Head and Eye Movements Reveal Compensatory Strategies for Acute Binaural Deficits During Sound Localization

Abstract

The present study aimed to define use of head and eye movements during sound localization in children and adults to: (1) assess effects of stationary versus moving sound and (2) define effects of binaural cues degraded through acute monaural ear plugging. Thirty-three youth (M_Age= 12.9 years) and seventeen adults (M_Age= 24.6 years) with typical hearing were recruited and asked to localize white noise anywhere within a horizontal arc from −60° (left) to +60° (right) azimuth in two conditions (typical binaural and right ear plugged). In each trial, sound was presented at an initial stationary position (L1) and then while moving at ∼4°/s until reaching a second position (L2). Sound moved in five conditions (±40°, ±20°, or 0°). Participants adjusted a laser pointer to indicate L1 and L2 positions. Unrestricted head and eye movements were collected with gyroscopic sensors on the head and eye-tracking glasses, respectively. Results confirmed that accurate sound localization of both stationary and moving sound is disrupted by acute monaural ear plugging. Eye movements preceded head movements for sound localization in normal binaural listening and head movements were larger than eye movements during monaural plugging. Head movements favored the unplugged left ear when stationary sounds were presented in the right hemifield and during sound motion in both hemifields regardless of the movement direction. Disrupted binaural cues have greater effects on localization of moving than stationary sound. Head movements reveal preferential use of the better-hearing ear and relatively stable eye positions likely reflect normal vestibular-ocular reflexes.

Introduction

The present study aimed to define the use of head and eye movements during sound localization in children and adults to: (1) assess effects of stationary versus moving sound on spatial hearing and (2) effects of degrading binaural cues through acute monaural ear plugging.

The binaural auditory system allows accurate spatial localization of sounds in the horizontal plane (Grothe & Pecka, 2014). This skill is already present in early life (Eklöf et al., 2022); indeed, infants look toward familiar voices and, as they begin to sit upright and gain head and neck stability, they can turn their head and eyes to a sound source. Latency of gaze toward sound location (Eklöf et al., 2022) decreases in the first years of life, indicating improving spatial hearing abilities, and sound localization amongst a closed set of loudspeakers/sound source positions is already highly accurate in young children (Grieco-Calub & Litovsky, 2012). Early spatial hearing can be compromised by hearing loss in childhood with significant developmental implications as shown by reduced language, academic and cognitive abilities in children with unilateral hearing loss (Arras et al., 2021; Bess et al., 1986; McSweeny et al., 2021; van Wieringen et al., 2019). Spatial hearing is difficult to restore in children with hearing loss through hearing devices such as hearing aids and cochlear implants (Bennett & Litovsky, 2020; Grieco-Calub & Litovsky, 2012; Killan et al., 2019). By contrast, individuals who lose their hearing during adulthood can regain sound localization. Adults who consent to receive a reversible conductive hearing loss through ear plugging, which introduces changes in interaural level differences (ILD) and interaural timing differences (ITD) cues (Lupo et al., 2011; Wightman & Kistler, 1997), show initially poor sound localization which improves through adaptation over time with training (Irving & Moore, 2011; Kumpik et al., 2010; Van Wanrooij & Van Opstal, 2004). Improved localization based on spectral cues (e.g., elevation) takes longer (weeks) (Hofman et al., 1998). Similarly, adults with post-lingual onset of deafness show improved sound localization with their cochlear implants (Isaiah et al., 2014) and can make good use of bimodal hearing (hearing aid in one ear and cochlear implant in the other) and bilateral cochlear implants when mismatched timing and level cues between the two devices are corrected (Angermeier et al., 2021; Potts et al., 2019).

Improved spatial hearing shown by adults with experimentally reversible or permanent hearing loss may be driven in part by their use of head and eye movements. In the vertical plane, pinna cues change with head position and then can be compared to assess whether a sound is occurring above or below (Kumpik & King, 2019) and sound localization in the horizontal plane is better when the head and eyes are allowed to move as compared to restrained conditions. Head movements help to distinguish between sounds coming from the front versus back of the head (Wightman & Kistler, 1989) and are “natural” behaviors involved during sound localization (Populin, 2006, 2008; Tollin et al., 2005). Common head movements during horizontal localization have been described as rotational or head-turning movements about a vertical axis and may also include forward–backward head tipping (Thurlow et al., 1967). In support, sound localization deteriorated in cats who were restricted to eye saccades through head restraint as compared to their abilities during unrestrained gaze shifts (movements of the head and eyes; Tollin et al., 2005). Similarly, rhesus monkeys showed decreased abilities to orient gaze toward sound locations when their heads were restrained compared to when their heads were freed (Populin, 2006).

Head movements may be used to create changes in binaural cues or signal-to-noise ratios at each ear which can reduce potential errors in sound localization (such as front-back confusions) or improve precision of sound localization which is best at the front of the head (Dobreva et al., 2012). They may also be used to visually scan the environment for confirmation of spatial position. Visual and auditory input are integrated in the superior colliculus (or optic tectum in birds; Brainard & Knudsen, 1993) and the superior colliculus is an important driver of sensory-guided gaze shifts (Zahler et al., 2021). Optical prisms in young barn owls disrupt this coordinated mapping of visual and auditory space, resulting in sound localization that is shifted to the direction of the visual field (Knudsen & Knudsen, 1990). This disruption has also been reported in humans; adults with amblyopia showed poor sound localization with the greatest error in the hemifield ipsilateral to the affected eye (Richards et al., 2017). These data are a reminder that spatial hearing in typical environments requires coordinated head and eye movements.

Typical environments also include moving sounds. Many studies focus on detection and sensitivity to moving sound using a just-noticeable difference between moving and stationary sound measured as movement detection thresholds (Phillips & Hall, 2001) or minimum audible movement angles (Harris & Sergeant, 1971). Only 2–5 degrees of movement are required to detect the presence of slow-moving sounds and larger degrees are needed when velocity is much faster (Grantham, 1986; Perrott & Musicant, 1977). Adults with normal hearing also accurately discriminate the direction of sound movement and can localize both the start and end point of the sound movement (Getzmann & Lewald, 2007; Moua et al., 2019). A more recent study assessing use of motion to detect and recognize one of three concurrent sound sources indicates that motion is first detected and then the direction and location of the moving sound are attended to (Cho & Kidd, 2022).

Head and eye motion appear to be natural responses to detecting and localizing moving sounds. Head movements accurately indicate localization of sound movement in both horizontal and vertical planes (Makous & Middlebrooks, 1990) and smooth pursuit eye movements can follow sound as it moves, revealing attention to the moving stimuli (Baumann & Greenlee, 2009; García-Uceda Calvo et al., 2021). Head motion helps us hear sounds coming from the external environment which does not occur similarly when listening under headphones (Brimijoin et al., 2013). These self-movements might also help comparisons between external sounds and self-position (Freeman et al., 2017). For example, sounds moving slowly in line with participant rotation are judged to be more still than sounds staying in one place (Freeman et al., 2017). Similarly, the minimum audible movement angle is smaller during self-motion compared to when the head remains stationary (Brimijoin & Akeroyd, 2014). These data indicate that the auditory world around us is perceived relative to our own spatial position and that there is an important role for head and eye movement for localizing moving sounds. What is not clear is whether head and eye movements during localization of moving sounds are disrupted by poor spatial hearing.

In the present study, we assess the role of head and eye movements on localization of both stationary and moving sounds. We test the hypotheses that unrestricted head and eye movements: (1) reflect accurate perception of stationary and moving sound location and (2) show a bias for monaural cues in the better hearing ear in the presence of monaural earplugging.

Methods

Participants

Fifty participants including 33 youth (M_Age = 12.9 years old and 0.95 95% CI) and 17 adults (M_Age = 24.6 years old and 5.56 95% CI) with typical hearing were recruited at The Hospital for Sick Children (SickKids) to participate in this research which was approved by the Research Ethics Board and adheres to the Tri-Council Policy on the Ethical Conduct for Research Involving Humans. Inclusion criteria were willingness to participate, normal hearing, and the ability to comprehend and perform task instructions. Hearing thresholds were screened for frequencies 250–8000 Hz at 25 dB HL in 44/50 (88%) participants. Testing could not be completed in the other six participants due to time constraints but they reported no hearing difficulties and no history of hearing loss.

Sound Localization Paradigm

Sound localization equipment was set up in a 2-m × 2-m × 2-m sound booth and consisted of a three-legged wooden frame supporting a 1-m long L-shaped arm which held a speaker fixed at the distal end 1.15 m from the floor. Previous studies of sound movement perception have employed multiple techniques to simulate sound movement such as varying level differences between signals presented by two speakers (Grantham, 1986); or vector-based amplitude panning over multiple speakers (Litovsky et al., 2019). In the present study, we used a more classic approach of moving the sound source along a horizontal arc. Harris and Sergeant (1971) used this approach by moving a loudspeaker on a trolley. Our setup similarly placed a small speaker on a moving speaker arm. The proximal end of the speaker arm was fixed to a silent-stepper motor (57BYGH420-2 Wantai Motor) allowing the speaker to move to any position within an arc of 120° in the horizontal plane from −60° (left) to +60° (right) azimuth.

Participants sat directly beneath the motor on a height-adjustable non-swivel stool. All equipment was visually hidden from participants by black acoustically transparent cloth. Participants were instructed to sit upright throughout testing and were told that they could move their head. White noise (125 to 6000 Hz) was calibrated to 65 dBA and roved in level by ±4 dB. The silent-stepper motor was active during entire blocks of stimulus presentation; it provided high sensitivity (1.8 steps/° per rotation) and was quiet (no measured change in stimulus presentation level with motor on versus off).

Stimuli were presented in 7–10 blocks which each consisted of six trials. This was repeated in two conditions (typical binaural and right ear plug). As shown in Figure 1(a), each trial had two components: (1) stimuli were delivered at an initial stationary position (L1) anywhere within ±60° for 3.0 s and then (2) the stimulus was delivered while moving until it reached a second position (L2) in five conditions (±40°, ±20°, or 0° movements). Moving sound was presented slowly at an average speed of 3.0°/s for 20° movements and 4.4°/s for 40° movements. Positions of both L1 and L2 were presented pseudo-randomly to ensure similar numbers of presentations in left and right hemifields and to ensure sound movement from L1-to-L2 remained in the same hemifield. As plotted in Figure 1(b), the duration of the moving sound varied from mean(SD) = 3.22(1.51) s to 9.03(0.17) s based on the position change from L1-to-L2 but was not dependent on the position of L1 [F_{(1, 15.8)}= 0.07, p = 0.79]. Participants were asked to locate L1 (“Where is the sound”) and L2 (“Where did the sound move to?”) with no visual cues. Participants indicated locations of the sound presentation at L1 and then after sound movement to L2 by moving a red laser dot projected onto the black curtain using a Logitech Gamepad F310 videogame controller (possible arc from −90° to +90°). The laser movement was controlled by a small silent-stepper motor (QS4218-51-10-049 Trinamic Motor).

Figure 1.

(a) Sound localization setup. The participant was seated in the center of the horizontal arc (−60° to +60° azimuth). Stationary sound presentation occurred at L1 for the first task (“Where is the sound”) and was then presented while moving to L2 for the second task (“Where did the sound move to?”). 7–10 blocks of 6 trials each were presented. (b) The duration of moving sound was randomized; duration was longer as L1-to-L2 degree change increased [mean duration by magnitude of movement: M₀= 3.2 s (1.5 SD), M₂₀= 6.7 s (0.4 SD), M₄₀= 9.0 s (0.16 SD)] but was not dependent on L1 position [F_{(1, 15.8)}= 0.07, p = 0.79].

Sound localization was measured in typical listening conditions (binaural) and under monaural (right) ear plugging. A TASCO™ SOFTSEAL33™ tapered polyurethane earplug (uncorded 200 PR; NRR of 33 dB) was inserted into the right ear and a AOSafety® Peltor® earmuff (estimated NRR of 30 dB) was placed overtop on the same ear.

Head and Eye Tracking During Sound Localization

The Pupil Labs Pupil Core eye tracking glasses and software were used to track pupil location overtime during the localization task (Kassner et al., 2014). Equipment calibration for each participant was done with the head position stabilized on a chin rest. The vertical and horizontal “gaze angles” were 13.21° and 22.39°, respectively. Participants were asked to look at dots presented at nine positions on a grid on a monitor (64.2 cm away). The eye positions and timestamps were synchronized with presentations. Pupil detection confidence (0–1) was monitored by the Pupil Labs software and data with confidence >0.7 were used for analyses.

Head movements were collected in the first 26 participants using the MbientLab MetaMotionR (MMR) motion tracker and in 25 participants using the EDTracker (ED) motion tracker (one participant was tested with both). Both devices use gyroscopic sensors to measure head movement in three degrees of freedom (Euler angles: yaw, pitch, and roll). The MMR had a sample rate of up to 800 Hz and the ED sampled data at 200 Hz. Data from 18/26 (69.2%) MMR users were excluded due to incomplete data capture and frequent artifacts. MMR artifacts were rapid changes in position unrelated to the participant. These artifacts were removed in eight participants using a combination of Euler corrections and linear detrend corrections in time windows constrained to each experimental block.

Analyses

Sound localization response accuracy for both L1 and L2 was assessed using linear mixed model regression which included speaker position (continuous variable) and condition (normal binaural, right ear plugged) as fixed factors and random intercepts and slopes by speaker position for each participant. Accuracy of detecting direction of sound movement was measured as the degree azimuth change from L1-to-L2 responses and assessed by linear mixed model regression using a fixed factor of change in degree azimuth of speaker position from L1-to-L2 (0°, ±20°, ±40°) and condition and random intercepts and slopes by degree of speaker position change for each participant. For linear mixed effects models, fits were computed using restricted maximum likelihood and estimates were computed using Satterthwaite's method. Direction perception was further measured by assigning a “0” when L2 responses were left of L1 responses and a “1” when L2 responses were right of L1 responses. The resulting binary variable was assessed using logistic regression models for each participant.

Head and eye tracking data were aligned in time with the onset of stimuli for each trial. Displacement waveforms were assessed from stimulus onset to response to L1 (stationary sound) and to L2 (moving sound). Displacement data were smoothed using a simple moving average window of variable window length depending on available data points by trial (windows ranged between 2 and 100 data points) and averaged across trials binned by speaker position. Responses to stationary sound were binned in 10° increments from −60° to +60° (12 positions) and responses to moving sound were binned by the five conditions of speaker movement (0°, ±20° and ±40°). Given 60 trials which each contain both stationary and moving sound presentations, five trials were averaged for stationary sounds at each of the 12 binned positions and 12 trials were averaged at each of the five sound movement conditions. Averaged displacement waveforms were assessed for: (1) peak latencies and amplitudes, (2) area under the curve of the waveform, and (3) path length (sum of the absolute values of the differentials of the waveform). Linear mixed model regression analyses were used to assess fixed effects of stimulus position or degree of change with random intercepts for each study participant.

Results

High Localization Accuracy Decreases with Monaural Plugging for Both Moving and Stationary Sounds

Responses to each stationary sound stimulus presentation at L1 are shown in Figure 2(a) along with the associated regression line for each participant. Localization of stationary sounds was highly accurate in the normal listening condition despite the absence of visual cues, as shown by regression lines falling close to the line of unity but was more variable during acute right ear plugging. The error across trials, shown in Figure 2(b), was low [M(SD) = 5.28°(3.06), max = 13.6°] in the normal binaural condition and similar in both left and right hemifields [t₍₄₉₎= 1.33, p = 0.19]. Error significantly increased in the plugged ear condition [F_{(1, 120)}= 13.6, p < 0.001], particularly in the right hemifield ipsilateral to the plugged ear which increased with age [F_{(1, 91.5)}= 8.86, p < 0.01]. Reaction time [M(SD) = 0.68 s(0.31), max = 1.40 s], shown in Figure 2(c), was similar between hemifields [t₍₄₉₎= 1.38, p = 0.17] in the normal binaural hearing condition and increased in the right ear-plugged condition [F_{(1, 81.7)}= 20.9, p < 0.001].

Responses to all moving sound presentation (from L1-to-L2) are shown in Figure 3(a). The left panel plots the five sound movement conditions against the degree of change in response positions for L1-to-L2. Error for the change in sound location was low in the normal binaural condition [M(SD) = 5.47(5.02)°] but increased [M(SD) = 23.2(13.2)°] with right ear plugging [F_(1,312.8)= 65.7, p < 0.001]. Moving sound perception was further assessed, as shown in Figure 3(b), by fitting a binomial logistic regression ( $E(Y_i)={\beta }_0+{\beta }_i$ ) of the proportion of responses in which the L2 response occurred to the right of L1 (rather than left) against the stimulus movement from L1-to-L2. The resulting curves for each participant (shown by individual lines) reveal high accuracy as measured by large model slopes [M(SD) = 1.02(0.40)] in the normal binaural condition which became more variable and shallower [t₍₁₅₎= 10.8, p < 0.001] during right ear plugging [M(SD) = 0.17(0.33)]. As shown in Figure 3(c), there was no effect of the hemifield of stimulus presentation on response error in either condition [F_(1,570.6)= 1.30, p = 0.25]. In Figure 3(d), mean response time (s) increased significantly during ear plugging [M(SD) = 0.87(0.36) s] compared to normal binaural listening [M(SD) = 0.70(0.33) s] [F_(1,85.8)= 45.9, p < 0.001].

Figure 2.

(a) Localization for all stationary sound presentations in the normal binaural (no-plug) listening condition (left) and in the right ear-plugged condition (right) are plotted with linear regression lines for each participant (n = 50). (b) RMSE (°) was similarly low in both left and right hemifields of stimulus presentation in the normal condition [M(SE) = 5.28°(0.43)] and increased significantly in the right ear-plugged condition (Estimate = 19.7, SE = 1.70, p < 0.001), particularly in the ipsilateral hemifield to the plug [F_(1,95.5)= 15.2, p < 0.001]. (c) Mean reaction time (±1SE) was similar in both hemifields in the binaural condition (Estimate = 0.03, SE = 0.03, p = 0.81) and increased in the right ear plugged condition (Estimate = 0.35, SE = 0.06, p < 0.001).

Head and Eye Movements During Normal Localization of Moving and Stationary Sounds

Head and eye movements were measured as displacement relative to the onset of sound presentation. Trials for stationary sound presentations were combined by stimulus position in 10° increments from −60° to −50° through to 50° to 60° resulting in displacement waveforms for 12 binned ranges for the head and eyes in each child as shown by the individual lines in Figure 4(a). Trials of moving sound for each participant are shown for each of the five position change conditions (0°, ±20°, ±40°) in Figure 4(b). Thicker waveform lines represent the group means. The location of the stationary sound presentation at L1 and of the final position of moving sound at L2 is shown by the blue horizontal lines. The vertical lines represent peaks of mean displacement at which both amplitude and latency were measured.

Figure 3.

(a) Response change from L1-to-L2 are plotted against the actual change in presentation locations from L1-to-L2 in the normal (left) and right ear-plugged (right) conditions. Changes for individual trials are shown by dots and box plots represent the distribution. (b) Proportion of responses in which L2 was right of L1 rather than left is plotted against the actual direction of moving stimulus presentation (° azimuth change L1–L2). Lines represent binomial logistic regression curves for each participant and symbols present the mean (±1 SE) proportion for each condition. (c) Response change RMSE (°) is compared to L1-to-L2 stimulus position change. (d) Mean reaction time (±1SE) increased for the right plug condition (Estimate = 0.27, SE = 0.04, p < 0.001) but was not statistically different between hemifields [F_(1,77.8)= 0.76, p = 0.39].

Peak latencies of head and eye displacements are shown in Figure 4(c) and (d) for individuals (dots) and across the group (mean ± 1SE = bars) data. Data in Figure 4(c) show that, in the normal binaural condition, peak eye displacement to stationary sound presentation at L1 was reached prior to peak head displacement [F_(1,1512)= 36.0, p < 0.001] and that both the head and eye displacements peaked at later times as L1 became closer to the maximum ±60° presentation locations [F_(11,1512)= 99.9, p < 0.001]. For moving sound to L2 (Figure 4(d)), peak eye displacement preceded head displacement [F_(1,608)= 52.6, p < 0.001] and peaks occurred at later times as the degree of sound movement change increased [F_(4,608)= 62.1, p < 0.001].

Amplitudes at displacement peaks are plotted in Figure 4(e) and (f) for both individual participants (dots) and across the group (M ± 1SE = bars). Negative values indicate maximum peaks in the left hemifield and positive values indicate maximum values in the right hemifield. Figure 4(e) shows that, in the normal binaural condition, both the eyes and the head moved in the correct direction and magnitude to L1 positions [F_(11,1479.9)= 65.9, p < 0.001]. Of note, peak eye displacement was largest to the stationary sounds presented at −40° to −50° (left hemifield between −50° and −60°: M = 19.7; right hemifield between 30° and 40°: M = 20.5) whereas head movements were largest at ±50°–60° (left hemifield between −50° and −60°: M = 23.2; right hemifield between 50° and 60°: M = 26.5). Similarly, Figure 4(f) shows correct direction and magnitude of eye and head displacements in the normal binaural condition to moving sound ending at L2 [F_(4,617)= 145.0, p < 0.001] with significant increases between maximum head and eye displacement at ±40° position change compared to ±20° position change (−20 compared to −40: Estimate = 6.6, SE = 1.3, p < 0.001; 20 compared to 40: Estimate = 9.3, SE = 1.3, p < 0.001).

Displacement was also assessed throughout the time of head and eye movement by measuring the area under the curve relative to each hemifield of activity. Individual (dots) and group (mean ± 1SE = bars) data are shown in Figure 4(g) and (h) for each hemifield for each stimulus condition (G:stationary, H:moving). Results are consistent with maximum amplitudes showing head and eye displacement in correct direction and magnitudes to the location of sound presentation in the normal binaural condition [stationary: F_(11,1403.8)= 61.0, p < 0.001; moving: F_(4,596.3)= 114.5, p < 0.001]. The area under the curve measure confirms that eye displacement to stationary sounds was maximal for −50° to −60° (M = 17.5) and 30° to 40° presentations (M = 19.2).

Figure 4(i) and (j) plots the pathlengths of head and eye movements for individual (dots) and group (M ± 1SE = bars) data. Pathlength represents the total distance traversed by each waveform and is calculated and plotted relative to each hemifield of activity for each condition. Data in Figure 4(i) shows that, in the normal binaural condition, there was considerable pathlength displacement of the eyes during stationary presentation of sound at L1 that occurred in both hemifields and which was greater than pathlength of head movements [F_(1,1356.3)= 173.4, p < 0.001]. Pathlength in response to moving sound to L2 (Figure 4(j)) was also larger for eye than head movements [F_(1,592.9)= 83.6, p < 0.001]. Pathlength in response to moving sound to L2 (Figure 4(j)) was also larger for eye than head movements [F_(1,592.9)= 83.6, p < 0.001].

Altered Head and Eye Movements During Right Ear Plugging

Measures of head and eye movements during right ear plugging are plotted in Figure 5 with left column plots showing responses to stationary sound presentation at L1 and right plots showing responses to sound moving from L1-to-L2. The data are in the same format as Figure 4. Waveforms shown in Figure 5(a) show that, during right ear plugging, the head moved very little to stationary sound presentations in left hemifield conditions with large right displacement in right hemifield conditions. Eye displacement was also relatively small. As shown in Figure 5(b), rightward head displacement occurred in all moving conditions with relatively small eye displacement.

Figure 4.

Displacement of head and eye during localization of stationary sound presented at L1 (left panel) and moving sound ending at L2 (right panel) during the binaural listening condition. Displacement waveforms for head and eye movements are shown for each participant in each stimulus condition in response to stationary (a) and moving (b) sound presentation. Grand average responses are plotted in thicker black lines. Vertical lines show stimulus offset for L1 (a) and L2 (b) stimulus presentations and horizontal bands represent binned stimulus positions (a) and speaker movement (b). (c–j) Magnitude and derived measures of displacement are plotted for individuals (dots) and group (mean ± 1SE = bars) for peak latency (c–d), peak amplitude (e–f), area under the curve (g–h), and pathlength (i–j). Data are shown for left (negative values) and right (positive values) hemifield positions separately for g–j. Overall, the head and eyes move to L1 or L2 with correct direction and magnitude. The eyes move to maximum positions at 30°–40° and show greater pathlengths than the head. Latencies of peak head and eye displacements (c–d) show earlier eye movements and increasing latency as L1 becomes closer to the maximum ±60° presentation locations and as movement change increases.

Peak eye displacements occurred slightly before peak head displacements for stationary sounds (Figure 5(c)) [F_{(1, 480)}= 36.0, p < 0.05] but not moving sounds (Figure 5(d)) [F_(1,202)= 0.008, p = 0.93] with little change over the different stimulus presentation conditions [stationary: F_(11,480)= 0.25, p = 0.99; moving: F_(4,202)= 3.89, p < 0.01].

Maximum amplitude measures from the plugged condition are plotted in Figure 5(e) and confirm larger peak head than eye displacement [F_(1,470)= 25.5, p < 0.001] and increasing rightward head displacement as stationary sound presentation moved rightward from the most extreme left position (−50°–60°) [F_(11,470)= 2.95, p < 0.001]. Similarly, peak head displacements were greater than peak eye displacement during moving sound localization in the plugged condition (Figure 5(f)) [F_(1,202)= 19.2, p < 0.001]. Peak head displacements were smallest for the largest leftward sound movement (−40° change) (M = 5.11, SE = 4.9) and greatest for the largest rightward sound movement (40° change) (M = 42.7, SE = 4.9).

Area under the curve also confirms larger head than eye displacement to both stationary and moving sound [stationary: F_(1,452.3)= 4.9, p < 0.05; moving sound: F_(1,183.2)= 14.0, p < 0.001] during right ear plugging as plotted in Figure 5(g). This overall measure of head displacement showed overall more movement in the rightward direction compared to the left (Left-Right Estimate = −16.4, SE = 2.16, p < 0.001). Both overall head and eye displacement, measured by area under the curve, were largest for the conditions of largest changes in moving sound (Figure 5(h)) (±40° change) [F_(4,182.7)= 5.9, p < 0.001].

Pathlength measures, shown in Figure 5(i) and (j), reveal similar overall head and eye movements for both stationary [F_(1,451.9)= 11.3, p < 0.001] and moving [F_(1,183.2)= 1.07, p = 0.30] sounds during right ear plugging. Head movements occurred more in the right than left hemifield across stationary [Estimate = 20.27, SE = 5.98, p < 0.01] and moving [Estimate = 62.42, SE = 6.14, p < 0.001] sound conditions whereas eye movements occurred to similar degrees in both hemifields (stationary: Estimate = 8.26, SE = 5.98, p = 0.51; moving: Estimate = 2.89, SE = 5.96, p = 0.96).

Discussion

In the present study, localization of stationary and moving sounds without visual cues was assessed in normal binaural and with acute right ear plugging conditions. Unrestrained head and eye movements were measured to assess how these are used in normal hearing during sound localization and to explore strategies for use of head and eye movements when binaural cues are altered. Findings are consistent with previous data showing that accurate sound localization of both stationary and moving sound is disrupted by acute monaural ear plugging and confirm hypotheses that head and eye movements are altered during monaural ear plugging to make increased use of the unaffected ear. The movement of the head to favor the unplugged left ear occurred for stationary sounds in the right hemifield, suggesting some perception of left versus right hemifield. Similar directions of head movements were measured in the sound motion perception task but occurred regardless of direction of movement, reflecting a greater impairment of disrupted binaural cues for this task. Results also showed that normal eye movements precede head movements for sound localization but that head movements are used more than eye movements during monaural ear plugging. The reliance on head movements in this condition reflect preferential use of the better hearing ear for sound localization and relatively stable eye positions which likely reflect normal vestibular-ocular reflexes.

Disruption of Sound Localization by Ear Plugging Affects Both Moving and Stationary Sounds

As shown in Figures 2 and 3, individuals with normal hearing have highly accurate abilities to locate both stationary sounds and to track sound movement with no visual cues. The error rate shown in Figure 2(b) in the binaural condition is consistent with prior studies of horizontal stationary sound localization (Bennett & Litovsky, 2020; Yost et al., 2013) even though sound was presented at over 120° positions along a 120° horizontal arc. This reflects the remarkable acuity of the normal binaural hearing system even when head position varies as shown previously (Goossens & van Opstal, 1999). All participants were able to correctly identify the direction of sound movement in all binaural hearing trials (Figure 3) and their response accuracy for the change in sound movement from start (L1) to finish (L2) was in line with previous studies (Moua et al., 2019) as shown in Figure 3(c).

Figure 5.

Displacement of head and eye during localization of stationary sound presented at L1 (left panel) and moving sound ending at L2 (right panel) during the right ear-plugged listening condition. Displacement waveforms for eye and head movements are shown for each participant in each stimulus condition for stationary (a) and moving (b) sound presentation. Grand average responses are plotted in thicker black lines. Vertical lines show stimulus offset for L1 (a) and L2 (b) stimulus presentations and horizontal bands represent binned stimulus positions (a) and speaker movement (b). (c–j) Magnitude and derived measures of displacement are plotted for individuals (dots) and group (mean ± 1SE = bars) for peak latency (c–d), peak amplitude (e–f), area under the curve (g–h), and pathlength (i–j). Data are shown for left (negative values) and right (positive values) hemifield positions separately for g–j. Overall, head displacement is greater than eye displacement and move more to the right as stimulus presentation becomes increasingly rightward. Latencies of peak head and eye displacements show earlier eye movements for stationary (c) but not moving sounds (d) and are not significantly affected by stimulus presentation condition.

Monaural plugging significantly reduced localization of both stationary (Figure 2) and moving sounds (Figure 3). These data confirm known effects of monaural ear plugging to reduce access to interaural level and timing cues and disrupt spatial hearing (Lupo et al., 2011; Kumpik et al., 2010). Testing occurred with acute plugging which meant that participants did not have time to adapt by reweighting interaural level cues which requires consistent plugging for several days to a week (Kumpik et al., 2010) and were aware of the asymmetric hearing condition. Linear regression analyses showed that all but two participants were able to judge stationary sounds as coming from the correct hemifield but that error rates were approximately doubled from normal binaural conditions (Figure 2b) and were consistent with previously reported effects of monaural earplugging (Slattery & Middlebrooks, 1994). The other two participants judged all sounds as coming from the left non-plugged side. Increased response times suggest that localization of stationary sound was more effortful during monaural ear plugging (Figure 2(c)) which is in keeping with data showing consequences of poor spatial hearing on selective attention (Dai et al., 2018) and cognition (McSweeny et al., 2021).

Monaural ear plugging also disrupted perception of moving sound. Changes in responses from starting position (L1) to end position (L2), as shown in Figure 3(c) revealed high errors in all conditions with larger errors as movement degree increased, suggesting that longer durations of stimulus presentation and/or larger degrees of movement did not help but rather further decreased perception of moving sound during monaural ear plugging. This was further confirmed by analysis of direction perception which, as shown in Figure 3(b), was significantly impaired (measured by shallower slopes). Given how good the participants were at this task in normal binaural conditions, these data highlight a dramatic breakdown in spatial hearing when binaural cues are acutely disrupted. The magnitude of the acute effect may be also surprising given that participants had unrestricted abilities to move their head and eyes which should have provided some indication of the external sound relative to their own spatial position (Brimijoin et al., 2013). It is possible that a period of training might have alleviated these problems (Isaiah et al., 2014; Kumpik et al., 2010; Majdak et al., 2013) but this has not been tested for localization of moving sounds to our knowledge. Movements of the head and eyes might provide some answers for why localization of stationary and moving sound is so impaired by monaural ear plugging.

Precise Eye Movements Normally Precede Head Movements for Accurate Localization of Moving and Stationary Sounds

Data collected in the normal binaural hearing condition, shown in Figure 4, reveal that the eyes start to move before the head to localize sound and reach peak displacement earlier than the head. The eyes can only rotate so far toward either side before head movement must be added to achieve gaze at the sound position targets. Peak latencies also confirm that more time is needed to make the larger eye and head displacements required for localizing stimuli presented at peripheral locations as compared to stimuli closer to center. Limits to peak amplitude eye movements occurred at ∼40° on either side whereas the largest peak head movement amplitudes were found at the 60° edges of the horizontal stimulus presentation arc. Area under the curve, an objective way of measuring amplitude (Polonenko et al., 2022; Vicente et al., 2022), revealed greater overall movements of the eyes than head. This might reflect the combination of smooth pursuits and saccades which would explain the large pathlength values measured for eye movements to both stationary and moving sound. By comparison, head movements do not feature saccadic excursions, appearing more fluid, which could explain the reduced area under the curve and pathlength. Overall, quantification of head and eye movements show that there is a natural tendency to look and then self-orient by moving the head toward a target stationary or moving sound when identifying its location.

Eye Movements Are Suppressed and Head Movements Favor the Unplugged Ear During Localization of Moving and Stationary Sounds

Data shown in Figure 5 reveal unique head and eye movements when binaural cues were disrupted by monaural ear plugging. There was no clear effect of stimulus position on latencies of peak displacements for stationary and moving sound although these peaks occurred slightly earlier in the control sound movement condition (0° movement). Peak amplitudes and area under the waveforms revealed a clear movement of the head for stationary sounds presented on the side of the right ear plug and toward the right for all moving sound conditions. Of note, the eyes moved relatively little in the monaural plug condition. This suggests both that the certainty of sound location was weakened from normal, disrupting the early instinct to move the eyes toward the sound, and that the eyes were being held steady during head movements. Pathlengths for eye movements were similar to pathlengths of head movements likely reflecting the tie between movements of the head and eyes during monaural plugging. The eye movements may be explained by the vestibulo-ocular reflex (VOR) in this situation. The VOR is known to be active even in the dark and would be expected to stabilize the eyes if the participants were intending to move their head specifically. If so, the participants were instinctively trying to localize sound by moving their heads. The main strategy appears to have been to use their better-hearing ear. During stationary sound localization, little head movement was needed when sounds were presented on the side of the left better hearing ear but the head moved right to orient the left ear rightward for sounds presented in the right hemifield. During sound movement, participants moved their heads to the right regardless of whether sound moved left or right. Using the same strategy for all conditions corresponds to the poor abilities that this group showed for detection of sound movement direction. This finding is counter to the idea that direction perception might be better on the side of the non-plugged ear and suggests that perception of moving sound is almost wholly dependent on access to accurate binaural cues. By contrast, preferential ear listening may reflect comparisons of sound between the ear, termed bilateral hearing (Litovsky & Gordon, 2016), and this strategy can be sufficient to detect whether stationary sound is coming from the left or right direction (Grieco-Calub & Litovsky, 2010).

Clinical Implications for Asymmetric Loss and Therapy

Results suggest that individuals with asymmetric hearing might rely on their better hearing ear through head movements when they try to localize stationary sounds and will struggle to hear moving sounds even when using this strategy. Data from clinical groups reveal that these deficits can improve over time if the hearing loss has an adult onset (Litovsky et al., 2010), if efforts are made to provide consistent and accurate binaural cues (Angermeier et al., 2021; Potts et al., 2019), and if there is a period of adaption or training (Isaiah et al., 2014). Such changes could reflect adaptation to ILDs, ITDs, and/or spectral cues and may be more useful for stationary than moving sound. Perception of moving sounds appears to be particularly poor in adults with bilateral cochlear implants (Moua et al., 2019) which is consistent with present results and our suggestion that this ability is less adaptable and thus requires access to accurate binaural cues and binaural processing.

Impaired spatial hearing in children with hearing loss is likely to be even more pronounced than in adults with post-lingual onset of hearing loss. It is clear that children with hearing loss in one or both ears have poor perception of binaural cues (Ehlers et al., 2017; Gordon et al., 2014; Polonenko et al., 2015) and localization of stationary sound (Grieco-Calub & Litovsky, 2010; Johnstone et al., 2010) suggesting that a sensitive period for binaural processing (Tillein et al., 2016) has been missed and/or spatial hearing cannot be promoted without greater attention to the binaural cues provided by the devices. Large gaps in perception of moving sounds have already been shown in children with single-sided deafness who receive cochlear implants (Gordon et al., 2023) and our studies in children with hearing loss in both ears are in progress. Data thus far suggest that efforts to provide accurate binaural cues in early development are critical.