Introduction
Spatial hearing is one of the essential aspects of human auditory processing because an ability to detect and recognize sounds from all surrounding directions can influence safety and communication. Spatial hearing allows for accurate identification of sound source location and enhanced recognition of a target speech in the presence of other competing sounds. Considering the importance of spatial hearing in daily life, the localization abilities of listeners with or without hearing loss have been assessed for clinical and research purposes.
For listeners with normal hearing (NH), it has been well established that their horizontal sound localization mainly relies on interaural time differences (ITD) and interaural level differences (ILD) of the sounds arriving at two ears, and the monaural spectral-shape cues additionally contribute to front-back and vertical localization [1]. According to the duplex theory of sound localization in the horizontal plane, the ITDs are dominant for localizing lower frequency signals, whereas the ILDs have dominant roles for localizing higher frequency signals (above 1,500 Hz) due to the frequency-dependent head shadow effects [2-4]. It has been shown that NH adults can localize broadband signals accurately in the frontal plane, and the localization performance of the frontal signals was better in the horizontal than in the vertical dimension [5,6].
Hearing loss (HL) distorts the delivery of binaural cues for spatial hearing. Thus, listeners with HL often struggle to localize sounds accurately and understand the target speech in complex noisy environments [6-8]. Noble and Collegues [7,8] compared the sound localization of listeners with different types of HL (conductive/mixed vs. sensorineural HL). They found more contribution of conductive aspects in the horizontal-plane localization, presumably due to impoverished low-frequency ITD cues. Even audiometrically slight HL can be associated with reduced binaural processing [9]. The usage of hearing assistive devices can improve sound localization, yet does not always restore localization performance to normal [10-13]. Cochlear implant (CI) users with combined electric-acoustic stimulation (EAS) had poorer-than-normal performance of localization, and their front-back detection was near chance level [11]. Another study found that CI users with good low-frequency hearing thresholds through the EAS showed poor localization performance, and an individual’s frequency-dependent hearing asymmetry across ears was a good predictor of the localization biases [12]. In individuals with single-side deafness across different clinics [13], unilateral CI did not restore binaural ability and sound localization. In addition to individual differences in hearing thresholds and asymmetry, an individual’s listening experience and learning strategy can also affect localization abilities. In an earlier study [14], some listeners with actual unilateral hearing loss (UHL) were better at localizing compared to UHL-simulated listeners, revealing listeners’ learning strategies to improve sound source localization. Another study found some UHL listeners achieved performance within the range of NH listeners on localization and speech-in-noise recognition tasks [15].
Since a large difference exists within a group of HL listeners on horizontal sound localization, several previous studies manipulated HL simulation to measure horizontal localization without learning compensatory strategies for binaural processing. For example, Parisa, et al. [16] simulated UHL by inserting an earplug into the ear canal and found that the horizontal localization of narrow-band signals was degraded by UHL simulation, particularly at higher frequencies. Asp, et al. [17] used the earplug and hearing protector for the UHL simulation and measured the horizontal localization and spatialized speech-in-noise recognition. They found the negative effects of the UHL simulation on the outcomes, emphasizing a need for intervention for mild-to-moderate UHL. The previous studies above [16,17] revealed deficits of horizontal localization based on a within-group comparison. Pittman, et al. [18] compared speech perception of children with actual HL (long-term effect) and HL-simulated NH children (immediate effect). In this study, HL simulation was manipulated through frequency-shaped broadband noise to elevate the hearing thresholds. The results of the between-group comparison indicated that long-term HL might give chances to develop compensatory strategies to reduce the effects of HL. It also showed that the HL simulation can be useful to separate the immediate and long-term effects of HL on auditory performance.
The present study examined the effects of bilateral hearing loss (BHL) versus UHL on horizontal localization accuracy and biases, as well as the spatial short-term memory span. We thus manipulated functional sensorineural HL simulation in NH adults bilaterally and unilaterally. The horizontal localization abilities and spatial memory were compared between actual BHL listeners and BHL-simulated NH listeners, as well as between BHL- and UHL-simulated conditions for NH listeners. The immediate effects of HL simulation would be advantageous to quantify the horizontal localization abilities since there would be fewer chances to develop compensatory strategies for the use of distorted localization cues. We hypothesized that the localization accuracy and biases would differ as a function of sound source direction between BHL and UHL listening conditions due to hearing asymmetry. Asymmetrical hearing induced from UHL simulation would introduce relatively greater variability and more distinct localization biases at specific stimulus locations.
Subjects and Methods
Subjects
Seventeen (12 males and 5 females) listeners with BHL participated in this study (mean age: 70.06±7.25 years). As an inclusion criterion for participation, each listener had bilaterally symmetrical sensorineural HL, with a mean absolute difference of ≤10 dB between the left and the right ear in four-frequency pure tone average (4fPTA) across 0.5, 1, 2, and 4 kHz. Hearing thresholds were obtained using a clinical audiometer (GSI AudioStar ProTM, Grason-Stadler, Eden Prairie, MN, USA) in a double-walled soundproof booth. According to the newly proposed World Health Organization’s hearing-impairment grading scale [19], each listener had mild to moderately severe degrees of hearing impairment. Their mean hearing thresholds were 22.47, 29.12, 35.59, 47.35, 55.88, and 70.00 dB HL (SD: 6.98, 9.23, 7.88, 6.87, 11.62, and 15.41 dB HL) for the better ear at octave-scale frequencies from 0.25 kHz to 8 kHz. The mean percent-correct sentence recognition scores in quiet were 91.33% (SD: 7.19%, range: 80%–100%) for the Korean standard sentence lists for adults [20], when the sentences were presented at a normal conversational level of 65 dB SPL. Their mean three-frequency PTA across 0.5, 1, and 2 kHz was 37.35 dB HL (SD: 7.00, range: 30–51.67 dB HL), and the mean 4fPTA across 0.5, 1, 2, and 4 kHz was 41.99 dB HL (SD: 7.20, range: 32.50–57.50 dB HL) for the better ear (on average, moderate hearing loss). Fig. 1 displays the individual and the mean hearing thresholds of actual BHL listeners (Fig. 1A) and simulated NH listeners (Fig. 1B) at octave-scale frequencies from 0.25 kHz to 8 kHz.
Seventeen (2 males, 15 females) NH listeners participated (mean age: 29.88±3.52 years), with normal hearing based on the thresholds of <20 dB HL at octave-scale frequencies between 0.25 kHz and 8 kHz for both ears. For the functional simulation of sensorineural HL, a spectrally shaped masking noise was generated by a one-third-octave band based on critical ratio estimation, as manipulated in the previous studies [21-24]. The previous studies [21-24] showed that the HL simulation using spectrally shaped masking noise can be useful because it might reflect poor frequency selectivity and loudness recruitment of impaired ears, as well as it might allow exploring the effects of a particular degree and configuration of hearing loss without adaptation or learning from long-term hearing loss. The simulation noise was delivered via insert earphone (ER-3A; Etymotic Research, Inc., Elk Grove Village, IL, USA) to the right ear for the UHL simulation or to both ears for the BHL simulation.
The NH group’s simulated (noise-masked) hearing thresholds were 21.47, 27.06, 39.41, 51.18, 58.53, and 57.94 dB HL (SD: 5.52, 5.61, 5.83, 7.19, 9.96, and 6.14 dB HL) on all octave frequencies from 0.25 kHz to 8 kHz. The mean three-frequency PTA across 0.5, 1, and 2 kHz was 39.33 dB HL (SD: 5.40, range: 30–48.33 dB HL), and the mean 4fPTA was 44.04 dB HL (SD: 5.11, range: 33.75–51.25 dB HL). An independent samples t-test revealed that the hearing thresholds of the actual BHL group did not differ from the elevated thresholds of the NH group at each octave frequency except 8 kHz.
All the participants were right-handed and had Type A tympanogram (Resonance R25C, Gazzaniga, Italy), and none reported a history of tinnitus, middle ear, neurological, or language dysfunctions. The Korean version of the Montreal Cognitive Assessment was administered to BHL participants to screen for cognitive impairment and verified that all had a passing score [25]. All the HL participants had no experience with hearing assistive devices in both ears. This study was approved by the Institutional Review Board of Hallym University of Graduate Studies (#IRB: HUGSAUD 681759). Informed consent was obtained from each participant before testing, and participants received compensation for their time at an hourly rate. For actual BHL listeners, it took less than an hour to complete testing, and for the NH listeners, it took approximately 60–90 minutes to complete three repeated listening conditions (no simulation, UHL, and BHL conditions). The order of conditions was counterbalanced across listeners, and all listeners were given breaks during testing.
Stimuli and procedure
Measure of horizontal sound localization performance
The horizontal localization abilities were tested using a set of 11 loudspeakers arrayed over a 150° localization arc, as shown schematically in Fig. 2. The 11 loudspeakers were positioned in a semicircle arc from -75° to +75° azimuth, with a spacing of 15°. Each loudspeaker was labeled with a number from 1 to 11 (from right to left). All loudspeakers were positioned at ear level and at a distance of 1.4 m from the participant. All the tests were measured in a double-walled soundproof room with internal dimensions of 3.6 m×3.6 m×2 m (W×L×H).
Stimuli consisted of a sequence of broadband noise (125–8,000 Hz) with 10 ms rise and fall times, and the stimulus had 500 ms in duration. Each stimulus was presented at a fixed level of 65 dB SPL. The output level of each loudspeaker was calibrated with a sound level meter (Brüel & Kjær 2250L, Brüel & Kjær, Nærum, Denmark). The amplifiers were balanced so that the sound levels emanating from the speakers at the different azimuths were within 1.5 dB at the participant’s position.
Before the test, a familiarization session was given to the listeners such that the noise was presented across each of the 11 loudspeakers. During the test, the participant was instructed to look straight ahead at the center loudspeaker (0° azimuth) and keep their heads fixed. After each stimulus presentation, each participant was asked to indicate the speaker corresponding to the perceived location of the sound. Each of the 11 locations was included once in a random order within a single block, and each block was repeated three times, resulting in 33 presented trials. NH listeners were tested in three different conditions (no, UHL, and BHL simulation conditions) in a counterbalanced order. All the experimental procedures, including loudspeaker selection, were controlled by a customized interface of MATLAB software (R2016b; The MathWorks, Inc., Natick, MA, USA). During the experiments, no feedback was given at any time.
Horizontal localization performance was quantified by three methods: percent-correct localization (%), root-mean-squared error (RMSE), and mean absolute error (MAE) as in the previous studies of localization [10,13,26,27]. Both RMSE and MAE measure the average magnitude of error, but MAE has an equal weight to all errors, whereas RMSE has relatively more weight for large errors because of the squared errors before averaging. The lower values of RMSE or MAE indicated greater accuracy of horizontal localization. This study presented 33 trials per test run, such that the smallest non-zero RMSE and MAE were 2.61° and 0.45°, respectively.
RMSE and MAE are calculated as follows:
where n is the number of trials (n=33), i is the individual trial number, the stimulus is the angle in degrees of the target speaker location, and the response is the angle in degrees of the speaker location selected by the participant.
Measure of spatial short-term memory span
For the spatial short-term memory span test, listeners were needed to remember a sequence of sound-source locations in the order presented (forward recall), one at a time. The participant was asked to hold their head still during the spatial span test. The 11-loudspeaker array over a 150° arc, which was used for the measure of horizontal localization (Fig. 2), was also used for the measure of spatial memory span.
As stimuli, a sequence of broadband noises was presented at 65 dB SPL with a 2-s interval, and each noise was randomly presented from one out of 11 speakers. The test began with two sound-source locations, increasing the series of locations to 9. The sequence of sound-source locations did not occur from ascending or descending adjacent loudspeakers (e.g., speaker #1-2-3 or speaker #8-7-6-5) as well as from the same locations. At the end of each sequence, the participants were asked to recall as many sound-source locations as possible in the presented order. The length of the sequence was incremented by one if the participant responded correctly to at least one of two trials of the same length. The test was completed when the participant failed both trials of the same length. The participant’s spatial span was the longest number of sequences accurately recalled.
Data analysis
The statistical analyses were performed using SPSS version 25.0 (IBM Corp., Armonk, NY, USA). Independent samples t-tests were used to compare the results of actual BHL listeners and the BHL-simulated NH listeners. One-way analyses of variance (ANOVAs) with repeated measures were then conducted to compare the results of NH listeners in three listening conditions (no, UHL, and BHL simulation). If sphericity was violated, the within-group comparisons of stimulation conditions were corrected with Greenhouse-Geisser estimates. Spatial memory span was also compared with the same comparisons, as used for the localization performance. The p-values <0.05 were considered statistically significant.
Results
Horizontal sound localization accuracy and errors
We measured the percent-correct localization accuracy and errors to evaluate horizontal localization abilities. Fig. 3 shows boxplots of the percent-correct localization score (%), the root-mean-square error (RMSE), and the mean absolute error (MAE) in actual BHL listeners and NH listeners under three listening conditions (no, BHL, UHL simulation).
As plotted in Fig. 3A, the mean localization accuracy was 85.38% (SD=17.28%, range: 39.40%–100%) in BHL listeners. For the NH listeners, the mean localization accuracy was 94.83% without simulation (SD=7.78%, range: 72.73%–100%), 76.47% with BHL simulation (SD=14.03%, range: 54.55%–96.97%), and 35.65% with UHL simulation (SD=23.04%, range: 9.09%–87.88%). Independent samples t-tests were conducted between the actual BHL listeners and the BHL-simulated NH listeners. The results showed no significant difference between the localization scores of the BHL listeners and the BHL-simulated NH listeners [t(32)=1.65]. The one-way ANOVAs with repeated measures were then conducted to compare the performances of NH listeners across three listening conditions (no, BHL, and UHL simulation). The results revealed a significant effect of the simulation condition on the localization accuracy [F(1.45, 23.24)=55.33]. Multiple comparisons with the Bonferroni adjustment revealed that all the percentcorrect localization accuracy differed from each other, showing the lowest localization accuracy from the UHL simulation and the highest localization accuracy in the no simulation condition.
Fig. 3B and C show the localization errors. The mean RMSE and MAE of the BHL listeners were 5.40° (SD=4.09°, range: 0°–17.32°) and 2.46° (SD=3.25°, range: 0°–12.73°). When no simulation was given to the NH listeners, their mean RMSE and MAE were 2.39° (SD=2.76°, range: 0°–7.83°) and 0.80° (SD=1.20°, range: 0°–4.09°). When the BHL simulation was applied to the NH group, the mean RMSE and MAE were 7.35° (SD=3.22°, range: 2.61°–15.45°) and 3.77° (SD=2.56°, range: 0.45°–9.55°). In the UHL simulation condition, their mean RMSE and MAE were 13.68° (SD= 3.89°, range: 5.22°–19.89°) and 10.91° (SD=4.32°, range: 1.82°–17.27°). The localization errors did not statistically differ between actual BHL listeners and BHL-simulated NH listeners [RMSE: t(32)=-1.55, MAE: t(32)=-1.31]. Among the repeated three listening conditions, a significant effect of the simulation condition was found in the localization errors of the NH group [RMSE: F(2, 32)=47.37; MAE: F(1.36, 21.82)=48.05], revealing the greater localization errors from the UHL simulation and the smaller errors in the no simulation condition.
Horizontal sound localization biases as a function of stimulus location
In addition to overall localization accuracy and errors, we also investigated the localization biases to compare response distributions as a function of speaker position. In the case of NH listeners without simulation, the localization score was near 100% across all the loudspeakers. Thus, we investigated the localization biases only for actual HL listeners and HL-simulated NH listeners.
Fig. 4 displays the mean localization accuracy as a function of stimulus location. As plotted, the actual BHL listeners and BHL-simulated NH listeners did not show distinct response bias across spatial locations. The localization scores ranged from 78% to 94% across speakers for actual BHL listeners (Fig. 4A) and from 70% to 84% for BHL-simulated NH listeners (Fig. 4B). Both groups did not show any pronounced performance at specific locations.
However, the UHL simulation substantially impaired the localization, especially when sound sources were presented to the sides of the simulated (right) ear (Fig. 4C). The localization score was very low even when the sound sources were coming from the front (0°) speaker (mean score: 29%), and the accuracy decreased more with right-position speakers (mean score ranging from 10% to 20%). It revealed that hearing asymmetry would disrupt sound-source localization not only at the side of the poorer hearing but also from the frontal speakers.
As reported earlier, overall localization accuracy was poorer and errors were larger in the UHL-simulated condition compared to the BHL-simulation or performance of actual BHL participants. The RMSE and MAE provide the magnitude of errors, but no information about the direction of errors. To further investigate the impact of hearing asymmetry in the localization biases, the distribution of horizontal localization responses was plotted as a function of stimulus location in Fig. 5. Note that the gray bubbles in Fig. 5 indicate perfect localization performance, and the size of the bubbles represents the proportion of correct responses for a given stimulus location (i.e., the larger bubbles reflect a greater number of correct responses). As shown, the majority of actual BHL participants were confused with adjacent speakers when localizing sound-source speakers, and there was no pronounced bias toward either the left or right side (Fig. 5A). This tendency was similar to the results of BHL-simulated NH listeners (Fig. 5B) because of symmetric hearing. However, with the UHL simulation (Fig. 5C), more distinct response biases on localization performance were found towards the non-simulated (left) ear, suggesting that hearing asymmetry (UHL simulation) resulted in distorted spatial cues on localization.
Spatial short-term memory span
The spatial memory span was determined by the number of loudspeakers that a subject could remember when a sequence of broadband noises was randomly presented from one out of 11 speakers. Fig. 6 displays the mean spatial memory span of BHL listeners and NH listeners under three conditions (no, UHL, and BHL simulation). The average spatial memory span of HL listeners was 4.06 (SD=2.38, range: 0–7). For NH listeners, the mean spatial memory span was 6.82 without simulation (SD=1.29, range: 4–9), 3.59 with BHL simulation (SD=2.21, range: 0–7), and 0.59 with UHL simulation (SD=1.70, range: 0–6), respectively. The spatial memory span did not statistically differ between actual HL listeners and BHL-simulated listeners [t(32)=1.65]. Results of oneway ANOVAs with repeated measures showed a significant main effect of the condition on the spatial memory span of NH listeners [F(2, 32)=67.21]. Multiple comparisons with Bonferroni adjustment revealed that the spatial span differed across conditions, showing the greatest spatial span with no simulation and the shortest spatial span from the UHL simulation.
Pearson correlation analyses were conducted to examine relations between localization scores (%) and spatial span for HL and NH groups separately. A correlation coefficient between 0.3 and 0.7 was considered to have a moderate-strength correlation, and a correlation coefficient greater than 0.7 indicated a strong positive correlation [28]. The results showed that the HL listeners with higher localization scores had significantly greater spatial span with moderate strength (r=0.58). The localization scores of NH listeners were significantly related to their spatial memory span with a moderate strength in the BHL simulation (r=0.59) and a strong strength in the UHL simulation condition (r=0.72). For the HL listeners, additional Pearson correlation analyses were subjected to examine whether the degree of their hearing loss would be related to their localization performance and spatial span. The three-frequency PTAs across 0.5, 1, and 2 kHz were significantly and strongly related to the localization score (r=-0.82), RMSE (r=0.78), the MAE (r=0.80), and moderately related to the localization score spatial span (r=-0.56) in the actual BHL participants. In summary, our results suggest that the localization deficits of HL listeners are related to the ability to memorize spatial positions, even when listeners have little difficulty with speech understanding in quiet. The deteriorated localization performance and spatial memory in the HL group appear to be associated with their hearing thresholds.
Discussion
In everyday environments, we often experience diverse sound signals arriving from multiple locations. In this complex listening situation, listeners can localize and segregate different sound sources spatially based on binaural cues and also selectively attend to the target sound. The present study compared the localization abilities between actual BHL listeners and BHL-simulated NH listeners as well as compared between BHL-simulated and UHL-simulated performance of NH listeners. We applied both the BHL and UHL simulations to NH adults in order to measure the effects of hearing asymmetry on horizontal localization abilities with fewer chances to develop compensatory strategies for monaural hearing.
When no simulation was given to NH listeners, the majority of NH participants were able to localize sound perfectly (mean localization score=94.83%, mean RMSE=2.39°), which is in line with the previous findings. The previous study [29], which used the same 11-loudspeaker array spanning a frontal 150° arc as we used, measured the horizontal localization of listeners with varying hearing thresholds and ages. They reported that the RMSE ranged from 0° to 5° for NH listeners whose PTA was <10 dB HL. Since the RMSE ranged from 0° to 7.83° for our NH listeners, the current data and the previous finding [29] appeared to be similar. Yost, et al. [30] measured the localization accuracy of NH adults using an 11-loudspeaker array within a frontal 150° arc. They found that the RMSE of NH adults was normally distributed with a mean of 6.2° (±1.79°), and filtering the stimulus noise did not significantly influence the localization performance. Dorman, et al. [31] compared the localization performance of young and old NH listeners, with a 13-loudspeaker array spanning a frontal 180° arc. They found that the mean RMSE was 6° and 5.4° for young and old NH listeners, showing no negative effect of age on horizontal localization. Taken together, the mean RMSE of NH adults across the studies above was less than 7° when 11 or more loudspeakers were used as an experimental setup.
This study compared the horizontal-plane localization abilities between actual BHL listeners and BHL-simulated NH listeners. As expected, their localization performance was significantly lower than the performance of non-simulated NH listeners. The mean localization scores were about 85% and 76%, and the mean RMSE values were 5.40° and 7.35° for the actual BHL listeners and BHL-simulated NH listeners, respectively. Although the mean localization performance of the actual BHL listeners was better than those of the BHL-simulated listeners, it did not reach statistical significance. Both groups also showed no distinct pattern of localization biases as a function of stimulus position. This might occur because our BHL listeners had bilaterally symmetrical sensorineural HL in high frequencies, with a mean absolute difference of ≤10 dB between ears based on 4fPTA across 0.5, 1, 2, and 4 kHz.
Similar to our finding, Lorenzi, et al. [32] measured the localization performance of listeners with bilaterally symmetrical sensorineural HL at high frequencies. Using an 11-loudspeaker array spanning a frontal 180° arc, the authors reported poorer localization performance in BHL listeners than in NH listeners. Like our study, HL listeners with high-frequency hearing loss in the previous study [32] had more audibility at lower frequencies and less audibility at higher frequencies, possibly making low-frequency cues more accessible compared to high-frequency cues. Using 13 loudspeakers separated by 15° within a 180° arc, Van den Bogaert, et al. [10] measured the horizontal localization ability of 10 BHL listeners. When unaided, their average RMSE was about 13° (range: 8.1°–18.2°) for broadband telephone signal. Considering the average RMSE of our BHL subjects was 5.40° (range: 0°–17.32°), the horizontal localization errors in the previous study [10] were larger than the current study. This could be associated with different types of stimuli or audibility at low frequencies since the subjects of Van den Bogaert, et al. [10] had poorer hearing at lower frequencies across 0.25, 0.5, and 1 kHz compared to our subjects. Among our 17 BHL listeners, 2 listeners showed localization scores of <50%, whereas 2 listeners showed perfect localization scores (100%). Two listeners who scored less than 50% had poorer hearing at low frequencies than other subjects. Also, the PTA across 0.5, 1, and 2 kHz was significantly and strongly related to the localization scores, the audibility at lower and mid frequencies would affect the horizontal-plane localization abilities for listeners with symmetric hearing loss.
The present study manipulated the UHL simulation to NH adults to measure the localization ability without any long-term exposure to hearing monaurally. With the UHL simulation, the mean localization score and RMSE were about 35.65% (range: 9.09%–87.88%) and 13.68° (range: 5.22°–19.89°), respectively, showing larger localization errors and greater variability in the UHL than in the BHL simulation. We also observed the localization responses were largely biased towards the better ear (non-simulated ear). Although this study applied functional sensorineural HL simulation, several previous studies used conductive HL simulation. For example, Parisa, et al. [16] applied the conductive type of UHL simulation to NH adults by placing earplugs. They measured the horizontal localization abilities with narrowband noise signals, and found poor horizontal localization accuracy and distinct localization biases on the occluded (simulated) ear, especially with higher-frequency stimuli. Another previous study [17] simulated mild and moderate levels of UHL by using earplugs alone and earplugs plus hearing protectors. The authors reported much worse localization accuracy in moderate levels of UHL than in mild levels of UHL simulation. Using 15 loudspeakers separated by 10° within a 140° arc, Firszt, et al. [15] compared the localization performance of the actual and simulated UHL listeners. They simulated UHL with an ear plug and hearing protector and found poorer performance in the UHL-simulated listeners than in the actual UHL listeners. Some listeners with severe to profound UHL performed within the range of non-simulated NH listeners’ performance, indicating that UHL listeners can develop their own strategies for better localization and spatial hearing. Other previous studies [33-38] reported not only reduced localization abilities but also higher handicaps, lower school performance, and more listening-related fatigue in UHL children or adults compared to the NH population. Given this, reduced audibility and hearing asymmetry would affect the spatial hearing abilities of the UHL population, yet their listening experience and learning strategy would be also associated with their auditory abilities, resulting in a large individual variability.
Spatial short-term memory might be involved in many everyday tasks, particularly when either wanted or unwanted sound sources are coming from diverse directions. We conducted the spatial short-term memory task, requiring listeners to keep a sequence of sound-source positions in their short-term memory and recall the series of stimulus locations as much as they could remember. Without hearing loss simulation, NH participants could also remember the locations of as many as six to seven loudspeakers (mean span=6.82, range: 4–9), even when the location of sound sources was uncertain from the speakers across a 150° localization arc. Considering that approximately 7 items can be recalled regardless of whether the stimulus is digits, letters, or words [39,40], a similar size of short-term storage appeared to be used for recalling sound-source locations. However, the mean spatial memory span was 3.59 (range: 0–7) and 0.59 (range: 0–6) in the BHL-simulated and UHL-simulated conditions, respectively. This suggests that substantial difficulties in localizing due to hearing asymmetry resulted in poor spatial recall of the locations of sound sources.
This study has several limitations. First, the actual UHL participants were not enrolled in this study, and the sample size of actual BHL listeners (n=17) was relatively small. Thus, our results should be interpreted carefully to understand spatial hearing in listeners with varying degrees and configurations of BHL or UHL. A larger sample size of actual UHL participants is needed in future studies. Second, we did not evaluate any subjective handicaps or difficulties in spatial hearing in daily life. Although horizontal localization and short-term spatial memory were assessed in this study, the impact of hearing loss on individual life would differ from the results of the objective test. Implementing a subjective questionnaire can be supportive in addition to an objective measure of spatial hearing.
In summary, this study measured the horizontal sound-source localization and spatial short-term memory span of actual and simulated HL listeners. Overall, the localization performances and the spatial span were similar between listeners with actual and simulated BHL. However, the localization performance with the UHL simulation was about two times worse than the localization with the BHL simulation. In addition to poor localization accuracy, the ability to recall the spatial positions encoded and stored in memory was also substantially disrupted in the UHL simulation. Due to asymmetric hearing, the localization responses were largely biased to the side of the intact ear even for the frontal sound signals, and the individual variability in responses was greater in the UHL-simulated ear. The substantial spatial deficits from asymmetric hearing suggest a need for clinical assessments of spatial hearing in addition to conventional hearing tests.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Print
