Exp Brain Res (2007) 178:37–48

DOI 10.1007/s00221-006-0709-y

RESEARCH ARTICLE

Auditory biofeedback substitutes for loss of sensory information in maintaining stance

Marco Dozza · Fay B. Horak · Lorenzo Chiari

Received: 22 August 2005 / Accepted: 7 September 2006 / Published online: 5 October 2006© Springer-Verlag 2006

Abstract The importance of sensory feedback forpostural control in stance is evident from the balanceimprovements occurring when sensory informationfrom the vestibular, somatosensory, and visual systemsis available. However, the extent to which also audio-biofeedback (ABF) information can improve balancehas not been determined. It is also unknown why addi-tional artiWcial sensory feedback is more eVective forsome subjects than others and in some environmentalcontexts than others. The aim of this study was todetermine the relative eVectiveness of an ABF systemto reduce postural sway in stance in healthy controlsubjects and in subjects with bilateral vestibular loss,under conditions of reduced vestibular, visual, andsomatosensory inputs. This ABF system used a thresh-old region and non-linear scaling parameters custom-ized for each individual, to provide subjects with pitchand volume coding of their body sway. ABF had thelargest eVect on reducing the body sway of the subjectswith bilateral vestibular loss when the environmentprovided limited visual and somatosensory informa-tion; it had the smallest eVect on reducing the sway ofsubjects with bilateral vestibular loss, when the envi-ronment provided full somatosensory information. Theextent that all subjects substituted ABF informationfor their loss of sensory information was related to the

extent that each subject was visually dependent orsomatosensory-dependent for their postural control.Comparison of postural sway under a variety of sen-sory conditions suggests that patients with profoundbilateral loss of vestibular function show larger thannormal information redundancy among the remainingsenses and ABF of trunk sway. The results support thehypothesis that the nervous system uses augmentedsensory information diVerently depending both on theenvironment and on individual proclivities to rely onvestibular, somatosensory or visual information to con-trol sway.

Keywords Posture · Center of pressure · Vestibular loss · Audio-biofeedback · Sensory integration · Sensory substitution · Posture control

Introduction

The control of postural sway depends on continuousfeedback of sensory information from the vestibular,somatosensory, and visual senses. The largest increasein postural sway in stance occurs when somatosensoryinformation is compromised (Nashner et al. 1982). Thenext largest increase occurs when vestibular informa-tion is lost, and the smallest, when vision is eliminatedby eye closure (Peterka and Black 1990; Macphersonand Inglis 1993; Horak and Macpherson 1996). Theseincreases in postural sway suggest that the central ner-vous system (CNS) relies primarily on somatosensoryinformation, less so on vestibular information, and evenless so on visual information to control postural swayduring quiet stance. In fact, a linear sensory interaction

M. Dozza · F. B. Horak (&)Neurological Sciences Institute, Oregon Health & Science University, 505 NW 185th Ave, Beaverton, OR 97006, USAe-mail: horakf@ohsu.edu

M. Dozza · L. ChiariDepartment of Electronics, Computer Science, and Systems, University of Bologna, Bologna, Italy

123

38 Exp Brain Res (2007) 178:37–48

model predicts such postural sway in adults duringstance by proposing a 70% dependence on somatosen-sory information from a Wrm surface, 20% on vestibu-lar information, and 10% on visual information(Peterka 2002). However, several studies support thenotion that the CNS re-weighs its relative dependenceon sensory information when the availability of infor-mation from diVerent senses changes (Jeka et al. 2000;Black and Nashner 1984; Horak and Hlavacka 2001).For example, when healthy subjects stand on an oscil-lating surface with eyes closed, they increasinglydepend on vestibular information and visual informa-tion and decrease dependence on somatosensory infor-mation from the surface as the amplitude of the surfacerotations increases (Peterka 2002).

It is as yet unknown the extent to which the CNS re-weighs its relative dependence on sensory informationin presence of augmented sensory information. Aug-mentation of sensory information, such as auditoryinformation, could be useful for rehabilitation of bal-ance in patients with sensory loss, especially if the CNSproportionately integrates this information with thenatural sensory information depending on the sensorydemands of the task.

One type of augmentation to reduce postural sway—auditory information in the form of biofeedback—hasreceived minimal investigation. When audio-biofeedback(ABF) was investigated, it was usually in conjunctionwith visual biofeedback (Nichols 1997; Barclay-Goddardet al. 2004). In studies of ABF and visual biofeedback,the sound constituting the ABF was a simple alarm sig-nal (Wong et al. 1997; Batavia et al. 2001) that was usedto augment the visual biofeedback. However, anothertype of ABF, able to represent a complex informationand not limited to an alarm signal, may be especiallyuseful to augment postural feedback since auditory cues:(1) are easy to integrate with the remaining senses insensory-impaired individuals, such as those with vestib-ular losses (Wickens and Hollands 2000), (2) do notinterfere with visual information, and (3) are capable ofsignaling spatial information (Vinge 1971; Nelson et al.1998). To illustrate this last point, humans use hearingfor spatial localization whenever we turn our heads tolocate the source of a sound. In addition, it has beenshown that novice pilots can learn how to Xy in a Xightsimulator using either visual information or auditorytracking for turns, bank angles, and tilt (Forbes 1946),and it was subsequently determined that healthy sub-jects can use auditory information nearly as accuratelyas visual information to detect body orientation andmotion in space (Vinge 1971).

Auditory and vestibular information are both trans-mitted to the brain via the VIII cranial nerve, which

projects to the temporal lobe. Auditory cues automati-cally (subconsciously) inXuence postural alignment,and postural alignment automatically alters the abilityto locate auditory cues in the environment (Lackner1974; Lackner and DiZio 2000). Even stationary audi-tory cues were found to reduce the body sway of con-trol and blind subjects when the cues were from stereospeakers in close proximity to both ears (Easton et al.1998).

Recently, it has been found that subjects with a loss ofvestibular information were able to use both ABF(Dozza et al. 2005b; Hegeman et al. 2005) and tactilebiofeedback (Tyler et al. 2003; Kentala et al. 2003) thatmap their body movement in order to reduce posturalsway. However, subjects with and without vestibular lossvaried widely in their ability to reduce sway with aug-mented sensory ABF and vibrotactile biofeedback. Thereasons for this inter-subject variability are unknown.However, similar inter-subject variability was also foundwhen subjects with and without vestibular loss relied ontheir three natural sources of sensory information(visual, vestibular, and somatosensory) to control pos-tural sway (Black and Nashner 1984; Lacour et al. 1997).For example, 50% of subjects with neuromas on theVIII cranial nerve increased their postural sway instance with eyes closed, but 50% decreased or did notchange their sway with eyes closed (Lacour et al. 1997).After surgery to remove the neuroma, the same sub-jects, who were visually dependent (i.e., relied more onvisual than on somatosensory information to maintainbalance) before the surgery, no longer increased theirsway with eyes closed, whereas those subjects who werenot visually dependent increased their sway with eyesclosed after surgery. Further, as people age or areexposed to weightlessness in space for a long time,many, but not all, increased their relative dependenceon visual and somatosensory information to maintainbalance (Paloski et al. 1992; Woollacott 1993). Sensorycompensation for pathological loss of sensory informa-tion has also been found to vary among subjects withprofound bilateral loss of vestibular information (BVL,bilateral vestibular loss). Fifty percent of these subjectswere able to signiWcantly reduce body sway during sur-face oscillations by opening their eyes, whereas theother Wfty percent could not (Buchanan and Horak2001). Studying BVL subjects using a custom-madeABF, Hegeman et al. (2005) reported balance improve-ments when they stood with eyes open on Wrm surfacebut not on foam surface or with eyes closed. However,Hegeman et al. (2005) did not perform any analysisaimed at understanding how and why individual subjectswere able or unable to use the ABF information toimprove their stability in the diVerent postural tasks.

123

Exp Brain Res (2007) 178:37–48 39

In the study described here, we investigated howindividual subjects’ relative dependence on a particularsensory channel inXuenced their ability to reduce pos-tural sway in stance when they used ABF informationto control body sway. The objectives of this researchwere (1) to determine the extent to which ABF infor-mation helps control postural sway given limitedvisual, vestibular, and surface somatosensory informa-tion and (2) to account for why the relative eVective-ness of ABF varies among individuals across sensoryenvironments. We used an ABF system, which wedesigned to mimic aspects of otolith vestibular infor-mation by monitoring accelerations in the transverseplane (Chiari et al. 2005).

Methods

Participants

Nine subjects, four men and Wve women, with pro-found BVL and nine age- and gender-matched, healthycontrol subjects participated in this study. There wereno signiWcant age, height, and weight diVerences(P > 0.05) between the BVL and control subjects,respectively: age 55 years (38–73) versus 55 years (33–72); height 171 cm (160–193) versus 167 cm (151–180);and weight 71 kg (51–115) versus 73 kg (65–86).Table 1 summarizes the BVL subjects’ pathologies,ages, duration of their vestibular loss, and their hori-zontal vestibulo-ocular reXex (VOR) gain at 0.05 Hz.Normal VOR gains range from 0.7 to 1 for the controlsubjects. All BVL subjects had a bilaterally absentresponse to warm and cold water on caloric tests and aVOR gain of less than 0.3 across a range of oscillationsbetween 0.01 and 0.1 Hz, indicating severe loss of ves-tibular function (Peterka and Black 1990). In addition,each BVL subject fell without an apparent posturalresponse soon after the start of surface sway-referenc-ing trials with eyes closed, consistent with their BVL(Nashner et al. 1982). All of the BVL and control sub-jects were free of hearing, orthopedic, and neurologicaldiseases or disorders, except the vestibular pathologyfor BVL subjects. Written informed consent wasobtained from all subjects prior to their participation.The rights of the participants were protected accordingto the 1964-Declaration of Helsinki.

Apparatus

For all experiments, the BVL and the control subjectswore a custom-made ABF system (Chiari et al. 2005)while standing on an AMTI OR6–6 force plate. The

ABF system provided auditory information to the sub-jects about their body sway while they stood on theforce plate.

The ABF system is comprised of three main parts:the sensory unit, the sensory processing unit, and theaudio output unit (Chiari et al. 2005). The sensory unitconsists of a small (1.5 £ 3 £ 3 cm3) sensor that ismounted on the subject’s back at L5 with a Velcro belt.The sensory unit uses 3031 Eurosensor accelerometers(range §0–50, resolution 2 £ 10¡4 g, noise 1 �v p-p,temperature error-zero ¡0.05 mV/°C) to sense the lin-ear accelerations along the anterior–posterior (AP)and medial–lateral (ML) directions near the bodycenter of mass. L5 was chosen because its position isminimally aVected by movement artifacts, such asrespiration, heartbeat, and voluntary head or limbmovement. The processing unit consists of a laptopcomputer (Intel Celeron 2.4 GHz) equipped with anA/D board (DAQCard NI 6024E). It acquires, records,and processes the AP and ML accelerations sensed bythe sensory unit and encode them into two analog sinewaves that constitute the ABF stereo sound. Theclosed-loop delay introduced by the processing wasestimated to be 5 ms. We developed the software forthe processing unit using Matlab© 6 R12 and MatlabData Acquisition Toolbox (Chiari et al. 2005). Theaudio output unit consists of an ampliWer (Fostex PH-5) that boosts the two sine waves provided by the com-puter so that the subjects are able to hear tonesthrough the earphones (Philips SBC HP-140), with thetones representing the degree and direction of thebody accelerations.

The force plate estimates body sway in the AP andML directions by recording forces and torques underthe subject’s feet. In certain testing conditions, theforce plate was covered with a 10 cm-thick, medium-density TemperTM foam (indentation force deXectionat 25%: 116 N, tensile strength: 125 kN/m2, elongation:109%, when temperature is 22.2°C and relative humidity

Table 1 Characteristics of BVL subjects including the vestibule-ocular reXex gain (VOR)

Subject’s ID

Age Diagnosis Duration of loss (years)

VOR

#1 46 Ototoxicity 7 0.030#2 50 Idiopathic 12 0.006#3 56 Idiopathic 14 0.005#4 60 Ramsey Hunt 3 0.020#5 61 Ototoxicity 10 0.047#6 38 Auto immune

disease7 0.140

#7 53 Ototoxicity 7 0.260#8 56 Ototoxicity 10 0.007#9 73 Ototoxicity 9 0.022

123

40 Exp Brain Res (2007) 178:37–48

is 50%) to reduce somatosensory information aboutbody sway from the feet. When a subject stands on thefoam, the distance between the subject’s feet and theforce plate continuously changes due to the complianceof the foam itself. As a consequence, the estimation ofthe center of pressure (COP) displacement was theo-retically not as accurate as without foam. However, theerror of estimation was calculated in post-process andfound to be smaller than 10%. Linear accelerationsfrom the sensory unit, as well as forces and torquesfrom the force plate, were acquired with a 100-Hz sam-ple rate.

Figure 1 shows, from a top-down perspective, fourdirections of sway and the relative ABF stereo soundchanges in each earphone, for each direction. The ABFleft–right balance and the volume in the earphoneschange according to ML body sway, and the pitch andvolume of the stereo sounds change according to APbody sway (Chiari et al. 2005). In this study, all soundswere dynamically adjusted for each subject based onunique deWnitions of: (1) the region of natural sway(Mayagoitia et al. 2002), and (2) the area of the sup-port base that is the region of a safe sway. Using aninverted-pendulum model (Gage et al. 2004), theregion of natural sway and the region of safe sway wereuniquely calculated for each subject. SpeciWcally, theregion of natural sway was determined by the range ofAP and ML accelerations compatible with an oscilla-tion of §1° around the vertical, which depended uponthe subject’s height. The region of the safe sway wasdetermined by the range of AP and ML accelerationscompatible with the subject’s COM projection on theground, not exceeding the subject’s base of foot sup-port. Thus, the region of natural sway and region ofsafe sway were used to customize and to optimize theABF tones for each subject.

The ABF system was designed so that the toneschanged, depending on the subject’s sway relative tothe calculated region of natural sway. When a subjectswayed within his or her calculated region of natural

sway in the ML and AP directions, the same constant,low-volume (20 dB-SPL), 400-Hz tone was fed back tothe subject through each earphone. However, when asubject swayed outside his or her region of naturalsway in the ML direction, the tones in the earphonessimultaneously became louder in the ear correspond-ing to the direction of body sway and quieter in theother ear. When the subject swayed outside the regionof natural sway in the anterior direction, the toneschanged equally in both ears and became louder (up to50 dB-SPL) in volume and higher in pitch (following alinear function up to 1000 Hz). When the subjectswayed outside the region of natural sway in the pos-terior direction, the tones changed equally in both earsand became louder in volume and lower in pitch (fol-lowing a linear function down to 150 Hz). When thesubject swayed outside the region of natural sway in anoblique direction, for example in the anterior-leftdirection, the tones became higher in pitch in bothears, louder in volume in the left ear, and quieter involume in the right ear. All the equations used to gen-erate the ABF sound using sigmoidal function arereported in detail in Chiari et al. (2005).

Procedure

Subjects stood on the force plate and kept their feet 15°externally rotated and their heels 1 cm apart (narrowstance position). They were instructed to maintainquiet stance throughout all testing when using and notusing the ABF device. Before the experimental proto-col began, subjects practiced with the ABF system for afew minutes on a Wrm surface with eyes open by volun-tarily swaying at diVerent angles and directions, and lis-tening to the corresponding changes in tones in theearphones until they understood how the trunk infor-mation was coded into the ABF sound. The subjectswere instructed to correct their body sway by using thetones, i.e., to maintain their sway within the region ofnatural sway by achieving a constant 400-Hz tone in

Fig. 1 ABF sound dynamics encoding postural sway. Pitch and volume change in the two earphones, depending on the direction of sway. The arrows in the middle of the force plate (outlined) indicate the direction of sway. The regions of natural sway (NS) and safe sway (SS) were customized for each subject

123

Exp Brain Res (2007) 178:37–48 41

each earphone. Once they understood how to changetheir body sway to achieve the constant 400-Hz tone,they performed three practice trials with eyes closedand without ABF, followed by three practice trials witheyes open on foam and without ABF. The purpose ofthe practice trials was for the subjects to gain conW-dence in standing with eyes closed or standing on thefoam-covered force plate without falling, and to mini-mize the initial eVects of standing on the foam. Datafrom these practice exercises and trials were not con-sidered in the analyses.

BVL subjects repeated a block of six conditionsthree times (18 trials total), and the control subjectsrepeated the same block of six conditions Wve times (30trials total). For each of these blocks, the six conditionswere presented in random order; three conditions werewith and three conditions were without ABF. Condi-tions one and two were: eyes closed on a Wrm surfacewithout ABF and with ABF. Conditions three and fourwere: eyes open on foam surface without ABF andwith ABF. Conditions Wve and six were: eyes closed onfoam surface without ABF and with ABF. We did nottest the eyes-open on Wrm-surface condition since, inthis condition, the sway of both the BVL and the con-trol subjects is expected to be inside the region of natu-ral sway, so there is no need for additional ABFinformation (Nashner et al. 1982). The BVL subjectsperformed fewer trials to limit fatigue. Each trial lasted1 min.

Data and statistical analysis

From the 2D, planar COP displacement, we quantiWedpostural sway with two independent parameters (Pri-eto et al. 1996; Rocchi et al. 2004; Maurer et al. 2004):the root-mean-square distance (COP-RMS) and thefrequency below which the 95% of the power of thesignal is included (F95%). From the 2D, planar accel-eration measured by the sensory unit, we computed theRMS (Acc-RMS). To determine the eVect on sway ofsubject groups, conditions, and ABF, we performed athree-way ANOVA, 2 groups (BVL and control) £ 3sensory conditions (vestibular, somatosensory, andvisual), repeated (eyes closed, eyes open on foam, andeyes closed on foam) £ 2 ABF conditions, repeated(ABF on and oV) for each parameter (COP-RMS,F95%, and Acc-RMS). The threshold for statistical sig-niWcance was P = 0.05.

To evaluate the correlation between severity of ves-tibular loss and the eVect of ABF on sway amplitude inthe eyes closed on foam condition, a robust regressioncorrelation analysis was performed between the VORgain and the percentage reduction in COP-RMS, with

and without ABF for BVL subjects. To assess whetherABF was eVective in helping subjects reduce bodysway in proportion to each subject’s level of depen-dency on visual and somatosensory information, arobust regression correlation analysis was performedbetween the levels of sensory dependency and theeVect of ABF on COP-RMS when only visual (eyesopen on foam with ABF condition) or only somatosen-sory information (eyes closed with ABF condition) wasavailable. The levels of visual dependency and somato-sensory dependency were estimated for each subject asthe percentage of the body sway reduction occurringwhen visual or somatosensory information was added(visual information, in the eyes open on foam condi-tion and somatosensory information, in the eyes closedcondition) and were compared to the reference eyesclosed on foam condition (when neither visual andsomatosensory information was available).

Results

Center of pressure displacement

For BVL and control subjects, body sway increased asnatural sensory information or ABF informationbecame absent or unreliable. Further, COP-RMS wassigniWcantly larger in the eyes open on foam conditionthan in the eyes closed condition (P < 0.05). COP-RMSwas also signiWcantly larger when eyes were closed thanwhen eyes were open while subjects stood on foamwithout ABF (P < 0.01). In the eyes closed, eyes openon foam, and eyes closed on foam conditions, BVL sub-jects’ COP-RMS was signiWcantly larger than the con-trol subjects’ COP-RMS (P < 0.001). Figure 2 shows theanterior–posterior versus lateral COP displacements ofone representative BVL subject (Fig. 2a) and one rep-resentative control subject (Fig. 2b), in all six condi-tions. Table 2 reports the COP-RMS values in the eyesclosed, eyes open on foam, and eyes closed on foamconditions for both subject groups.

In the three ABF conditions, both groups beneWtedfrom ABF. That is, ABF signiWcantly decreased COP-RMS for both the BVL and control groups (P < 0.05).The percentage of changes in COP-RMS due to ABF isshown in Table 3. No signiWcant interaction was foundbetween the groups and the conditions tested sinceCOP-RMS was larger in BVL subjects than in controlsubjects in every condition. In addition, there was nosigniWcant interaction between the groups and ABF asboth groups improved in the conditions tested. A sig-niWcant interaction was found between the conditionfactor and the ABF factor (P < 0.001) due to ABF

123

42 Exp Brain Res (2007) 178:37–48

decreasing COP-RMS more in the eyes closed on foamcondition than in the eyes closed or eyes open on foamcondition (Table 3). For the BVL subjects in the eyesclosed on foam condition, a signiWcant interaction wasfound among all three ANOVA factors (P < 0.001)due to ABF decreasing COP-RMS the most in the eyesclosed on foam condition for all BVL subjects.

Figure 3 shows the average COP-RMS reductionwhen BVL and control subjects used ABF on foamwith eyes closed. As shown in Fig. 3, all but one of the

BVL subjects able to perform the eyes closed on foamcondition beneWted from ABF in this condition. Inaddition, BVL subject #2 fell a few times in the eyesclosed on foam condition, but she never fell in this con-dition while using ABF. BVL subject #1 fell consis-tently in the eyes closed on foam condition but alsonever fell in this condition while using ABF. BVL sub-ject #8 beneWted from ABF, although minimally whencompared to the other BVL subjects. BVL subject #5(Fig. 3) was not able to stand in the eyes closed onfoam condition, with or without ABF, although he ben-eWted from ABF in the other conditions (eyes closedand eyes open on foam). Also as shown in Fig. 3, allcontrol subjects beneWted from ABF in the eyes closedon foam condition.

Frequency spectrum

For BVL and control subjects, the amount of posturalcorrections (indicated by the parameter F95%)decreased as natural sensory information becameavailable or reliable and increased when ABF informa-tion was available. SpeciWcally, the frequency spectrumcomponents of the COP were signiWcantly aVected bythe diVerent test conditions, with the power at thehigher frequencies increasing when visual and/orsomatosensory sensory information was reduced(P < 0.001). F95% was higher in the eyes closed onfoam condition than in the eyes open on foam condi-tion (P < 0.05), and higher in the eyes open on foamcondition than in the eyes closed condition (P < 0.05).F95% was also higher for the BVL subject group thanfor the control group in all conditions (P < 0.001).Table 2 reports F95% values in the three conditionstested without ABF for the BVL and control subjects.The use of ABF signiWcantly increased F95% for boththe BVL and control subjects in all conditions(P < 0.001). Table 3 shows the percent of increase inF95% when controls and BVL subjects used ABF ineach condition. There was a signiWcant interaction(P < 0.05) between the condition tested and the pres-ence of a vestibular deWcit, with F95% increasing in the

Fig. 2 COP displacement in the horizontal plane is representedfor a one representative bilateral vestibular loss subject and b onecontrol subject in the three conditions: (1) eyes closed (EC), (2)eyes open on foam (EOF), and (3) eyes closed on foam (ECF),without (black) and with (gray) ABF

a

b

Table 2 Mean values and standard deviation (in parenthesis) of postural parameters for bilateral vestibular loss (BVL) and control sub-jects in the three conditions tested without audio-biofeedback (ABF)

Root mean square distance (RMS) is reported for the center of pressure displacement (COP) and for the acceleration sensed at trunklevel (Acc). Also, the values of frequency, below which the 95% of the power of the COP signal is included, are reported

Parameter Eyes closed Eyes open and foam Eyes closed and foam

BVL Control BVL Control BVL Control

COP-RMS (mm) 13.82 (8.9) 8.31 (2.8) 14.01 (9.7) 9.34 (1.2) 24.66 (7.58) 14.92 (3.7)F95% (Hz) 1.85 (0.55) 1.31 (0.15) 1.87 (0.52) 1.39 (0.19) 2.51 (0.31) 1.59 (0.18)Acc-RMS (mm/s2) 14.12 (8.07) 12.61 (2.4) 16.79 (9.50) 13.48 (1.9) 56.09 (19.13) 21.84 (5.60)

123

Exp Brain Res (2007) 178:37–48 43

BVL subject group more than in the control group,particularly in the eyes closed on foam condition.

Sensory substitution

Subjects beneWted from ABF information in relation tothe lack of natural sensory information. For most BVLsubjects, the extent that they reduced their body swaywith ABF in the eyes closed on foam condition corre-lated with the extent of their vestibular loss (r = 0.76;

P < 0.05). Table 1 shows the VOR gains and percent-age of improvement in sway for all of the subjects usingABF. One subject with very low VOR gain (#9) couldonly stand with the ABF in this condition so the per-centage of improvement could not be calculated.

For both the BVL and control groups, the eVective-ness of ABF in reducing body sway was related to howdependent each subject was on visual or somatosen-sory information, but not on the amount of sway in thebaseline eyes closed on foam condition. Somatosen-sory-dependent subjects beneWted the most from ABFwhen somatosensory information was missing, andvision-dependent subjects beneWted the most fromABF when visual information was missing. Figure 4shows the relative dependence of each subject onvisual or somatosensory information versus theamount of beneWt that each received from ABF underconditions in which visual or somatosensory informa-tion was limited (i.e., the eyes closed and eyes open onfoam conditions). A linear relationship for the BVLsubjects and the control subjects was found betweenthe degree of beneWt from ABF and their dependenceon visual and somatosensory information, shown bythe greater number of circles in the top-right and bot-tom-left quadrants of Fig. 4. The circles in the top-rightquadrant represent the subjects who were somatosen-sory-dependent and beneWted the most from ABFwhen somatosensory information was missing. The cir-cles in the bottom-left quadrant represent subjects whowere vision-dependent and beneWted the most fromABF when visual information was missing.

Discussion

ABF eYcacy in reducing sway is related to the availability of sensory information

Results from this study show that the amount that ABFcompensates for missing sensory information dependson the extent of sensory loss. When somatosensoryinformation was reduced (the eyes open on foam

Fig. 3 The percentage of COP-RMS reduction using ABF is re-ported for each bilateral vetibular loss (a) and control (b) subjectin the condition eyes closed on foam. Data were ordered by per-centage improvement using ABF. Subject numbers indicatematching subjects between the groups. 9 BVL Subject 2 fell twicewithout ABF but never fell during trials using ABF. ‡ BVL Sub-ject 9 fell repeatedly with and without ABF. § BVL Subject 1could stand only with the help of ABF. Black, dashed lines repre-sent the mean reduction using ABF. Gray, shadowed areas repre-sent the standard error of the reduction using ABF

b

a

Table 3 Mean percentage diVerence of each postural parameter with and without audio-biofeedback (ABF) for bilateral vestibular loss(BVL) and control subjects

Root mean square distance (RMS) is reported for the center of pressure displacement (COP) and for the acceleration sensed at trunklevel (Acc). Also, the values of frequency, below which the 95% of the power of the COP signal is included, are reported

Parameter Eyes closed Eyes open and foam Eyes closed and foam

BVL Control BVL Control BVL Control

COP-RMS ¡3.24 ¡10.87 ¡9.98 ¡5.42 ¡23.07 ¡15.90F95% 21.90 23.01 10.54 18.89 8.38 9.28Acc-RMS ¡20.82 ¡35.24 ¡27.38 ¡40.56 ¡46.18 ¡32.15

123

44 Exp Brain Res (2007) 178:37–48

condition) and the more that BVL and control subjectswere somatosensory-dependent, the more they beneWtedfrom ABF and were able to reduce their sway. Whenvisual information was not available (the eyes closedcondition) and the more that BVL and control subjectswere visually dependent, the more they also beneWtedfrom ABF and were able to reduce their sway. Whenboth somatosensory information and visual informationwere limited (the eyes closed on foam condition), bothBVL and control groups showed the most beneWt fromABF. Thus, we hypothesize that the degree to whichsubjects beneWt from ABF to reduce postural swaydepends on their degree of visual, somatosensory andvestibular loss (Lacour et al. 1997; Kluzik et al. 2005).

Our results also showed a trend in which the moresevere the vestibular loss, the more subjects beneWted

from ABF. This trend needs further testing with moresubjects in order to show statistical signiWcance. OurWndings are consistent with other studies that alsoreported that control and BVL subjects were able toreduce postural sway with visual, tactile, and audio-biofeedback (Wall et al. 2001; Hegeman et al. 2005).However, our study, for the Wrst time, has identiWed apotential relationship between beneWts from ABFinformation and limitation of sensory information.

Both BVL and control subjects’ postural swayincreased when sensory information was limited, con-Wrming the commonly held hypothesis that the controlof postural sway depends on the amount of availablesensory feedback that is available (Horak et al. 2002;Dickstein et al. 2001; Peterka 2002). Our BVL subjectsshowed signiWcantly larger sway than did our controlsubjects in all conditions tested, in agreement withother studies (Hufschmidt et al. 1980; Black and Nash-ner 1984; Gagey and Toupet 1991). However, the BVLsubjects’ degree of sway reduction via ABF wheneither visual information or somatosensory informa-tion was available was not related to the extent of theirvestibular loss. This Wnding may be due to the sub-jects’ hesitance to rely on novel sensory information(available via ABF) when ordinary sensory informa-tion normally and extensively used to compensate theloss of vestibular information (Lacour et al. 1997) wasalso available. However, this Wnding may also beexplained by the ABF information not yet being inte-grated with the subjects’ existing somatosensory andvisual information since they used ABF for only15 min or less during testing. This lack of integration isalso supported by another study in which we foundthat the use of ABF requires a larger number of rapidpostural corrections (Dozza et al. 2005a). Lack of inte-gration may be the consequence of the subjects’ pay-ing excessive attention to the ABF, thus interferingwith the attention paid to other sensory information.It has been shown how dual-task interferencedecreases with practice over time when tasks becomequasi-automatic (Schumacher et al. 2001). Conse-quently, it may be possible for ABF information tobecome more integrated with other sensory informa-tion as when ABF is used after a longer period of timethan just the few minutes in our study (Dault andFrank 2004).

Attention to natural sensory information may have limited ABF eYcacy in BVL subjects

Although BVL subjects reduced their sway more thanthe control subjects did in the eyes closed on foamcondition, they did not in the eyes open or in the Wrm

Fig. 4 Subjects in terms of their vision and somatosensory depen-dency. There is a correlation between the use of ABF in the eyesclosed and eyes open on foam conditions, and visual and somato-sensory dependency. Each subject’s tendency to rely, more on vi-sion or somatosensory information is reported on the horizontalaxis. Negative values imply a dependency on vision more than onsomatosensory information, whereas positive values imply adependency on somatosensory more than on vision information(a zero value on the horizontal axis indicates a subject who relieson vision as much as on somatosensory information to maintainbalance in stance). The vertical axis shows the eVect of ABF foreach subject. Positive values imply ABF reduces sway more whensomatosensory information is made unreliable by standing onfoam, negative values imply ABF reduces sway more when visualinformation is missing (a zero value on the vertical axis indicatesa subject who, when using ABF, reduces sway when vision infor-mation is limited as much as when somatosensory information isinadequate). The Pearson coeYcient for the regression line isr = 0.57 comprising data from both group and is statistically sig-niWcant (P < 0.05). The Pearson coeYcients reported in the Wgurefor the two groups of subjects separated (r = 0.62 and r = 0.65 forbilateral vestibular loss and control subjects, respectively) are notstatistically signiWcant (P > 0.05), however they are close to statis-tical signiWcance P = 0.06

123

Exp Brain Res (2007) 178:37–48 45

surface conditions. In contrast, Hegeman et al. (2005)found that BVL subjects reduced sway in stance usingABF only with eyes open on a Wrm surface, but notwith eyes closed and/or when on foam. This diVerenteVect of ABF may be related to diVerences in: (1) thedesign of the ABF systems, (2) the use of trunk angularvelocity instead of linear acceleration that was fed backto the subjects, (3) the linear algorithm chosen to maptrunk movement into sound, (4) subject selection, and(5) how postural sway was measured and quantiWed. Inour study, the high degree of attention that BVL sub-jects normally pay to visual and somatosensory infor-mation in the eyes closed and eyes open on foamconditions may have limited their ability to use ABFsince the initial use of ABF requires some a degree ofattention to the tones in the earphones (Dozza et al.2005a). Indeed, during the rehabilitation period ofBVL subjects, they are taught to pay more voluntaryattention to visual and somatosensory informationthan would be the case if they did not have the BVL, tocompensate for the vestibular loss (Shumway-Cookand Horak 1990; Shumway-Cook et al. 1996). Conse-quently, focusing more on visual information andsomatosensory information available in the eyes openon foam and eyes closed conditions, may have inter-fered with their ability to concentrate on the ABF(Shumway-Cook et al. 1997; Redfern et al. 2004).However, in the eyes closed on foam condition, whenvisual information and somatosensory informationwere limited, subjects could focus their attention onthe ABF. Another explanation for subjects’ decreasingtheir sway with ABF is that their use of ABF and theheadphone equipment inXuenced them to pay moreattention to their sway. However, in studies in whichsubjects were instructed to deliberately focus theirattention on their body sway and to increase their con-trol of posture, they did not reduce their sway(Vuillerme and Nafati 2005). Thus, we believe that thelarge sway reduction induced by ABF in BVL subjectswas not likely only due to the subjects’ paying moreattention to their sway.

Use of ABF reduced BVL subjects’ inter-subject variability

We found a high inter-subject variability among BVLsubjects for all the parameters analyzed, which agreeswith Wndings from many other studies (Hufschmidtet al. 1980; Black et al. 1988; Gagey and Toupet 1991).Indeed, two of the nine subjects did not beneWt fromABF in the eyes closed condition. Some of this vari-ability may be explained in terms of how individualBVL subjects compensate for the vestibular loss, which

is by increasing reliance on either visual or somatosen-sory information (Zacharias and Young 1981; Lacouret al. 1997). If inter-subject variability depends on thedegree of visual or somatosensory dependency, wemay expect inter-subject variability to decrease whenvisual information and somatosensory information arelimited (the eyes closed on foam condition). Indeed,we found a consistent decrease in inter-subject vari-ability in this condition, when BVL subjects exhibitedrelatively smaller standard deviations (Table 2),although their sway was larger than in the eyes open onfoam and eyes closed conditions (Black et al. 1999).

Our BVL subjects showed signiWcantly higher fre-quency of postural corrections (F95%) than did ourcontrol subjects in all conditions tested. This result sug-gests that BVL subjects were using a diVerent mode ofcontrolling their balance than were the control subjects(Creath et al. 2005). However, without kinematic mea-sures, we cannot distinguish between ankle and hipsway strategies, as it was done by Creath et al. (2005).The higher frequency of postural corrections that theBVL subjects exhibited may also be related to thehigher sensory noise due to the vestibular loss thatBVL have compared to control subjects.

ABF redundancy with sensory information was higher for BVL than control subjects

In order to better highlight the diVerence in the use ofABF information between BVL and control subjects,we performed a meta-analysis which combined theresults from BVL and control subjects in all the condi-tion presented in this study in terms of sensory infor-mation redundancy using Venn diagrams. Redundancyof sensory information occurs when the same informa-tion is provided by more than one sensory channel.Sensory integration for balance is driven by—that is, isdependent on—redundancy of natural sensory infor-mation from somatosensory, visual, and vestibularchannels (Creath et al. 2002). Extensive redundancy ofsensory information provides persons with a betterestimate of body segment position and kinematics,which results in smaller postural sway (Kuo et al. 1998;van der Kooij et al. 2001).

To quantify sensory redundancy among the naturalsensory information and ABF, we averaged the swayreduction occurred in the conditions tested (when nat-ural and ABF sensory information was available) andrepresented these averages using Venn diagrams.Figure 5 shows two Venn diagrams (one for the BVLsubjects and one for the control subjects) that representthe contributions when all or some of the sensory infor-mation channels were contributing sensory information to

123

46 Exp Brain Res (2007) 178:37–48

control sway. The size of each diagram and theirpercentages represent the percent of COP sway reduc-tion occurred from a condition in which ABF, somato-sensory, and visual information are all limited (byturning oV the ABF device, by using foam, by closingthe eyes, respectively; i.e., the eyes closed on foamcondition without ABF) and a condition when onlyone of these information is available.

The redundancy between the ABF contribution inreducing sway and the contribution from each of theother sensory information was larger for BVL subjects(Fig. 5a) than for control subjects (Fig. 5b). For BVLsubjects, ABF reduced sway 46% (4, 11, and 31%)compared to 32% (12, 7, and 13%) for control subjects(each of the three percentages in parenthesis is theamount of redundancy between ABF information andvisual, somatosensory, and both visual and somatosen-sory, respectively). From these analyses, for BVL sub-jects, the redundancy among somatosensory, visual,and ABF information was higher (31%) than for con-trol subjects (13%). The greater redundancy in BVLthan control subjects suggests that compensating forvestibular loss depends on more extensive sensoryredundancy between visual and somatosensory infor-mation. Figure 5 shows that ABF information can alsobe redundant with visual and somatosensory sensoryinformation, suggesting that the CNS may treat ABFinformation similarly to natural sensory information.Also, since redundancy between ABF information andother sensory information is greater for BVL subjectsthan for control subjects, BVL subjects may beneWtmore from the ABF information than may control sub-jects, especially in sensory-deprived situations. In fact,with more practice, ABF information may also facili-tate a more accurate integration and calibration of

sensory information, induced by the CNS continuallycomparing natural sensory information to ABF infor-mation.

The use of foam to limit somatosensory informationmay have limited in the accuracy of sensory redun-dancy estimation. In fact, when determining the rolethat the somatosensory information plays in reducingsway (Fig. 5), we did not include all somatosensoryinformation that the CNS received from the entirebody but only the somatosensory information from thesubject’s feet which was restricted by using the foam.Even with these qualiWcations, Fig. 5 provides newinsight into the mechanisms of sensory redundancy andsensory re-weighing during human stance.

Conclusions

We found that the BVL and the control subjects usedABF information about their trunk acceleration tocontrol sway, in proportion to the extent that theirother sensory information was reduced. In addition, allsubjects used ABF diVerently, depending on their indi-vidual proclivities to rely on vestibular, somatosensory,or visual information in order to control sway. Redun-dancy between sensory information from diVerent sen-sory channels and ABF information was larger in BVLsubjects than in control subjects, suggesting that ABFinformation may help subjects compensate for vestibu-lar loss by facilitating the CNS’s integration of sensoryinformation.

Acknowledgments We would like to thank Dr. Conrad Walland Dr. Angelo Cappello for stimulating inspiration, Dr. SandraOster for English editing and education, Andrew Owings and Dr.Charles Russell for technical support, Triana Nagel-Nelson forrecruitment of subjects, and all of our subjects for donating theirtime. We also thank Dr. Velio Macellari and Dr. Daniele Gian-santi for having made available the portable measurement sys-tem. This study was supported by NIH grants DC01849,DC04082, and DC06201.

References

Barclay-Goddard R, Stevenson T, Poluha W, MoVatt M, TabackS (2004) Force platform feedback for standing balance train-ing after stroke. Cochrane Database Syst RevCD004129

Batavia M, Gianutsos JG, Vaccaro A, Gold JT (2001) A do-it-yourself membrane-activated auditory feedback device forweight bearing and gait training: a case report. Arch PhysMed Rehabil 82:541–545

Black FO, Nashner LM (1984) Vestibulo-spinal control diVers inpatients with reduced versus distorted vestibular function.Acta Otolaryngol Suppl 406:110–114

Black FO, Paloski WH, Reschke MF, Igarashi M, Guedry F,Anderson DJ (1999) Disruption of postural readaptation byinertial stimuli following space Xight. J Vestib Res 9:369–378

Fig. 5 a, b In the form of Venn diagrams the contributions ofsomatosensory (SOM yellow/lighter-colored circle), visual (VISblue/darker-colored circle), and (ABF orange/dark-gray diagram)information in reducing COP-RMS during quiet stance for bilat-eral vestibular loss and control subjects, respectively. Percentagesindicate the size of the diVerent areas and represent the COP-RMS reduction experienced by the subjects when that informa-tion was available. Overlapping areas represent redundancy ofinformation across the sensory systems

123

Exp Brain Res (2007) 178:37–48 47

Black FO, Shupert CL, Horak FB, Nashner LM (1988) Abnormalpostural control associated with peripheral vestibular disor-ders. Prog Brain Res 76:263–275

Buchanan JJ, Horak FB (2001) Vestibular loss disrupts control ofhead and trunk on a sinusoidally moving platform. J VestibRes 11:371–389

Chiari L, Dozza M, Cappello A, Horak FB, Macellari V, Gian-santi D (2005) Audio-biofeedback for balance improvement:an accelerometry-based system. IEEE Trans Biomed Eng52:2108–2111

Creath R, Kiemel T, Horak F, Jeka JJ (2002) Limited controlstrategies with the loss of vestibular function. Exp Brain Res145:323–333

Creath R, Kiemel T, Horak F, Peterka R, Jeka J (2005) A uniWedview of quiet and perturbed stance: simultaneous co-existingexcitable modes. Neurosci Lett 377:75–80

Dault MC, Frank JS (2004) Does practice modify the relationshipbetween postural control and the execution of a secondarytask in young and older individuals? Gerontology 50:157–164

Dickstein R, Shupert CL, Horak FB (2001) Fingertip touch im-proves postural stability in patients with peripheral neuropa-thy. Gait Posture 14:238–247

Dozza M, Chiari L, Chan B, Rocchi L, Horak FB, Cappello A(2005a) InXuence of a portable audio-biofeedback device onstructural properties of postural sway. J NeuroengineeringRehabil 2:13

Dozza M, Chiari L, Horak FB (2005b) Audio-biofeedback im-proves balance in patients with bilateral vestibular loss. ArchPhys Med Rehabil 86:1401–1403

Easton RD, Greene AJ, DiZio P, Lackner JR (1998) Auditorycues for orientation and postural control in sighted and con-genitally blind people. Exp Brain Res 118:541–550

Forbes TW (1946) Auditory signals for instrument Xying. J Aero-naut Soc 255–258

Gage WH, Winter DA, Frank JS, Adkin AL (2004) Kinematicand kinetic validity of the inverted pendulum model in quietstanding. Gait Posture 19:124–132

Gagey PM, Toupet M (1991) Orthostatic postural control in ves-tibular neuritis: a stabilometric analysis. Ann Otol RhinolLaryngol 100:971–975

Hegeman J, Honegger F, Kupper M, Allum JH (2005) The bal-ance control of bilateral peripheral vestibular loss subjectsand its improvement with auditory prosthetic feedback.J Vestib Res 15:109–117

Horak FB, Buchanan J, Creath R, Jeka J (2002) Vestibulospinalcontrol of posture. Adv Exp Med Biol 508:139–145

Horak FB, Hlavacka F (2001) Somatosensory loss increasesvestibulospinal sensitivity. J Neurophysiol 86:575–585

Horak FB, Macpherson JM (1996) Postural equilibrium andorientation. In: Rowell RB, Shepherd JT (eds) Publishedfor the American Physiology Society by Oxford UniversityPress, New York, pp 255–292

Hufschmidt A, Dichgans J, Mauritz KH, Hufschmidt M (1980)Some methods and parameters of body sway quantiWcationand their neurological applications. Arch Psychiatr Nerve-nkr 228:135–150

Jeka J, Oie KS, Kiemel T (2000) Multisensory information for hu-man postural control: integrating touch and vision. ExpBrain Res 134:107–125

Kentala E, Vivas J, Wall C III (2003) Reduction of postural swayby use of a vibrotactile balance prosthesis prototype in sub-jects with vestibular deWcits. Ann Otol Rhinol Laryngol112:404–409

Kluzik J, Horak FB, Peterka RJ (2005) DiVerences in preferredreference frames for postural orientation shown by after-

eVects of stance on an inclined surface. Exp Brain Res162:474–489

van der Kooij H, Jacobs R, Koopman B, van der HF (2001) Anadaptive model of sensory integration in a dynamic environ-ment applied to human stance control. Biol Cybern, pp. 103–115

Kuo AD, Speers RA, Peterka RJ, Horak FB (1998) EVect of al-tered sensory conditions on multivariate descriptors of hu-man postural sway. Exp Brain Res 122:185–195

Lackner JR (1974) The role of posture in sound localization. Q JExp Psychol 26:235–251

Lackner JR, DiZio PA (2000) Aspects of body self-calibration.Trends Cogn Sci 4:279–288

Lacour M, Barthelemy J, Borel L, Magnan J, Xerri C, Chays A,Ouaknine M (1997) Sensory strategies in human posturalcontrol before and after unilateral vestibular neurotomy.Exp Brain Res 115:300–310

Macpherson JM, Inglis JT (1993) Stance and balance followingbilateral labyrinthectomy. Prog Brain Res 97:219–228

Maurer C, Mergner T, Peterka RJ (2004) Abnormal resonancebehavior of the postural control loop in Parkinson’s disease.Exp Brain Res 157:369–376

Mayagoitia RE, Lotters JC, Veltink PH, Hermens H (2002)Standing balance evaluation using a triaxial accelerometer.Gait Posture 16:55–59

Nashner LM, Black FO, Wall C III (1982) Adaptation to alteredsupport and visual conditions during stance: patients withvestibular deWcits. J Neurosci 2:536–544

Nelson WT, Hettinger LJ, Cunningham JA, Brickman BJ, HaasMW, McKinley RL (1998) EVects of localized auditory infor-mation on visual target detection performance using a hel-met-mounted display. Hum Factors 40:452–460

Nichols DS (1997) Balance retraining after stroke using forceplatform biofeedback. Phys Ther 77:553–558

Paloski WH, Reschke MF, Black FO, Doxey DD, Harm DL(1992) Recovery of postural equilibrium control followingspaceXight. Ann N Y Acad Sci 656:747–754

Peterka RJ (2002) Sensorymotor integration in human posturalcontrol. J Neurophysiol 88:1097–1118

Peterka RJ, Black FO (1990) Age-related changes in human pos-ture control: sensory organization tests. J Vestib Res 1:73–85

Prieto TE, Myklebust JB, HoVmann RG, Lovett EG, MyklebustBM (1996) Measures of postural steadiness: diVerences be-tween healthy young and elderly adults. IEEE Trans BiomedEng 43:956–966

Redfern MS, Talkowski ME, Jennings JR, Furman JM (2004)Cognitive inXuences in postural control of patients with uni-lateral vestibular loss. Gait Posture 19:105–114

Rocchi L, Chiari L, Cappello A (2004) Feature selection ofstabilometric parameters based on principal componentanalysis. Med Biol Eng Comput 42:71–79

Schumacher EH, Seymour TL, Glass JM, Fencsik DE, LauberEJ, Kieras DE, Meyer DE (2001) Virtually perfect time shar-ing in dual-task performance: uncorking the central cognitivebottleneck. Psychol Sci 12:101–108

Shumway-Cook A, Horak FB (1990) Rehabilitation strategies forpatients with vestibular deWcits. Neurol Clin 8:441–457

Shumway-Cook A, Horak FB, Yardley L, Bronstein AM (1996)Rehabilitation of balance disorders in the patient with ves-tibular pathology. In: Bronstein AM, Brandt T, WoollacottMH (eds) Clinical disorders of balance, posture and gait.Arnold London, pp 213–220

Shumway-Cook A, Woollacott M, Kerns KA, Baldwin M (1997)The eVects of two types of cognitive tasks on postural stabil-ity in older adults with and without a history of falls. J Ger-ontol A Biol Sci Med Sci 52:M232–M240

123

48 Exp Brain Res (2007) 178:37–48

Tyler M, Danilov Y, Bach-y-Rita P (2003) Closing an open-loopcontrol system: vestibular substitution through the tongue.J Integr Neurosci 2:159–164

Vinge E. Human operator for aural compensatory tracking. Pro-ceedings of the 7th Annual Conference on Maual Control(NASA SP-281). 1971. U.S. Government Printing OYce.

Vuillerme N, Nafati G (2005) How attentional focus on bodysway aVects postural control during quiet standing. PsycholRes, pp. 1–9

Wall C III, Weinberg MS, Schmidt PB, Krebs DE (2001) Balanceprosthesis based on micromechanical sensors using vibrotac-tile feedback of tilt. IEEE Trans Biomed Eng 48:1153–1161

Wickens CD, Hollands JG (2000) Engineering Phychology andHuman Performance. New Jersey

Wong AM, Lee MY, Kuo JK, Tang FT (1997) The developmentand clinical evaluation of a standing biofeedback trainer.J Rehabil Res Dev 34:322–327

Woollacott MH (1993) Age-related changes in posture and move-ment. J Gerontol 48 Spec No: 56–60

Zacharias GL, Young LR (1981) InXuence of combined visualand vestibular cues on human perception and control ofhorizontal rotation. Exp Brain Res 41:159–171

123