors for head sway in the anterior-posterior direction (Hy) Fig. 6 Top Maximum correlations and standard errors for CPx and Hx. Bottom Associated time lags and standard errors for CPx and Hx. Negative time lags mean that CPx leads Hx temporally the two-speaker and sonar conditions resulted in reduced anterior-posterior head sway (F1,48=5.3, P<0.025 and F1,48=3.5, P<0.06, respectively). indicate that center-of-foot pressure was leading head sway. An ANOVA comparing blind and sighted subjects across sensory input conditions revealed no main or interactive effects. The high CPx and Hx correlations and negative time lags mean that head movements were strongly related but temporally behind center-of-pressure movements. Cross-correlations between CPx and Hx Maximum cross-correlations between CPx and Hx, as well as the time lags associated with the correlations are presented for both groups of subjects in Fig. 6. As can be seen, the maximum correlations for both groups of subjects were high across all conditions, averaging about 0.80. A mixed two-way ANOVA comparing subject groups across sensory input conditions revealed no main or interactive effects. The time lags corresponding to the correlations (Fig. 6, bottom) were uniformly negative for both groups of subjects across all conditions, ranging from about 50 to 100 ms. The negative time lags Cross-correlations between CPy and Hy Maximum cross-correlations between CPy and Hy, as well as the time lags associated with the correlations, are presented for both groups of subjects in Fig. 7. Again the correlations for both groups were high across all conditions, averaging in this case about 0.65. An ANOVA comparing subject groups across sensory input conditions revealed no main or interactive effects. 547 For the most part, maximum mean power occurred at about 0.15 Hz for both CPx and Hx across the sensory input conditions. The most striking difference in the power spectra for sighted compared with blind people appears in Hx displacement, where blind people evidenced substantially more power than sighted subjects across the frequency spectrum for all four sensory input conditions. This finding, of course, is related to the greater overall Hx sway for blind subjects. Separate 210 ANOVAs (groupfrequency bin) were conducted on Hx for each sensory input condition, and in each case the effect of group as well as the interaction was significant. As can be seen, the interaction was attributable to proportionately greater Hx power by blind people at the lower frequency bins. Overall, for all four dependent measures, CPx, CPy, Hx, and Hy, the anterior-posterior components generally mirrored those of the lateral components. Given that lateral sway is the primary component of body sway in the tandem Romberg stance, the discussion of the results will be focused on the lateral center-of-pressure and head sway. Discussion Fig. 7 Top Maximum correlations and standard errors for CPy and Hy. Bottom Associated time lags and standard errors for CPy and Hy. Negative time lags mean CPy leads Hy temporally The time lags corresponding to these correlations(Fig. 7, bottom) were uniformly negative for both groups of subjects across all sensory input conditions, ranging from 25 to 125 ms. An ANOVA comparing blind and sighted subjects across sensory input conditions revealed a significant interaction between groups and input conditions (F3,48=3.02, P<0.04). The interaction was attributable to significantly greater time lags for blind subjects for the no-sound and sonar conditions. In general, however, like CPx and Hx relations, the high CPy and Hy correlations and negative time lags indicate that head movement is strongly related but temporally behind center-of-pressure movements. Mean power spectra for CPx and Hx The mean power spectra in each frequency bin are presented in Figs. 8 and 9 for CPx and Hx, respectively. The primary goal of this research has been to determine whether auditory information regarding a stable environment can, like visual and haptic information, enhance postural stability. The effect of auditory information on postural stability was assessed in sighted as well as congenitally blind people, who lack a visual reference system to encode spatial information. Using a tandem Romberg (heel-to-toe) stance, it was found that body sway was significantly attenuated when two stationary speakers, one placed adjacent to each ear, were present for both sighted and blind people. Neither a single speaker placed in the median plane directly in front of the subject nor a headmounted binaural sonar device attenuated body sway for either group of subjects. Blind people evidenced greater overall head movement than sighted people; but both groups of subjects exhibited significantly less head sway in the presence of either the two-speaker or sonar auditory information. In discussing these results, it is useful to compare the magnitude of the effects to the magnitude of visual and haptic influences on postural stability. As can be seen in Fig. 4, center-of-pressure or body sway was reduced by about 10% for the two-speaker condition relative to the control condition. Recall from the results that sway in sighted subjects with eyes open averaged 0.25 cm. Thus vision attenuates sway about 60%, far greater than auditory information. In our prior work assessing the effects of haptic cues on postural stability of blind and sighted subjects (Jeka et al. 1996), we found that the most stabilizing condition consisted of subjects holding a cane in a slanted orientation (relative to the body) against the floor while applying a very small force (<0.4 N of applied force), a condition which does not afford mechanical support, indicating the attenuation of body sway must be sensory in or- 548 Fig. 8 Means and standard errors for CPx power spectra across sensory cue conditions Fig. 9 Means and standard errors for Hx power spectra across sensory cue conditions 549 igin. The attenuation of sway under these conditions was found to be about 40% relative to a sighted control condition. Thus, while auditory information from a stable environment can significantly improve stability, its effects are small compared with those of vision and haptics.2 This pattern of sensory influences parallels the pattern found using the sensory dominance or bias paradigm, for which a modality precision hypothesis has been proposed to account for a sensory control hierarchy for the perception of spatial position (e.g., Pick et al. 1969). Likewise, the customary precision of processing spatial information in a modality may determine the influence different sensory information will have on perceived self-position and stabilization of posture. The relatively small influence of auditory information on postural stability is also attested to by the fact that neither the one-speaker nor sonar condition attenuated sway. The greater influence of the two-speaker condition relative to the one-speaker condition is probably explained in terms of the sound intensity changes at the ears which occur with lateral sway and signal body position through distance change. According to the inverse square law, sound intensity changes would be particularly pronounced for the 5-cm speaker distance used in this condition (Ashmead et al. 1995, p. 241). Given that the mean head sway magnitude under the auditory conditions was 1±1.5 cm (see Figs. 1, 4), quite substantial sound intensity changes would occur for even these small head movements. In contrast, the 1- to 1.5-cm head sway did not serve to generate sufficient interaural difference information about object to self direction (i.e., intensity, time-of-arrival, and phase) for a single speaker located 30 cm in front of the subject.3 The lack of an effect for the sonar device relative to the control condition appears surprising, as the Trisensor sonar has been found to improve stability on the stabiliometer platform task by approximately 30% (Easton 1992). However, two differences between the present study and the earlier assessment of sonar input on postural stability should be noted. First, blind subjects in the stabilometer study received extensive training in spatial localization with the sonar (12 h of distributed practice). Second, information regarding spatial position is quite pronounced with the sonar in the stabilometer task, as large side-toside head and body movements (about the knees) occur 2 In a related finding it has recently come to our attention that ambient auditory clicks fail to reduce the variance in sway when gastrocnemius muscles are vibrated (Petersen et al. 1995, 1996) 3 It is important to note another factor that could contribute to the stabilizing effect of the two-speaker compared with the one-speaker condition. The one-speaker condition yields inherently ambiguous auditory information regarding lateral head translation compared with rotation (about its vertical axis). That is, for the one-speakercondition, either a head translation or a rotation would yield similar interaural differences, whereas for the two-speaker condition the two types of head movement are uniquely specified, with translation resulting in an intensity increase in one ear and a decrease in the other ear, and rotation resulting in an intensity decrease in each ear. As head translation is the crucial variable to stabilize in the tandem Romberg stance, the two-speaker condition would unambiguously specify its occurrence. which yield large interaural amplitude differences regarding object to self direction relative to stable objects in front of the subject (i.e., three, 2.54-cm-diameter poles separated by30.48 cm). By contrast, for the Romberg stance, balance is irretrievably lost after the occurrence of much smaller head and body movements than occur on the stabilometer. Thus, either a lack of practice with sonar information or the relative lack of information about body position, given that only relatively small head movements can occur in the Romberg stance, could account for the lack of a sonar influence on postural stability. Sonar input did significantly affect head stabilization, however, as did the two-speaker condition, for both groups of subjects. Although, with the sonar, head sway was reduced for both blind and sighted people, centerof-pressure or body sway was not reduced. Thus while the sonar had a small stabilizing effect on head sway it was not accompanied by reduced center-of-pressure sway. The analysis of center-of-pressure and head sway cross-correlations and associated time lags can be used to suggest coordination strategies used by subjects to achieve postural stability. As will be recalled (Fig. 6), the maximum correlations were similar and high for both groups of subjects, and the associated time lags were negative and relatively small. At the small amplitude and low frequency of sway observed in the present test situation, it is reasonable to assume that all body segments were essentially moving together as a single unit in blind and sighted subjects (i.e., an inverted pendulum), even though the head movements of the blind subjects were larger than those of sighted subjects. Indeed, for the blind subjects, Hx sway was about twice as large as their CPx sway, which would be predicted by an inverted pendulum model. It is as though blind participants, perhaps because of their inherently unstable head posture (see Jeka et al. 1996), ªlockº the head to the body to form a single rigid unit in order to achieve center-of-pressure stability. It is also worth noting that blind subjects evidenced head sway that was twice as large as body sway only in the lateral and not in the anterior-posterior direction, which would be precisely what was needed to stabilize posture in the tandem Romberg stance. Sighted subjects performance, in contrast, may not have so closely resembled an inverted pendulum, perhaps because of greater head stability fostered by reflexive or learned gaze stabilization mechanisms (Goldberg and Peterson 1986; Outerbridge and Melvill Jones 1971). This analysis suggests that blind peoples inherently less stable head position requires them to adopt a more passive body stabilization strategy than sighted people. Overall, however, the general effects of auditory information on postural stability in this experiment are strikingly similar for blind and sighted subjects. This pattern of results is clearly not consistent with a deficit model of auditory perception, which would have predicted superior performance by sighted people. The results are also inconsistent with a strong version of the compensation model, which would predict superior performance by 550 blind subjects, but are consistent with a weaker version of the compensation model, which would predict simply that blind subjects are capable of compensating nonvisually for a lack of vision (see Rauschecker 1995). The present findings are consistent with a recent comprehensive assessment of spatial hearing in blind and sighted subjects that also found no striking differences between the groups on a wide range of auditory tasks, including minimum audible angle and distance, reaching to sounds, walking to sounds and perceiving the path of a moving sound source (Ashmead et al. 1997). It should be noted, however, that, consistent with a strong version of the compensation model, Schiff and Oldak (1990) have reported that blind people are more accurate at judging time to arrival of an approaching vehicle using auditory information than are sighted subjects and are comparable with sighted subjects accuracy on the task using visual information. Even though judging vehicle approach, using auditory information, and the stabilizing effect of the two-speaker condition on postural stability in the present study both involve using auditory intensity changes regarding object or self distance, blind people may not customarily use auditory information for precise postural control as extensively as for vehicle or object approach. On the other hand, Jeka et al. (1996) report no differences between sighted and blind subjects for the effects of a hand-held cane on postural stability, a task with which blind people have substantially more experience. In sum, the present results clearly indicate that a deficit model of auditory spatial perception in blind people is untenable. Additional research is needed to assess the strong version of the compensation model of auditory perception in blind humans. One possible means of doing so would be to provide blind people with extensive spatial orientation training over an extended period in an environment rich in reliable, high-energy acoustic information (Easton and Bentzen 1997), in order to determine whether they can develop superior auditory perception. Acknowledgements The research reported has been supported in part by NIH grants EY04907 and EY05887 to R. D. Easton and by AFOSR grant F49620-95-1±0390, and NASA grants NAGW 4374 and NAGW 4375 to J. R. Lackner. We are grateful to Eli Rabin for assistance in data collection and analysis, and Beezy Bentzen for discussion regarding the perceptual world of blind people and for assistance in blind subject recruitment. References Ashmead DH, Davis DL, Northington A (1995) The contribution of listeners approaching motion to auditory distance perception. J Exp Psychol Hum Percept Perform 21:239±256 Ashmead DH, Wall RS, Ebinger KA, Hill M, Yang X, Eaton S (1997 Manuscript submitted for publication) Spatial hearing in children with visual disabilities. Brabyn JH (1985) A review of mobility aids. In: Warren D, Strelow E (eds) Electronic spatial sensing for the blind. Martinus-Nijhoff, Dordrecht, pp 13±27 Dichgans J, Brandt T (1978) Visual-vestibular interaction: effects on self-motion perception and posture control. In: Held R, Leibowitz HW, Teuber HL (eds) Handbook of sensory physiology. Springer, Berlin Heidelberg New York, pp 755±804 DiZio P, Jeka JSJ, Horak FB, Lackner JR, Krebs D (1997) Reduction of postural sway due to precision contact of the finger tip in patients with bilateral vestibular loss. (Manuscript submitted for publication) Easton RD (1992) Inherent problems of attempts to apply sonar and vibrotactile sensory aid technology to the perceptual needs of the blind. Optom Vis Sci 69:3±14 Easton RD, Bentzen BL (1997) The effect of spatial localization training in an acoustically rich environment on spatial updating in the blind. (Manuscript submitted for publication) Goldberg J, Peterson BW (1986) Reflex and mechanical contributions to head stabilization in alert cats. J Neurophysiol 56: 857±875 Hill EW, Ponder P (1976) Orientation and mobility techniques: a guide for the practitioner. American Foundation for the Blind, New York Holden M, Ventura J, Lackner JR (1994) Stabilization of posture by precision contact of the index finger. J Vestib Res 4:285± 301 Jeka JJ, Lackner JR (1994) Fingertip contact influences human postural control. Exp Brain Res 100:495±402 Jeka JJ , Lackner JR (1995) The role of haptic cues from rough and slippery surfaces in human postural control. Exp Brain Res 103:267±276 Jeka JJ, Easton RD, Bentzen BL, Lackner JR (1996) Haptic cues for orientation and postural control in sighted and blind individuals. Percept Psychophys 58:409±423 Kay L (1974) A sonar aid to enhance spatial perception of the blind: engineering design and evaluation. Radio Electron Eng 44:605± 627 Kay L (1985) Sensory aids to spatial perception for blind persons: their design and evaluation. In: Warren D, Strelow E (eds) Electronic spatial sensing for the blind. Martinus-Nijhoff, Dordrecht, pp 125±139 Lackner JR (1977) Induction of illusory self-rotation and nystagmus by a rotating sound field. Aviat Space Environ Med 48:129±131 Lee DN (1978) The functions of vision. In: Pick HL, Saltzman E (eds) Modes of perceiving and processing information. Erlbaum, Hillsdale, NJ, pp 159±170 Lee DN (1990) Getting around with light and sound. In: Warren R, Wertheim AH (eds) Perception and control of self-motion. Erlbaum, Hillsdale, NJ, pp 487±505 Outerbridge JS, Melvill Jones G (1971) Reflex control of head movement in man. Aerosp Med 42:935±940 Petersen H, Magnusson M, Johansson R, ¾kesson M, Fransson P-A (1995) Acoustic cues and postural control. Scand J Rehab Med 27:99±104 Petersen H, Magnusson M, Johansson R, Fransson P-A (1996) Auditory feedback regulation of perturbed stance in stroke patients. Scand J Rehab Med 28:217±223 Pick HL (1974) Visual coding of nonvisual spatial information. In: MacLeod RB, Pick HL (eds) Perception: essays in honor of James J. Gibson. Cornell University Press, Ithaca, NY, pp 153±165 Pick HL, Warren DH, Hay JC (1969) Sensory conflict in judgments of spatial direction. Percept Psychophys 6:203±205 Rauschecker JP (1995) Compensatory plasticity and sensory substitution in the cerebral cortex. Trends Neurosci 18:36±43 Schiff W, Oldak R (1990) Accuracy of judging time to arrival: effects of modality, trajectory, and gender. J Exp Psychol Hum Percept Perform 21:303±316 Senders V (1958) Measurement and statistics. Oxford University Press, New York Warren DH (1970) Intermodality interactions in spatial localization. Cognitive Psychol 1:114±133 Warren DH (1984) Blindness and early childhood development. American Foundation for the Blind, New York Warren DH, Pick HL (1970) Intermodality relation in blind and sighted people. Percept Psychophys 8:430±432 Welch RB, Warren DH (1980) Immediate perceptual response to intersensory discrepancy. Psych Bull 88:638±667