Spatial Release From Informational and Energetic Masking in Bimodal and Bilateral Cochlear Implant Users

Abstract

Purpose

Spatially separating speech and background noise improves speech understanding in normal-hearing listeners, an effect referred to as spatial release from masking (SRM). In cochlear implant (CI) users, SRM has often been demonstrated using asymmetric noise configurations, which maximize benefit from head shadow and the potential availability of binaural cues. In contrast, SRM in symmetrical configurations has been minimal to absent in CI users. We examined the interaction between two types of maskers (informational and energetic) and SRM in bimodal and bilateral CI users. We hypothesized that SRM would be absent or “negative” using symmetrically separated noise maskers. Second, we hypothesized that bimodal listeners would exhibit greater release from informational masking due to access to acoustic information in the non-CI ear.

Method

Participants included 10 bimodal and 10 bilateral CI users. Speech understanding in noise was tested in 24 conditions: 3 spatial configurations (S₀N₀, S₀N_45&315, S₀N_90&270) × 2 masker types (speech, signal-correlated noise) × 2 listening configurations (best-aided, CI-alone) × 2 talker gender conditions (different-gender, same-gender).

Results

In support of our first hypothesis, both groups exhibited negative SRM with increasing spatial separation. In opposition to our second hypothesis, both groups exhibited similar magnitudes of release from informational masking. The magnitude of release was greater for bimodal listeners, though this difference failed to reach statistical significance.

Conclusions

Both bimodal and bilateral CI recipients exhibited negative SRM. This finding is consistent with CI signal processing limitations, the audiologic factors associated with SRM, and known effects of behind-the-ear microphone technology. Though release from informational masking was not significantly different across groups, the magnitude of release was greater for bimodal listeners. This suggests that bimodal listeners may be at least marginally more susceptible to informational masking than bilateral CI users, though further research is warranted.

A recent study of a single-site clinical population in the United States revealed that approximately 80% of adult recipients are using a bimodal hearing configuration (cochlear implant [CI] plus contralateral hearing aid) and 20% of adult CI recipients pursue bilateral CIs (Holder et al., 2018). Thus, most modern CI recipients hear with two ears, which is known to yield significant advantages over unilateral hearing in most listening environments. The benefits of two-eared listening, however, can vary depending on listening modality—bimodal versus bilateral CI.

Head Shadow

The most common and robust two-eared effect observed for bimodal and bilateral CI populations is head shadow (Buss et al., 2008; Gifford et al., 2014, 2018; Kokkinakis, 2018; Kokkinakis & Pak, 2014; Litovsky et al., 2006; Pyschny et al., 2014; Schleich et al., 2004; Sheffield et al., 2015, 2019; van Hoesel & Tyler, 2003). Head shadow, often referred to as better ear listening, arises primarily from interaural level differences (ILDs) resulting from the physical barrier of the head. One can derive benefit from head shadow even with just one functioning ear, provided that the distracter is oriented to the side of the poorer hearing ear. Both bimodal and bilateral CI users exhibit significant speech perception benefit from head shadow; however, there are a number of studies demonstrating interaural asymmetry in aided speech recognition and/or the range of audibility for bimodal listeners, thus reducing the benefit of head shadow on the side with the poorer hearing ear (Ching et al., 2004; Dunn et al., 2005; Gifford et al., 2014, 2018; Morera et al., 2005; Potts et al., 2009; Pyschny et al., 2014). In contrast, the bilateral CI population typically exhibits less interaural asymmetry in audibility and speech recognition, thus resulting in similar benefit from head shadow regardless of masker position (Gifford & Dorman, 2019; Gifford et al., 2014, 2018); however, even among bilateral CI users, differences in benefit from head shadow with different masker position have been observed, relating to interaural frequency mismatch or speech recognition differences (Culling et al., 2012; Goupell, Stakhovskaya, & Bernstein, 2018; Goupell, Stoelb, et al., 2018; Litovsky et al., 2006, 2009).

Spatial Release From Masking

Spatial release from masking (SRM) is also a benefit afforded by two-eared hearing; however, as for head shadow, two functioning ears are not required for SRM benefit (Davis & Gifford, 2018; Gifford et al., 2014; Loizou et al., 2009; Plant et al., 2016; Pyschny et al., 2014; Sheffield et al., 2015; Williges et al., 2019). The reason is that SRM is thought to arise from a combination of head shadow and interaural comparison of binaural cues, namely, ILDs and interaural time differences (ITDs; though additional cues such as spatial attention and better ear listening can also play a role; Goupell et al., 2016). Thus, even a listener with one hearing ear can exhibit SRM benefit provided that the noise masker originates from the opposite side of the head.

The majority of studies investigating SRM in clinical populations have utilized asymmetric noise configurations (maskers positioned asymmetrically across hemifields) with the target at 0° azimuth and noise source(s) ranging from 10° to 90° (Culling et al., 2012; Davis & Gifford, 2018; Gifford et al., 2014; Litovsky et al., 2006; Loizou et al., 2009; Pyschny et al., 2014; Schleich et al., 2004; van Hoesel & Tyler, 2003). Loizou et al. (2009) assessed SRM in a cocktail party setting for bilateral CI users using three different noise sources, and all experimental noise configurations were asymmetric, such that the target was at 0° azimuth and the three noise sources originated from the following three azimuthal conditions: (a) 60°, 90°, and 330°; (b) 30°, 60°, and 90°; and (c) all three noises at 90° (Loizou et al., 2009). Loiselle et al. (2016) investigated speech understanding for bimodal and bilateral CI users using a cocktail party paradigm with different noise sources originating from 90° and 270° in a symmetrical noise configuration (maskers positioned symmetrically across hemifields). However, there was not a comparison condition in which speech and noise were colocated, thereby precluding any analysis of SRM and the relative benefit of spatial separation (Loiselle et al., 2016). Such asymmetric noise configurations stand to maximize the benefit from head shadow and the potential availability of binaural cues, which are primarily ILDs for this clinical population. Furthermore, because the output for behind-the-ear (BTE) CI processors and hearing aids is higher for sounds originating from ± 90° (e.g., Aronoff et al., 2011; Dwyer et al., 2020; Kolberg et al., 2015), the use of head shadow and ILD-based cues can be exacerbated further, thereby resulting in higher speech recognition performance in asymmetric conditions where the speech stimulus originated from 90° (Dwyer et al., 2020; Kolberg et al., 2015).

Unlike asymmetrical configurations, symmetrical noise conditions can effectively minimize the effects of head shadow and ILD-based cues, thereby allowing for an examination of SRM that depends largely on binaural timing comparisons related to their interaural correlation across the two ears (Bernstein et al., 2015). When symmetrical noise configurations are used, current literature suggests that SRM for CI users is minimal or negative (Misurelli & Litovsky, 2012, 2015). Since BTE microphones yield higher output for sounds originating from 90°, as compared to 0°, it is possible that performance may decrease with increasing spatial separation merely due to a decrease in signal-to-noise ratio (SNR; i.e., lower SNR due to ear-level masker increase at ± 90°). In other words, the decrease is not a function of signal processing limitations but, rather, may be due to an inherent consequence of microphone location. Because binaural cues such as ITDs are not relayed well to bilateral and bimodal patients, our study sought to further investigate how speech recognition differs among these two groups when symmetrically placed maskers are utilized.

Informational and Energetic Masking

In environments for which there are multiple talkers, both energetic and informational masking can significantly impact speech understanding. Energetic masking is generally considered a peripheral auditory phenomenon involving spectral overlap of target and masker(s), thereby interfering with the audibility of the target signal. In contrast, informational masking is generally considered a more central auditory phenomenon occurring when both target and masker(s) are clearly audible, but speech understanding is negatively impacted due to perceptual similarities between target and distracter(s), stimulus uncertainty, and linguistic content present in the distracter(s) (Arbogast et al., 2002; Brungart, 2001; Durlach et al., 2003; Freyman et al., 2007; Hawley et al., 2004; Ihlefeld & Shinn-Cunningham, 2008; Kidd et al., 2005; Leek et al., 1991; Lutfi, 1990).

In multitalker scenarios, listeners can use fundamental frequency (F0) as a cue to segregate target speech from multitalker babble. In the presence of one to three distracters, F0 as a source segregation cue is strongest and becomes an even more prominent cue with greater differences in F0 between target and distracter(s), resulting in greater target speech understanding (Brokx & Nooteboom, 1982; Darwin et al., 2003; Deroche & Culling, 2013; Mackersie et al., 2011). Because bimodal listeners often have acoustic access to F0 from the nonimplanted ear, bimodal listeners potentially have an advantage over bilateral CI users for speech understanding in cocktail party scenarios (Loizou et al., 1998). This was the scientific motivation for the second aim of the current study, in which we investigated the effects of energetic and informational masking for bimodal and bilateral CI users using a cocktail party experimental design with symmetrical maskers.

The current study examined SRM in bimodal and bilateral CI recipients using both informational and energetic maskers. Performance differences across groups were examined, in addition to the effects of masker type, listening configuration, and talker gender. Our primary hypothesis was that neither group would demonstrate SRM benefit for speech understanding in symmetrical noise configurations, and in fact, we predicted that, as noise location was changed to the far-left and far-right hemifield, listeners would exhibit poorer speech recognition as compared to the colocated condition. Secondarily, we hypothesized that bimodal listeners would exhibit significantly greater release from informational masking as compared to bilateral CI users due to greater acoustic F0 access from the non-CI ear.

Method

Participants

Participants included 10 adult bimodal listeners (M _age = 58 years, range: 32–79 years; five men, five women) and 10 bilateral listeners (M _age = 48 years, range: 32–65 years; four men, six women). All participants had at least 6 months of CI experience. Additional demographic information for the bimodal and bilateral groups is shown in Tables 1 and 2, respectively.

Table 1.

Demographic information for the bimodal group.

Participant	Age (years)	Manufacturer	CI ear
1	32	AB	L
2	61	AB	R
3	62	AB	L
4	72	MED-EL	L
5	61	AB	L
6	79	Cochlear	R
7	54	Cochlear	R
8	68	Cochlear	L
9	52	Cochlear	R
10	43	Cochlear	R
M	58
SD	14

Note. CI = cochlear implant; L = left; R = right.

Table 2.

Demographic information for the bilateral group.

Participant	Age (years)	Manufacturer	Test ear (CI-alone condition)
1	52	Cochlear	—
2	33	AB	L
3	59	Cochlear	R
4	40	AB	R
5	32	AB	L
6	65	AB	L
7	48	AB	L
8	44	MED-EL	R
9	52	MED-EL	L
10	53	MED-EL	L
M	48
SD	11

Note. The CI-alone test ear was inadvertently not recorded for Participant 1. CI = cochlear implant; L = left; R = right.

A hearing evaluation was completed on all participants prior to testing using a Grason-Stadler GSI 61 audiometer. The non-CI ear of bimodal participants was tested via an ER-3A insert earphone. Audiometric thresholds for the non-CI ear of the bimodal participants are shown in Figure 1. CI-aided thresholds were tested in the sound field, and thresholds for all qualifying participants were between 20 and 30 dB HL from 250 to 6000 Hz.

Figure 1.

Audiometric thresholds for the non–cochlear implant ear of bimodal participants. Individual data are represented by light gray lines. Group mean is represented by the dark line.

All procedures were explained to the participants prior to their participation, and written informed consent was obtained. Participants were compensated for their participation following completion of the study.

CI Programming

The CI settings used for testing were those used by the participants in everyday listening. The CI microphone mode for each respective manufacturer was set as follows: Advanced Bionics, omnidirectional, 100% T-mic; Cochlear, standard directional; MED-EL, omnidirectional.

Hearing Aid Fitting

A Phonak Bolero V90 BTE hearing aid with noncustom comply tip coupling was used for bimodal testing. Gain for 55, 65, and 75 dB SPL input levels was verified on-ear using the Audioscan Verifit version of the NAL-NL2 prescriptive formula. Features including acclimatization, frequency lowering, and noise reduction were deactivated. Microphone settings were set to Phonak's Real Ear Sound. The Real Ear Sound setting is designed to restore the natural directionality contribution of the pinna, which is otherwise lost due to BTE microphone placement.

Test Environment

All testing was completed in a single-walled sound-attenuated booth using a five-speaker layout in the front hemifield from 270° to 90°, with each speaker separated by 45° and positioned 24 in. (60 cm) from the participant's head similar to the test environment described by Loiselle et al. (2016). A schematic of this setup is shown in Figure 2.

Figure 2.

Schematic diagram of the loudspeaker configuration used for testing. This figure has been adapted with permission from Revit et al. (2002).

Stimuli

Speech stimuli used for testing were lists of IEEE sentences (IEEE Recommended Practice for Speech Quality Measurements, 1969). The IEEE sentence test battery consists of four different talkers—two female and two male talkers. The noise maskers used for testing consisted of continuous speech (IEEE sentences, male talkers), resulting in both informational and energetic masking, as well as signal-correlated noise (SCN), resulting in energetic masking. ¹ In all cases, the noise maskers were played continuously throughout the entire list presentation, such that there were no periods of silence in between target sentences.

Procedure

For all trials, target speech (S) originated from 0° azimuth. The noise maskers (N) were presented in three speaker configurations (0°, 45° & 315°, and 90° & 270°), thus creating three symmetrical spatial conditions (S₀N₀, S₀N_45&315, and S₀N_90&270, respectively). Target speech was always presented at 60 dB SPL (A weighted), and the masker noise level was determined individually by finding the SNR at which listeners scored approximately 50% correct in the CI-alone S₀N₀ condition, using the speech masker. The 50% correct point in the CI-alone condition was chosen to allow for a broad range of potential improvement in the remaining listening conditions and to prevent possible ceiling effects. Individual SNR was determined first by assessing speech recognition at +5 dB SNR for a single IEEE list of 10 sentences. SNR was then increased or decreased accordingly until the individual scored within ± 10 percentage points of 50%. The sentences used for this procedure were considered practice and not used for actual testing.

For all spatial conditions, participants were tested in two listening configurations: “best-aided” (bilateral or bimodal) and “CI-alone” (better ear for the bilateral CI users). For bimodal participants, the non-CI ear was plugged with an earplug for all CI-alone testing. Target speech was either female or male, and distracter sentences were always male, thus creating a “different-gender” condition and a “same-gender” condition.

In total, participants were tested in 24 conditions: 3 spatial configurations (S₀N₀, S₀N_45&315, S₀N_90&270) × 2 masker types (speech, SCN) × 2 listening configurations (best-aided, CI-alone) × 2 talker gender conditions (different-gender, same-gender). The different-gender conditions have larger differences in F0, compared to the same-gender conditions. For each condition, two randomly selected lists of IEEE sentences were presented as the target speech. Presentation order, spatial configuration, and masker type were randomized.

Data Analysis

Data analysis focused on both SRM and release from informational masking. Planned data analysis focused on both talker gender condition and each listening configuration (best-aided, CI-alone). The IBM SPSS Statistics Version 25 and GraphPad Prism 7.0 software programs were used for all statistical analyses.

SRM

SRM (in percent correct) was defined as the difference in performance achieved when the target and maskers were spatially separated (S₀N_45&315 and S₀N_90&270) as compared to the colocated condition (S₀N₀), or as compared to S₀N_45&315. Thus, the focus of our SRM analyses was on the following three comparisons: S₀N_45&315–S₀N₀, S₀N_90&270–S₀N₀, and S₀N_90&270–S₀N_45&315. A mixed-model analysis of variance (ANOVA) was performed for each respective masker type, with the between-groups factor of group (bimodal and bilateral) and the repeated-measures factor of spatial configuration (S₀N₀, S₀N_45&315, S₀N_90&270). In all analyses of SRM, the SRM data were utilized for analysis (as opposed to overall percent correct scores). It was predicted that no differences between groups would be observed and that both would exhibit negative SRM due to symmetrically positioned maskers.

Release From Informational Masking

Release from informational masking was defined as the improvement in performance achieved with the SCN masker compared with the speech maskers, for a given spatial configuration. For example, for the S₀N_45&315 spatial configuration, the bimodal group exhibited a mean sentence recognition score of 79.46% in the SCN condition and 68.22% in the speech masker condition. Thus, release from informational masking is calculated as 11.24% for the bimodal group for this spatial configuration. There were three primary goals with respect to our release from informational masking analyses. In all analyses of release from informational masking, the release from informational masking data were utilized for analysis (as opposed to overall percent correct scores). First, the authors examined the effect of group (bimodal vs. bilateral) as a function of spatial configuration. For this analysis, a mixed-model ANOVA was performed, with the between-groups factor of group (bimodal and bilateral) and the repeated-measures factor of spatial configuration (S₀N₀, S₀N_45&315, S₀N_90&270). Second, the effect of talker gender was examined using a two-way repeated-measures ANOVA, with the two factors being spatial configuration (S₀N₀, S₀N_45&315, S₀N_90&270) and talker gender condition (different-gender, same-gender). The authors predicted that greater release from informational masking would occur for the same-gender condition. Third, the effect of listening condition (best-aided vs. CI-alone) on release from informational masking was examined using a two-way repeated-measures ANOVA, with the two factors being spatial configuration (S₀N₀, S₀N_45&315, S₀N_90&270) and listening condition (best-aided, CI-alone). The primary purpose was to reveal the impact of the second ear on release from informational masking.

Overall Performance: Effect of Listening Condition

The effect of listening condition (best-aided vs. CI-alone) on speech recognition was examined using a two-way repeated-measures ANOVA, with the two factors being spatial configuration (S₀N₀, S₀N_45&315, S₀N_90&270) and listening condition (best-aided, CI-alone). In all analyses of overall performance, the overall percent correct scores were utilized for analysis. The purpose was to examine the effect of the second ear on overall performance. It was predicted that listening with two ears would result in better performance, even if the input from the second ear was of a different modality.

Overall Performance: Effect of Masker Type

The effect of masker type was analyzed using a two-way repeated-measures ANOVA, with the two factors being spatial configuration (S₀N₀, S₀N_45&315, S₀N_90&270) and masker type (speech, SCN). The purpose was to examine the effect that both masker types have on overall performance. It was predicted that overall performance would be significantly better with the SCN (energetic) masker, as opposed to speech (informational and energetic), for both groups.

Results

SRM: Combined Informational and Energetic Masking (Speech Maskers)

Best-Aided Condition: Bimodal and Bilateral CI

Analyses of SRM were completed separately for both gender conditions. Figure 3A shows the mean SRM for the different-gender condition as a function of spatial configuration. All results for the combined informational and energetic masker (speech) are indicated using circles. A mixed-model ANOVA revealed no significant effect of group, F(1, 18) = 0.01, p = .91, η_p ² = .01, or spatial configuration, F(1.29, 23.15) = 2.22, p = .15, η_p ² = .11. The interaction between group and spatial configuration was also not significant, F(1.29, 23.15) = 0.02, p = .93, η_p ² = .01.

Figure 3.

Mean and individual spatial release from masking (SRM; in %) as a function of spatial condition. Error bars represent ± 1 SEM. Bimodal and bilateral cochlear implant (CI) recipients are represented by black and gray symbols, respectively. Unfilled symbols represent speech maskers, and filled symbols represent signal-correlated noise maskers. (A) Best-aided SRM for the different-gender condition. (B) Best-aided SRM for the same-gender condition. (C) CI-alone SRM for the different-gender condition. (D) CI-alone SRM for the same-gender condition.

Figure 3B shows the mean SRM for the same-gender condition as a function of spatial configuration. A mixed-model ANOVA revealed no significant effect of group, F(1, 18) = 1.61, p = .22, η_p ² = .08, or spatial configuration, F(1.31, 23.54) = 3.05, p = .08, η_p ² = .15. The interaction between group and spatial configuration was also not significant, F(1.31, 23.54) = 0.39, p = .60, η_p ² = .02. Thus, when tested in the best-aided condition, there was no difference in SRM magnitude between groups or across spatial configuration.

CI-Alone Condition

Figure 3C shows the mean SRM for the different-gender condition as a function of spatial configuration. A mixed-model ANOVA revealed no significant effect of group, F(1, 18) = 1.18, p = .29, η_p ² = .06, or spatial configuration, F(1.15, 20.72) = 1.45, p = .25, η_p ² = .08. The interaction between group and spatial configuration was also not significant, F(1.15, 20.72) = 0.09, p = .80, η_p ² = .01.

Figure 3D shows the mean SRM for the same-gender condition as a function of spatial configuration. A mixed-model ANOVA revealed no significant effect of group, F(1, 18) = 0.14, p = .72, η_p ² = .01, although the effect of spatial configuration was significant, F(1.34, 24.17) = 12.34, p = .01, η_p ² = .41. The interaction between group and spatial configuration was not significant, F(1.34, 24.17) = 0.88, p = .39, η_p ² = .05. Follow-up pairwise comparisons on the significant effect of spatial configuration revealed a significant difference in SRM magnitude for the S₀N_90&270–S₀N_45&315 versus S₀N_45&315–S₀N₀ comparison (p = .01) and the S₀N_90&270–S₀N_45&315 versus S₀N_90&270–S₀N₀ comparison (p = .01).

Thus, among both groups, when tested CI-alone, there was no difference in SRM magnitude between groups or across spatial configuration for the different-gender condition. For the same-gender condition, there was no difference in SRM magnitude between groups, but there was a difference for spatial configuration (S₀N_90&270–S₀N_45&315 versus S₀N_45&315–S₀N₀ and S₀N_90&270–S₀N_45&315 versus S₀N_90&270–S₀N₀). Furthermore, compared to the colocated condition, negative SRM was observed across groups. These findings were unexpected and raised concern that the nonimplanted ear may not have been fully occluded (despite the use of an earplug), thereby resulting in greater noise interference. To account for that possibility, the relationship between individual SRM and unaided low-frequency pure-tone average (LFPTA) was examined. LFPTA was defined as the average of thresholds at 125, 250, and 500 Hz. There was no significant correlation between LFPTA and SRM for the different-gender condition in the presence of speech maskers (r = .24, p = .51) or between LFPTA and SRM for the same-gender condition in the presence of speech maskers (r = −.40, p = .25).

SRM: Energetic Masking (SCN Maskers)

Best-Aided Condition: Bimodal and Bilateral CI

Figure 3A shows the mean SRM for the different-gender condition as a function of spatial configuration. All results for the energetic masker (SCN) are indicated using squares. A mixed-model ANOVA revealed no significant effect of group, F(1, 18) = 0.01, p = .95, η_p ² = .01, or spatial configuration, F(1.271, 22.881) = 1.55, p = .23, η_p ² = .08. The interaction between group and spatial configuration was not significant, F(1.271, 22.881) = 0.02, p = .94, η_p ² = .01.

Figure 3B shows the mean SRM for the same-gender condition as a function of spatial configuration. A mixed-model ANOVA revealed no significant effect of group, F(1, 18) = 0.01, p = .97, η_p ² = .01, although the effect of spatial condition was significant, F(1.167, 21.001) = 4.27, p = .05, η_p ² = .19. The interaction between group and spatial condition was not significant, F(1.167, 21.001) = 0.00, p = .99, η_p ² = .00. Follow-up pairwise comparisons on the significant effect of spatial configuration revealed a significant difference in SRM magnitude for the S₀N_90&270–S₀N_45&315 versus S₀N_90&270–S₀N₀ comparison (p = .02).

Thus, when tested in the best-aided condition, there was no difference in SRM magnitude between groups or across spatial configuration for the different-gender condition. For the same-gender condition, there was no difference in SRM magnitude between groups, but there was for spatial configuration (S₀N_90&270–S₀N_45&315 vs. S₀N_45&315–S₀N₀).

CI-Alone Condition

Figure 3C shows the mean SRM for the different-gender condition as a function of spatial configuration. A mixed-model ANOVA revealed no significant effect of group for the CI-alone condition, F(1, 18) = 0.86, p = .37, η_p ² = .05, although the effect of spatial configuration was significant, F(1.417, 25.511) = 12.95, p = .01, η_p ² = .42. The interaction between group and spatial configuration was not significant, F(1.417, 25.511) = 0.16, p = .78, η_p ² = .01. Follow-up pairwise comparisons on the significant effect of spatial configuration revealed a significant difference in SRM magnitude for the S₀N_90&270–S₀N_45&315 versus S₀N_45&315–S₀N₀ comparison (p = .01) and the S₀N_90&270–S₀N_45&315 versus S₀N_90&270–S₀N₀ comparison (p = .01).

Figure 3D shows the mean SRM for the same-gender condition as a function of spatial configuration. A mixed-model ANOVA revealed a significant effect of group for the CI-alone condition, F(1, 18) = 8.33, p = .01, η_p ² = .32, as well as spatial configuration, F(1.177, 21.188) = 4.56, p = .04, η_p ² = .20. The interaction between group and spatial configuration was not significant, F(1.177, 21.188) = 0.73, p = .42, η_p ² = .04. Follow-up pairwise comparisons on the significant effect of spatial configuration revealed a significant difference in SRM magnitude for the S₀N_90&270–S₀N_45&315 versus S₀N_90&270–S₀N₀ comparison (p = .01).

Thus, among both groups, when tested CI-alone, there was no difference in SRM magnitude between groups, but there was across spatial configuration (different-gender condition: S₀N_90&270–S₀N_45&315 vs. S₀N_45&315–S₀N₀ and S₀N_90&270–S₀N_45&315 vs. S₀N_90&270–S₀N₀; same-gender condition: S₀N_90&270–S₀N_45&315 vs. S₀N_90&270–S₀N₀). Furthermore, compared to the colocated condition, negative SRM was observed across groups. As previously mentioned, it is possible that the observed decrement in performance with increasing spatial separation was the result of the acoustic ear not being fully occluded (despite the use of an earplug). To account for that possibility, the relationship between performance decrement and unaided LFPTA was examined. There was no correlation between LFPTA and SRM for the different-gender condition in the presence of SCN (r = −.19, p = .60) or between LFPTA and SRM for the same-gender condition in the presence of SCN (r = −.01, p = .98).

Release From Informational Masking

Best-Aided Condition: Bimodal and Bilateral CI

Best-aided release from informational masking (in percentage points) for the different-gender condition as a function of spatial configuration is shown in Figure 4A. Mean release from informational masking in the different-gender condition was 11.30 percentage points (SD = 10.25) for the bimodal group and 10.06 percentage points (SD = 8.25) for the bilateral group. A mixed-model ANOVA revealed no significant effect of group, F(1, 18) = 0.16, p = .70, η_p ² = .01, or spatial condition, F(2, 36) = 0.27, p = .77, η_p ² = .02. The interaction between group and spatial configuration was also not significant, F(2, 36) = 0.36, p = .70, η_p ² = .02.

Figure 4.

Mean and individual release from informational masking (in %) as a function of spatial condition. Error bars represent ± 1 SEM. Bimodal and bilateral cochlear implant (CI) recipients are represented by black and gray symbols, respectively. Unfilled symbols represent the best-aided condition, and filled symbols represent the CI-alone condition. (A) Release from informational masking for the different-gender condition. (B) Release from informational masking for the same-gender condition.

Best-aided release from informational masking (in percentage points) for the same-gender condition as a function of spatial configuration is shown in Figure 4B. Mean release from informational masking in the same-gender condition was 22.58 percentage points (SD = 10.87) for the bimodal group and 18.05 percentage points (SD = 11.72) for the bilateral group. A mixed-model ANOVA revealed no significant effect of group, F(1, 18) = 1.18, p = .29, η_p ² = .06, although the effect of spatial configuration was significant, F(2, 36) = 4.93, p = .01, η_p ² = .22. The interaction between group and spatial condition was not significant, F(2, 36) = 0.99, p = .38, η_p ² = .05. Follow-up pairwise comparisons on the main effect of spatial configuration revealed a significant difference between S₀N_45&315 and S₀N_90&270 (p = .03) and S₀N₀ and S₀N_90&270 (p = .03).

Importantly, there was no difference across groups for release from informational masking when tested in the best-aided condition. For the same-gender condition, there was an effect of spatial configuration, such that greater release from informational masking was exhibited in the S₀N_90&270 condition when compared to the colocated condition and S₀N_45&315.

CI-Alone Condition

CI-alone release from informational masking (in percentage points) for the different-gender condition as a function of spatial configuration is shown in Figure 4A. Mean release from informational masking in the different-gender condition was 16.07 percentage points (SD = 11.31) for the bimodal group and 13.51 percentage points (SD = 11.05) for the bilateral group. A mixed-model ANOVA revealed no significant effect of group, F(1, 18) = 0.48, p = .50, η_p ² = .03, or spatial configuration, F(2, 36) = 0.69, p = .51, η_p ² = .04. The interaction between group and spatial configuration was also not significant, F(2, 36) = 0.02, p = .98, η_p ² = .00.

CI-alone release from informational masking (in percentage points) for the same-gender condition as a function of spatial configuration is shown in Figure 4B. Mean release from informational masking in the same-gender condition was 26.62 percentage points (SD = 13.01) for the bimodal group and 25.04 percentage points (SD = 19.53) for the bilateral group. A mixed-model ANOVA revealed no significant effect of group, F(1, 18) = 0.07, p = .80, η_p ² = .00, although the effect of spatial configuration was significant, F(2, 36) = 3.21, p = .05, η_p ² = .15. The interaction between group and spatial configuration was not significant, F(2, 36) = 0.92, p = .41, η_p ² = .05. Follow-up pairwise comparisons on the main effect of spatial configuration revealed a significant difference between S₀N₀ and S₀N_45&315 (p = .01).

Thus, there was no difference across groups for release from informational masking when tested in the CI-alone condition. For the same-gender condition, there was an effect of spatial configuration, such that greater release from informational masking was exhibited in the S₀N_45&315 condition when compared to the colocated condition.

Release From Informational Masking: Effect of Talker Gender

Best-Aided Condition: Bimodal and Bilateral CI

Effect of talker gender was analyzed using a two-way repeated-measures ANOVA, with the two factors being spatial configuration (S₀N₀, S₀N_45&315, S₀N_90&270) and talker gender condition (different-gender, same-gender). For the bimodal group, analysis revealed a significant effect of spatial configuration, F(2, 18) = 3.99, p = .04, η_p ² = .31, and talker gender condition, F(1, 9) = 11.45, p = .01, η_p ² = .56. The interaction between spatial configuration and talker gender condition was not significant, F(2, 18) = 1.92, p = .18, η_p ² = .18. Follow-up pairwise comparisons on the main effect of spatial configuration revealed a significant difference between S₀N₀ and S₀N_90&270 (p = .05). For the bilateral group, analysis revealed no significant effect of spatial configuration, F(2, 18) = 1.09, p = .39, η_p ² = .11, but there was a significant effect of talker gender condition, F(1, 9) = 6.86, p = .03, η_p ² = .43. The interaction between spatial configuration and talker gender condition was not significant, F(2, 18) = 0.35, p = .71, η_p ² = .04.

When tested in the best-aided condition, both groups demonstrated significantly greater release from informational masking for the same-gender condition compared with the different-gender condition. For the bimodal group, there was also a significant effect of spatial configuration, such that greater release from informational masking was exhibited in the S₀N_90&270 condition when compared to the colocated condition.

CI-Alone Condition

For the bimodal group, analysis revealed no significant effect of spatial configuration, F(2, 18) = 1.38, p = .28, η_p ² = .13, but there was a significant effect of talker gender condition, F(1, 9) = 7.83, p = .02, η_p ² = .47. The interaction between spatial configuration and talker gender condition was not significant, F(2, 18) = 1.36, p = .28, η_p ² = .13. For the bilateral group, analysis revealed no significant effect of spatial configuration, F(2, 18) = 0.60, p = .56, η_p ² = .06, but there was a significant effect of talker gender condition, F(1, 9) = 13.58, p = .01, η_p ² = .60. The interaction between spatial configuration and talker gender condition was not significant, F(2, 18) = 3.66, p = .05, η_p ² = .29. When tested CI-alone, both groups demonstrated significantly greater release from informational masking for the same-gender condition compared with the different-gender condition.

Release From Informational Masking: Effect of Listening Condition

Bimodal Versus CI-Alone

The effect of listening condition (best-aided vs. CI-alone) on release from informational masking was examined using a two-way repeated-measures ANOVA, with the two factors being spatial configuration (S₀N₀, S₀N_45&315, S₀N_90&270) and listening condition (best-aided, CI-alone). For the different-gender condition, analysis revealed no significant effect of spatial configuration, F(2, 18) = 0.07, p = .93, η_p ² = .01, or listening condition, F(1, 9) = 2.75, p = .13, η_p ² = .23. The interaction between spatial configuration and listening condition was also not significant, F(2, 18) = 0.63, p = .55, η_p ² = .07. For the same-gender condition, analysis revealed no significant effect of spatial configuration, F(2, 18) = 2.15, p = .15, η_p ² = .19, or listening condition, F(1, 9) = 1.55, p = .24, η_p ² = .15. However, the interaction between spatial configuration and listening condition was significant, F(2, 18) = 4.72, p = .02, η_p ² = .34. Follow-up pairwise comparisons on the significant interaction effect revealed a significant difference between S₀N₀ and S₀N_90&270 (p = .01) and between S₀N_45&315 and S₀N_90&270 (p = .05) in the bimodal condition, and between S₀N₀ and S₀N_45&315 (p = .01) in the CI-alone condition. Furthermore, for the S₀N_45&315 spatial configuration, there was a significant difference between listening condition (p = .01).

Thus, for the same-gender condition, there was a significant effect of spatial configuration, though this was dependent upon listening condition. Furthermore, release from informational masking in the same-gender condition was greater for CI-alone listening, as compared to bimodal, for the S₀N_45&315 spatial configuration specifically.

Bilateral Versus CI-Alone

For the different-gender condition, analysis revealed no significant effect of spatial configuration, F(2, 18) = 0.72, p = .50, η_p ² = .07, or listening condition, F(1, 9) = 2.45, p = .15, η_p ² = .21. The interaction between spatial configuration and listening condition was also not significant, F(2, 18) = 0.05, p = .95, η_p ² = .01. For the same-gender condition, analysis revealed no significant effect of spatial configuration, F(2, 18) = 1.61, p = .23, η_p ² = .15, or listening condition, F(1, 9) = 2.25, p = .17, η_p ² = .20. The interaction between spatial configuration and listening condition was also not significant, F(2, 18) = 1.62, p = .23, η_p ² = .15.

Overall Performance: Effect of Listening Condition (Speech Maskers)

Bimodal Versus CI-Alone

The effect of listening condition (bimodal vs. CI-alone) on speech recognition was examined using a two-way repeated-measures ANOVA, with the two factors being spatial configuration (S₀N₀, S₀N_45&315, S₀N_90&270) and listening condition (bimodal, CI-alone). Figure 5A shows the mean speech recognition scores for both listening conditions (bimodal, CI-alone) for the different- and same-gender conditions using the speech masker. For the different-gender condition, analysis revealed no significant effect of spatial configuration, F(2, 18) = 1.20, p = .32, η_p ² = .12, but the effect of listening condition was significant, F(1, 9) = 11.53, p = .01, η_p ² = .56. The interaction between spatial configuration and listening condition was not significant, F(2, 18) = 2.40, p = .12, η_p ² = .21. For the same-gender condition, analysis revealed a significant effect of spatial configuration, F(2, 18) = 11.32, p = .01, η_p ² = .56, and listening condition, F(1, 9) = 12.25, p = .01, η_p ² = .58. Follow-up pairwise comparisons on the significant effect of spatial configuration revealed a significant difference between S₀N₀ and S₀N_45&315 (p = .01) and between S₀N₀ and S₀N_90&270 (p = .03). The interaction between spatial configuration and listening condition was not significant, F(2, 18) = 2.83, p = .09, η_p ² = .24.

Figure 5.

Mean and individual speech recognition scores as a function of spatial condition for the different- and same-gender conditions. Error bars represent ± 1 SEM. Different-gender condition is represented in black; same-gender condition is represented in gray. Unfilled symbols represent the best-aided condition, and filled symbols represent the cochlear implant–alone condition. (A) Bimodal group scores using the speech masker. (B) Bilateral group scores using the speech masker. (C) Bimodal group scores using the signal-correlated noise (SCN) masker. (D) Bilateral group scores using the SCN masker.

Thus, the addition of the acoustic hearing ear resulted in significantly better performance when compared to CI-alone listening for both gender conditions in the presence of the speech masker. For the same-gender condition, there was also a significant effect of spatial configuration, such that poorer performance was exhibited in the S₀N_45&315 and S₀N_90&270 conditions when compared to the colocated condition.

Bilateral Versus CI-Alone

The effect of listening condition (bilateral vs. CI-alone) on speech recognition was examined using a two-way repeated-measures ANOVA, with the two factors being spatial configuration (S₀N₀, S₀N_45&315, S₀N_90&270) and listening condition (bilateral, CI-alone). Figure 5B shows the mean speech recognition scores for both listening conditions (bilateral, CI-alone) for the different- and same-gender conditions using the speech masker. For the different-gender condition, analysis revealed no significant effect of spatial configuration, F(2, 18) = 1.94, p = .17, η_p ² = .18, but the effect of listening condition was significant, F(1, 9) = 19.91, p = .01, η_p ² = .69. The interaction between spatial configuration and listening condition was not significant, F(2, 18) = 1.80, p = .19, η_p ² = .17. For the same-gender condition, analysis revealed no significant effect of spatial configuration, F(2, 18) = 3.42, p = .06, η_p ² = .28, but the effect of listening condition was significant, F(1, 9) = 14.26, p = .01, η_p ² = .61. The interaction between spatial configuration and listening condition was not significant, F(2, 18) = 0.78, p = .47, η_p ² = .08. Thus, the addition of the second CI ear resulted in significantly better performance when compared to CI-alone listening for both gender conditions in the presence of the speech masker.

Effect of Listening Condition (SCN Maskers)

Bimodal Versus CI-Alone

The effect of listening condition (bimodal vs. CI-alone) on speech recognition was examined using a two-way repeated-measures ANOVA, with the two factors being spatial configuration (S₀N₀, S₀N_45&315, S₀N_90&270) and listening condition (bimodal, CI-alone). Figure 5C shows the mean speech recognition scores for both listening conditions (bimodal, CI-alone) for the different- and same-gender conditions using the SCN masker. For the different-gender condition, analysis revealed no significant effect of spatial configuration, F(2, 18) = 2.30, p = .13, η_p ² = .20, but the effect of listening condition was significant, F(1, 9) = 7.44, p = .02, η_p ² = .45. The interaction between spatial configuration and listening condition was not significant, F(2, 18) = 2.01, p = .16, η_p ² = .18. For the same-gender condition, analysis revealed a significant effect of spatial configuration, F(2, 18) = 8.71, p = .01, η_p ² = .49, and listening condition, F(1, 9) = 22.79, p = .01, η_p ² = .72. Follow-up pairwise comparisons on the significant effect of spatial configuration revealed a significant difference between S₀N₀ and S₀N_45&315 (p = .01) and between S₀N₀ and S₀N_90&270 (p = .01). The interaction between spatial configuration and listening condition was not significant, F(2, 18) = 2.85, p = .08, η_p ² = .24.

Thus, the addition of the acoustic hearing ear resulted in significantly better performance when compared to CI-alone listening for both gender conditions in the presence of the SCN masker. For the same-gender condition, there was also a significant effect of spatial configuration, such that poorer performance was exhibited in the S₀N_45&315 and S₀N_90&270 conditions when compared to the colocated condition.

Bilateral Versus CI-Alone

The effect of listening condition (bilateral vs. CI-alone) on speech recognition was examined using a two-way repeated-measures ANOVA, with the two factors being spatial configuration (S₀N₀, S₀N_45&315, S₀N_90&270) and listening condition (bilateral, CI-alone). Figure 5D shows the mean speech recognition scores for both listening conditions (bilateral, CI-alone) for the different- and same-gender conditions using the SCN masker. For the different-gender condition, analysis revealed a significant effect of spatial configuration, F(2, 18) = 8.94, p = .01, η_p ² = .50, and listening condition, F(1, 9) = 13.25, p = .01, η_p ² = .60. Follow-up pairwise comparisons on the significant effect of spatial configuration revealed a significant difference between S₀N₀ and S₀N_45&315 (p = .01) and between S₀N₀ and S₀N_90&270 (p = .04). The interaction between spatial configuration and listening condition was not significant, F(2, 18) = 1.03, p = .38, η_p ² = .10. For the same-gender condition, analysis revealed no significant effect of spatial configuration, F(2, 18) = 1.95, p = .17, η_p ² = .18, but the effect of listening condition was significant, F(1, 9) = 16.09, p = .01, η_p ² = .64. The interaction between spatial configuration and listening condition was not significant, F(2, 18) = 0.43, p = .66, η_p ² = .05.

Thus, the addition of the second CI ear resulted in significantly better performance when compared to CI-alone listening for both gender conditions in the presence of the SCN masker. For the different-gender condition, there was also a significant effect of spatial configuration, such that poorer performance was exhibited in the S₀N_45&315 and S₀N_90&270 conditions when compared to the colocated condition.

Overall Performance: Effect of Masker Type

Best-Aided Condition: Bimodal

Figures 6A and 6B show the mean sentence recognition scores for the speech and SCN masker types in the different- and same-gender conditions, respectively. Effect of masker type was analyzed using a two-way repeated-measures ANOVA, with the two factors being spatial configuration (S₀N₀, S₀N_45&315, S₀N_90&270) and masker type (speech, SCN). For the different-gender condition, analysis revealed a significant effect of spatial configuration, F(2, 18) = 5.33, p = .02, η_p ² = .37, and masker type, F(1, 9) = 19.46, p = .01, η_p ² = .68. The interaction between spatial configuration and masker type was not significant, F(2, 18) = 0.28, p = .76, η_p ² = .03. Follow-up pairwise comparisons on the main effect of spatial configuration revealed a significant difference between S₀N₀ and S₀N_90&270 (p = .01). For the same-gender condition, analysis revealed a significant effect of spatial configuration, F(2, 18) = 9.02, p = .01, η_p ² = .50, and masker type, F(1, 9) = 54.81, p = .01, η_p ² = .86. The interaction between spatial configuration and masker type was also significant, F(2, 18) = 6.62, p = .01, η_p ² = .42. Follow-up pairwise comparisons on the significant interaction effect revealed a significant difference across masker types at all three spatial configurations (p = .01). Furthermore, for both masker types, there was a significant decrease in performance between S₀N₀ and S₀N_45&315 (p = .05) and between S₀N₀ and S₀N_90&270 (p = .03).

Figure 6.

Mean and individual speech recognition scores as a function of spatial condition for both masker types. Error bars represent ± 1 SEM. Scores with the speech masker are represented in black; scores with the signal-correlated noise masker are represented in gray. Unfilled symbols represent the best-aided condition, and filled symbols represent the cochlear implant–alone condition. (A) Bimodal group scores for the different-gender condition. (B) Bimodal group scores for the same-gender condition. (C) Bilateral group scores for the different-gender condition. (D) Bilateral group scores for the same-gender condition.

Thus, for both gender conditions, the bimodal group performed significantly better with the SCN masker than with speech. Furthermore, regardless of gender condition, performance decreased with increasing spatial separation when compared to the colocated condition.

Best-Aided Condition: Bilateral CI

Figures 6C and 6D show the mean sentence recognition scores for the speech and SCN masker types in the different- and same-gender conditions, respectively. For the different-gender condition, analysis revealed no significant effect of spatial configuration, F(2, 18) = 3.61, p = .05, η_p ² = .29, but the effect of masker type was significant, F(1, 9) = 32.03, p = .01, η_p ² = .78. The interaction between spatial configuration and masker type was not significant, F(2, 18) = 0.35, p = .71, η_p ² = .04. Follow-up pairwise comparisons on the main effect of spatial configuration revealed a significant difference between S₀N₀ and S₀N_90&270 (p = .01). For the same-gender condition, analysis revealed no significant effect of spatial configuration, F(2, 18) = 2.18, p = .14, η_p ² = .50, but the effect of masker type was significant, F(1, 9) = 40.03, p = .01, η_p ² = .82. The interaction between spatial configuration and masker type was not significant, F(2, 18) = 0.94, p = .41, η_p ² = .09. Thus, for both gender conditions, the bilateral group performed significantly better with the SCN masker than with speech.

CI-Alone Condition: Bimodal

Figures 6A and 6B show the mean sentence recognition scores for the speech and SCN masker types in the different- and same-gender conditions, respectively. For the different-gender condition, analysis revealed no significant effect of spatial configuration, F(2, 18) = 1.67, p = .22, η_p ² = .16, but the effect of masker type was significant, F(1, 9) = 55.19, p = .01, η_p ² = .86. The interaction between spatial configuration and masker type was not significant, F(2, 18) = 0.31, p = .74, η_p ² = .03. For the same-gender condition, analysis revealed a significant effect of spatial configuration, F(2, 18) = 12.12, p = .01, η_p ² = .57, and masker type, F(1, 9) = 107.68, p = .01, η_p ² = .92. The interaction between spatial configuration and masker type was not significant, F(2, 18) = 2.08, p = .15, η_p ² = .19. Follow-up pairwise comparisons on the main effect of spatial configuration revealed a difference between S₀N₀ and S₀N_45&315 (p = .01) and S₀N₀ and S₀N_90&270 (p = .01).

Thus, for both gender conditions, the bimodal group when tested CI-alone performed significantly better with the SCN masker than with speech. Furthermore, performance decreased with increasing spatial separation when compared to the colocated condition for the same-gender condition, but not for the different-gender condition.

CI-Alone Condition: Bilateral

Figures 6C and 6D show the mean sentence recognition scores for the speech and SCN masker types in the different- and same-gender conditions, respectively. For the different-gender condition, analysis revealed a significant effect of spatial configuration, F(2, 18) = 5.32, p = .02, η_p ² = .37, and masker type, F(1, 9) = 20.19, p = .01, η_p ² = .69. The interaction between spatial configuration and masker type was not significant, F(2, 18) = 0.45, p = .65, η_p ² = .05. Follow-up pairwise comparisons on the main effect of spatial configuration revealed a difference between S₀N₀ and S₀N_45&315 (p = .04). For the same-gender condition, analysis revealed no significant effect of spatial configuration, F(2, 18) = 2.18, p = .14, η_p ² = .20, but the effect of masker type was significant, F(1, 9) = 20.75, p = .01, η_p ² = .70. The interaction between spatial configuration and masker type was not significant, F(2, 18) = 2.05, p = .16, η_p ² = .19.

Thus, for both gender conditions, the bilateral group when tested CI-alone performed significantly better with the SCN masker than with speech. Furthermore, performance decreased with increasing spatial separation when compared to the colocated condition for the different-gender condition, but not for the same-gender condition.

Discussion

SRM depends on both binaural and monaural mechanisms. Bilateral CI users do not have access to the binaural cues necessary for precise across-ear comparisons; in addition, the performance ability of each ear can be very different. This is also true for bimodal patients; however, bimodal listeners stand to benefit from greater acoustic access to F0, whereas bilateral CI users do not. In the current study, differences between bimodal and bilateral speech recognition performance were examined in the context of two phenomena: SRM and release from informational masking. Test variables included three spatial configurations (S₀N₀, S₀N_45&315, S₀N_90&270), two masker types (speech, SCN), two listening conditions (best-aided, CI-alone), and two talker gender conditions (different-gender, same-gender). Results will be discussed with respect to group differences on these phenomena and the various conditions of study.

SRM in Best-Aided (Bimodal or Bilateral) and CI-Alone Listening Conditions

We hypothesized that bimodal and bilateral CI users would observe no SRM benefit for speech understanding in the symmetrical noise configurations utilized here and would, in fact, exhibit poorer speech recognition as the maskers approached 90° (e.g., Aronoff et al., 2011; Dwyer et al., 2020; Kolberg et al., 2015; Misurelli & Litovksy, 2012). In support of our hypothesis, neither group demonstrated SRM for either the best-aided or CI-alone conditions; rather, both groups exhibited a decrease in performance with increasing spatial separation between the target and symmetrically positioned maskers. This finding was exhibited for each talker gender condition (different-gender, same-gender), each masker type (speech, SCN), and each listening condition (best-aided, CI-alone).

SRM is believed to result predominantly from head shadow and binaural ILD and ITD cues. Thus, with spatial separation, normal binaural processing allows listeners to compare differences in the timing and intensity of sounds originating from different azimuths. These binaural cues thereby help the listener distinguish target from masker. This is in contrast to the processing afforded via current CI technology. Both bilateral CI and bimodal fittings lack binaural coordination between devices (e.g., lack of coordination re: sampling time, onset time), and consequently, access to binaural cues is inherently limited (Aronoff et al., 2010; Kan & Litovsky, 2015; Laback et al., 2015; Misurelli & Litovsky, 2012). Furthermore, as CIs utilize envelope-based strategies, fine structure information, which is useful for relaying low-frequency ITDs, is discarded. While head shadow cues remain robust since they are monaural in nature, ILDs are typically reduced due to across-ear differences in amplitude processing (Aronoff et al., 2010; Laback et al., 2015; Lenssen et al., 2011; Loiselle et al., 2016; Seeber & Fastl, 2008; van Hoesel & Tyler, 2003), and envelope ITDs (Francart et al., 2014; Laback et al., 2004) are minimal to nonexistent due to the lack of across-ear timing coordination. Furthermore, most ear-level processors have microphone placement at the top of the pinna, which has been shown to yield higher output for sounds originating from the side (e.g., 90°) as compared to the front (0°; Dwyer et al., 2020; Kolberg et al., 2015). As such, microphone placement for BTE hearing aids and CI processors has been shown to result in poorer outcomes for noise originating from the side of the head as compared to the front (Festen & Plomp, 1986). Thus, with symmetrical maskers, it is likely that the fixed SNR utilized here actually resulted in a lower (i.e., poorer) effective SNR in spatially separated conditions, as compared to the colocated condition.

In the CI literature, most of the existing SRM studies have utilized asymmetrical noise configurations (Culling et al., 2012; Davis & Gifford, 2018; Gifford et al., 2014; Litovsky et al., 2006; Loizou et al., 2009; Pyschny et al., 2014; Schleich et al., 2004; van Hoesel & Tyler, 2003). When the masker(s) is positioned asymmetrically, monaural cues may dominate (e.g., head shadow). As such, head shadow cues are considered largely responsible for the observed SRM reported in these studies (Loizou et al., 2009; Misurelli & Litovsky, 2015)—though the utility of additional factors such as spatial attention and better ear listening has also been shown, particularly when the target and interferer are presented to opposite ears (Goupell et al., 2016). It is also possible that the bimodal listeners—with acoustic access to F0 and temporal fine structure—could have benefited from better ear glimpsing (e.g., Brungart & Iyer, 2012; Cooke, 2006; Glyde et al., 2013; Hu et al., 2018). While asymmetrical configurations have been used most commonly in the CI literature, their ecological validity is limited. That is, noise presented to only one side of the listener or to discrete yet asymmetrical locations around the head is somewhat contrived; in real-world acoustic environments, background noise is largely expected to surround the listener, even if not symmetrically placed.

This study sought to utilize a more realistic configuration, in which noise maskers were presented symmetrically. In contrast to an asymmetrical presentation, symmetrical noise places greater weight on ITD cues since level-related cues are minimal, if not completely eliminated. Given the current CI processing limitations and their effect on binaural processing as well as BTE processor microphone location, the negative SRM exhibited in this study was not unexpected. This finding is also consistent with the existing literature, of which there are only two studies to our knowledge. Misurelli and Litovsky (2012, 2015) examined SRM in pediatric bilateral CI recipients using both asymmetrical and symmetrical noise configurations. SRM was present among bilateral CI users in the asymmetrical configurations (albeit reduced compared to normal-hearing listeners), but similar to our findings, SRM in the symmetrical configurations was minimal to absent, and in fact, several participants exhibited negative SRM consistent with the current study.

Misurelli and Litovsky (2012, 2015) did not examine bimodal listeners, and thus, our bimodal data extend the existing body of literature. Our data suggest that bimodal recipients likewise exhibit negative SRM and, in fact, perform similarly—both in magnitude and direction—to bilateral CI users with respect to SRM in the symmetrical noise configurations utilized in this study. As with bilateral CI recipients, bimodal listeners have a similar lack of binaural coordination across devices (e.g., re: sampling time, onset time), thus precluding the ability for precise binaural comparisons (Aronoff et al., 2010; Kan & Litovsky, 2015; Laback et al., 2015; Misurelli & Litovsky, 2012). To that end, it is likely that the mechanisms responsible for the pattern of SRM demonstrated here are similar across groups.

Taken together, when monaural head shadow cues are able to dominate (asymmetrical noise configurations), SRM has been demonstrated among CI users. However, when listeners are forced to make binaural comparisons (symmetrical noise configurations), not only is SRM nonexistent but spatial separation may actually degrade performance. In the current study, both bimodal listeners and bilateral CI recipients exhibited negative SRM as speech recognition performance was notably poorer with spatial separation of signal and maskers as compared to the colocated condition. This finding is consistent with both the limitations of current CI signal processing, the mechanistic contributors to SRM in this population (Misurelli & Litovsky, 2012, 2015), and known microphone effects for BTE hearing technology (e.g., Aronoff et al., 2011; Dwyer et al., 2020; Kolberg et al., 2015). If the necessary binaural cues are not accessible, SRM cannot be achieved.

Release From Informational Masking

Both groups of listeners exhibited significant release from informational masking in the best-aided condition. As expected, the magnitude of release from informational masking was significantly greater for the same-gender condition as compared to the different-gender condition, with increasing spatial separation in both groups. Recall that we hypothesized that bimodal listeners would exhibit significantly greater release from informational masking compared to bilateral CI users. The reason is that the nonimplanted ear of bimodal listeners typically affords access to acoustic F0 and temporal fine structure; thus, bimodal listeners are potentially more susceptible to informational masking. In contrast, our data indicate that bimodal and bilateral listeners exhibited similar magnitudes of release from informational masking, and this was true for both same- and different-gender conditions. It is worth noting that the magnitude of release from informational masking was greater for bimodal listeners (M = 17.4, 95% CI [14.4, 20.4]) as compared to bilateral recipients (M = 14.1, 95% CI [11.3, 16.9]) in all three spatial configurations and for both gender conditions, though the difference between groups failed to reach statistical significance. These findings suggest that bimodal listeners may be at least marginally more susceptible to informational masking than bilateral CI users, though further research is warranted.

This between-groups finding for similar rates of release from informational masking was not consistent with the results reported by Pyschny et al. (2014), who showed significantly greater release from informational masking for bimodal listeners as compared to bilateral CI users. However, there were notable differences across studies. First, Pyschny et al. used asymmetric noise configurations, resulting in greater better ear listening effects and, hence, greater overall speech recognition with spatial separation, irrespective of masker type. Second, the speech maskers used by Pyschny et al. were randomly selected strings of words and thus not continuously repeated blocks of sentences as used in this study (and similar to Loizou et al., 2009). Given that the current experimental design eliminated at least some of the stimulus uncertainty for the speech maskers, such as those in Pyschny et al., it is possible that our listeners may have learned at least some of the competing sentences and thus been better able to ignore the linguistic content of the speech maskers. We recognize this as a limitation in the current study as related to clinical relevance for bimodal and bilateral CI users in everyday listening scenarios for which there is most likely more stimulus uncertainty with speech maskers. Further evidence that there is greater potential for release from informational masking for bimodal listeners is gleaned from the CI-alone data. In the CI-alone condition, both listener groups exhibited similar magnitudes of release from informational masking in both the same- and different-gender conditions (bimodal mean = 21.3, 95% CI [17.8, 24.9]; bilateral mean = 19.3, 95% CI [14.9, 23.6]). Thus, there was no inherent difference in susceptibility to informational masking as a function of group (bimodal vs. bilateral) when listening condition was the same across groups (CI-alone).

Effect of Talker Gender

An additional focus of the current study was on the effect of talker gender. The maskers utilized here were always male, and the target talker was either female or male, allowing for performance in different- and same-gender conditions to be compared. Both bimodal and bilateral CI listeners demonstrated significantly greater release from informational masking in the same-gender condition as compared with the different-gender condition. This effect was also observed among both groups when tested CI-alone. These findings suggest that release from informational masking is influenced by talker gender, such that the magnitude of release is significantly greater when talker–masker conditions are more similar.

Effect of Listening Condition

The benefit of “two-eared” listening was examined with respect to speech recognition across spatial configuration. For both bimodal and bilateral CI listeners, the addition of the second ear resulted in significantly better performance when compared to CI-alone listening for both gender conditions and for both masker types.

The effect of listening condition was also examined with respect to release from informational masking. Among the bimodal group, release from informational masking was greater for CI-alone listening, as compared to the bimodal configuration—but only for the same-gender condition in the S₀N_45&315 spatial configuration. Among the bilateral group, no differences between bilateral and CI-alone performance were observed.

Effect of Masker Type

Finally, the effect of masker type was examined. Both bimodal and bilateral CI users performed significantly better with the SCN masker than with the speech masker, for both gender conditions. Thus, regardless of listening modality—that is, even if the input across the two ears was different—performance was best in the presence of the energetic masker only. This effect was also observed among both groups when tested CI-alone. This finding was not unexpected as a speech masker (informational and energetic) results in greater similarity and stimulus uncertainty between target and masker than does an SCN masker (energetic; Arbogast et al., 2002; Brungart, 2001).

Additionally, regardless of gender condition, bimodal performance decreased with increasing spatial separation when compared to the colocated condition; bilateral CI users did not exhibit this same finding.

Clinical Implications and Future Directions

Neither bimodal nor bilateral CI users demonstrate a typical pattern of SRM with increasing spatial separation between target and masker. This finding highlights the importance of devoting continued efforts to improving binaural coordination across devices. Furthermore, it is possible that the level effects associated with BTE microphone placement have a negative impact on performance as spatial separation increases, especially in symmetrical noise configurations. Future research is warranted to investigate whether level effects associated with BTE microphones are at least partly responsible for the negative SRM exhibited by both bimodal and bilateral CI users in the current study—the implications of which are far reaching, as the majority of noise in the “real-world” surrounds the listener.

Conclusions

In summary, the results of our study add to the existing literature on SRM and informational masking in the bimodal and bilateral CI population. Several studies have examined SRM using asymmetrical noise configurations; however, such listening configurations allow head shadow cues to dominate. When more realistic symmetrical configurations are utilized, SRM is nonexistent, or even negative—a finding reasonably attributed to both the lack of coordinated across-ear timing cues and microphone effects for BTE hearing technology. Though release from informational masking was not significantly different across bimodal and bilateral CI listeners, the magnitude of informational masking release was greater for bimodal listeners, which is consistent with the theory that CI users with acoustic hearing may be more susceptible to informational masking. Future research is warranted to further examine the relationship between residual acoustic hearing and susceptibility to informational masking.

Funding Statement

This research was supported by the National Institute on Deafness and Other Communication Disorders (NIDCD R01 DC009404, PI: Gifford) and the Vanderbilt Institute for Clinical and Translational Research (NIH UL1 TR000445).