Systematic distortions of auditory space perception following prolonged exposure to broadband noise

Systematic distortions of auditory space perceptionfollowing prolonged exposure to broadband noise

Simon Carlilea)

Auditory Neuroscience Laboratory, Department of Physiology, and Institute for Biomedical Research,University of Sydney, Sydney NSW 2006, Australia

Stephanie Hyams and Skye DelaneyDepartment of Physiology, University of Sydney, Sydney NSW 2006, Australia

~Received 10 August 1999; revised 16 October 2000; accepted 3 April 2001!

Perceptual distortions referred to as aftereffects may arise following exposure to an adapting sensorystimulus. The study of aftereffects has a long and distinguished history@Kohler and Wallach, Proc.Am. Philos. Soc.88, 269–359~1944!# and a range of aftereffects have been well described insensory modalities such as the visual system@Barlow, inVision: Coding and Efficiency~CambridgeUniversity Press, Cambridge, 1990!#. In the visual system these effects have been interpreted asevidence for a population of cells or channels specific for certain features of a stimulus. Howeverthere has been relatively little work examining auditory aftereffects, particularly in respect of spatiallocation. In this study we have examined the effects of a stationary adapting noise stimulus on thesubsequent auditory localization in the vicinity of the adapting stimulus. All human subjects in thisstudy were trained to localize short bursts of noise in a darkened anechoic environment. Adaptationwas achieved by presenting 4 min of continuous noise at the start of each block of trials and wasmaintained by a further 15-s noise burst between each trial. The adapting stimulus was located eitherdirectly in front of the subject or 30° to the right of the midline. Subjects were required to determinethe location of noise burst stimuli~150 ms! in the proximity of the adapting stimulus following eachinterstimulus period of adaptation. Results demonstrated that following adaptation there was ageneral radial displacement of perceived sound sourcesaway from the location of the adaptingstimulus. These data are more consistent with a channel-based or place-based process of soundlocalization rather than a simple level-based adaptation model. A simple ‘‘distribution shift’’ modelthat assumes an array of overlapping spatial channels is advanced to explain the psychophysicaldata. © 2001 Acoustical Society of America.@DOI: 10.1121/1.1375843#

PACS numbers: 43.66.Qp, 43.66.Pn, 43.66.Ba@DWG#

ionpndepoutoesuetiourtio

io

retat-

ctionhicn off theenceeu-

owst

enphe-ed

dis-mon-

ofel

alulusberderthe

u-ing

ma

I. INTRODUCTION

Sound localization depends on a variety of localizatcues present at each ear. Monaural cues arise from the stral modifications of incoming sound by the external ear abinaural cues arise as a result of differences in the sounthe two ears. At the level of the cochleae, the sensorythelia encode frequency and project into the auditory nervsystem as tonotopic maps of frequency. However, audilocalization depends on higher-order computational procing that extracts the monaural and binaural location cfrom this ordered frequency representation. The generaof a neural representation of space, and hence the accperception of auditory space, requires a complex integraof both monaural and binaural inputs~see Carlile, 1996a forreview!.

Physiological investigations of the neural representatof auditory space in the barn owl~Knudsen and Konishi,1978a, b; Knudsenet al., 1987! and the mammalian superiocolliculus ~e.g., guinea pig: King and Palmer, 1983; ferrKing and Hutchings, 1987; Carlile and King, 1994a, b; cMiddlebrooks and Knudsen, 1984! have demonstrated neu

a!Author to whom correspondence should be addressed; [email protected]

416 J. Acoust. Soc. Am. 110 (1), July 2001 0001-4966/2001

Downloaded 18 Mar 2013 to 152.3.102.242. Redistribution sub

ec-dati-s

rys-sn

aten

n

::

rones whose response rates vary systematically as a funof the spatial location of the sound source. The topograparrangement of these neurons is consistent with the notioa neural map of auditory space at least at these levels ocentral nervous system. Furthermore, there is some evidfor lateral interactions between these ‘‘space-mapped’’ nrons ~Knudsen and Konishi, 1978a, b; Kinget al., 1990!. Inthe study reported here we were interested in examining hthis kind of ‘‘space-selective’’ processing might be manifepsychophysically in human subjects.

One way in which psychophysical studies have beused to examine complex sensory processes is via thenomena of adaptation and aftereffects. Following prolongexposure to a constant stimulus, a variety of perceptualtortions have been described. Such effects have been destrated in early experiments examining the perceptionsound location using a range of adapting stimuli. Flug~1921! and Bartlett and Mark~1922! described an auditorylocalization aftereffect following exposure to both binaurand monaural sound sources. In these early studies, stimlevels were manipulated by adjusting the lengths of rubtubing that delivered the sound stimuli to each ear. Unsuch stimulus conditions, sounds were lateralized withinhead rather than localized in the free-field. The sound stimlus was adjusted by the subjects both prior to and followil:

/110(1)/416/9/$18.00 © 2001 Acoustical Society of America

ject to ASA license or copyright; see http://asadl.org/terms

too

laom

ueinteulme

arfre

tethlua.ttth

inr-t

udaao

dionpson

a.’’Jaertinaa

ecth

,e

ctrestntin

gni-pt-

ndheces

thetionvelb-owca-tral

hisliith

gester-ent

on

riorbe-entedheting

ddycal-

ofg

weden-

b-andhe

racy

k,ing

arysys-zi--ndhesys-f

sualon-

stimulation by ‘‘fatiguing’’ tones so that it was perceivedbe located in the center of the head. Following stimulationone ear by an adapting tone, there was a systematic dispment of subsequent binaural stimuli to locations away frthe ear to which the adapting tone was presented.

Lateralization experiments by Elfner and Perrott~1966!using headphone presentation of three different tone stim~i.e., 700, 1000, and 3000 Hz! demonstrated shifts in thauditory image following prolonged exposure to binauraltensity mismatches of up to 126 min. The results indicathat in most cases, the perceived location of the test stimwas shifted toward the ear receiving the less intense stilus. However, in some cases the shift was toward thewhere the adapting stimulus was more intense. These vtions might be explained by the mismatch between thequency of the adapting stimulus and the test stimulus.

Studies by Krauskopf~1954! and Taylor ~1962! con-firmed the initial observations by Flugel. Both these laexperiments were conducted using free-field stimuli. Infirst study, subjects adjusted the location of a sound stimuso that it appeared to lie on the median plane both beforeafter a 2-min exposure to an adapting broadband stimulusthe second experiment, subjects indicated the location oftest stimulus using a spatial scale. Prolonged exposure toadapting stimulus on one side of the median plane shiftedsubjects’ perception of the midline away from the adaptstimulus. Taylor~1962! measured the magnitude of this peceived displacement when the separation between theand adapting stimulus locations was varied. The magnitof the displacement increased as a function of the increingly lateral location of the adapting stimulus and peaked3° when the adapting stimulus was located at around 30°the midline. Both free-field and headphone-based stuyield qualitatively similar results. This supports the notithat the headphone-based studies are examining similarcesses as those measured in free-field studies, despitedifferences in the perceptual experience under each cotion. The results of Curthoys~1968!, Elfner and Perrott~1966!, and Thurlow and Jack~1973! all confirm the earlierfindings of a displacement of perceived target location awfrom the adapting stimulus: a so-called ‘‘repulsion effectIn a series of headphone-based studies, Thurlow and~1973! demonstrated that offsets of interaural time and intaural level differences were equally successful in generathis aftereffect, although the results were less straightforwwhen interaural time and level difference were traded offtest and adapting stimuli, respectively.

Comparison of the absolute magnitude of the aftereffacross previous studies is complicated by the fact thatduration of the adapting stimuli varied considerably~from 10s to a few minutes!, the acoustic environments varied~head-phone, free-field, anechoic! and in some free-field studiesfree head movement was allowed. However, despite thdifferences there is a general finding of a ‘‘repulsion’’ effeCurthoys proposed that the auditory spatial aftereffectsulted from a level-dependent adaptation of the ear closethe adapting stimulus that was greater than the adaptatiothe further ear. He suggested that this unequal adaptawould lead to a shift of subsequently perceived locatio

J. Acoust. Soc. Am., Vol. 110, No. 1, July 2001


fce-

li

-dusu-aria--

res

ndInhehee

g

estes-tff

es

ro-me

di-

y

ck-g

rds

te

se.-toin

ons

away from the more adapted ear. The changes in the matude of the effect with increasing displacement of the adaing stimulus from the midline observed by both Taylor aCurthoys is in fact consistent with what is known about tlocation-dependent changes in interaural level differen~Carlile and Pralong, 1994! and the resulting difference inthe adaptation of the two ears. This overall explanation ofauditory spatial aftereffect is dependent on the assumpthat the sensitivity of the auditory system to interaural ledifferences is sufficiently dominant to explain the effects oserved. However, such an explanation only explores hadaptation to ILD cues affects perception of the target lotion and ignores the potential roles of the ITD and speccues. Indeed, the work of Thurlow and Jack~1973! also in-dicate a role for interaural time cues in the generation of taftereffect. In addition, in experiments using VAS stimuwhere the interaural time cues have been set in conflict wother localization cues, there is some evidence to sugthat perception of spatial location is dominated by the intaural time difference cue when low frequencies are pres~Wightman and Kistler, 1992!.

However, a number of predictions can be made basedthe adaptation model discussed previously~Curthoys, 1968!.If the adapting stimulus is located to one side of the antemidline, all the target positions, whether they be locatedtween the adapting stimulus location and midline or betwethe adapting stimulus and interaural axis, would be predicto shift toward the anterior midline and therefore toward tlocation of the less adapted ear. Furthermore, an adapstimulus located directly in front of the subject~where inter-aural differences are zero! would adapt both ears equally anshould result in no auditory spatial aftereffect. In this stuwe examined these aftereffects using a head pointing loization task for stimuli presented in anechoic free space~Car-lile et al., 1997! where the subject had no preconceptionsthe potential locations of the stimuli. In addition to testinthe above-mentioned predictions, this approach also alloan examination of the adaptation aftereffect over the dimsions of azimuth and elevation.

II. METHODS

We examined the ability of a total of fifteen human sujects to localize a sound source or a visual target beforeafter exposure to a free-field adapting auditory stimulus. Tbehavioral methods used in assessing localization accuhave been described in detail previously~Carlile et al.,1997!. Briefly, all training and testing took place in a daranechoic chamber. Within the chamber, a robot arm carrya stimulus speaker and a light emitting diode~LED! could bemoved to almost any location on the surface of an imaginsphere surrounding the subject. A single-pole coordinatetem was employed to describe stimulus position with amuth 0° and elevation 0°~0°/0°! directly ahead on the audiovisual horizon of the subject. Upward elevation arightwards azimuth increased positively. The position of thead was tracked using an electromagnetic positioningtem~Polhemus, Isotrak!. The receiver was mounted on top othe subject’s head using an adjustable head strap. A vireference system, placed directly in front of the subject c

417Carlile et al.: Auditory spatial aftereffect


n-nsataiza-ts.

i-esiondof

enandof

n-r

theng

FIG. 1. The localization performance of four represetative subjects prior to adaptation for the test locatioaround the azimuth 0°/elevation 0° location. These dhave been chosen to demonstrate the range of localtion biases seen in our population of fifteen subjecEach plot represents a segment of the frontal hemsphere with the lines of azimuth spaced at 30°, the linof elevation spaced at 15°, and the azimuth 0°/elevat0° location directly at the center. The small closecircles indicate the target locations and the centroidsthe localization estimates are indicated by the opsquares. The dashed line between the closed circlesopen squares indicates the magnitude and directionthe localization bias for each test location. Each cetroid is calculated from nine repeat localizations foeach target location, and the standard deviations ofdistributions are indicated by the ellipses surroundieach centroid.

ct

tht

poexsro

dengsacaa

a-

s

tafoev

thanda

zapctoo

re

-ch

po-b-re-ester

a

ri-bout

rject

ri-cale.cts

imu-

vehe

enng

sisting of red and green colored LEDs, worked in conjuntion with the head tracker and could be used to indicatethe subjects the position of his or her head relative tostimulus coordinate system. This system was used to aidsubject in placing his or her head in the calibrated startsition at the beginning of each trial. In a supplementaryperiment, the visual reference system was also used to athe subject in keeping his or her head stationary during plonged exposure to the adapting stimulus. Subjects unwent a period of training to assist them in reliably pointitoward the perceived target location with the nose. Onceisfied he or she was correctly indicating the perceived lotion of the target, the subject pressed a hand-held buttonthe position of his or her head was recorded~for details seeCarlile et al., 1997!.

Following the period of training, the baseline localiztion accuracy of each subject was determined~see Carlileet al., 1997!. Subjects were required to localize a 150-mnoise burst (200 Hz–14 kHz63 dB; 60 dB SPL! from 76 testpositions. Subjects completed 4 such tests prior to undering the adaptation experiments, and the localization permance of each subject was within the normal range prously reported by this laboratory~Carlile et al., 1997!.

Four adaptation experiments were carried out incourse of this study: two primary adaptation experimentstwo supplementary experiments, the details of which arescribed below. In all experiments the testing procedure wdivided into preadaptation and adaptation blocks of localition trials. Localization accuracy was determined over a stial area surrounding the adapting speaker: Twelve test lotions were arranged as two concentric rings centered onlocation of the adapting stimulus with a spherical radiusaround 6° and 17°, respectively. Stimuli were locatedeach ring with roughly equal circumferential spacing~see

418 J. Acoust. Soc. Am., Vol. 110, No. 1, July 2001


-oehe--

sist-r-

t--

nd

k-r-i-

ed

e-s-

a-a-hefn

Fig. 1!. In two separate experiments, adapting stimuli welocated directly in front at 0°/0°~test positions: 0°/215°;215°/28°; 15°/28°; 0°/25°; 25°/22°; 5°/22°; 0°/0°;25°/2°; 5°/2°; 0°/5°;215°/8°; 15°/8°; 0°/15°! and 30° to the rightof the midline~test positions: 30°/215°; 15°/28°; 45°/28°;30°/25°; 25°/22°; 35°/22°; 30°/0°; 25°/2°; 35°/2°; 30°/5°;15°/8°; 45°/8°; 30°/15°!. The location of the adapting stimulus was also included in the set of test locations. In eablock, stimuli were presented 3 times at each of the 13sitions giving a total of 39 localization trials per block. Sujects completed 3 blocks for each test condition. Thissulted in 9 repeat localization trials for each location and tcondition, for a total of 117 localization trials per subject ptest condition. Localization judgments were plotted usingspherical plotting and analysis package~SPAK: Leong andCarlile, 1998!. We chose to illustrate these data using sphecal means as this avoids the need to make assumptions athe type of coordinate system to employ~e.g., a single ordouble pole projections for azimuth and elevation!. Themean~or centroid! of the cluster of localization estimates foeach test location was calculated and plotted for each sub~see Fig. 1! along with the standard deviation of each distbution. Further methodological, graphical, and statistianalyses are described in the following where appropriat

In the two primary adaptation experiments, ten subjewere tested using the adaptation stimulus directly ahead~0°/0°! and eight subjects were tested using the adaptation stlus located at 30° to the right of the subject~30°/0°!. Bothpositions were located on the audio-visual horizon and fiof the subjects were common to both experiments. Tadapting stimulus was a continuous broadband noise~similarto the target stimuli!, initially presented for 4 min at thebeginning of each block of trials and then for 15 s betweeach localization trial. During exposure to the adapti

Carlile et al.: Auditory spatial aftereffect


p

le

taoioimhoolu

tsec

eatiorsge

on

zer

llre

wjeprth

inin

lt oomh

rdthhetaa

hifon

ntars

ad-re-

ctly

nsepta-ataere

lcu-avetedter-

tes

ur-theon.pt-urecu-apt-

ta-estu-thatpt-

, a

lustop-on-ker

f

stimulus, the subject was requested to remain in a fixedsition facing directly ahead~0°/0°! during which time a LEDat 0°/0° was illuminated to assist in this task. In the suppmentary adaptation experiments~see the following!, the headposition indicator was also used to help the subject mainhead position for the duration of the adapting stimulus. Flowing exposure to the adapting stimulus, the localizattest was repeated with the exception that the adapting stlus was presented for 15 s between each test stimulus. Tsubjects localized 39 positions per block and each block tbetween 20 and 25 min of testing. The order of the stimupositions was randomized within each block. There wasleast a 24-h break between each block of adaptation tesminimize the effects of any prolonged adaptation aftereff~see also Curthoys, 1968!.

III. RESULTS

A. Principal adaptation experiments

Prior to any adaptation, each subject’s localization pformance was measured using the 13 test locations. Esubject displayed some localization bias in the preadaptatests ~Fig. 1! although the overall magnitude of the errowas within the range of localization errors seen in a larpopulation of subjects~see Carlileet al., 1997!. To illustratethe range of subject-related localization bias, example ctrol localization results are plotted for four subjects for soulocations around 0° azimuth, 0° elevation~Fig. 1!. For onesubject illustrated, most targets were consistently localias slightly too high@Fig. 1~a!# whereas for another, particulalocations were systematically localized as too [email protected]., Fig.1~b!#. In a third subject the localizations were systematicatoo right-ward@Fig. 1~d!# whereas in a fourth, the errors wemore randomly directed@Fig. 1~c!#.

For the purpose of plotting the adaptation resultswere concerned that pooling localization data across subwith different systematic biases would obscure changesduced by the experimental manipulation. Figure 2 plotscentroid of the control responses~open squares! for the sub-ject shown in Fig. 1~b! indicating a general downward biasthis subject’s preadaptation localization response. Followpreadaptation testing, the subjects were then exposed toadapting stimulus as described previously and his or hercalization performance remeasured. The principal effecthe adaptation was to displace the localization away fradapting speaker location as can be seen by the radial sin the adaptation responses~open circles! compared to thecontrol responses~open squares! ~Fig. 2!. However, in caseswhere the underlying control localization bias was towathe location of the adapting speaker then the extent ofeffect would be underestimated if plotted in terms of tactual location of the target. In addition, combining daacross subjects with the opposite directional biases wouldto cancel any effect due to adaptation. To correct for tbias, all the subjects’ responses were normalized in thelowing way. The centroids for both the preadaptation aadaptation responses were shifted by an identical amouremove the bias in the preadaptation response at each tposition. This process is equivalent to subtracting the bia



o-

-

inl-nu-us,ksattot

r-chn

r

n-d

d

y

ectso-e

gtheo-f

ifts

is

ctsl-dtogetin

the preadaptation response from both preadaptation andaptation response centroids at each target location, andsults in the corrected preadaptation centroids falling exaon their respective target locations@cf. Figs. 2 and 3~b!#.Thus, the deviation of the centroid of the adapted respofrom the actual location represents the effects of the adation with the underlying systematic bias removed. These dwere calculated for each subject, the shifted responses wthen pooled and variations across the population were calated for each test location. In Fig. 3 the adaptation data hbeen plotted for the same four individual subjects illustrain Fig. 1 to indicate again the range of the adaptation afeffect.

The centroids of the normalized localization estimaare plotted for the adapting speaker located at 0°/0°~Fig. 4!pooled from all ten subjects tested at this location. The srounding ellipses represent the standard deviation ofpooled normalized localization estimates for each locatiThe perception of spatial location in the vicinity of the adaing source was systematically distorted following exposto the adapting stimulus. We have quantified this by callating the spherical angle between the location of the ading stimulus ~at the center of the array! and each of thecentroids of the location judgments for both the preadaption and adaptation conditions. With the exception of the tlocation corresponding to the location of the adapting stimlus the differences between these two angles indicatedthere was a shift in perceived location away from the adaing stimulus following adaptation~mean 2.6°65.4°; mean6s.d.!.

When the adapting stimulus was located at 30°/0°similar pattern of distortion was evident~Fig. 5!. Stimuluslocations between the midline and the adapting stimuwere localized toward the anterior midline, and locationsthe right of the adapting stimulus were shifted mainly uward or downward. The average extent of the adaptatiinduced shifts for the 30°/0° location of the adapting spea

FIG. 2. The mean localization estimates for preadaptation~open squares!and adaptation~open circles! conditions for one subject. The distributions othe estimates have been omitted for clarity@see Fig. 1~b!#. All other detailsare as for Fig. 1.



me2,al-tes

d-in

FIG. 3. The normalized adapted responses for the safour subjects as in Fig. 1. As demonstrated in Fig.these data have been normalized by shifting each locization estimate so that the preadaptation estimaoverlie the actual target location~see the text for de-tails!. This procedure removes the individualized preaaptation localization bias. All other details are asFig. 1.

folu

niree-tinah°/ly-

the-of

ctsof

ub-at

ead

suac

tioth

ightachom-ridianthe

was 3.7°67.9. Compared to 0°/0°, the scatter of the dataany one test location was greater for the adapting stimulocation of 30°/0°.

To test if there was a significant radial displacemeproduced by exposure to the adapting stimulus, a pat-test compared~a! the spherical angle between the pradapted localization centroid and the location of the adapspeaker with~b! the spherical angle of the adapted localiztion centroid and the location of the adapting speaker. Twas carried out for both the adapting speaker locations 0and 30°/0°~Table I! and in both cases there was a high

FIG. 4. The mean normalized adaptation localization responses for tenjects for an adapting stimulus location of azimuth 0°/elevation 0°. Epoint is the centroid of 90 localization judgments~small open circles!. Allother details are as per Fig. 1. Note that, with the exception of the locaat azimuth 0°/elevation 0°, there is a clear radial shift in the location ofperceived test location.



rs

td

g-is0°

significant increase in the spherical angle as a result ofadapting stimulus (p,0.01). This indicates a significant radial shift in the perceived location away from the locationthe adapting stimulus.

B. Supplementary adaptation experiments

1. Supplementary experiment 1

In the two experiments reported previously the subjewere requested to keep facing forward during the courseexposure to the adapting stimulus. To aid this process sjects were provided with a visual reference point located0°/0°. However, we were concerned as to whether small h

b-h

ne

FIG. 5. The mean normalized adaptation localization responses for esubjects for an adapting stimulus location of azimuth 30°/elevation 0°. Esmall open circle indicates the centroid of 72 localization judgements. Cpared to the preceding figures, the plot has been shifted so that the meat 30° occupies the center of the plot with the midline represented toright. All other details are as per Fig. 1.



prlaa-

nsthth

sonidit

lite

itiotothwa

acin

enth

ititi

tioiv

fi-su

-a

nt

rteeeth

ina

l-a

Weveddis-ac-e

m-thede-

rib-

ctsheingut aa-hear-ual

ieduites.

fol-is-

vi-his

edbye-on-thew-ntsred

undednitydialpt-edtheer-

endch

onsth

ar-

thif

ea

/0°

l

movements may have occurred during the course of thelonged exposure to the initial 4-min adapting stimulus. Retive changes in the location of the adapting stimulus mhave the potential to ‘‘dilute’’ or reduce the locationdependent adaptive effect. To test this we devised a meausing the head tracking system to simultaneously monitorexact position of the head at 10 Hz during the course ofadapting stimulus presentation.

Using this apparatus we studied the head movementfour additional subjects during adaptation experiments cducted at 0°/0°. Three of these subjects were also provwith feedback at 10 Hz as to the location of the head wrespect to the stimulus coordinate system~see Sec. II: Whenthe head was properly aligned a central green LED wasotherwise, the direction of any misalignment was indicaby the appropriately placed red LED!. This monitoring indi-cated that there were continual small changes in the posof the head, particularly during the 4-min initial exposurethe adapting stimulus. The mean deviation recorded inone subject who did not receive head-position feedback4.5° with a peak deviation of 17.7°. By contrast, mean hedeviation for the three subjects who did receive feedbwas in general less than 1° with peak deviations rangfrom 2.0° to 2.4°.

To examine the potential effect of small head movemduring exposure to the adapting stimulus we comparedadaptation of the three subjects who received head posfeedback with the ten subjects tested at the same locawho did not. Those subjects who received head posifeedback demonstrated a mean displacement of the percelocations of the test stimuli by 3.7°. This shift was signicantly larger than the displacement observed in thosejects who received no head position feedback~2.6°; Studentst-test, p,0.01; see Table I!. These data indicate that feedback to the subject about relative head position playsimportant role in head stabilization under these conditioConsequently, the adaptation-induced displacement inperceived location observed in the two experiments repoabove might well be an underestimate of the adaptationfect. That is, the measured displacement may have blarger had head position information been provided tosubjects during adaptation.

2. Supplementary experiment 2

Two previous studies indicate the potential for a shiftthe perceptual frame of reference following exposure toadapting or concurrent auditory stimulus~Chandler, 1961! or

TABLE I. The spherical angle between the localization estimate andadapting speaker for preadapted and adapted conditions. The mean dences are shown for each experiment together with the statistic forcomparison using a paired-t. Each condition was significant atp,0.01.Both supplementary experiments employed an adapting stimulus at 0°

Expt 1: 0°/0° Expt 2: 30°/0°Sup 1:

Auditory Sup 2: Visua

Mean difference 2.6°65.4° 3.767.9° 3.766.3° 0.663.6°Paired-t stat. 39.3 40.87 22.4 18.9df 1169 1040 350 350



o--y

ofee

of-

edh

t;d

n

easdkg

te

ononned

b-

ns.hedf-ene

n

to some posturally related stimuli~Lackner, 1974!. For ex-ample Chandler~1961! demonstrated that a binaural imbaance in dichotically presented tonal stimuli could result inmisperception of the visual vertical by up to one degree.were interested to determine if the adaptation effect obserhere was the result of an auditory sensory adaptation, aturbance of spatial reference, or a reduction in motor capity to indicate a specific position in space. To identify thextent of the contribution of any nonauditory sensory coponent to the displacement of the perceived location ofauditory targets we carried out the same experimentscribed previously~supplementary 1! with the exception thatthe subjects were required to indicate the location of avisualtarget both before and after exposure to an adaptingauditorystimulus. Four subjects were used, two of whom had contuted data to the principal adaptation experiments.

As in the first supplementary experiment, these subjewere also provided with head position information during tcourse of the adapting stimulus. The auditory adaptstimulus was located at 0°/0° and each subject carried ototal of 117 visual localization trials over the 13 test loctions for the preadaptation condition and 117 trials for tadaptation condition. As before, the experiments were cried out in the darkened anaechoic chamber but the visstimulus was a 150-ms flash emitted from a red LED carrby the robot arm. As would be expected, subjects were qaccurate in indicating the position of the visual stimuluWhen pooled across the four subjects, visual localizationlowing auditory adaptation was found to be radially dplaced by a mean of 0.6° compared with the preadaptedsual localization. A paired t-test indicated that although tdifference was small, it was statistically significant (p,0.01) and of similar magnitude to the auditory-inducdisplacement of the visual vertical previously reportedChandler~1961!. These data indicate that a small displacment effect can be attributable to cross-modal or other nsensory spatial distortion. However, when compared tomagnitude of the displacement of the auditory targets folloing exposure to an adapting auditory stimulus this accoufor less than 16% of the observed auditory effect measuunder the same conditions.

IV. DISCUSSION

This study indicates that exposure to a broadband sofor a number of minutes results in a shift in the perceivlocation of subsequent sound sources located in the viciof the adapting stimuli. There was a general pattern of radisplacement of stimuli away from the location of the adaing stimuli. This was particularly marked for sounds locatdirectly ahead, and less so for sources 30° to the right ofmidline where there was a more pronounced shift in the pceived elevation of the test stimuli.

These findings are generally consistent with and extthe previous studies of auditory spatial aftereffects. Sustudies have shown perceptual effects for stimulus locatiabout the anterior midline following adaptation using bofree-field ~e.g., Taylor, 1962; Curthoys, 1968! andheadphone-based stimuli~e.g., Thurlow and Jack, 1973!. Ageneral finding was a shift of the perceived judgment of t

efer-ch

.



ane

-held

-teh

fio

-dtamte

0othasrd-Inseea0tethzideanpt

tais,ue

ene

heamlielda

;le

in

aceof

atednsi-eiro-ap

eu-cific

of

ingca-tialpicta-

isedialdtheen-

us-eu-tedmas a

ofna-

wnlusromory

eft

ightpula-tput

lt ofmu-ther

get stimuli placed around the midline as located further awfrom the adapting stimulus located off the median pla~Curthoys, 1968! or toward the middle of the head for thlateralized stimuli~Thurlow and Jack, 1973!. Thurlow andJack~1973! also report a small ‘‘repulsion’’ effect for a midline adapting stimulus using headphone-based stimuli. Tis consistent with the results reported here for the free-fiadapting stimulus at 0° azimuth. In two previous free-fiestudies~Taylor, 1962; Curthoys, 1968!, the peak adaptationinduced shift was 3° when the adapting stimulus was locaat 30° degrees off the midline. This is within the range of taverage shifts noted in this study~2.6°–3.7°!.

Previously such shifts had been explained in terms osimple level-based adaptation model resulting in adaptatinduced changes in apparent interaural level differences~inparticular see Curthoys, 1968!. That is, if the adapting stimulus resulted in greater adaptation in the near ear comparethe far ear, the effective ILDs caused by the subsequentget stimuli would be distorted to favor the ear further frothe adapting stimulus and the location of the perceivedstimulus would be shifted toward the further~less adapted!ear.

For location judgments made following adaptation atazimuth/0° elevation, the simple adaptation modelCurthoys predicts that there should be no differences inperceived location of target stimuli following adaptationthe level of the adapting stimulus was equal in both eaHowever, the data in this study demonstrated a clear rashift in perceived locations for all target stimuli in the vicinity of the adapting stimulus at azimuth 0°/elevation 0°.addition, the distortion of perceived location encompasboth dimensions of azimuth and elevation, an entirely nobservation. This upward shift in perceived location wmost pronounced for adapting stimuli located at azimuth 3elevation 0°, particularly in the case of the targets localateral of the adapting stimulus where the direction ofmean shift was rotated more vertically. Again, for the amuth 30°/elevation 0° location a simple adaptation mowould have predicted a straightforward shift toward theterior midline for all target locations surrounding the adaing speaker.

Clearly, mislocalization produced by differential adaption of each ear is insufficient to explain the pattern of mlocalization seen in the subjects in this study. In additionsimple level-based adaptation model fails to account for csuch as interaural time difference cues~ITDs! and spectralcues. Unambiguous localization is most likely to be depdent on the integration of all these different cue typ~Middlebrooks, 1992; see Carlile, 1996a, b for review!. Thisis consistent with the neurophysiological findings in tmammalian auditory system. In the deep layers of the mmalian superior colliculus, the map of auditory space reon neurones with relatively restricted spatial receptive fiewhich are shaped by the neurones selectivity for a smrange of binaural and spectral cues~Palmer and King, 1982King and Palmer, 1983; Carlile and Pettigrew, 1987; Middbrooks, 1987; Kinget al., 1990; Carlile and King, 1994a b!.These physiological data, together with similar findingsthe owl ~see, e.g., Knudsenet al., 1987; Olsenet al., 1989!,



ye

isld

de

an-

tor-

st

°fe

s.ial

dws°/de-l--

--as

-s

-ssll

-

indicate that these neural representations of auditory spare dependent on the integration of two or more typescues. Such neurons may provide the basis for segregprocessing of different regions of space. That is, their setivity to a limited range of convergent cues results in thbeing ‘‘tuned’’ to a specific location in space and the topgraphic organization of the neurons results in a neural mthat is isomorphic with two-dimensional auditory space.

In the visual system there are psychophysical and nrophysiological data which suggest the presence of spedetectors or channels which account for the perceptionparticular features of the visual field~see, e.g., Barlow,1990!. It is not unreasonable to postulate that neurons codfor discrete regions of auditory space might also be inditive of some form of channel processing of auditory spaperception. In other words, in the absence of a receptotorepresentation of auditory space, it is likely that a computionally based representation of auditory space is comprof a large number of neurons coding for different spatlocations~see Knudsenet al., 1987!. In such a scheme anparticularly where the receptive fields are relatively large,perception of auditory space might be dependent on thesemble output of this neural array. Such a concept is illtrated in Fig. 6 where the topographic arrangement of nrones with spatially restricted receptive fields is represenby a mosaic of overlapping receptive fields. Borrowing frothe visual literature, each neurone can be conceived ofseparate processing channel ‘‘tuned’’ to a particular areaspace. Such a model may also provide a plausible explation of the effects of auditory spatial adaptation. The doregulation of an individual channel by a prolonged stimuexposure would change the resulting balance of outputs fthe remaining channels, which in turn results in the sensaftereffect observed~see Fig. 6; Mather, 1980; Wade, 1994!.

FIG. 6. Simple model of channel processing in auditory localization. Lpanel: Array of overlapping channels with a zone of excitation~light grayoverlying circle! and zone of previous adaptation~gray underlying circle!produced by different stimuli centered on particular spatial channels. Rupper panel: Ensemble output in the unadapted state with the mean potion output indicated by the solid arrow. Right lower panel: Ensemble ouwhere a subpopulation of channels has been adapted~bottom!. Note that thelocus of mean activity indicated by the solid arrow has shifted as a resuthe down regulation of a subpopulation of channels by the adapting stilus. Such a model also provides a powerful heuristic for a range of furauditory localization experiments.



neffeuee

nia

isvcam

D.burre

feI

enthhioaboesiohoaio

-presinns

oel

aypleake

stir

e

rew

athng

pli-

e in

alons.in-tionnalcificp-

ofion

un-ncecs,

c

al-

itory

litory

-,’’ J.

fJ.

n

nal

.

dAm.

c.

l.

Thus the down regulation of the auditory spatial chancentered on the adapting stimulus could subsequently athe balance of outputs from the channel array for subseqadjacent locations and result in the outward shift in the pception of nearby locations.

The assumptions of the model are straightforward abiologically plausible: namely, that localization processingmediated through a large mosaic of overlapping spatial chnels ~receptive fields! and that the perception of locationderived from the output of the array. As discussed preously, examples of neurones that could form the biologisubstrate of such a system have been found in the mamlian and avian superior colliculus and in the avian MLSuch models have previously been referred to as ‘‘distrition shift’’ models and provide a powerful framework foexamining the characteristics of individual channels orceptive fields~see in particular Mather, 1980!. The psycho-physical data demonstrating the auditory spatial afterefare consistent with this kind of stimulus cue processing.this model, the down regulation of the spatial channel ctered on the adapting stimulus would subsequently affectbalance of outputs from the channel array and result in a sin the perception of nearby locations away from the locatof the adapting stimulus. From the current data we are unto determine if the adaptation is likely to occur at the pointhigh-level convergence of the various localization cuwithin the underlying processing streams, or a combinatof the two. If the adaptation was occurring at the level of tprocessing of individual localization cues, it would also beconsiderable interest to determine if each cue was equadapted or if there was some form of differential adaptatacross ITDs, ILDs, and spectral cues.

Such ‘‘distribution’’ shift models of auditory spatial aftereffects also provide a consistent basis for interpretingvious studies of auditory spatial aftereffects. It is of interthat this conception is also related to the early ideas behthe Kohler–Wallach theory of figural aftereffects whetranslated into a place code of auditory localization proceing ~see in particular Krauskopf, 1954! and are also related tthe place models developed to explain localization ‘‘funning’’ ~see Thurlowet al., 1965!.

Such a ‘‘channel’’ model of localization processing malso be useful in interpreting experiments using multiconcurrent stimuli. In two recent brief reports, the localiztion of speech sounds with concurrent distracters or mashave been described~Hawley et al., 1999; Hartung andBraasch, 1999!. In the case where noise has been used aconcurrent distracter, a displacement of the apparent locaof the target speech away from the distracter has beenported ~Hartung and Braasch, 1999!. By contrast, whenspeech has been used as a distracter, a low incidencmislocalization toward the distracter is reported~Hawleyet al., 1999!. On the other hand, where nonspeech concursounds were used as targets and distracters the effectsreported as being highly subject dependent~Wightman andKistler, 1997!. Of interest, the latter short report notes thone consistent effect of the concurrent distracter was ondisruption of the elevation component of the localizatioUnfortunately, the significant differences in the methodolo



lctnt

r-

dsn-

i-la-

-

-

ctn-e

iftnlef,nefllyn

e-td

s-

-

-rs

aone-

of

ntere

te

.y

between these previous studies and the current study comcate any test of the proposed ‘‘channel’’ model.

Studies of the neural representation of auditory spacthe owl ~Knudsen and Konishi, 1978a, b! and ferret~Kinget al., 1990! have also provided some neurophysiologicevidence of lateral interactions between space-tuned neurThis begs the question of the potential effects of lateralteractions between the proposed hypothetical localizachannels. Such interactions could have important functioconsequences by enhancing the contrast between spespatial auditory locations. Contrast in this context might aply to separating out foreground sounds of interest~e.g.,speech! from background noise. This kind of streaminginformation has also been shown to be reliant on localizatprocessing~Plomp and Mimpen, 1981!.

ACKNOWLEDGMENTS

This work was supported by Australian Research Cocil Grant Nos. A79905421 and A49700049. The assistaof Anna Corderoy and Craig Jin with some of the statistidata collection, and figures is warmly acknowledged.

Barlow, H. B. ~1990!. ‘‘A theory about the functional role and synaptimechanism of visual aftereffects,’’ inVision: Coding and Efficiency,edited by C. Blakemore~Cambridge University Press, Cambridge!,pp. 363–375.

Bartlett, F. C., and Mark, H.~1922!. ‘‘A note on local fatigue in the auditorysystem,’’ Br. J. Psych.11, 105–134.

Carlile, S.~1996a!. ‘‘Auditory space,’’ in Virtual Auditory Space: Genera-tion and Applications, edited by S. Carlile~Austin, TX!, Chap. 1.

Carlile, S.~1996b!. ‘‘The physical and psychophysical basis of sound locization,’’ in Virtual Auditory Space: Generation and Applications, editedby S. Carlile~Austin, Texas!, Chap. 2.

Carlile, S., and King, A. J.~1994a!. ‘‘The integration of monaural andbinaural cues in the generation of a neural representation of audspace,’’ Proc. Aust. Neurosci. Soc.5, 31.

Carlile, S., and King, A. J.~1994b!. ‘‘Monaural and binaural spectrum levecues in the ferret: Acoustics and the neural representation of audspace,’’ J. Neurophysiol.71, 785–801.

Carlile, S., Leong, P., and Hyams, S.~1997!. ‘‘The nature and distribution oferrors in the localization of sounds by humans,’’ Hear. Res.114, 179–196.

Carlile, S., and Pettigrew, A. G.~1987!. ‘‘Distribution of frequency sensi-tivity in the superior colliculus of the guinea pig,’’ Hear. Res.31, 123–136.

Carlile, S., and Pralong, D.~1994!. ‘‘The location-dependent nature of perceptually salient features of the human head-related transfer functionAcoust. Soc. Am.95, 3445–3459.

Chandler, K. A. ~1961!. ‘‘The effect of monaural and binaural tones odifferent intensities on the visual perception of verticality,’’ Am.Psychol.74, 260–265.

Curthoys, I. S.~1968!. ‘‘The effect of binaural adaptation and masking oauditory localisation,’’ Ph.D. thesis, Monash University~unpublished!.

Elfner, L., and Perrott, D. R.~1966!. ‘‘Effect of prolonged exposure to abinaural intensity mismatch on the locus of a dichotically produced toimage,’’ J. Acoust. Soc. Am.39, 716–719.

Flugel, J. C.~1921!. ‘‘On local fatigue in the auditory system,’’ Br. J. Psych11, 105–134.

Hartung, K., and Braasch, J.~1999!. ‘‘Localization of distracted speech annoise in reverberant and anechoic environments,’’ J. Acoust. Soc.105, 1149.

Hawley, M. L., Litovsky, R. Y., and Colburn, H. S.~1999!. ‘‘Speech intel-ligibility and localization in a multi-source environment,’’ J. Acoust. SoAm. 105, 3436–3448.

King, A. J., Carlile, S., and Chevassut, T. J. T.~1990!. ‘‘Spatial organisationof auditory receptive fields in the ferret superior colliculus,’’ Br. J. Audio24, 198.



ft:

-se

f

l.

’

-

ft

eJ

o

’ J.

e

eption

n

st.

m.

st.

King, A. J., and Hutchings, M. E.~1987!. ‘‘Spatial response properties oacoustically responsive neurons in the superior colliculus of the ferremap of auditory space,’’ J. Neurophysiol.57, 596–624.

King, A. J., and Palmer, A. R.~1983!. ‘‘Cells responsive to free-field auditory stimuli in guinea pig superior colliculus: Distribution and responproperties,’’ J. Physiol.~London! 342, 361–381.

Knudsen, E. I., and Konishi, M.~1978a!. ‘‘Center-surround organization oauditory receptive fields in the owl,’’ Science202, 778–780.

Knudsen, E. I., and Konishi, M.~1978b!. ‘‘A neural map of auditory spacein the owl,’’ Science200, 795–797.

Knudsen, E. I., Lac, S., and Easterly, S. D.~1987!. ‘‘Computational maps inthe brain,’’ Annu. Rev. Neurosci.10, 41–65.

Krauskopf, J.~1954!. ‘‘Figural aftereffects in auditory space,’’ Am. Psycho67, 278–287.

Lackner, J. R.~1974!. ‘‘Changes in auditory localisation during body tilt,’Acta Oto-Laryngol.77, 19–28.

Leong, P. H. W., and Carlile, S.~1998!. ‘‘Methods for spherical data analysis and visualisation,’’ J. Neurosci. Methods80, 191–200.

Mather, G. ~1980!. ‘‘The movement aftereffect and a distribution-shimodel for coding the direction of visual motion,’’ Perception9, 379–392.

Middlebrooks, J. C.~1987!. ‘‘Binaural mechanisms of spatial tuning in thcat’s superior colliculus distinguished using monaural occlusion,’’Neurophysiol.57, 688–701.

Middlebrooks, J. C.~1992!. ‘‘Narrow-band sound localization related texternal ear acoustics,’’ J. Acoust. Soc. Am.92, 2607–2624.



A

.

Middlebrooks, J. C., and Knudsen, E. I.~1984!. ‘‘A neural code for auditoryspace in the cat’s superior colliculus,’’ J. Neurosci.4, 2621–2634.

Olsen, J. F., Knudsen, E. I., and Easterly, S. D.~1989!. ‘‘Neural maps ofinteraural time and intensity differences in the optic tectum of the owl,’Neurosci.9, 2591–2605.

Palmer, A. R., and King, A. J.~1982!. ‘‘The representation of auditory spacin the mammalian superior colliculus,’’ Nature~London! 299, 248–249.

Plomp, R., and Mimpen, A. M.~1981!. ‘‘Effects of the orientation of thespeakers head and the azimuth of a noise source on the speech recthreshold for sentences,’’ Acustica48, 325–328.

Taylor, M. M. ~1962!. ‘‘The distance paradox of the figural aftereffect iauditory localization,’’ Can. J. Phys.16, 278–282.

Thurlow, W. R., and Jack, C. E.~1973!. ‘‘Some determinants oflocalization-adaptation effects for successive auditory stimuli,’’ J. AcouSoc. Am.53, 1573–1577.

Thurlow, W. R., Marten, A. E., and Bhatt, B. J.~1965!. ‘‘Localizationaftereffect with pulse-tone and pulse-pulse stimuli,’’ J. Acoust. Soc. A37, 837–842.

Wade, N. J.~1994!. ‘‘A selective history of the study of visual motionaftereffects,’’ Perception23, 1111–1134.

Wightman, F. L., and Kistler, D. J.~1992!. ‘‘The dominant role of low-frequency interaural time-differences in sound localization,’’ J. AcouSoc. Am.91, 1648–1661.

Wightman, F. L., and Kistler, D. J.~1997!. ‘‘Sound localization in the pres-ence of multiple distracters,’’ J. Acoust. Soc. Am.101, 3105.



Documents

Systematic distortions of auditory space perception following prolonged exposure to broadband noise