Abstract
This study aimed to determine hearing benefits and challenges for children using cochlear implants (CIs) in one ear and acoustic hearing in the other (bimodal hearing) through hearing aids (HA-CI) or normal hearing (single-sided deafness/SSD-CI). Participants were 34 children with CIs [MAge (SD)=12.2 (3.2) years] and 6 peers with typical hearing/controls [MAge (SD)=13.8 (1.7) years]. Side of better hearing (aural preference) was measured by word recognition in quiet and noise, and spondee-word recognition thresholds in noise co-located/0° or separated (left/right 90°). Self-reported hearing was measured using the Speech, Spatial and Qualities of Hearing Scale (SSQ). Localization of stationary and moving sound (and unrestricted head movements) and sensitivity to interaural level and timing cues were measured (separate controls [n=5, MAge (SD)=14.0 (1.6) years]). Speech perception using the CI alone was similar in children with SSD-CI and HA-CI (p=0.84) but SSD-CI users had greater aural preference in their acoustic ear (p<0.01). CI users had high errors localizing stationary sound and poor moving sound detection (p<0.001) which was associated with earlier onset of hearing loss, and poor access to interaural cues (p<0.001) compared to controls. Head movements in children with CIs tended to favour their non-implanted ear (p<0.05). Results indicate that children with asymmetric hearing gain speech perception with their CI but have aural preference for the non-implanted ear, particularly in the presence of good residual hearing and later onset of hearing loss. Bimodal listeners had poor access to binaural cues, poor spatial hearing with slightly better results in children with later onset of hearing loss and ineffective gaze movements.
Keywords
Introduction
Several decades of research have demonstrated that children with unilateral hearing loss have challenges in multiple areas of development relative to their peers with normal hearing. These children show compromised speech understanding in noise (Jensen et al., 1989; Lieu et al., 2013) which might reflect, in part, poor spatial hearing due to inabilities to access interaural timing difference (ITD) and interaural level difference (ILD) cues (Murphy et al., 2011; Gordon et al., 2023). These hearing problems extend to delays in speech-language development (Bess & Tharpe, 1984, 1986; Klee & Davis-Dansky, 1986; Lieu et al., 2010; McSweeny et al., 2021), increased cognitive load (Klee & Davis-Dansky, 1986; McSweeny et al., 2021), and reduced IQ (Lieu et al., 2013). Children with unilateral hearing loss have a higher risk of repeating a grade than their peers with normal hearing (Bovo et al., 1988; Lieu, 2004). They also report that their hearing loss is a significant handicap (Gordon et al., 2023; Newman et al., 1997) and that it reduces their quality of life (Umansky et al., 2011).
Cochlear implantation has emerged as a promising option to support development in children with severe-to-profound deafness in one ear by providing them with bilateral hearing. Despite the very different input provided by electrical stimulation from a cochlear implant (CI) in one ear and acoustic stimulation through a hearing aid (HA) or normal hearing in the contralateral ear, there may be redundancy of bilateral input that supports speech perception in the bilateral condition in adults (Ching et al., 2005; Grossmann et al., 2016; Jakob et al., 2022; Mok et al., 2006; Peters et al., 2021) and children (Arndt et al., 2024; Arras et al., 2022; Gordon et al., 2023; Polonenko et al., 2018b). Benefits are particularly clear in adults for speech perception in noise (Arndt et al., 2011). On the other hand, the application of bimodal stimulation for bilateral hearing has raised questions regarding potential interference between the acoustically hearing ear and implanted ear, particularly in the context of adult single-sided deafness (SSD) (Bernstein et al., 2020), and limited abilities to promote spatial hearing in children with SSD (Gordon et al., 2023).
Although the CI can help reduce the head shadow from the deaf ear in adults (Mertens et al., 2017; Thomas et al., 2017), bimodal devices or CIs in SSD have provided minimal improvements in sound localization in children (Gordon et al., 2023; Thomas et al., 2017). Sound localization after CI is marginally better in adults with SSD (Arndt et al., 2011; Firszt et al., 2018; Grossmann et al., 2016; Jakob et al., 2022). A major limitation to supporting spatial hearing is the large interaural mismatches created by bimodal input in: 1) cochlear place-of-stimulation (Bernstein et al., 2018), 2) timing of auditory nerve and brainstem responses (Polonenko et al., 2015; 2018a), and 3) level differences (Blanks et al., 2008; Holtmann et al., 2020). In adults with post-lingual onset of hearing loss, delaying the CI stimulation to reduce interaural timing mismatches can improve both localization accuracy (Seebacher et al., 2019; Zirn et al., 2019) and localization bias (Angermeier et al., 2021). Efforts to improve access to ILDs by matching automatic gain control between devices improved speech recognition in noise in adult bimodal listeners but effects on sound localization were not clear (Holtmann et al., 2020; Veugen et al., 2016). These findings suggest potential of bimodal input to stimulate existing pathways that were once engaged in binaural processing with remaining questions about gains in spatial hearing.
Sound localization may be more challenging to establish in children with early onset deafness in one ear than adults who acquire deafness in one ear after having normal binaural hearing as children. Binaural processing in development can be disrupted as shown by reweighting of excitatory inputs to cortical neurons in unilaterally deaf white cats (Kral et al., 2013). These changes may underlie the “Aural Preference Syndrome” described in children with unilateral deafness which include reorganization along the bilateral auditory pathways and corresponding asymmetries in speech perception (Gordon et al., 2015). Such effects of unilateral hearing on the developing auditory system and hearing have been mitigated to some degree in children with pre-lingual deafness in both ears by providing bilateral CIs without delays to surgery (Gordon et al., 2013). By contrast, delayed CI surgery is more common in children with unilateral deafness because many, particularly those with SSD, were not considered candidates for CI until recently (Cushing et al., 2019). The change in candidacy was informed by evidence that reducing delays to CI can avoid neural plasticity that strengthens input from one ear compared to the other and asymmetric hearing (Gordon et al., 2015) in these children (Polonenko et al., 2018a), which could potentially support spatial hearing.
It is not clear how hearing history affects access to interaural cues (ITDs, ILDs) and spatial hearing in children with bimodal hearing. It is possible that later onset of unilateral deafness would have allowed some development of binaural processing which might be exploited by providing bilateral input through bimodal hearing. Yet, the effectiveness of bimodal hearing, even after normal binaural hearing in early life, could be limited by remaining asymmetries between the ears in the bilateral auditory pathways and in the input provided which can create an aural preference for either the CI or the non-implanted ear. The extent to which an aural preference develops could depend on the history and degree of asymmetric hearing including when the CI was received. The use of a hearing aid in the non-implanted ear is also an important factor as shown by more consistent use in children with bimodal devices than children who use a CI alone for SSD (Gao et al., 2026).
Benefits and challenges for spatial hearing in children with bimodal hearing were explored in the present study using age-appropriate behavioral paradigms. Behavioral measures of both speech perception and binaural hearing were considered. Benefits of spatial separation between speech and noise on word recognition thresholds were explored because this measure has been sensitive to aural preference in previous cohorts of children with and without hearing loss (Gordon et al., 2023; Polonenko et al., 2018b). The Speech, Spatial and Qualities of Hearing Questionnaire (SSQ) was included as it has reliably characterized hearing challenges in similar groups (Alemu et al., 2025; Arndt et al., 2011; Jakob et al., 2022; Peters et al., 2021).
Access to ILDs and ITDs, delivered by headphones and/or direct CI input, was measured because these binaural cues are fundamental for spatial hearing and are likely severely affected in bimodal hearers. A binary lateralization task was used as previously reported in participants with hearing loss (Kan et al., 2013; Salloum et al., 2010), revealing particular vulnerability to ITDs in children who use bilateral hearing aids (Gorodensky, Alemu, et al., 2019) or bilateral CIs (Alemu et al., 2025; Kan et al., 2013; Salloum et al., 2010; Steel et al., 2015). Sound localization was tested both for stationary and moving sound delivered by a loudspeaker. Head and eye movements were allowed to provide children with the best chance to locate the sound (Brimijoin & Akeroyd, 2014).
Head and eye movements (combined into gaze) were monitored to assess whether participants use expected movements to favour their better hearing ear (Alemu et al., 2024b). These strategies were shown upon acute ear plugging in children and adults with normal hearing (Alemu et al., 2024b). It is possible that these strategies have not been developed in CI users as previously shown in children using bilateral CIs (Alemu et al., 2025). Whereas abnormalities in stationary sound localization have been reported in CI users (as reviewed above), localization of moving sound with CIs has only been recently studied (Alemu et al., 2025; Moua et al., 2019). Significant impairments localizing the direction of moving sound could have real world consequences for children as they navigate dynamic environments. In normal hearing, duration and distance cues are more heavily weighted for tracking moving sound (when available) but velocity can be utilized if the former cues are unavailable (Carlile and Best, 2002).
The main objective of the current study was to determine hearing benefits and challenges in children using CIs in one ear and acoustic hearing in the other (bimodal hearing) through either hearing aids or normal hearing. The study added to the existing literature by including children with a wide range of hearing in the non-implanted ear. The first hypothesis sought to confirm previous findings by testing the hypothesis that, in children and adolescents with asymmetric hearing loss, aural preference for the acoustic hearing ear after CI increases relative to hearing history including the onset of asymmetric hearing and degree of residual hearing in the non-implanted ear. The larger cohort also provided an opportunity to address previous speculations regarding binaural hearing in children with bimodal hearing. To do this, the study tested the hypotheses that, in children with bimodal hearing, sound localization after CI is poorer than normal, related to sensitivity to binaural cues, and better for children whose onset of deafness occurred later in childhood. Further, a novel question about how children with bimodal hearing use head movements during sound localization testing was examined; the hypothesis was that gaze movements by children with bimodal hearing favour their acoustic hearing ear more than the CI during spatial hearing tasks.
Methods
Participants
Demographic and Hearing Outcomes for Children including 34 Bimodal Listeners and 11 Peers with Typical Hearing are Summarized Below.
Outcome Measures
Behavioral Measures of Aural Preference and Hearing
Speech Perception
Speech perception in children with CIs was measured in quiet (speech only) in unilateral listening conditions (i.e. either to the acoustic ear or the CI ear) using age-appropriate tests during routine clinical appointments. One of three open-set tests were chosen based on the child’s abilities: the Phonetically Balanced Kindergarten (PBK) test (Haskins, 1949), the Glendonald Auditory Screening Procedure (GASP) word test (Erber, 1982) or the Multisyllabic Lexical Neighborhood Test (MLNT) (Kirk et al., 2022). In most cases (73%), the PBK test was used. Word stimuli were presented from a loudspeaker positioned at 0°, in front of the child, either through a recording or monitored live voice at a presentation level of 45 or 55 dB HL. The unilateral CI only condition was measured by attenuating the sound to the acoustic ear either with an ear plug and muff or with speech noise presented through an insert earphone, or by delivering the speech stimulus through the Cochlear Wireless Mini Microphone (through its line-in port) paired to the CI. There were no significant effects between these methods on accuracy scores in a larger dataset from children with bimodal hearing that included the present study cohort (Bartels et al., 2026). The rationalized arcsine unit (RAU) transformation was applied to speech scores and used as the dependent variable in parametric statistical analyses (Hoen, 2015; Sherbecoe & Studebaker, 2004) to assess asymmetric function between ears (aural preference). Speech tests were available in 27 of the 34 children with bimodal hearing (15 HA-CI, 12 SSD-CI).
Spatial release from masking asymmetry (SRMA) was calculated using speech reception thresholds (SRTs) in response to spondee target speech and speech-weighted noise presented at 45 dB HL in 30 out of 34 children with bimodal hearing (16 HA-CI, 14 SSD-CI), and 4 of the 6 peers with typical hearing who completed the sound localization paradigm. SRTs were measured by delivering speech in-front (S0°) and noise in three positions: co-located (S0°N0°) and with 90° of spatial separation to the left side (S0°N-90°) and to the right side (S0°N+90°). SRTs were measured by decreasing speech presentation level from a starting level of 55 dB HL in step sizes of 4 dB following correct repetitions and increasing presentation level in step sizes of 2 dB following incorrect repetitions. SRTs were measured across two trials by listening configuration and were averaged across trials. SRMA (dB) was the difference in benefit of moving the noise from the front to the left versus right side in children with normal hearing or between the acoustic and CI side in bimodal listeners: SRMA = (SRT(S0°N0°) – SRT(S0°N-90° or acoustic)) – (SRT(S0°N0°) – SRT(S0°N+90° or CI)). Effects of hearing history on aural preference measured by SRMAs were assessed.
Self-Reported Hearing
A modified version of the Speech, Spatial and Qualities of Hearing Scale (SSQ) was administered in paper and digital (REDCap) versions in 30 of 34 children with bimodal hearing (12 HA-CI, 18 SSD-CI) (Galvin & Noble, 2013; Gatehouse & Noble, 2004). SSQ tests were not completed in 4 children due to time constraints. The SSQ was administered as an adolescent version to children who were able to comprehend and respond to questions and as a parental version to parents/caregivers of children who were unable to independently comprehend or respond to the questions. The adolescent version of the SSQ was administered to 20 children (9 HA-CI, 11 SSD-CI) and the parental version of the SSQ was administered to 15 children (10 SSD-CI, 5 HA-CI). There were five children including three SSD-CI users and two HA-CI users who had both versions of the SSQ completed. SSQ scores were assessed for differences between sub-tests and groups. Data from 12 of these participants have been reported previously (Gordon et al., 2023).
Measures of Binaural Hearing
Localization of Stationary and Moving Sound
A horizontal (azimuthal) sound-source localization task was completed by all 34 bimodal listeners (16 HA-CI, 18 SSD-CI) and 6 age-matched peers with typical hearing. Sound localization data from 12 SSD-CI users and the 6 control participants have been previously published (Alemu et al., 2024b; Gordon et al., 2023). Bandpass-filtered white noise (125 to 8000 Hz) was presented by a speaker fixed to an L-shaped arm that could start and move anywhere along a horizontal arc of 120° [0° to ±60°, i.e. leftward (-) or rightward (+)]. Each trial consisted of stationary sound presented at a random location (L1) for a duration of three seconds (instruction: “where is the sound?”) followed by sound presented while moving to a second location (L2) (instruction: “where did the sound move to?”). A test block included five conditions of sound movement including: 0° (no movement) [MDuration (SD) = 3.2 (1.5) s; MVelocity = 0.0 °/s], 20° leftward (-) or rightward (+) from L1 [MDuration (SD) = 6.7 (0.4) s; MVelocity = 3.0 °/s] and 40° leftward (-) or rightward (+) from L1 [MDuration (SD) = 9.0 (0.2) s; MVelocity = 4.4 °/s]. The speaker and arm were controlled from the main testing PC using a customized testing program as described previously (Alemu et al. 2024b, 2025). Participants used a videogame controller operating a red laser to point to their perceived location of L1 and L2. User response was recorded as azimuthal position (responses between ±90°) at button press and response time was recorded as the start of controller movement. Trials were organized in blocks of six trials each with one stationary presentation and one moving presentation per trial. Prior to testing, a block of practice presentations was presented with feedback provided to help participants learn the task. In the bimodal listeners, 5-10 blocks were completed in a bimodal condition (CI-on) and 5-10 blocks in a CI-off condition. In the control group, 7-10 blocks were completed with typical hearing and 7-10 blocks with the right ear attenuated using an ear plug with an earmuff overtop (Alemu et al., 2024b). Root mean square errors (RMSE) were calculated for stationary stimuli and assessed for effects of hemifield of stimulus presentation, side of the CI, listening condition, group, and response time. Perception of moving sound direction was measured by logistic slope to assess effects of the same factors.
Behavioral Lateralization of Interaural Level and Timing Differences
Participants completed a lateralization task in response to bilaterally delivered sound with varying interaural level differences (ILDs) or interaural timing differences (ITDs). Lateralization data were available in 17 of 34 bimodal listeners (13 HA-CI, 4 SSD-CI) and a control group of 5 participants whose data were previously published (Alemu et al., 2024a; Gorodensky, Alemu, et al., 2019). Acoustic stimuli (250-Hz click train of 36 ms presented at 1-Hz bilaterally through ER3A insert earphone) were presented bilaterally in children with normal hearing and in the non-implanted ear in CI users. CI stimulation [biphasic pulses from an apical electrode (#20) at 250 pps for 36 ms] were delivered through a research processing interface that held ITD = 0 constant. ITD = 0 was defined as simultaneous presentation of the CI stimulus and click by the insert earphone transducer. Presentation levels (dB in the acoustic hearing ear and CU in the CI ear) were determined independently in each ear based on thresholds and maximum comfort levels to these stimuli. A further bracketing procedure was used to achieve balance (ILD = 0 defined as 50:50 left:right lateralization) as described in Gorodensky, Alemu et al. (2019). ILDs weighted to the left and ITDs leading from the left are indicated by negative values and right weighted/leading stimuli are indicated by positive values. Similar stimuli were used in prior studies of children with hearing loss (Salloum et al., 2010; Gordon et al., 2014; Steel et al., 2015; Easwar et al., 2017a, 2017b; Gorodensky, Alemu et al., 2019; Alemu et al. 2024a, 2025; Fung et al., 2025). In bimodal users, ILDs were created by a change in dB (acoustic side) with an equal but opposite change in CU levels (CI side) (eg. ILD = +6 dB/CU indicates a 3 dB increase acoustic right side with a 3 CU decrease on the left CI side = +6 dB). Stimuli were delivered in blocks in which the following conditions were presented in random order: six ILDs [±1.5, ±3 and ±4.5 dB in children with normal hearing and a wider range of ±3, ±6 and ±9 dB/CU in children with CIs], six ITDs [±0.2, ±0.4 and ±1 ms in children with normal hearing and a wider range of ±1, ±2 and ±3 ms in children with CIs]. For each set of ILDs or ITDs there were also two unilateral presentations (left or right) and one condition of ILD = 0 and ITD = 0. Children completed six blocks for a total of 108 trials overall. Instructions were to choose if the sound was heard on the “left” or “right” side of the head. Responses and reaction times were recorded for each trial. Two control participants completed an earlier version of the stimuli set in which ITD/ILD trials were presented in separate blocks and included an additional ITD of ±0.6 ms but these trials were removed in the present analysis. Participants were excluded from further analysis if the proportion of accurate responses for unilateral trials fell below 75% across trials. Of the 17 bimodal listeners tested, 14 (12 HA-CI, 2 SSD-CI users) met inclusion criteria. Sensitivity to ILDs and ITDs were measured by the logistic slopes of proportion of “right” responses across the conditions tested. Relationships between binaural cue sensitivity and stationary sound perception (RMSE) and perception of moving sound direction (logistic slope) were assessed.
Head Tracking During Sound Localization
Head movements were collected in real-time during the horizontal sound-source localization task and were available in 32 of 34 bimodal listeners (16 HA-CI, 16 SSD-CI users) and 5 of the 6 peers with typical hearing (control group) who completed the sound localization paradigm. Head movements were tracked using the EDTracker Pro wireless head tracker at a sample rate of 125 Hz along three degrees of freedom (i.e. pitch, yaw and roll) as previously described (Alemu et al., 2024b, 2025). Another device called the MbientLab Metasensor was used for head tracking in four participants (three bimodal listeners and one control listener) but displacement drifts were encountered which were corrected in post-processing (Alemu, 2020). Eye movements were also tracked using the Pupil Labs Pupil Core eye tracking glasses in a subset of children due to equipment breakdowns (14 HA-CI, 13 SSD-CI users, and 5 peers with typical hearing). Participants were allowed to move their head and eyes without restriction but were instructed to maintain a seated position and to avoid moving their legs or torso. Head and eye displacement were calculated from real-time position data in the horizontal plane relative to the time of stimulus onset. Displacement was binned by speaker azimuthal position or movement condition within a time window starting from onset of stimulus presentation through to stimulus offset. Displacement responses were averaged to the nearest 10th millisecond and binned relative to stimulus onset. For stationary sound presentation, displacement was binned in 10° increments from -60° to +60° [12 increments in total across the 120° range]. For moving sound presentation, displacement was binned by speaker movement conditions including 0, ±20° and ±40°. Afterwards a smoothing routine was applied using a simple moving average filter as described in Alemu (2020) particularly for eye tracking, to interpolate across gaps created by removal of low-quality data (e.g. eye blinks). Head movements were calculated by area under the curve (AUC) per stimulus trial and assessed for effects of hemifield of stimulus presentation, side of the CI, listening condition, and group. Data from six of these participants has been reported previously (Alemu et al., 2024b).
Statistical Analyses
Statistical analyses and graphs were produced using the R programming language (Version 4.0.3) and the R-Studio IDE (Version 1.3.1093) for statistical computing and graphics (RStudio Team, 2020). Specific statistical models and tests are detailed in Appendix A. Linear mixed-effects regression models were used in the case of repeated measures per participant across test conditions with further testing of the model by type III analysis of variance using Satterthwaite’s method and multiple comparisons using estimated marginal means. Simple linear regression was used to assess associations between outcome measures and demographic factors characterized by the Principal Component Analysis (PCA). For conciseness, only the estimated values and/or associated p-values have been included in the results section.
Analyses of Demographic Factors in the Bimodal Group
PCA [prcomp function from R’s base stats package] was performed to potentially reduce seven demographic measures and measures of hearing from all 34 children with bimodal hearing into a smaller set of uncorrelated variables while retaining much of the original variance. The measures used in the PCA included age at test, age at diagnosis of hearing loss, age at cochlear implantation, duration of deafness (by side), and pre-implant unaided pure-tone averages (PTAs) calculated from pure-tone audiometric thresholds measured at 500, 1000 and 2000 Hz (by side). Data were scaled for unit variance (correlation matrix used) prior to performing the PCA. Results of the PCA characterized the main differences between the whole group of bimodal users (HA-CI and SSD-CI) and were used to assess variability in outcome measures in the bimodal group.
Analyses of Aural Preference and Hearing Measures
Statistical testing evaluated aural preference measures (speech perception, spatial release of masking) and self-reported hearing (SSQ) for differences between children with bimodal hearing and children with normal hearing. Characteristics of bimodal users, defined by the PCA, were used to explore variability in these outcomes. Differences between the HA-CI and SSD-CI subgroups were also assessed in separate models for particular outcome measures. In these cases, group included three levels (control, HA-CI, SSD-CI) for each outcome measure.
Analyses of Binaural Hearing Measures
Statistical testing evaluated effects of group, listening condition, and side of sound presentation on outcome measures of sound localization, lateralization, and head displacement. Characteristics of bimodal users, defined by the PCA, were used to explore variability in these outcomes. Differences between the HA-CI and SSD-CI subgroups were also assessed in separate models for particular outcome measures. In these cases, group included three levels (control, HA-CI, SSD-CI) for each outcome measure.
Results
Characteristics of Children With Bimodal Hearing
Principal Demographic Components
Contributions from each measure on principal components (PCs) 1-3, accounting for 82.7 % of the variability in the data, are shown in Figure 1A. PC1 (38.9 % of the variance) was associated with the onset of hearing loss (age at diagnosis of hearing loss, age at cochlear implantation and pre-implant unaided PTA on the acoustic side), PC2 (26.3 % of the variance) was associated with hearing on the acoustic side (duration of deafness on the acoustic side, age at test, and pre-implant unaided PTA on the acoustic side), and PC3 (17.5 % of the variance) was associated with hearing on the implanted side (pre-implant unaided PTA on the deaf side and duration of deafness on the deaf side). In Figure 1B–D, PCs 1-3 are compared across participants and ellipses are provided for HA-CI and SSD-CI subgroup data. Clear distinctions between these subgroups are shown in PC1 which largely reflected a wider range of onset of hearing loss and better hearing in the acoustic ear in the SSD-CI than the HA-CI group. PC2 values were higher in the SSD-CI group, reflecting their better acoustic hearing, and six children in the SSD-CI subgroup showed particularly negative values in PC3, indicating poorer hearing thresholds in the CI ear. Principal Component Analysis (PCA) was performed on seven demographic measures and measures of hearing including age at test (years), age at hearing loss diagnosis (years), age at CI (years), duration of deafness (by ear) [years], and pre-implant unaided PTA (dB HL) based on pure tone audiometric thresholds measured at 500, 1000 and 2000 Hz (by ear) (
Aural Preference Measured by Speech Perception Tests
Aural preference in the bimodal group was assessed by speech perception accuracy and speech detection in noise in a spatial release from masking (SRM) test. Speech perception accuracy using the CI alone is plotted against speech perception accuracy when using the acoustic hearing ear alone in Figure 2A. Speech perception accuracy with the acoustic hearing ear tended to be better than the CI side (p = 0.05) which reflected particularly asymmetric speech perception in the SSD-CI group (p <0.001). Speech reception thresholds (dB) [SRTs] at three noise positions are shown for the control and bimodal groups (symbols delineate HA-CI and SSD-CI subgroups) in Figure 2B. SRTs were lower (better) in the control group than the CI users (p < 0.05). There was significant improvement in SRTs when noise was spatially separated from speech compared to the co-located condition [Noise left: Estimate (SE) = -7.1 (1.5), p < 0.001; Noise right: Estimate (SE) = 5.9 (1.5), p < 0.001]. In the bimodal group, there was no effect of side of implantation (p = 0.46). Secondary analyses confirmed there were no significant differences between HA-CI and SSD-CI subgroups on the SRTs (Estimate (SE) =1.05 (2.24), p = 0.88). Benefits of spatial separation are plotted in Figure 2C; there was no significant difference in SRM between bimodal listeners and controls (p = 0.60). Aural preference measured by speech perception scores between sides and SRM asymmetry (SRMA) [dB] are plotted against the PCA components (from Figure 1) in Figure 2D and E, respectively. Speech perception accuracy switches from being better in the CI (- values) to better in the acoustic ear (+ values) with further increases in preference for the acoustic hearing ear as PC1 values increased (p < 0.001). This reflected greater aural preference for the acoustic ear with later onset of hearing loss. There were no significant effects of PC2 (hearing on the acoustic side) or PC3 (hearing on the CI side) on asymmetry in speech perception scores. SRMA indicated increased release of masking when noise was to the CI side (aural preference for the acoustic side, - values) or when noise was to the CI side (aural preference for acoustic ear + values). None of the PCs were significantly associated with this measure but SSD-CI users had a large and significant preference for their normal hearing ear [MSRMA (dB) = -4.5, 95% confidence interval (-7.56, -1.44)] whereas many HA-CI users showed a preference for the CI ear [MSRMA (dB) = 1.59, 95% confidence interval (-0.29, 3.47)]. Secondary analyses of bimodal subgroup confirmed that SRMA was significantly more positive (preference for acoustic hearing ear) in the SSD-CI than HA-CI subgroups (Estimate (SE) = 4.9 (1.7), p < 0.05). (
Self-Reported Hearing Challenges
Self-reported hearing scores from the Speech, Spatial and Qualities of Hearing Scale (SSQ) are shown for the bimodal group in Figure 3A where higher scores indicate better self-reporting hearing. SSQ scores were significantly reduced for the spatial hearing subtest compared to other sub-tests [F(2, 58) = 12.67, p < 0.001]. As shown in Figure 3B, scores on the spatial subsection increased with PC1 (later onset of hearing loss) (p < 0.01). There was a trend for increased scores on the speech subsection with PC1 (p = 0.05) but no significant effects of PC2 or PC3 on these subsections. None of the PCs were significantly associated with the qualities of hearing subsections. Secondary analyses of bimodal subgroup showed that that SSQ scores were significantly higher in the SSD-CI compared to the HA-CI subgroup [Estimate (SE) = -1.26 (0.42), p < 0.01]. There was no significant interaction between subgroup and SSQ category [F(2, 54) = 0.86, p = 0.43]. Self-reported hearing scores from the Speech, Spatial and Qualities of Hearing Scale (SSQ). As shown in (
Impaired Sound Localization in Children With Bimodal Hearing
Sound localization in bimodal users was hypothesized to be poorer than normal, related to sensitivity to binaural cues, and better for children whose onset of deafness occurred later in childhood.
Impaired Stationary Sound-Source Localization
Stationary sound localization responses in children with typical hearing with and without right ear plugging are plotted by stimulus location in Figure 4A. The root-mean-square error (RMSE) [°] in Figure 4B confirms significantly greater error in bimodal listeners than peers with typical hearing [Estimate (SE) = -16.9 (4.5), p < 0.001] but no significant differences between HA-CI and SSD-CI subgroups [Estimate (SE) = 3.8 (3.5), p = 0.54]. Errors in bimodal listeners were similar to the control group when monaurally plugged [Estimate (SE) = -5.8 (5.1), p = 0.67]. For bimodal listeners, RMSE was significantly greater when sound was presented from the hemifield ipsilateral to their deaf side when their CI was off [Estimate (SE) = 12.3 (2.6), p < 0.001] but reduced on this side, becoming more symmetric between hemifields, when the CI was on [Estimate (SE) = -0.67 (2.6), p = 1.0]. This asymmetry was also found on the plugged side in the control group [Estimate (SE) = -10.1 (6.3), p = 0.75]. The benefit of removing the monaural earplug in the control group was compared to CI benefit in bimodal listeners in Figure 4C, revealing greater effects for sounds on the affected side (CI/plug) [Difference = -9.5, p < 0.01] and larger effects of monoaural plugging in the control group than of CI use in the bimodal listeners [Estimate (SE) = -22.3, p < 0.001]. There was no significant difference in errors with or without CI use between HA-CI and SSD-CI subgroups [Estimate (SE) = -4.7 (-13.6), p = 0.41]. As shown in Figure 4D, there was a significant increase in RMSE with increasing reaction time (mean across trials) for bimodal listeners (p < 0.05) but not controls (p > 0.05). Localization responses to stationary sound for bimodal listeners and peers with typical hearing (sound localization subgroup). Localization responses are plotted by stimulus location in (
Impaired Perception of Moving Sound Direction and Position
Perception of sound movement direction is plotted by stimulus location change (°) using logistic regression curves by participant in Figure 5A. Mean (SE) responses per condition are shown by the open circles. The bimodal listeners had significantly reduced direction perception acuity compared to peers with typical hearing [Difference = -0.87, p < 0.001] with no significant difference between HA-CI and SSD-CI subgroups [Estimate (SE) = 0.03 (0.3), p = 0.97]. RMSE was calculated for response position change from L1 to L2 compared to the actual position change. As shown in Figure 5B there were greater errors in bimodal listeners compared to controls [Estimate (SE) = -20.2 (5.7), p < 0.01] and no significant difference between HA-CI and SSD-CI subgroups [Estimate (SE) = 0.51 (4.5), p = 0.99]. There was no significant improvement in accuracy with/without CIs for bimodal listeners [Estimate (SE) = 0.2 (2.0), p = 0.99]. As shown in Figure 5C, there was a significant increase in RMSE of L1-L2 associated with increasing mean response time for bimodal listeners (p < 0.05) but not controls (p > 0.05). Localization of moving sound responses are plotted for bimodal listeners and controls (sound localization sub-group). (
Effects of Binaural Cue Sensitivity
Binaural cue sensitivity to interaural level difference (ILD) and interaural timing difference (ITD) cues were measured by logistic regression curves fitted for each participant. Curve slopes across ILDs, shown in Figure 6A, were reduced in CI users relative to controls [MControl (SD) = 3.3 (5.7) proportion/dB; MBimodal (SD) = 0.09 (0.09) proportion/dB]. The group differences create an association between higher ILD slopes and reduced error in stationary sound localization (p < 0.05), shown in Figure 6B, and increased slopes for perception of sound movement direction (p < 0.001), shown in Figure 6C but do not hold for the CI group alone (stationary: p = 0.25; moving: p = 0.49). ITDs, shown in Figure 6D, were significantly impaired in CI users [t(13.0) = 14.8, p < 0.001]. The group differences account for higher ITD slopes to be associated with reduced error in stationary sound localization [F(1, 17) = 18.7, p < 0.001], shown in Figure 6E, and increased slopes for perception of sound movement direction [F(1, 17) = 69.7, p < 0.001], shown in Figure 6F, but these do not hold for the CI group alone (stationary: p = 0.51; moving: p = 0.07).
Factors Contributing to Impaired Spatial Hearing in Children With Bimodal Hearing
Demographic measures and measures of hearing, reduced using PCA, in bimodal CI users were compared to: 1) errors in stationary sound localization, and 2) slope (β) of moving sound direction perception, as plotted in Figure 7. Reduced error in stationary sound localization (p < 0.05) and more acute perception of movement direction (p < 0.05) was associated with later onset of hearing loss (PC1). Aural preference for acoustic hearing, measured by SRMA, tended to increase with better hearing in the non-implanted ear (PC2) [p = 0.09], consistent with findings shown in Figure 3D. Duration of deafness in the implanted ear (PC3) was not significantly associated with any of the spatial hearing measures (p > 0.05). Principal component analysis (PCA), used to reduce demographic and hearing measures, was used to assess errors in localizing stationary sound RMSE (°) [“CI on” listening condition] and perception of moving sound direction 
Bimodal Listeners Show Abnormal Head Movements During Sound Localization
Effects of aural preference (hearing asymmetry) on sound localization were explored by testing the hypothesis that, in bimodal listeners, gaze movements favour the acoustic hearing ear more than the CI during spatial hearing tasks.
Head displacement waveforms averaged by participant are plotted in response to stationary (see Figure 8A) and moving (see Figure 8B) sound presentation. Head displacement, summarized as area under the curve (AUC) [product of displacement and time; °*s] in response to stationary (see Figure 8C) and moving (see Figure 8D) sound presentation, revealed movements to the correct hemifield of stationary sound presentation (p < 0.001) or direction of sound movement in the non-plugged control group (p < 0.01); the estimate (SE) range of head displacement was AUC = 18.27 (2.86) °*s for stationary sound and AUC = 118.6 (34.7) °*s for moving sound. Upon right ear plugging, these estimated (SE) ranges were reduced [stationary: 7.05 (2.82) °*s; moving: 5.90 (35.70) °*s], reflecting trends for less head displacement to stationary sound in the left (unplugged) hemifield (p = 0.07) and to sound moving to the left (p < 0.05). This suggests decreased tendency to use the right plugged ear for sound localization. The CI users moved their heads to favour their acoustic hearing ear for both stationary (p < 0.05) and moving sound (p < 0.05) but head displacement was not proportional to stationary sound position (p = 0.70) or degree of moving sound (p = 0.44) and there was no effect of CI use (stationary: p = 0.86; moving: p = 0.10). The waveforms in children with CIs were also less smooth than in the control group. To capture these more frequent changes, head displacement path length (PL) was calculated for stationary (Figure 8E) and moving (Figure 8F) sound presentation. PL tended to be larger in children with bimodal hearing than peers with typical hearing for both stationary [Estimate (SE) = 9.6 (9.1), p = 0.29] and moving sound [Estimate (SE) = 25.3 (19.5), p = 0.20] but there were no significant group differences. PL appeared to increase between plugged and non-plugged conditions in the control group but this change was not statistically significant [Estimate (SE) = 22.0 (10.3), p = 0.14]. Bimodal listeners showed little change in PL with their CI on versus off [stationary: Estimate (SE) = 3.6 (1.9), p = 0.23; moving: Estimate (SE) = 5.3 (4.2), p = 0.57]. There were no significant differences between HA-CI and SSD-CI subgroups on PL for either stationary (p = 0.43) or moving sound (p = 0.22). Head displacement is shown for bimodal users and peers with typical hearing (sound localization sub-group) in response to stationary (
Gaze (head + eye) displacement was also assessed with similar findings. In the control group, the normal gaze estimate (SE) range was 30.26 (4.11) °*s for stationary sound and 216.7 (67.1) °*s for moving sound which became reduced to 8.32 (4.10) °*s and 21.5 (66.1) °*s, respectively. The effect of side of acoustic ear in bimodal listeners also affected their gaze for moving [F(1, 25) = 5.70, p < 0.05] but not stationary [F(1, 25) = 0.24, p = 0.63] sound.
Discussion
The present study examined the degree of asymmetric hearing, referred to here as the presence of an aural preference, and spatial hearing in children who have a CI in one ear and acoustic hearing in the other (bimodal hearing). Results showed persistent aural preference and modest benefits of CI use for sound localization in children and adolescents with asymmetric hearing loss. Sound localization, as hypothesized, was poorer than normal, related to binaural cue sensitivity, and improved when the onset of deafness occurred later in childhood. Despite allowing these children to complete the spatial hearing tasks with unrestricted head and eye movements, the hypothesis that they would utilize gaze movements favouring their acoustic hearing ear was not supported. The spatial hearing impairments in children listening bimodally with CIs were not significantly different from those measured during acute monaural ear plugging in age-matched children with normal hearing and not significantly different between HA-CI and SSD-CI subgroups, suggesting abnormal strategies for spatial hearing when interaural cues are disrupted in both groups.
The cohort of children studied had variable onset and degree of asymmetric hearing loss with limited duration of deafness in the ear that was implanted (PCA analyses Figure 1). Aural preference for the acoustic hearing ear, measured by better speech perception scores in that ear than in the CI ear, increased with later onset of hearing loss as indicated by PC1 (Figure 2). Positive PC1 values mostly reflected children with late onset single-sided deafness (SSD), suggesting that this subgroup of bimodal listeners had a particularly strong preference for their hearing ear over the CI. Prior reports from larger cohorts of children with bimodal hearing (Bartels et al., 2026; Polonenko et al., 2018b) indicate that the CI provides benefits in the deaf ear for perception of speech both in quiet and noise. On the other hand, asymmetric hearing persists and is particularly evident in the subgroup of children with late onset SSD. This could be one reason that school-aged children with SSD show particular challenges wearing their CI consistently (Gao et al., 2026; Wener et al., 2023). Daily hours of CI use also decline with better pure tone average thresholds in the non-implanted ear (Wener et al., 2023) and are worse in children with late than early onset SSD (Gordon et al., 2023). There may be an additional effect of device use as children with SSD used their CI for ∼1 hour less per day than their peers who used a hearing aid for mild to moderate loss in the non-implanted ear (Gao et al., 2026). Despite these differences, the bimodal users (both SSD-CI and HA-CI) continue to report hearing challenges relative to data in children with normal hearing of similar ages (Alemu et al., 2025), particularly in spatial hearing on the SSQ (Figure 3). Challenges in the SSQ were also significantly related to PC1 which further demonstrated increased difficulties reported by children with late onset SSD.
Sound localization was poor for stationary sound in the CI users and improved marginally on the side of the CI when the CI was worn (Figure 4), reflecting some relief of poorer access to sound in the hemisphere of the deaf ear. Similar impairments in stationary sound localization were also found upon acute monaural ear plugging in a small group of age-matched children with normal hearing, although the benefits of removing the plug in that group were larger than the benefits of adding the CI in the bimodal listeners (Figure 4). On the other hand, the bimodal group found the task more effortful as shown by higher response times in bimodal listeners than the monaurally plugged peers which increased with increasing localization error. These results are consistent with previous reports by our group in a smaller cohort of children with SSD (Gordon et al., 2023). There have been several studies showing improved stationary sound localization with CI use in children with SSD but these paradigms often use a smaller set of speakers that are not hidden from view (Arndt et al., 2024; Arras et al., 2022). The reported spatial hearing benefits of CI in children with SSD in prior studies could thus reflect improved strategies to discriminate sound locations rather than to recognize the target location itself. This could also be true of children using bilateral CIs as they also show poorer stationary sound localization than controls when tested using the same set up as in the present study [M(SD) = 26.8° (13.1)] (Alemu et al., 2025); the descriptive data suggest that the bilateral CI users have slightly lower errors relative to the present cohort of bimodal listeners (Figure 4B: M(SD) = 32.1° (12.9)]. We note that the relationship between the experimental sound localization and lateralization paradigms used to test spatial hearing in children with CIs in the present study may not reflect their real-world spatial hearing challenges.
Prior work has shown that adults with normal hearing can reduce their spatial hearing errors in the presence of a monaural plug over a three day period (Kumpik et al., 2010; Sanchez Jimenez et al., 2023). This is consistent with evidence of binaural cue reweighting in animals with normal hearing after experimentally induced unilateral hearing loss (Keating et al., 2013, 2015). Yet, such adaptation was not explored in the present study. Testing was done with acute ear plugging in children with normal hearing, and in children with asymmetric hearing showing spatial hearing deficits at a single measurement period even after years of CI hearing experience. Particularly poor spatial hearing in bimodal listeners was shown by the smaller benefit of CI for error reduction compared with the benefit of removing the earplug in children with normal hearing (Figure 4). The limited spatial hearing in children with CIs shown here might have implications for training efforts that have reportedly had some success for promoting improved sound localization in adults (Isaiah et al., 2014).
Another concerning finding was that children with bimodal hearing with a CI showed particular challenges distinguishing between leftward and rightward moving sound (Figure 5). Direction of moving sound perception was also impaired in children with monaural ear plugs but the children with CIs had significantly higher errors which were not improved by CI use and which scaled with longer reaction times. These results are consistent with data reported from children using bilateral CIs; despite lower errors for stationary sound than the present cohort, the bilateral CI group showed similar errors to the present bimodal group when attempting to localize the direction of moving sound in the same paradigm [M(SD) = 33.7° (21.6)] (Alemu et al., 2025). Adults with bilateral CIs have also found perception of moving sound to be effortful, potentially reflecting abnormal integration of changing binaural cues as sound moves in space (Litovsky et al., 2019).
Abnormal binaural processing can occur due to poor fine temporal processing by the CI (Zirn et al., 2016) and changes to the bilateral auditory pathways when hearing loss occurs in early development (Litovsky et al., 2010) as reviewed in the introduction. Bimodal listeners have trouble accessing both ILDs and ITDs due to large mismatches in the acoustic input from one ear and the CI electrical input from the other ear (Bernstein et al., 2018; Blanks et al., 2008; Holtmann et al., 2020; Polonenko et al., 2015) and may rely on selective listening between ears as shown in adults with SSD using CIs (Bernstein & Goupell, 2025). The present cohort showed poor sensitivity to both ILDs and ITDs (Figure 6). The importance of binaural hearing during childhood was confirmed by the finding that later onset of hearing loss, measured as PC1 in the principal component analysis, afforded some reduction in stationary and moving sound localization errors (Figure 7). This aspect of hearing history had a less clear effect on aural preference. There was no relationship between the PC1 and the spatial release of masking task but PC1 was associated with more asymmetric speech perception scores (Figure 2). Thus, it appears that having some binaural processing early in life does not ensure symmetric hearing from both ears through bimodal listening but, on the other hand, is the most important factor for later spatial hearing with a CI. The lack of significant difference between HA-CI and SSD-CI subgroups on spatial hearing outcomes further supports the relative importance of early binaural hearing. This is also consistent with findings of better spatial hearing with bilateral CIs in the presence of post-lingual rather than pre-lingual onset of deafness (Litovsky et al., 2012) and of abnormal binaural processing in early deafness (Tillein et al., 2016) which is not restored by bilateral CIs (Easwar et al., 2017a; 2017b, 2018).
Analyses of head displacement quantified the normal ability to orient toward stationary or moving sound (Figure 8). In the control group, acute monaural earplugging disrupts this by reducing head turns that require more exposure of the plugged ear. This is consistent with previous findings in a larger cohort of children and adults with normal hearing who showed clear head turns to favour their non-plugged ear across both hemifields (Alemu et al., 2024b). By contrast, bimodal listeners showed longer head displacement and pathlengths than normal (Figure 8). These movements were not different for either hemifield or position of stationary sound or for direction or degree of movement for moving sound, suggesting increased uncertainty during the task with no clear strategy for improvement of sound localization through head movement. Thus, it appears that head movements in bimodal users were not providing increased access to binaural cues with or without the CI.
Overall, spatial hearing in children receiving CIs for unilateral deafness was poor with decreasing outcomes in children with early onset hearing loss. This might reflect the very limited access to interaural cues (Figure 6) from early in life. Slopes of lateralization responses to both ILDs and ITDs were reduced in children with bimodal hearing compared to the control group. These challenges likely also reflect the large mismatches in place of cochlear stimulation, timing, and level between the two ears as already reported in bimodal hearing (Bernstein et al., 2018; Blanks et al., 2008; Holtmann et al., 2020; Polonenko et al., 2015). These mismatches in input are not typically addressed during CI programming which focuses, rather, on providing audible and comfortable sound. Although suggestions for programming CIs in the presence of acoustic hearing have been made (Dillon et al., 2022; Gifford et al., 2022; Zirn et al., 2019), there are no clear protocols for setting CIs in children to improve access to binaural hearing.
Limitations of the Present Study
The paradigms in the present study identified outcomes of bimodal hearing including speech perception in quiet and noise, self-reported hearing challenges, spatial hearing, and access to binaural cues. Some limitations were noted. ILDs were created by equal changes in dB on the acoustic side as CU changes on the CI side, reflecting the difficulty matching levels along the dynamic range in bimodal users but also potentially reducing ILD sensitivity. Even so, impairments in all measures were found which might be affecting motivation and willingness by children with bimodal hearing to use their CIs as previously reported (Gordon et al., 2023). Inconsistent device use can, in turn, affect outcomes (Park et al., 2019; Phan et al., 2023). This was not addressed in the present study but these effects will need to be explored further as well as related questions regarding the “real world” benefits of CI in these children. In addition, the children in the present study used their own speech processors during most of the outcome measures. These were programmed clinically with the primary intention of providing audible and comfortable input rather than binaural/spatial hearing.
Conclusions
Data from the present study demonstrate that children with unilateral deafness achieve speech perception with their CI but hearing remains better in the non-implanted ear with significant residual hearing and late onset of hearing loss. Despite small advantages of the CI on the implanted side, spatial hearing remains challenging for both stationary and moving sound in children with SSD and children who wear a HA in the non-implanted ear. Marginally reduced errors occur with later onset of unilateral deafness, reflecting the importance of normal binaural hearing in early development. Further efforts to improve spatial hearing in children through CIs in children with unilateral deafness including SSD should focus on programming that provides better access to binaural cues and consideration of etiology and onset of deafness.
Supplemental Material
Supplemental Material - Challenges of Cochlear Implant Use for Spatial Hearing in Children and Adolescents With Asymmetric Hearing Including Single Sided Deafness
Supplemental Material for Challenges of Cochlear Implant Use for Spatial Hearing in Children and Adolescents With Asymmetric Hearing Including Single Sided Deafness by Robel Z. Alemu, Alan Blakeman, Jaina Negandhi, Hanne Bartels, Blake C. Papsin, Sharon L. Cushing, and Karen A. Gordon in Trends in Hearing.
Footnotes
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
