Two electrode arrays, a lateral and a perimodiolar CI, were inserted in scala tympani (Figure 1) of a finite element model based on high-resolution human cochlea micro-CT from our previous work (Potrusil et al., 2020). The geometry included reconstructed pathways of type I ANFs. In the first simulation step, the stimulating voltage profiles along the ANFs were computed for active CI electrode contacts using COMSOL Multiphysics (version 5.5¹). In the second step, excitations of investigated ANFs were simulated with a multicompartment model (Rattay et al., 2001b). Rattay’s multicompartment model considered the lengths of the nodes of Ranvier, peripheral terminal compartment, and peripheral and central internodes are 2.5, 10, 250, and 500 μm, respectively, except for the last peripheral internode. In addition, the diameter of soma is 20 μm, and the length of the active pre- and post-somatic regions are 100 and 5 μm, respectively. More information can be found in Rattay et al. (2001b) and Potrusil et al. (2020).

The investigated ANFs, named target neurons (TNs), were selected based on the closest distance from their peripheral terminal site to the center of the electrodes. Table 1 represents the angles measured from the round window with respect to the modiolus axis for the six investigated electrodes (three from each CI array system) and TNs in different cochlear turns (basal, middle, and upper–middle). Our previous clinical study (Heshmat et al., 2020) investigated critical morphometry parameters such as diameter and myelination thickness of the peripheral parts of ANFs for several human cochleae at various ages with different hearing loss levels. According to observations of different distributions in diameter and myelination thickness, the study suggested a degeneration pattern for peripheral parts of human ANFs based on hearing loss levels. Therefore, the peripheral process morphometry was divided into six groups indicating different neural health statuses (Table 2) according to the different varieties of the peripheral parts of human ANFs.

Monophasic, pseudomonophasic, triphasic, and biphasic pulses were applied for both polarities, CAT and ANO (Figure 2). The phase duration was 100 μs, except for the second phase of pseudomonophasic pulse, in which the duration was increased by a factor of four, and its amplitude decreased by the same factor to keep the charge balanced (Macherey et al., 2011). All multicompartment model evaluations were performed in MATLAB (version R2020a²).

Moreover, four CI stimulation strategies, which vary depending on the return electrode (ground electrode) position, were used in this study: (i) a monopolar strategy (MP), with the return electrode located outside the cochlea, usually behind the ear; (ii) a bipolar strategy (BP), with the return electrode placed next to the stimulating electrode; (iii) a tripolar strategy (TP) that the return electrodes are two neighboring electrodes, each electrode taking half of the current delivered to the stimulating electrode; and (iv) a partial tripolar strategy (PTP) with a combined configuration of MP and TP, in which part of the stimulating current (defined by the coefficient σ, 0 < σ < 1) returns to the two neighboring electrodes and the rest to the common ground (Figure 3). The investigated PTP configuration was considered with 75% of the current returning to the neighboring electrodes and 25% to the ground, according to Jolly et al. (1996); Litvak et al. (2007), and Mesnildrey et al. (2020). These four CI stimulation strategies (MP, BP, TP, and PTP) were evaluated using COMSOL (Figure 3) for both CI systems.

Two geometrical parameters are important in order to compare the lateral and the perimodiolar CI: (i) the center-to-center distance of adjacent channels and (ii) the distance to the modiolus axis. To perceive the impact of each of these parameters, first, we used the CI systems with features as reported by the manufacturer. Next, we applied a similar distance between channels in both CI arrays and compared the results with the original arrays. For this aim, in perimodiolar CI, we considered the second basal adjacent channels (Figure 3B), whereas, in lateral CI, an extra electrode was set between two investigated electrodes each time, which was used as the new ground electrode (Figure 3C).

Figure 4 summarizes excitation thresholds of three investigated TNs in the basal, middle, and upper–middle turn for the perimodiolar CI with pulse shapes: monophasic, pseudomonophasic, triphasic, and biphasic in the MP and three multipolar strategies: BP, TP, and PTP. In all panels, six ANF statuses based on degeneration levels (Table 2) were considered with a circle (intact), rectangle, triangle, cross, diamond, and plus, for AS1, AS2, AS3, AS4, AS5, and AS6, respectively.

Figure 4A shows the excitation thresholds of the monophasic pulse for three TNs, B11 (blue), B7 (green), and M9 (red), in four stimulation strategies. The MP strategy needs the lowest currents in both polarities. The deviation between ANO and CAT thresholds is small for each degeneration case. This is obvious as most cases accumulated over the black dotted line where the polarity ratio (ANO/CAT) is 1. In addition, the threshold increases in both polarities by increasing the degeneration levels of TNs except in M9, which has almost the same ANO threshold for all degenerated levels because, in all status cases, AP initiation site is the central process.

Moreover, in contrast to MP strategy, the ANO threshold ranges increase negligibly in BP, TP, and PTP; however, the CAT threshold ranges increase significantly. The investigated TNs have almost equal ANO thresholds for different ANF degenerated cases, which indicates the AP initiates from nearly the same site of the central process of TNs. In contrast, the CAT threshold variations are substantial and increased significantly in each level of degeneration, especially in AS5 and AS6 cases. Finally, the behaviors of the fibers in TP and PTP strategies are similar to each other. To summarize, using monophasic pulse in four stimulation strategies shows that all threshold cases are densely accumulated in MP, whereas, in multipolar strategies, each TN individually forms a cluster that can be distinguished by degeneration levels.

Figure 4B represents the excitation thresholds for the pseudomonophasic pulse. The ANF behavior is almost similar to monophasic pulse (Figure 4A) for all strategies. Furthermore, the ANO threshold variations in the BP, TP, and PTP are minimal between degeneration cases of individual TNs compared with MP strategy, and again, the clustering behavior is formed in multipolar strategies compared with MP. Interestingly, the CAT threshold ranges are decreased by half compared with monophasic pulse, as in Figure 4A. This threshold difference can be understood from the voltage membrane changes induced by the two pulse shapes, as shown in Figure 5A. The pseudomonophasic pulse has a lower negative slope at the onset of the CAT-leading phase compared with the monophasic pulse represented with the pink rectangle (Figure 5A). Therefore, the pulse makes the transmembrane voltage less negative (20 vs. 55 mV hyperpolarized). The second charge-balanced phase of the pseudomonophasic pulse has a positive effect, causing the transmembrane voltage to depolarize more and faster to the excitation threshold level. For a better understanding, the voltage membrane changes by applying a monophasic pulse are displayed over time for a progressively degenerated fiber (Figure 5B) and a healthy fiber (Figure 5C).

Figure 4C demonstrates the excitation thresholds with the triphasic pulse. The ANF behavior is almost comparable to previous pulse shapes in all stimulation strategies, and again, the clustering behavior is formed in multipolar strategies compared with MP. Moreover, in multipolar strategies, the CAT threshold ranges of the triphasic pulse are approximately 50% higher than pseudomonophasic and 50% lower than the monophasic pulse.

Figure 4D displays the excitation thresholds with biphasic pulse. The behavior in the MP strategy is similar to the previous pulse shapes but with less variability and polarity ratio almost close to 1. However, the cluster form based on degeneration levels among individual TNs is not visible anymore. The TN thresholds mostly accumulate over the black dotted line, resulting from the same AP initiation site of TNs in different degeneration levels in both polarities. As a result of losing the CAT threshold variations in multipolar strategies by applying the biphasic pulse, it is challenging to distinguish TNs response based on different degeneration levels, which differs from the previous pulse shapes (Figures 4A–C).

Figure 6 shows excitation thresholds as Figure 4 but for a lateral CI. Subsequently, Figure 6A displays the excitation thresholds of monophasic pulse for the three investigated TNs in the four stimulation strategies. The MP strategy needs significantly lower currents, especially in CAT thresholds, compared with the other strategies. In addition, the ANO and CAT thresholds do not change substantially by increasing degeneration levels. However, both ANO and CAT thresholds increase significantly in multipolar strategies compared with the MP strategy. The ANO threshold variations between the TNs are small in the BP strategy compared with TP and PTP strategies. However, the cluster behavior is not formed in multipolar strategies in lateral CI, as they formed in the perimodiolar CI (Figure 4A).

Figure 6B represents the excitation thresholds with the pseudomonophasic pulse. The trend is similar to monophasic pulse (Figure 6A) for all strategies, although CAT thresholds in the BP are significantly lower and more concentrated compared with monophasic pulse in severe degenerated cases (Figure 6A, BP strategy). Surprisingly, the similar CAT threshold ranges between monophasic and pseudomonophasic pulses (Figures 6A vs. 6B) are in contrast to the CAT threshold behavior in the perimodiolar CI case (Figures 4A vs. 4B).

Figure 6C displays the excitation thresholds with the triphasic pulse with a similar structure as Figures 6A,B. The ANF behavior is comparable to previous asymmetric pulses, especially with pseudomonophasic. The CAT threshold ranges of triphasic pulses are approximately 25% larger than monophasic and pseudomonophasic pulses.

Figure 6D demonstrates the excitation threshold with the biphasic pulse. The ANO and CAT threshold ranges are almost doubled compared with the perimodiolar case (Figure 4D). The threshold characteristics in both polarities are comparable in all four stimulation strategies for all TNs, except in some severe degenerated cases.

Figures 7–10 summarize the polarity sensitivity defined by the ANO/CAT threshold ratios of three investigated TNs for the perimodiolar CI (left panels) and lateral CI (right panels) with pulse shapes: monophasic (Figure 7), pseudomonophasic (Figure 8), triphasic (Figure 9), and biphasic (Figure 10), in the MP and the three multipolar strategies for the six degeneration levels (Table 2). The trend ANO/CAT < 1, indicating the lower ANF threshold for anodic stimulation, is related to the degeneration level and the modiolus distance of the stimulating electrode. Polarity sensitivity is much more pronounced for monophasic (Figure 7), pseudomonophasic (Figure 8), and triphasic (Figure 9) versus biphasic (Figure 10) stimulation.

Figure 7 illustrates the polarity threshold ratios with monophasic pulse for two CI systems. In the MP strategy (Figure 7A), ANO sensitivity (ANO/CAT < 1) in the perimodiolar CI occurred more frequently except in four cases that can be related to nearer-distance electrode to modiolus wall. A declining trend in ratios appears from intact to more degenerated cases in both CI array systems only for the upper middle TN, M9.

Figures 7B–D display the polarity ratios in multipolar strategies for both CI systems. Interestingly, a declining trend in ratios always happens for all three TNs in the case of the perimodiolar array. More degenerated cases (AS4–AS6) become ANO sensitive in the lateral system, whereas CAT sensitivity happens in intact and less degenerated cases for all three multipolar strategies. However, in the perimodiolar array, regardless of degeneration level, most cases show ANO sensitivity.

Figure 8 shows the polarity threshold ratios for pseudomonophasic pulse. The behavior is very similar to monophasic pulse (Figures 7A–D). Interestingly, again the decreasing trend can be observed in the perimodiolar array for all TNs in the three investigated multipolar strategies. In contrast, in the lateral array system, ANO versus CAT sensitivity occurs mainly in AS4–AS6 degenerated cases versus intact, slight, and moderate degenerated cases (AS1–AS3).

Figure 9 demonstrates the polarity threshold ratios for triphasic pulse. The behavior is similar to monophasic and pseudomonophasic pulses (Figures 7, 8). Again, the decreasing trend can be noticed in the perimodiolar array for all TNs in the three investigated multipolar strategies. In contrast, in the lateral array system, ANO versus CAT sensitivity occurs mainly in AS4–AS6 vs. AS1–AS3.

Figure 10 shows the polarity threshold ratios with biphasic pulse for both CI systems. Contrary to the previous pulses, for the biphasic pulse in the lateral array, the ANO sensitivity of the high degeneration versus CAT sensitivity for intact and less degenerated cases is not observed anymore. In most cases of the perimodiolar array, the polarity ratio is close to 1; hence, the declining behavior in perimodiolar CI from intact to progressive degeneration does not occur in the biphasic pulse.

Figure 11 displays AP initiation sites at ANO and CAT threshold levels for the three investigated TNs in the basal, middle, and upper–middle turn for the perimodiolar CI (blue) and the lateral CI (red) with pulse shapes: monophasic (A), pseudomonophasic (B), triphasic (C), and biphasic (D) in the MP and the three multipolar strategies. In all panels, y axis represents the six ANF statuses based on the degeneration levels, as in Table 2.

Figure 11A shows AP initiation sites when a monophasic pulse is applied. In the lateral CI, AP mostly initiates at the peripheral sites regardless of the pulse polarity, stimulation strategy, and degeneration levels, except for progressive cases, which are considered without peripheral process. However, in the primordial CI, the AP initiates in the periphery when CAT pulse is applied except for the progressive cases, and in the case of ANO pulses, the initiation sites mainly occur in the central processes.

Figures 11B–C display the AP initiation sites when pseudomonophasic and triphasic pulse is applied. The same behavior in AP initiation sites can be observed in the lateral array as in Figure 11A, although by using the perimodiolar array, initiation sites differ for ANO versus CAT thresholds mostly when TP and PTP strategies are used.

By applying a symmetric biphasic pulse (Figure 11D), the lateral CI keeps the AP initiation sites in the peripheral sites, whereas in the perimodiolar array, the AP initiation sites are mostly the same in both CAT and ANO thresholds regardless of TN degeneration levels.

Differences in threshold characteristics between perimodiolar and lateral CIs result from the electrode to modiolus distance and the distance between channels. To investigate the impact of distance between channels, one channel gap is considered in perimodiolar CI by increasing the distance between channels from 0.7 to 1.4 mm (Figure 3B). In addition, for the lateral CI, we decreased the distance between channels from 2.4 to approximately 1.3 mm (Figure 3C).

The effect of doubled channel separation is evident for multipolar configurations in perimodiolar array; compare Figures 4, 12, left panels. For monophasic pulses, the ANO and CAT threshold ranges are reduced by a factor of 2 when the distance of returning electrode is increased (Figures 12A, left panel vs. 4A). The clustering behavior is also noticeable in all three multipolar strategies. The investigated TNs have individually similar ANO thresholds in all degenerated cases; in contrast, the CAT thresholds are increased in each level of degeneration, similar to that in Figure 4A but with lower thresholds (Figure 12A, left panel).

The left panel of Figure 12B shows similar characteristics of pseudomonophasic pulses as in monophasic pulse but with lower CAT thresholds dropped by half (Figures 12A vs. B, left panels), as Figures 4A,B for multipolar strategies. The same clustering behavior can be observed as in Figure 4B.

Triphasic pulse (Figure 12C, left panel) displays similar behavior with previous pulses with the almost same CAT threshold ranges as in monophasic pulse (Figure 12A, left panel) and higher CAT thresholds than pseudomonophasic pulse (Figure 12B, left panel). In addition, the ANO and CAT threshold ranges are reduced by a factor of 2 when the distance of returning electrode is increased compared with Figure 4C.

For biphasic pulse again, the ANO and CAT threshold ranges in the multipolar strategies are decreased by a factor of about two (Figures 12D, left panel vs. 4D). However, the cluster form based on degeneration levels among individual TNs is not visible anymore, and the threshold ratios are mostly close to one, as in Figure 4D. Overall, both polarity threshold ranges are decreased by increasing the returning electrode distance in all multipolar strategies, which are related directly to current shunting between electrodes.

The effect of halving channel distance is studied and shown in Figure 12, right panels for multipolar strategies in the lateral CI. As displayed in the right panels of Figures 12A–D, the threshold characteristics are very similar to multipolar strategies compared with Figures 6A–D, except for the threshold ranges, which are significantly increased. The ANO/CAT ratios are again close to 1 (Figures 6D, 12D, right panel) for the biphasic pulse, and no cluster is formed regardless of pulse shape and stimulus strategies as before.

Figure 13A represents the polarity ratios for monophasic stimulation. The polarity ratio in perimodiolar CI significantly decreased by doubling the distance between channels compared with Figures 7B–D, left panels. Interestingly, the declining trend is also again visible between intact and different degenerated cases, similar to Figures 7B–D (left panels). In contrast, the polarity ratio substantially increased by halving the distance between channels in the lateral CI (Figure 13A, right panels). However, the polarity behavior of all TNs in intact and different degenerated levels is similar to Figures 7B–D (right panels).

Figure 13B illustrates the polarity ratios with pseudomonophasic stimulation. The polarity ratio is significantly decreased by doubling the distance between channels in perimodiolar CI (Figure 13B, left panels) compared with Figures 8B–D (left panels). The declining trend is again visible between intact and most degenerated cases, similar to Figures 8B–D (left panels). On the other hand, lateral CI shows similar polarity behavior as monophasic pulse (Figure 13A, right panels) and Figures 8B–D, right panels.

Figure 13C demonstrates the polarity ratios for triphasic stimulation. In perimodiolar CI (Figure 13C, left panels), the polarity ratio is significantly decreased by doubling the distance between channels compared with Figures 9B–D (left panels). Again, the declining trend is noticeable between intact and most degenerated cases, similar to Figures 9B–D, left panels. Lateral CI represents similar polarity behavior as previous pulses (Figures 13A,B, right panels; Figures 9B–D, right panels).

Figure 13D shows the polarity ratios for biphasic stimulation. In contrast to the previous pulse shapes, by applying the biphasic pulse in the lateral CI array (Figure 13D, right panels), the polarity effect is lost, similar as in Figures 10B–D, right panels. The declining trend in perimodiolar CI (Figure 13D, left panels) from intact to progressive degeneration does not occur by applying this pulse shape, same as in Figures 10B–D, left panels.

Biphasic symmetric charge balance pulse is widely used as a standard pulse shape to stimulate the surviving ANFs in CIs. According to numerous clinical studies, pulse shape is a critical factor for studying neural status (Undurraga et al., 2010; Bahmer and Baumann, 2012a,b, 2013; Hughes et al., 2018; Luo J. et al., 2020). Consistent with several clinical studies, which reported that asymmetric pulse shapes are more effective than symmetric biphasic pulse for evaluating polarity sensitivity in CI users (Macherey et al., 2006, 2008, 2017; Carlyon et al., 2013; Guérit et al., 2018; Jahn and Arenberg, 2019a), our results showed that when a symmetric biphasic pulse is applied, recognition of peripheral changes based on polarity sensitivity was not feasible. We found threshold variations, and consequently, polarity sensitivity between degeneration levels could not be detected with symmetric biphasic pulses. As a symmetric biphasic pulse consists of phases with equal amplitude and duration but opposite polarities, the contribution of the extracellular gradient is almost the same for both phases; consequently, it may not be possible to estimate which polarity affects the ANFs mainly.

In agreement with the presented results, several studies reported that symmetric biphasic pulses did not provide much evidence for polarity sensitivity (Undurraga et al., 2013; Guérit et al., 2018; Hughes et al., 2018; Xu et al., 2020). In addition, Spritzer groups found opposite results of polarity, which was consistent with (Macherey et al., 2006), and also, they could not find polarity sensitivity for peak electrode locations (Spitzer and Hughes, 2017; Spitzer et al., 2019). They suggested that using symmetric biphasic pulse is not an excellent choice to study polarity sensitivity (Spitzer et al., 2019). In line with findings by Undurraga et al. (2013) who reported the same ANO and CAT threshold detection using biphasic pulse, we found ANO/CAT ratios mostly close to one in perimodiolar array when a biphasic pulse was applied to the active electrodes.

In contrast to symmetric pulses, several studies demonstrated promising results in polarity investigations by using asymmetric pulse shapes, for example, pseudomonophasic and triphasic (Macherey et al., 2006, 2008, 2017; Carlyon et al., 2013, 2018; Undurraga et al., 2013; Bahmer et al., 2017; Mesnildrey et al., 2017; Jahn and Arenberg, 2019a). However, some of these studies could not provide direct evidence to prove the role of the polarity effect in estimating neural health, which might result from various factors such as stimulation strategy and CI systems.

Despite using triphasic pulse, Carlyon et al. (2018) and Jahn and Arenberg (2019a) reported that up to 70% of investigated electrodes showed anodic sensitivity and could not prove polarity behavior as an indicator for estimating neural survival. Similar to their results, we found more anodic sensitivity using asymmetric pulses, such as pseudomonophasic and triphasic pulses, in the perimodiolar array. However, this anodic sensitivity occurred not only in degenerated cases but also in intact case; because the perimodiolar array is mostly close to the modiolus axis, anodic pulses may exhibit more efficiency as they active central parts of ANFs (Figure 11). On the other hand, we found a reduction trend in polarity ratios from approximately 1 in intact to approximately 0.5 in the most degenerated case. As anodic sensitivity is observed more frequently regardless of neural health status, we suggest that considering the ANO/CAT ratios and comparing them across the array might shed light on estimating neural health in future investigations.

Several clinical investigations demonstrated that MP strategy induces uniform thresholds across the CI and develops outcomes with poor tonotopic representation due to a broad electric field (Bierer and Middlebrooks, 2002; Bierer et al., 2015a), yet MP stimulation strategy is widely used as a standard strategy in commercial CI systems. In contrast, multipolar strategies, such as BP, TP, and PTP, produce sharper electric fields compared with the MP strategy (Kral et al., 1998; Bierer and Middlebrooks, 2002; Snyder et al., 2004; Nelson et al., 2008; Bierer and Faulkner, 2010; Landsberger et al., 2012; Srinivasan et al., 2013; Long et al., 2014). Evidence from numerous psychophysical and physiological studies suggests that multipolar strategies increase selectivity, reduce channel interaction, and cause larger variation of detection thresholds between electrodes across CI array, and it is beneficial for diagnostic purposes, particularly for investigating the electrode neuron interface (Pfingst and Xu, 2004; Bierer, 2007; Bierer and Faulkner, 2010; Landsberger et al., 2012; Srinivasan et al., 2013; Bierer and Nye, 2014; Long et al., 2014; Schvartz-Leyzac et al., 2020). Undurraga et al. (2012) and Chatterjee and Kulkarni (2014) have reported that multipolar strategies cause changes in AP initiation sites leading to threshold variation. In agreement with the findings mentioned previously, we found the obtained threshold values were more concentrated and in the same range in MP stimulation, whereas by applying multipolar strategies, threshold values created clusters based on different degeneration levels as well as cochlear turn.

While degeneration of ANFs occurs mostly in the peripheral part (Fayad and Linthicum, 2006), peripheral processes may remain partly intact. However, depending on hearing loss levels, reduction in diameter and myelination thickness happens (Spoendlin and Schrott, 1989; Heshmat et al., 2020). These findings suggest that variation in peripheral processes is imminent; thus, peripheral excitation results in behavioral differences based on degeneration status across the array.

Various clinical studies had suggested that polarity effects were observed more significantly when multipolar strategies were applied (Macherey and Carlyon, 2010; Macherey et al., 2011; Undurraga et al., 2012; Jahn and Arenberg, 2019a; Mesnildrey et al., 2020). Miller and Chatterjee have reported that multipolar strategies excite more peripheral processes of ANFs and demonstrate better local ANF activation and degeneration patterns (Miller et al., 2003; Chatterjee and Kulkarni, 2014). In alignment with those studies, our results showed that with changing the level of degenerations in the peripheral process, at least for one polarity, threshold values were changed based on peripheral degeneration levels when multipolar strategies were used. Consequently, in the perimodiolar CI, the polarity threshold ratios significantly differed and followed a declining trend for intact to progressive degenerated cases, whereas by using MP stimulation, the declining trend was lost. In addition, in the lateral CI, the anodic versus cathodic sensitivity occurred mainly in severe to progressive degenerated cases versus intact to moderate cases by applying only multipolar strategies. However, according to our results, applying symmetric pulse shape regardless of stimulation strategy led to similar threshold ranges in polarity, i.e., polarity threshold ratios close to 1 (Figure 10), which agrees with clinical studies (Macherey et al., 2008, 2006; Undurraga et al., 2013). Furthermore, Macherey group reported that using the BP strategy with symmetric pulse could not provide enough evidence for polarity behavior, whereas with applying asymmetric pulse, polarity effects were observed (Macherey and Carlyon, 2010; Macherey et al., 2011).

Commercially available CI arrays are generally produced in two types: lateral and perimodiolar arrays. In a lateral CI system, electrodes are located farther from the modiolus wall than in a perimodiolar array. To investigate the impact of electrode modiolus wall distance, we used both CI systems in a realistic model as reported in manufacturer data. Consistent with Schvartz-Leyzac et al. (2020), who observed that thresholds are directly proportional to electrode modiolus axis distance, our results indicate that lateral array needed larger threshold values compared with perimodiolar array.

However, owing to the fact that in a lateral array, the distance between channels is larger than in a perimodiolar array, we placed an additional electrode between two investigated existed electrodes. In addition, in perimodiolar system, we used the multipolar strategies with one gape in between channels (BP + 1, TP + 1, and PTP + 1) to have the channel distances as similar as possible in both CIs. Our result showed that reducing the channel distance in lateral CI affected only the threshold ranges and not the overall behavior of the thresholds. Consequently, we observed a significant increase in the threshold values due to the current shunt effect. Subsequently, increasing the channel distances in the perimodiolar CI led to a significant decrease in threshold values. Interestingly, in perimodiolar array, the threshold variation between cochlear turns slightly decreased due to wider electric field by setting one electrode gap between channels, which is in line with previous studies that reported variation in thresholds across the CI strongly depends on the expanse of the electric field (Bierer, 2007; Bierer et al., 2011; Bierer and Nye, 2014). Therefore, our results suggest that while the distance between electrodes alters the threshold ranges, the modiolus wall distance strongly affects threshold variations and consequently polarity ratios, not only across the array but also between degeneration levels when multipolar strategies are applied, as some clinical studies reported that the distance between electrode to modiolus wall impacts detection thresholds in multipolar strategies (Long et al., 2014; Mesnildrey et al., 2020).

On the other hand, similar to Long et al. (2014), who reported no significant difference in threshold ranges across the array in MP strategy, we also found analogous threshold ranges in both CIs, not only across the array but also between different degeneration levels when applying MP strategy. Consequently, we could not find any effect of electrode modiolus distance on polarity behavior in MP strategy, as also seen in the study by Jahn and Arenberg (2019a) that used MP strategy and reported that electrode modiolus wall distance does not affect polarity behavior.

Our study investigated the impact of pulse types, stimulation strategies, and electrode distance to modiolus on thresholds, spike initiation sites, and polarity sensitivity for normal and degenerated ANFs. However, the perceptual efficacy of the artificially generated spiking patterns depends essentially also on the population size of spiking ANFs per active channel. Frequency information transmitted by a small group of ANFs may be lost already during the first neural processing in the cochlear nucleus. Therefore, the focus on a stimulation strategy with sharp frequency selectivity is sometimes successfully replaced by sending the same (combined) input to two channels because of a poor ANF density in a specific frequency region of a patient or by using an implant with a low channel number (Saleh et al., 2013; Bierer and Litvak, 2016; Zhou, 2017). Electrically evoked compound APs recorded by modern CIs are of help to estimate the population size for each channel and intensity (Cullington, 2002; Abbas et al., 2004; Nehmé et al., 2014; DeVries et al., 2016). In addition, the amplitudes of the electrically evoked auditory brainstem response (Bierer and Faulkner, 2010; Bierer et al., 2011; Causon et al., 2019) reflect the frequency responses of each channel along the first processing centers. Both techniques can be applied to adapt our findings for individual patient data.