A cochlear implant (CI) restores hearing for profoundly deaf patients by transmitting sound to an array of electrodes that stimulates the inner ear. The small number of frequency bands and limited transmission of temporal fine structure affects the music perception. The present work investigates the pleasantness of chords and chord sequences in adults using such electric hearing. In the first task, participants compared chord types according to their perceived pleasantness. Normal-hearing listeners judged the major chord and the minor chord as the most pleasant ones compared to other chord types. CI users appraised the major chord as more consonant than other chord types. The second task used four-chord sequences, half of which ended on an authentic V-I cadence. In the other presentations, the final tonic was replaced either by a transposed major chord or by a dissonant chord. The participants had to judge whether the ending was conclusive. While normal-hearing listeners preferred authentic cadences, all but one CI user assessed the modified cadences as similarly satisfying. The results indicate that CI users appreciated consonance of isolated chords to a certain extent similar to normal-hearing listeners. Nevertheless, the majority of CI users fail to register the musical syntax in the harmonic progression of cadences.

For profound neurosensory hearing loss, cochlear implant (CI) devices have proved to be a highly successful neural prosthesis in order to largely restore speech communication (Wilson & Dorman, 2008). As a consequence of the success of such devices, the enjoyment of music has become another important factor in the quality of life for the majority of CI users (Lassaletta et al., 2007). However, the sound information is inevitably degraded on several stages. Because the number of stimulating electrodes is up to 22 in present devices (Limb & Roy, 2014), the spectral information is considerably reduced with respect to the action of about 3500 inner hair cells in normal hearing listeners (cf. e.g., Zeng, Tang, & Lu, 2014). Since the perception of musical pitch and timbre relies critically on precise spectral information, this strongly affects the sound of music (Limb & Roy, 2014). Furthermore, most devices use fixed pulse trains for the electrical stimulation of the auditory nerve, limiting the perception of temporal pitch to about 300 Hz (Kong, Deeks, Axon, & Carlyon, 2009). In general, this omits the information of temporal fine structure of the sound waveform that is also important for pitch perception (Moore, 2008), although some recent sound processing strategies (used also by the subjects in the present study) to some extent transmit temporal fine structure at low frequencies (see Wouters, McDermott, & Francart, 2015). Finally, pathological changes in the auditory pathway can interfere with the transmission of musically relevant information through a CI (for a review, see Looi, 2008; Limb & Roy, 2014).

In most music-related detection and discrimination tasks, CI users typically show a larger variability of individual performance than normal-hearing listeners (Gfeller et al., 2008, Looi, 2008). For instance, few CI users are able to determine the higher tone in a pair of tones separated by only a semitone. Most CI users require differences of more than three semitones (Kang et al., 2009) and some may require more than one octave in this task (Ping, Yuan, & Feng, 2012). CI users are also significantly less accurate than normal-hearing listeners in the discrimination and recognition of melodies and instrument timbres (Brockmeier et al., 2011; Cooper, Tobey, & Loizou, 2008; Kang et al., 2009). Their ability to follow a melodic contour depends on, among other factors, the instrument timbre, whereby the piano is more difficult to follow than, for example, the violin or the organ (Galvin, Fu, & Oba, 2008). Only the registration of rhythm and meter seems to be largely unaffected by the sound processing, presumably due to the sufficiently good timing precision of the stimulation (cf. Limb & Roy, 2014).

Little work has been devoted to CI users’ perception of polyphonic music. Galvin, Fu, and Oba (2009) modified their measurement of melodic contour identification by adding another masking instrument that played a fixed tone. In the presence of this masker, the performance of the CI users decreased by more than 15%, whereas the performance of normal-hearing participants remained unchanged. Thus, for CI users, a competing instrument makes it more difficult to follow a melody. CI users are also less accurate than normal-hearing listeners in identifying the number of tones played simultaneously (Donnelly, Guo, & Limb, 2009).

Musical harmony is a parameter of Western tonal music that has hardly been addressed in research on CI users, although it may crucially contribute to the understanding of music. Harmony is generally understood as the “combining of notes, simultaneously, to produce chords, and successively, to produce chord progressions” (Dahlhaus, 1980). This definition implies a vertical dimension of simultaneous tones, and a horizontal dimension, comprising sequences of tones, dyads, or chords (Tramo, Cariani, Delgutte, & Braida, 2001). According to Terhardt's model, the vertical consonance of simultaneously sounding tones relies on sensory contributions and on acquired, music-specific factors (Terhardt, 1984). Sensory consonance of a chord played in isolation originates from the signal itself, like the roughness that may rise from the interaction of components of the tones (Johnson-Laird, Kang, & Leong, 2012; Plomp & Levelt, 1965). Acquired, non-sensory factors contributing to consonance for simultaneously presented tones include relation to a diatonic scale and the harmonicity of the resulting frequency spectrum, although the latter may also include sensory contributions (Johnson-Laird et al., 2012; McDermott, Lehr, & Oxenham, 2010). In triads, Johnson-Laird et al. (2012) noted that sensory consonance alone fails to explain the perceived consonance of various triads, and thus extended their model by the principles of scale membership, the predominance of the major chord, and the role of thirds. In another model, Cook and Fujisawa (2006) included a malus for triads with symmetric intervals, such as the augmented triad that is built from two major thirds.

In the horizontal dimension, the relations of successive chords contribute substantially to the musical syntax (Patel, 2003). The melody and sequence of chords establish a framework with a tonal center, the root of the key (Dahlhaus, 1980). Phrases are expected to return to the tonal center, and deviation from this expected end produces harmonic tension (Patel, 2003). Since the late 16th century, cadences have been established as a standard for harmonic closure music. In “Le istitutioni harmoniche” (1558), Zarlino compared the cadence to the role of the “period in speech.” (“Onde la Cadenza é di tanto valore nella Musica, quanto il Punto nella Orazione; & si puó veramente chiamare Punto della Cantilena.”) About two centuries later, Rameau postulated in his “Traite´ de l'harmonie” (1722) that the authentic cadence is a “satisfying conclusion after which one does not miss anything.” (“L'On appelle Cadence parfaite, une certaine conclusion de chant, qui satisfait de façon, que l'on n'a plus rien à desirer aprés une telle Cadence.”) The authentic or perfect cadence ends on the consonant root chord (the tonic, I), which is preceded by a chord on the fifth scale degree (the dominant, V) that guides the harmony to the tonic (Rockstro, Dyson, Drabkin, & Powers, 1980). Even if played in isolation, these two chords elicit a plausible phrase (Bharucha & Stoeckig, 1986; Rosner & Narmour, 1992), in particular when in the second inversion of the dominant the treble acts as a leading tone to the octave of the tonic.

Deviations from the authentic cadence are used in order to generate tension, or to shift the harmony to a different tonal center. This may be achieved by ending with a sensory-dissonant chord (such as a seventh chord; Bigand, Parncutt, & Lerdahl, 1996), or by a vertically consonant chord with a different harmonic function. For example, half cadences end on the vertically consonant dominant and do not elicit complete conclusiveness in the average listener (Sears, Caplin, & McAdams, 2014). Deceptive cadences, featuring another vertically consonant chord, even seem to be considerably less conclusive to musical experts than to laypersons. In contrast, for example, jazz harmony often adds a vertically dissonant major seventh or ninth to the final tonic chord, which may still be perceived conclusive. Therefore, such deviations affect the conclusiveness in a progressive way. Bigand and coauthors (Bigand & Pineau, 1997; Bigand, Poulin, Tillmann, Madurell, & D'Adamo, 2003) used sequences of chords corresponding to the V-I cadence, which return to the tonal center. In half of the presentations, this expectation was violated by conclusion with a I-IV cadence, thus ending on the less stable subdominant. When asked to decide whether the final chord successfully completed the sequence, participants were less secure and slower in their decision if the final chord was not the tonic one. This confusion indicated that the harmonic alteration affected how conclusive the sequence was.

Deficits in the individual perception of chords compromise the listener's understanding of the musical syntax. Harmonic syntax can be learned explicitly, and implicitly by regular exposure to a musical system, starting already at an early stage in life (Bigand & Poulin-Charronnat, 2006; Jentschke, Friederici, & Koelsch, 2014). Some implicit familiarity with the syntactical role of cadences was found in five-year-old children already (Corrigall & Trainor, 2010). Their reactions to consonant sounds are accelerated compared to dissonant sounds, and the same is found in modification of the closure of chord sequences (i.e., when the sequence ends on the subdominant). Such an implicit “grammar” may be understood and successfully used even in the absence of an explicit ability to formulate the underlying rules. Taken together, the authentic cadence is a standard element of musical harmony, and violations of it are expected to stand out, even for listeners with little explicit musical knowledge.

Böckmann-Barthel, Ziese, Rostalski, Arens, and Verhey (2013) found that the ability to discriminate chords seems to depend to a small degree on the coding strategy of the CI device but that CI users were in general less accurate in discriminating chords than NH listeners. A poorer performance in chord discrimination of CI user than NH listeners was also found in Brockmeier et al. (2011). In addition to chord discrimination, they asked their listeners to rate the consonance of different chords. CI users responded similar to the NH participants but less accurately for different levels of consonance. With the same test, no differences were found between prelingually deafened children using a CI and NH peers (Stabej et al., 2012). To date, only one study investigated how CI users process progressions of chords (Koelsch, Wittforth, Wolf, Müller, & Hahne, 2004). These authors studied event-related electroencephalographic (EEG) potentials in postlingually deafened CI users listening to common cadences. The appearance of a functionally irregular (Neapolitan sixth) chord elicited a similar but significantly reduced component as compared to normal-hearing subjects. They postulated that this potential was the correlate of implicit knowledge about musical harmony gained before the occurrence of the hearing loss. However, they did not ask the listeners about their experience of the sounds (i.e., how sensitive they were to the tested irregularities).

The present study investigates to what extent CI users experience fundamental harmonic elements such as vertical chord consonance and authentic cadences in the context of their peripheral constraints in hearing. Two psychoacoustic tasks are used: In the first task, participants selected the more harmonically pleasant item from pairs of different chords. It is hypothesized that, due to the signal degradation of the CI, listeners are less pronounced in the judgment of chord types than the normal-hearing listeners. In the second task, short sequences of chords were presented, half of which ended with a V-I cadence. In the remaining presentations, the final chord was modified into a deviating ending. The task was to decide whether the final chord ends the phrase conclusively. In two conditions, the sequence was modified by either using a dissonant chord or by a consonant chord outside the frame of the tonal center as the final chord. These two conditions test if the CI listeners use only one of these cues for solving the task. It is hypothesized that normal-hearing listeners judge the perfect cadence as conclusive whereas modified final chords provoke a resolution and therefore are judged as inconclusive.

Method

PARTICIPANTS

Fifteen CI users aged 24 to 75 years (median = 61 years) participated in the study. All participants were bilaterally postlingually deaf and listened through implant systems by MED-EL AG (Innsbruck, Austria) with ten to twelve active electrodes steered by the OPUS 2 sound processor. The frequency range of the acoustic signals used by the CI devices was 100 Hz to 8500 Hz in all participants except for CI2, whose upper limit was 7000 Hz. Demographic data including the sound processing strategies are provided in Table 1. All participants were using a coding strategy transmitting temporal fine structure to the two (FSP) or four (FS4) most basal electrodes (cf. the 5th column of Table 1). This sound coding appears to be slightly advantageous in musical tasks, especially discriminating chords (Böckmann-Barthel et al., 2013). Only participant CI 11 had FSP available in the lowest electrode only. All had a minimum experience of 11 months with the device. All CI users passed the standard Freiburg monosyllabic speech test (i.e., their word recognition score was better than 30%). Speech perception in noise was assessed with the Oldenburg matrix sentence test, using a closed-set design with the speech level fixed at 65 dB and adaptive variation of the noise level (Wagener & Brand, 2005). The individual speech reception threshold (SRT) values are given in Table 1. No unilaterally implanted CI user had a usable residual hearing on the contralateral side. Bilateral CI users were asked to turn off the device on their less experienced side. All participants were Caucasoid, native German speakers, and had visited a German school. Noteworthy, in Germany, obligatory music classes in school generally do not include individual training of an instrument or singing, but rather knowledge of folk songs and theoretical background. Musical experience was assessed with an excerpt from a questionnaire developed for CI listeners (Brockmeier et al., 2007). Based on those data, none of the participants reported any music training before or after implantation apart from obligatory classes. However, one participant (CI11) had taken up individual keyboard training recently.

TABLE 1.
Demographic Data and Cochlear Implant Devices of the CI Users Participating in the Present Study
ID SexAge
(years)
Duration of CI use (years; months)Processing strategyEtiologySide of ImplantImplant typeActive
electrodes
SRT
(dB)
speech in quiet (%)
CI1  m 64 12;8 FSP Blast trauma  Right C40+ 11 +6.9 75 
CI2  f 74 9;10 FSP Progressive hearing loss  Left C40+ 10 +3.4 45 
CI3  m 55 2;11 FSP Progressive hearing loss  Right SONATA ti100 12 +0.9 87 
CI4  m 72 0;11 FS4 Progressive hearing loss  Left CONCERTO 12 -2.0 65 
CI5  f 75 3;5 FSP Progressive hearing loss  Left SONATA ti100 11 +1.5 60 
CI6  m 66 1;11 FSP Progressive hearing loss  Left SONATA ti100 12 +0.1 76 
CI7  m 58 6;0 FS4 Gentamicin therapy  Left PULSAR ci100 11 +0.7 85 
CI8  f 71 13;7 FSP Progressive hearing loss  Right C40+ 11 +2.9 55 
CI9  f 24 5;0 FS4 Hereditary hearing impairment  Right PULSAR ci100 12 -3.6 50 
CI10  f 68 3;0 FSP Progressive hearing loss  Left SONATA ti100 12 +8.0 46 
CI11  f 61 9;0  FSP 1 Hereditary hearing impairment  Left C40+ 11 +9.3 52 
CI12  m 47 1;5 FS4 Hereditary hearing impairment  Right CONCERTO 12 +2.7 52 
CI13  f 57 4;8 FSP Progressive hearing loss  Left PULSAR ci100 11 -1.5 61 
CI14  m 69 1;4 FS4 Progressive hearing loss  Left CONCERTO 12 +6.9 50 
CI15  m 70 1;6 FS4 Progressive hearing loss  Right CONCERTO 11 -1.5 80 
ID SexAge
(years)
Duration of CI use (years; months)Processing strategyEtiologySide of ImplantImplant typeActive
electrodes
SRT
(dB)
speech in quiet (%)
CI1  m 64 12;8 FSP Blast trauma  Right C40+ 11 +6.9 75 
CI2  f 74 9;10 FSP Progressive hearing loss  Left C40+ 10 +3.4 45 
CI3  m 55 2;11 FSP Progressive hearing loss  Right SONATA ti100 12 +0.9 87 
CI4  m 72 0;11 FS4 Progressive hearing loss  Left CONCERTO 12 -2.0 65 
CI5  f 75 3;5 FSP Progressive hearing loss  Left SONATA ti100 11 +1.5 60 
CI6  m 66 1;11 FSP Progressive hearing loss  Left SONATA ti100 12 +0.1 76 
CI7  m 58 6;0 FS4 Gentamicin therapy  Left PULSAR ci100 11 +0.7 85 
CI8  f 71 13;7 FSP Progressive hearing loss  Right C40+ 11 +2.9 55 
CI9  f 24 5;0 FS4 Hereditary hearing impairment  Right PULSAR ci100 12 -3.6 50 
CI10  f 68 3;0 FSP Progressive hearing loss  Left SONATA ti100 12 +8.0 46 
CI11  f 61 9;0  FSP 1 Hereditary hearing impairment  Left C40+ 11 +9.3 52 
CI12  m 47 1;5 FS4 Hereditary hearing impairment  Right CONCERTO 12 +2.7 52 
CI13  f 57 4;8 FSP Progressive hearing loss  Left PULSAR ci100 11 -1.5 61 
CI14  m 69 1;4 FS4 Progressive hearing loss  Left CONCERTO 12 +6.9 50 
CI15  m 70 1;6 FS4 Progressive hearing loss  Right CONCERTO 11 -1.5 80 

Note: All sound processor strategies transmitted temporal fine structure to four (FS4), two (FSP), or a single basal electrode (FSP 1). SRT: Speech reception threshold in noise, determined with the Oldenburg sentence test. Speech perception in quiet determined with the Freiburg monosyllabic speech test.

Nineteen participants aged 22 to 68 years (median = 60 years) with age-related normal hearing (NH, according to DIN EN ISO 7029) served as a control group. These listeners were selected with respect to a similar age distribution and level of music education as the CI participants. All participants were Caucasoid, native German speakers, and had visited a German school. Eight of the NH participants reported individual music training (instrumental or singing) during their youth, six of those for more than three years. Three of the latter reported ongoing musical activity.

Written informed consent to the study was given by each individual participant before the measurement. The local institutional review board approved the study to fulfill the declaration of Helsinki.

APPARATUS

The experiments were conducted in a large sound-attenuated room. Sounds were presented through a single frontal active monitor loudspeaker (Reveal 6D, Tannoy Ltd., Coatbridge, UK). A sound level comfortable to the CI listeners (of about 70 dB SPL) was adjusted in a preliminary test, and kept fixed for all participants. While good audibility for all participants was ensured, the absolute level sound is of little information, since in all CI devices an automatic gain control maps the original sound dynamics to a very narrow range below 20 dB (Wilson & Dorman, 2008). Instructions were provided and responses were given on a touchscreen monitor display in front of the participant. MATLAB (The Mathworks Inc., Natick MA, USA) was used for stimulus generation, presentation, and response collection. CI users require a rather constant and spectrally simple signal to perceive a clear pitch, whereas the normal-hearing listeners might be confounded by the unusual sound of a chord made from pure tones. Therefore, the chords were constructed synthetically from harmonic complex tones of only five partials with frequencies 1 f0 to 5 f0 and a rather steep decay of 6 dB per partial. Each chord was constructed from four such complex tones. The equal temperament scale was used throughout for the fundamentals of the tones. The chords were arranged in open harmony to increase the spectral distance between the tones. Each chord was gated using 50 ms raised-cosine ramps at onset and offset.

HARMONIC PREFERENCE TASK

Six typical musical chord types were used: major, minor, augmented, diminished, suspended fourth, and diminished fifth. All consisted of four tones and differed from the major chord in root position by shifting one or two of the tones by one semitone. Figure 1A displays the respective musical scores of each of the chords. The root note was either 147 Hz or 185 Hz, corresponding to a musical note D or F♯, respectively. The chords were presented pairwise. Each chord had a duration of 1600 ms and the chords of a pair were separated by a 1600 ms silence interval. All possible combinations of the chord types were used so that each chord type occurred 20 times. Within a trial, the root note was the same for both items. After the presentation of a pair, the participants were asked by a graphical user interface to indicate which of the two chords sounded harmonically more pleasant (in German: “harmonischer”). The choice options were “sound 1” or “sound 2.” A repeated listening to a pair was not allowed.

FIGURE 1.

Example scores of the chord sequences used in the experiment. In panel A, the six different chords used in the Harmonic Preference task are shown (in D major scale): major (maj), minor (min), augmented (aug), diminished (dim), suspended fourth (sus4), and diminished fifth (b5). Panels B to D illustrate the cadence judgment task. Panel B shows an authentic cadence that was expected to be judged as conclusive. Panel C shows the modified sequences of the shifted condition 1, where the whole final chord was transposed by one or two semitones upward (sh+1 and sh+2, respectively) or downward (sh-1 and sh-2, respectively). The four alternatives are shown in C. In the substituted condition, shown in panel D, the final chord was replaced by an augmented (aug1 or aug2) or a diminished chord (dim1 or dim2).

FIGURE 1.

Example scores of the chord sequences used in the experiment. In panel A, the six different chords used in the Harmonic Preference task are shown (in D major scale): major (maj), minor (min), augmented (aug), diminished (dim), suspended fourth (sus4), and diminished fifth (b5). Panels B to D illustrate the cadence judgment task. Panel B shows an authentic cadence that was expected to be judged as conclusive. Panel C shows the modified sequences of the shifted condition 1, where the whole final chord was transposed by one or two semitones upward (sh+1 and sh+2, respectively) or downward (sh-1 and sh-2, respectively). The four alternatives are shown in C. In the substituted condition, shown in panel D, the final chord was replaced by an augmented (aug1 or aug2) or a diminished chord (dim1 or dim2).

CADENCE JUDGMENT TASK

Authentic cadence sequences (i.e., tonic – subdominant – dominant – tonic) of four-note chords were constructed, with the final chord identical to the major chord in the harmonic preference task. An example of authentic cadence sequences is given in Figure 1B. To avoid habituation, the key was selected randomly from D/D♯ /E/F/F♯. Each of the first three chords was 1000 ms long. The chords were separated by 200-ms silence intervals. The duration of the final chord was 1600 ms.

In 50% of the presentations, an authentic cadence was presented. In the other 50% of the presentations, the final chord of the authentic cadence was replaced by a music-syntactically irregular chord in two different conditions. Examples of these modified cadences are shown in Figures 1C and 1D. In the “shifted” condition 1, the modified chord was the tonic transposed by one or two semitones upward (denoted as sh+1 and sh+2 in Figure 1C, respectively) or downward (sh-1 and sh-2). Thus, although it was a simple major chord, it no longer matched the harmonic center defined by the first three chords of the cadence. Therefore, simple notion of a final dissonance did not enable the listener to detect the distractor version. In the “substituted” condition 2, the final chord was replaced by an augmented or diminished chord. A listener only using sensory dissonance of the final chord as a cue would succeed in this condition but fail in the first condition. For each of these, two transposed variants were used: Either the root note corresponded to the expected concluding major chord (denoted as dim1 and aug1 in Figure 1C), or the fifth (for the dim2 and aug2 chord) matched the expected chord (Figure 1D). The melody within the soprano part or the bass part was identical to the authentic example and could not serve as a cue to detect the modified version. Thus, a listener who notices a modification of the soprano or bass part but does not experience the sensory dissonance would perform at a lower level in the “substituted” condition than in the “shifted” condition. For each condition, 16 authentic cadences and 16 deviant cadences were presented.

After the presentation of each cadence, the participants were asked to decide whether the ending was harmonically conclusive (i.e., with no expectancy for continuation) or inconclusive (i.e., with expectancy for continuation) by selecting the German terms “schlüssig” or “offen” on a graphical user interface. Responses were given on a MATLAB graphics user interface. A training sequence of seven examples was provided before the measurement. In six of the CI listeners (and none of the control listeners) these examples were repeated to make sure the task was understood.

DATA ANALYSIS

In the harmonic preference task, the responses of each individual participant were converted into a preference score for each chord type. Similar to Tufts, Molis, and Leek (2005), this score was obtained by counting the number of pairs in which this chord type was judged more pleasant minus the number of pairs in which this chord type was judged less pleasant. Thus, a score range of -20 to 20 was available for each chord type. Note that since all chords were presented with equal number, the sum of the scores of the six chord types was zero by definition for each subject. Within each group a repeated-measures analysis of variance (ANOVA) on the factor “chord type” (six stages) tested the null hypothesis that the different chord types were equally preferred. Greenhouse-Geisser corrections were applied when appropriate. Separate independent sample t-tests tested the null hypothesis that a chord type received different ratings by the two groups of participants. In the case of significant effects, effect sizes were calculated as partial eta-squared (partial ηp2) for the ANOVAs and correlation coefficient r for post hoc pairwise comparisons.

In the cadence judgment task, hit rates and false alarm rates were determined as follows: “Satisfactory” responses were scored as hits if given in response to an authentic cadence, and as false alarm if given in response to a deviant cadence. The individual hit rate (HR) was calculated as the number of hits divided by the total number of authentic cadences presented. In the same way, the individual false alarm rate (FR) was the ratio of the number of false alarms and the total number of deviant cadences presented. The sensitivity index d’ = z(HR) − z(FR) according to signal detection theory was calculated for each participant and condition. Here, perfect rates of 1 or 0 must be corrected in order to avoid infinite d’ values. Therefore, false alarm rates of 0 were replaced by 1/(2n), where n represented the number of deviant cadences presented. In the same way, hit rates of 1 were replaced by 1 - 1/(2m), m being the number of authentic cadences presented (cf. Stanislaw & Todorov, 1999). With this correction, perfect performance results in d’ = 3.725. The null hypothesis that CI participants were as accurate as NH participants in classifying cadences into satisfactory or unsatisfactory, and that the deviant chords in both conditions were as easily detected was assessed by a mixed-design ANOVA with a within-subjects factor “condition” (two stages) and a between-subjects factor “group” (CI users or NH controls).

Results

HARMONIC PREFERENCE TASK

Figure 2 shows the results of the first task, where the listeners were asked to choose the more harmonically pleasant chord of a pair. The figure displays the total scores of each chord type averaged for the CI users and the NH control group. Generally, NH participants selected the major chord and the minor chord as the most pleasant chord types. The group of CI participants only assigned a positive score to the major chord. Separate repeated-measures ANOVAs revealed a significant, medium-sized effect of the within-subjects factor “chord type” on the NH group, F(5, 90) = 14.29, p < .001, ηp2 = .44. Within the CI group, the differences were also significant, however with a small effect size, F(5, 70) = 5.32 p < .01 ηp2 = .28.

FIGURE 2.

Total preference scores of different chord types for the CI participants (left) and the NH control participants (right). The group mean is shown. Error bars indicate standard deviations.

FIGURE 2.

Total preference scores of different chord types for the CI participants (left) and the NH control participants (right). The group mean is shown. Error bars indicate standard deviations.

Post hoc t-tests (with Bonferroni's correction) of the chord type ratings within the NH group revealed no significant difference between the minor and the major chord. These chord types were rated significantly different from all other chord types at a p <.05 level. This result indicates that, as expected, major and minor chords were appraised as the most consonant ones. The CI participants rated the major chord as significantly more consonant than the minor and the augmented chord at a p < .05 level. No other differences were significant. This indicates that, for the CI users, the major chord stands out in harmonic pleasantness from other investigated chords (i.e., also the minor chord). The correlation coefficients of these two chord pairs show that these differences are substantial and the small overall effect size found above is due to the fact that only these two pairs are different. Details of the post hoc tests are found in Table 2. In independent-sample t-tests comparing the respective scores of the two groups for each chord type, only the minor chord was rated less pleasant by the CI participants, t(32) = 4.13, p <.001, r = .59. No other differences between the groups were found.

TABLE 2.
Post hoc t-test Results of the Individual Chord Pairs From the Harmonic Preference Task
CI(df = 14)NH(df = 18)
Chord pair   t p r   t p r 
maj – min   3.99 < .05 .73   0.72 ns — 
maj – aug   3.80 < .05 .71   5.80 < .001 .81 
maj – dim   3.48 ns —   4.66 < .01 .74 
maj – sus4   2.67 ns —   5.49 < .001 .79 
maj – b5   2.93 ns —   4.26 < .01 .71 
min – aug   0.55 ns —   4.10 < .001 .70 
min – dim  −0.58 ns —   3.56 < .05 .64 
min – sus4  −1.17 ns —   5.24 < .05 .78 
min – b5  −0.71 ns —   4.74 < .01 .75 
sus4 – aug   1.58 ns —   0.96 ns — 
sus4 – dim   0.76 ns —  −0.79 ns — 
sus4 – b5   0.58 ns —   0.29 ns — 
aug – dim  −1.02 ns —  −1.80 ns — 
aug – b5  −1.27 ns —  −0.67 ns — 
dim – b5  −0.16 ns —   1.24 ns — 
CI(df = 14)NH(df = 18)
Chord pair   t p r   t p r 
maj – min   3.99 < .05 .73   0.72 ns — 
maj – aug   3.80 < .05 .71   5.80 < .001 .81 
maj – dim   3.48 ns —   4.66 < .01 .74 
maj – sus4   2.67 ns —   5.49 < .001 .79 
maj – b5   2.93 ns —   4.26 < .01 .71 
min – aug   0.55 ns —   4.10 < .001 .70 
min – dim  −0.58 ns —   3.56 < .05 .64 
min – sus4  −1.17 ns —   5.24 < .05 .78 
min – b5  −0.71 ns —   4.74 < .01 .75 
sus4 – aug   1.58 ns —   0.96 ns — 
sus4 – dim   0.76 ns —  −0.79 ns — 
sus4 – b5   0.58 ns —   0.29 ns — 
aug – dim  −1.02 ns —  −1.80 ns — 
aug – b5  −1.27 ns —  −0.67 ns — 
dim – b5  −0.16 ns —   1.24 ns — 

Note: Rows 2–4: CI listeners, row 5–7: NH listeners, respectively. p significance levels with Bonferroni correction. Correlation coefficients r are given in case of significant differences.

CADENCE JUDGMENT TASK

Figure 3 shows the result of the second task, where the participants were asked to point out whether a cadence of four harmonies ended conclusively or inconclusively. In general, participants of the NH control group showed a good performance in the task with high hit rates and low false-alarm rates, resulting in a mean d’ = 2.95 (SD = 0.94) in the “shifted” condition 1 and d’ = 3.29 (SD = 0.63) in the “substituted” condition 2. A one-sample two-sided t-test of the d’ values against d’ = 3.73 showed that these values were still significantly below perfect performance in both conditions: “shifted” condition, t(18) = −3.58 p < .01; “substituted” condition, t(18) = −2.41 p < .05. Only one participant (NH10) obtained a sensitivity index d’ smaller than one in both conditions (i.e., this participant was unable to do this task). No significant difference between the two conditions was found in a paired t-test, t(18) = −1.35, p > .05.

FIGURE 3.

Individual results of the cadence judgment task. The sensitivity indices (d’) for conditions 1 (“shifted”) and 2 (“substituted,” black and grey circles, respectively) were calculated from the hit rates HR (authentic cadences identified as conclusive) and false alarm rates FR (modified cadences judged as conclusive). The top panel shows the results of the CI participants, and the bottom panel shows those of the NH participants. Except for one participant (CI04) they show significant lower indices than NH participants.

FIGURE 3.

Individual results of the cadence judgment task. The sensitivity indices (d’) for conditions 1 (“shifted”) and 2 (“substituted,” black and grey circles, respectively) were calculated from the hit rates HR (authentic cadences identified as conclusive) and false alarm rates FR (modified cadences judged as conclusive). The top panel shows the results of the CI participants, and the bottom panel shows those of the NH participants. Except for one participant (CI04) they show significant lower indices than NH participants.

In contrast to the NH participants, the CI users had high false alarm rates and hit rates near chance level. This resulted in d’ values close to or below one (top panel of Figure 3). Only participant CI4 reached a d’ value greater than one in both conditions. On average, the group of CI participants reached a mean score of 0.2 (SD = 1) in the “shifted” condition 1 and of 0.3 (SD = 0.6) in the “substituted” condition 2 (Figure 4). In a one sample two-sided t-test, these results were significantly below the upper border value of unreliable performance (d’ = 1) in both conditions; “shifted” condition, t(14) = −3.06 p < .01; “substituted” condition, t(14) = −4.87 p < .01. The difference of the two conditions was not significant, t(14) = −0.26 p > .05.

FIGURE 4.

Average sensitivity index values of the cadence judgment task for CI and NH participants in condition 1 (“shifted”) and 2 (“substit” (uted), black and grey circles, respectively).

FIGURE 4.

Average sensitivity index values of the cadence judgment task for CI and NH participants in condition 1 (“shifted”) and 2 (“substit” (uted), black and grey circles, respectively).

The differences between the two groups of participants were verified by a mixed-model ANOVA with the within-subject factor “condition” (1 or 2) and the “group” factor (NH or CI participants). The ANOVA revealed a significant and large effect of the factor “group” was significant, F(1, 32) = 95.98, p < .001, ηp2 = .68, whereas the factor “condition” was not significant, F(1, 32) = 1.09, p > .05, and neither was the interaction of both factors, F(1, 32) = 0.39, p > .05.

Discussion

The present study investigated the consonance of single chords (vertical dimension of harmony) and perception of cadences (horizontal dimension of harmony). The results of the two experiments are in the corresponding two following sections of this Discussion. The last section of the Discussion focusses on individual differences.

HARMONIC PREFERENCE TASK

In this task, isolated chords were compared in pairs with respect to their harmonic pleasantness. Here, the NH participants judged the major chord and the minor chord as the most pleasant types and the other chords like the augmented or diminished chords at a lower pleasantness level. This is in agreement with musicological expectation and with previous experiments on the vertical consonance of different chord types (Cook, Fujisawa, & Konaka, 2007; Johnson-Laird, Kang, & Leong, 2012; Roberts, 1986). The CI participants on average judged the major chord with highest pleasantness, consistent with the NH users. Unlike NH participants, the minor chord was rated as less pleasant than the major chord and on a level with the remaining chords. Previous work with CI users has addressed the pleasantness of chords only with the Mu.S.I.C. Perception Test, using piano chords (Brockmeier et al., 2011, Stabej et al., 2012). Both studies concluded some ability to recognize various dissonances but reported only averaged ratings without specifying the used chord types. Brockmeier et al. (2011) also presented pairs of piano chords and showed that, to some degree, the CI participants discriminated different chords. Notably, most of the items were not of a musically functional chord type (such as a minor, augmented, or sept chord), and the pair items could differ by number of contained tones as well as their pitch range. The present investigation extends the previous studies by showing explicitly that the major chord is judged as significantly more pleasant than other musical chord types by the CI users.

The results of the CI users may be influenced by sensory factors as well as acquired, non-sensory factors. It has been argued that sensory consonance of two tones in NH participants arises from beatings between the partials falling into a common auditory filter (Plomp & Levelt, 1965) or by the regularity of the spectrum (McDermott et al., 2010). Because the filter widths of cochlear implants are significantly larger than the critical bandwidth of NH listeners, and the CI sound processing discards accurate frequency information of tone complexes (Moore, 2008), it is unlikely that the harmonic regularity is a cue for CI users. However, a contribution of roughness to dissonance is possible since temporal amplitude fluctuations are encoded in the CI system, also without special coding of temporal fine structure (Drennan, Won, Nie, Jameyson, & Rubinstein, 2010). In this context, low rating of the minor chord may be explained by roughness as follows: The fifth partial of the root note is, in the higher of the keys used, at 925 Hz. Together with the fundamental frequency of the third at 880 Hz, it could generate audible beating with a frequency of 45 Hz, in particular as both components fall on the same electrode for the present electrode mappings. For the major chord, the fundamental frequency of the third is at 932 Hz and with the fifth partial of the root note would produce beatings of 7 Hz (i.e., a frequency associated with fluctuation strength rather than with roughness; Fastl & Zwicker, 2007). In the other four chord types, several interacting frequency pairs are found that could generate a roughness impression. This argument is also applicable to the lower of the keys in the task. It is likely that other factors also influence perception of the chords, such as the non-physiological filter shapes of the sound processors and increased roughness sensation due to pathological changes at the stage of the auditory nerve. It also should be pointed out that the present experiment does not allow disentangling of the sensory and non-sensory factors of vertical consonance. By addressing the pleasantness of chords, our experiment also relies on an aesthetic experience of the participants. More data on how CI listeners perceive roughness and amplitude fluctuations are required to clarify the role of this sensation on pleasantness ratings of chords.

Existing models of chord consonance combine the factors of beatings between the components, regular harmonic structure (Johnson-Laird et al., 2012; Mc-Dermott et al., 2010), and the interval symmetry of the chords (Cook et al., 2007). The application to the present results of CI listeners would ask for an adaptation of these models to the fixed filters of the CI processor. Since some dependence of the resulting consonance on the chord position has been reported (Cook et al., 2007, Johnson-Laird et al., 2012), future experiments should also consider investigating different positions.

CADENCE JUDGMENT TASK

The interplay of harmonic tension and release is mediated by harmonic relations like the authentic cadence that generally elicits a feeling of a phrase ending (cf. Bigand et al., 1996; Patel, 2003). Therefore, the second experiment tested how convincing different chords served as the conclusive end of a conventional cadential phrase of Western harmony. Half of the cadences were authentic, ending with the major tonic. In the other half, the final chord was replaced by either a transposed major chord (“shifted” condition 1) or a dissonant chord (“substituted” condition 2). As expected, most of the NH participants reliably judged perfect cadences as conclusive endings and modified cadences as inconclusive ones. The high sensitivity index d’ indicates that these participants differentiated authentic from modified cadence sequences with ease. Those listeners showing a somewhat reduced performance (with d’ < 3 in both conditions) had all reported no individual music training in youth.

The results of the CI users differ considerably from those of the NH participants. One participant (CI4) achieved nearly perfect performance with a high d’ in in the “shifted” condition 1. The same participant and one other (CI5) achieved at least d’ > 1 in the “substituted” condition 2. All other participants gained d’ < 1 in both conditions and responded near chance level with d’ values below unity. Thus, only a single CI user performed on a level with the NH participants (in the “shifted” condition) or slightly below (in the “substituted” condition).

Low values of the sensitivity index could be related to a substantial bias of the criterion (i.e., a tendency either to tolerate even the modified cadences as satisfactory) or to reject even the authentic cadences as unsatisfactory. For detailed insight, the hit rates were plotted versus the false alarm rates in Figure 5 (cf. Figure 2 of Stanislaw & Todorov, 1999). Data points of perfect performance (i.e., hit rates of one and false alarm rates of zero) are found in the upper left corner. A very tolerant criterion would generate data points near the upper right corner, and a very intolerant criterion near the lower left corner. The data of the CI participants are found in the center of the plot, thus indicating no general bias to one or the other side. The data points of the NH users, in contrast, cluster around the upper left corner of perfect performance.

FIGURE 5.

Hit rates (HR) are shown as a function of false-alarm rates (FR) for CI users (top panels) and NH participants (bottom panels) in condition 1 (“shifted,” left panels, black circles) and 2 (“substituted,” right panels, grey circles) of the cadence judgment task. Larger circles indicate multiple occurrence of the data point. The large grey circle in the bottom right panel represents ten occurrences. It hides two adjacent data points of single participants.

FIGURE 5.

Hit rates (HR) are shown as a function of false-alarm rates (FR) for CI users (top panels) and NH participants (bottom panels) in condition 1 (“shifted,” left panels, black circles) and 2 (“substituted,” right panels, grey circles) of the cadence judgment task. Larger circles indicate multiple occurrence of the data point. The large grey circle in the bottom right panel represents ten occurrences. It hides two adjacent data points of single participants.

The data indicate that most CI users have great difficulty in judging the conclusiveness of a cadence (i.e., judging horizontal harmony). It is likely that the poor recognition of authentic cadences in CI participants is due, in part, to the lack of clarity in pitch perception. Numerous studies have demonstrated reduced performance of CI users in tasks requiring the recognition of musical pitch, such as in melody recognition or discrimination tasks (cf. Limb & Roy, 2014). As mentioned at the beginning of this paper, one study found electrophysiological correlates of the processing of irregular chords within a sequence in their group of CI users (Koelsch et al., 2004). These data indicate that, to some extent, harmonic information is available to CI users. In the present study, it was shown in a behavioral experiment that only one CI listener was able to identify deviations from the conventional harmonic. This indicates that the majority of the CI users are insensitive to these deviations. This finding seems to be at odds with Koelsch et al. (2004) who found a similar component in the EEG responses to an irregular chord as in NH subjects. This component was reduced but present in the CI listeners. A possible explanation for the seemingly contradicting results is that changes of the final chord in the present experiment are less salient than the occurrence of other chords, such as the Neapolitan chord as used by Koelsch et al. (2004). Because CI users judged all other chord types as less harmonically pleasant than the major chord in the harmonic preference task, one might have expected solid performance in the “substituted” condition 2 of the cadence judgment task, and especially low false alarm rates. However, the top panels of Figure 5 show that the false alarm rates are as high as in the “shifted” condition 1. This is also seen in the two participants CI4 and CI5 who succeeded at least in one condition (Figure 6A). No evidence is found that high false alarm rates in the “substituted” condition 2 explain part of the performance. The results suggest that the evaluation of cadences represents a more complex task for the CI participants than a rating of the vertical consonance of a single chord. Because authentic cadences can be learned implicitly by exposure (Bigand & Poulin-Charronnat, 2006), the CI users may at least have had too little exposure to musical phrases before or after implantation to form a standard model for a cadence. It is possible that explicit training of musical phrases could make harmonic syntax more available to CI users.

FIGURE 6.

Panel A displays individual hit rates (HR) and false alarm rates (FR) of the two individual participants CI4 and CI5 with the best results. Black bars show results for the “shifted” condition 1, grey bars those of the “substituted” condition 2. Panel B shows the pattern of chord scores from the harmonic preference task of participant CI4.

FIGURE 6.

Panel A displays individual hit rates (HR) and false alarm rates (FR) of the two individual participants CI4 and CI5 with the best results. Black bars show results for the “shifted” condition 1, grey bars those of the “substituted” condition 2. Panel B shows the pattern of chord scores from the harmonic preference task of participant CI4.

The beneficial influence of acoustic experience on musical listening is bolstered with the finding that prelingually deafened children were less accurate than their postlingually deafened peers as well as postlingually deafened adults in melody recognition (Olszewski, Gfeller, Froman, Stordahl, & Tomblin, 2005). More experiments on how prelingually deafened listeners perceive complex musical sounds through a cochlear implant are required to yield more insights into these questions.

WHAT FACTORS AFFECT THE PERFORMANCE OF INDIVIDUAL CI USERS?

A single participant, CI4, showed outstanding performance in the cadence judgment task (whereas another participant, CI5, performed at least slightly better in one condition but not the other). The performance of participant CI4 raises the question of what might distinguish the participant from the rest of the CI group. First, the technical and audiological prerequisites were optimal in this case: He had suffered progressive deafness at an advanced age only, and notwithstanding rather short device experience, had one of the best speech perception thresholds in noise. All of the twelve electrodes of the device were active. This is not generally the case in CI users. The participant also used the most recent stimulation mode FS4, which enabled the transmission of the sound's temporal fine structure to four electrodes representing the lowest frequency bands. Melody performance at higher pitches seems to benefit from increased numbers of electrodes (Singh, Kong, & Zeng 2009). On the other hand, even the participants using the FSP stimulation could use temporal fine structure to some degree, and no study up to now has reported a reliable difference in musical tasks between those stimulation strategies (cf. Riss et al., 2014). The harmonic preference scores of CI4 are similar to the average of the CI participants, with the exception that the diminished fifth chord is given the highest rating (see Figure 6B). Nevertheless, he was still able to differentiate the sound of the chords and judged the major chord as the most harmonic one.

Gfeller et al. (2008) investigated different parameters that can improve performance in different musical tasks. Among those, good speech perception in noise was beneficial, for example, for recognition of instrument timbre, and so was music training. Participant CI4 had indeed good speech perception in noise. In the questionnaire used to assess the participants’ musical experience (Brockmeier et al., 2007), participant CI4 reported, however, no special music training before or after implantation, no distinctive attitude to music, and no involvement in musical activities, and so did all other participants. Thus the good technical predispositions and high speech perception performance of this CI user are probably the most important parameters for succeeding in the recognition of authentic cadences.

Note that the present results should not be taken as evidence that for most CI users the enjoyment of polyphonic music through a CI is impossible. Numerous reports have found that CI users enjoy listening to music at home, attend concerts, and go dancing (e.g., Lassaletta et al., 2007). Many of them are well aware of the advantage in comparison to when they could hardly perceive any sound. Other cues such as rhythm or melody also add to the perception of musical structure (Boltz, 1993) but were avoided in the present experiment. Notably, the prolonged duration of a final chord should help CI users to identify the end of a phrase. However, the apparently degraded perception of harmony in electric hearing abates the emotional as well as the structural information within a musical piece. It remains to be seen in future research if a specific music training overcomes the problems observed in the experiments of the present study.

Summary and Conclusion

The present work investigated two aspects of harmony perception in patients suffering from severe hearing loss who hear by using electrical stimulation of the auditory nerve via a cochlear implant (CI). The results indicate that the substantial reduction of sound information has a larger impact on the vertical dimension of harmony (for single chords), and, more pronounced, on the horizontal dimension of harmony (for chord sequences). Isolated major chords are perceived as more consonant than chords that are expected to be musically dissonant. Interestingly, this is not true for minor chords. Thus, only to a certain extent the vertical consonance of different chord types is preserved in electric hearing. The results of the cadence judgment task indicate that, within a cadence, harmonic modifications of the final chord of the cadence are easily distinguished from the authentic cadence by normal-hearing participants but not differentiated by CI participants (i.e., CI users seem to have great difficulties to perceive the horizontal dimension of harmony). Overall, the results may substantially add to understanding the limited enjoyment of music perception with a CI. A promising result of the present study is that in general, the CI users judge—as normal-hearing listeners do—the major chord as harmonically more pleasant than the augmented, diminished, suspended fourth, and diminished fifth chords. This indicates that some information on harmony is already available to CI users. Based on this result, and on the good result of one CI user in the cadence judgment task, future research could examine whether focused music training would enable more CI users to recognize authentic cadences and improve the processing of musical syntax.

References

References
Bharucha, J. J., & Stoeckig, K. (
1986
).
Reaction time and musical expectancy: Priming of chords
.
Journal of Experimental Psychology
,
12
,
403
410
.
Bigand, E., Parncutt, R., & Lerdahl, F. (
1996
).
Perception of musical tension in short chord sequences: The influence of harmonic function, sensory dissonance, horizontal motion, and musical training
.
Perception and Psychophysics
,
58
,
125
141
.
Bigand, E., & Pineau, M. (
1997
).
Global context effects on musical expectancy
.
Perception and Psychophysics
,
59
,
1098
1107
.
Bigand, E., Poulin, B., Tillmann, B., Madurell, F.,& D'adamo, D. A. (
2003
).
Sensory versus cognitive components in harmonic priming
.
Journal of Experimental Psychology
,
29
,
159
171
.
Bigand, E., & Poulin-Charronnat, B. (
2006
).
Are we “experienced listeners”? A review of the musical capacities that do not depend on formal musical training
.
Cognition
,
100
,
100
130
.
Böckmann-Barthel, M., Ziese, M., Rostalski, D., Arens, C., & Verhey, J. L. (
2013
).
Melody and chord discrimination of cochlear implant users in different pitch ranges
.
Cochlear Implants International
,
14
,
246
251
.
Boltz, M. G. (
1993
).
The generation of temporal and melodic expectancies during musical listening
.
Perception and Psychophysics
,
53
,
585
600
.
Brockmeier, S. J., Fitzgerald, D., Searle, O., Fitzgerald, H., Grasmeder, M., Hilbig, S., et al (
2011
).
The MuSIC perception test: A novel battery for testing music perception of cochlear implant users
.
Cochlear Implants International
,
12
,
10
20
.
Brockmeier, S. J., Grasmeder, M., Passow, S., Mawmann, D., Vischer, M., Jappel, A., et al (
2007
).
Comparison of musical activities of cochlear implant users with different speech-coding strategies
.
Ear and Hearing
,
28
,
49S
51S
.
Cook, N.D., & Fujisawa, T. (
2006
).
The psychophysics of harmony perception: Harmony is a three-tone phenomenon
.
Empirical Musicology Review
,
1
,
106
126
.
Cook, N.D., Fujisawa, T.X., & Konaka, H. (
2007
).
Why not study polytonal psychophysics?
Empirical Musicology Review
,
2
,
38
44
.
Cooper, W. B., Tobey, E., & Loizou, P. C. (
2008
).
Music perception by cochlear implant and normal hearing listeners as measured by the Montreal Battery for Evaluation of Amusia
.
Ear and Hearing
,
29
,
618
626
.
Corrigall, K.A., & Trainor, L. J. (
2010
).
Musical enculturation in preschool children: Acquisition of key and harmonic knowledge
.
Music Perception
,
28
,
195
200
.
Dahlhaus, C. (
1980
). Harmony. In S. Sadie (Ed.),
The new Grove's dictionary of music and musicians
(Vol.
8
, pp.
175
177
).
London, UK
:
Macmillan
.
Donnelly, P.J., Guo, B.Z., & Limb, C. J. (
2009
).
Perceptual fusion of polyphonic pitch in cochlear implant users
.
Journal of the Acoustical Society of America
,
126
,
EL128
EL133
.
Drennan, W. R., Won, J.H., Nie, K., Jameyson, E., & Rubinstein, J. T. (
2010
).
Sensitivity of psychophysical measures to signal processor modifications in cochlear implant users
.
Hearing Research
,
262
,
1
8
.
Fastl, H., & Zwicker, E. (
2007
)
Psychoacoustics – Facts and models
(pp.
259
261
).
Berlin
:
Springer
Galvin, J. J. III, Fu, Q. J., & Oba, S. (
2008
).
Effect of instrument timbre on melodic contour identification by cochlear implant users
.
Journal of the Acoustical Society of America
,
124
,
EL189
EL195
.
Galvin, J. J. III, Fu, Q.J., & Oba, S. (
2009
).
Effect of a competing instrument on melodic contour identification by cochlear implant users
.
Journal of the Acoustical Society of America
,
125
,
EL98
EL103
.
Gfeller, K., Oleson, J., Knutson J. F., Breheny P, Driscoll, V., & Olszewski, C. (
2008
).
Multivariate predictors of music perception and appraisal by adult cochlear implant users
.
Journal of the American Academy of Audiology
,
19
,
120
134
.
Jentschke, S., Friederici, A.D., Koelsch, S. (
2014
).
Neural correlates of music-syntactic processing in two-year-old children
.
Developmental Cognitive Neuroscience
,
9
,
200
208
.
Johnson-Laird, P.N., Kang, O.E., & Leong, Y. C. (
2012
).
On musical dissonance
.
Music Perception
,
30
,
19
35
.
Kang, R., Nimmons, G.L., Drennan, W., Longnion, J., Ruffin, C., Nie, K., et al (
2009
).
Development and validation of the University of Washington Clinical Assessment of Music Perception Test
.
Ear and Hearing
,
30
,
411
418
.
Koelsch, S., Wittfoth, M., Wolf, A., Muller, J., & Hahne A. (
2004
).
Music perception in cochlear implant users: An event-related potential study
.
Clinical Neurophysiology
115
,
966
972
.
Kong, Y.Y., Deeks J. M., Axon P. R., & Carlyon R. P. (
2009
).
Limits of temporal pitch in cochlear implants
.
Journal of the Acoustical Society of America
,
125
,
1649
1657
.
Lassaletta, L., Castro, A., Bastarrica, M., Pérez-Mora, R., Madero, R., De Sarriá, M., & Gavilan, J. (
2007
).
Does music perception have an impact on quality of life following cochlear implantation?
Acta Otolaryngologica
,
127
,
682
686
.
Limb, C.J., & Roy, A. T. (
2014
).
Technological, biological, and acoustical constraints to music perception in cochlear implant users
.
Hearing Research
,
308
,
13
26
.
Looi, V. (
2008
).
The effect of cochlear implantation on music perception: A review
.
Otorinolaringologia
,
58
,
169
190
.
Mcdermott, J. H., Lehr, A.J., & Oxenham, A. J. (
2010
).
Individual differences reveal the basis of consonance
.
Current Biology
,
20
,
1
7
.
Moore, B.C.J. (
2008
).
The role of temporal fine structure processing in pitch perception, masking, and speech perception for normal-hearing and hearing-impaired people
.
JARO
,
9
,
399
406
.
Olszewski, C., Gfeller, K., Froman, R., Stordahl, J., & Tomblin, B. (
2005
).
Familiar melody recognition by children and adults using cochlear implants and normal hearing children
.
Cochlear Implants International
,
6
,
123
140
.
Patel, A. D. (
2003
).
Language, music, syntax and the brain
.
Nature Neuroscience
,
6
,
674
681
.
Ping, L., Yuan, M., & Feng, H. (
2012
).
Musical pitch discrimination by cochlear implant users. Annals of Otology
,
Rhinology and Laryngology
,
121
,
328
336
.
Plomp, R., & Levelt, W. J. M. (
1965
).
Tonal consonance and critical bandwidth
.
Journal of the Acoustical Society of America
,
38
,
548
560
.
Rameau, J.-P. (
1722
)
Traité de l'harmonie
.
Paris
:
J. B. C. Ballard
.
Riss, D., Hamzavi, J. S., Blineder, M., Honeder, C., Ehrenreich, I., Kaider, A., et al (
2014
).
FS4, FS4-p, and FSP: A 4-month crossover study of 3 fine structure sound-coding strategies
.
Ear and Hearing
,
35
,
e272
e281
.
Roberts, L. A. (
1986
).
Consonance judgements of musical chords by musicians and untrained listeners
.
Acustica
,
62
,
163
171
.
Rockstro, W., Dyson, G., Drabkin, W., & Powers, H.S. (
1980
). Cadence. In S. Sadie (Ed.),
The new Grove's dictionary of music and musicians
(Vol.
3
, pp.
582
586
).
London, UK
:
Macmillan
.
Rosner, B.S., & Narmour, E. (
1992
).
Harmonic closure: Music theory and perception
.
Music Perception
,
9
,
383
410
.
Sears, D., Caplin, W.E., & Mcadams, S. (
2014
).
Harmonic closure: Music theory and perception
.
Music Perception
,
31
,
397
417
.
Singh, S., Kong, Y.-Y., & Zeng, F.-G. (
2009
).
Cochlear implant melody recognition as a function of melody frequency range, harmonicity, and number of electrodes
.
Ear and Hearing
,
30
,
160
168
.
Stabej, K.K. Smid, L., Gros, A., Zargi, M., Kosir, A., & Vatovec, J. (
2012
).
The music perception abilities of prelingually deaf children with cochlear implants
.
Otorhinolaryngology
,
76
,
1392
1400
.
Stanislaw, H., & Todorov, N. (
1999
).
Calculation of signal detection theory measures
.
Behavior Research Methods, Instruments, and Computers
,
31
,
137
149
.
Terhardt, E. (
1984
).
The concept of musical consonance: A link between music and psychoacoustics
,
Music Perception
,
1
,
276
295
.
Tramo, M.J., Cariani, P.A:, Delgutte, B., & Braida, L.D. (
2001
).
Neurobiological foundations for the theory of harmony in Western tonal music
.
Annals of the New York Academy of Sciences
,
930
,
92
116
.
Tufts, J.B., Molis, M. R., & Leek, M. R. (
2005
).
Perception of dissonance by people with normal hearing and sensorineural hearing loss
.
Journal of the Acoustical Society of America
,
118
,
955
967
.
Wagener, K.C., & Brand, T. (
2005
).
Sentence intelligibility in noise for listeners with normal hearing and hearing impairment: Influence of measurement procedure and masking parameters
.
International Journal of Audiology
,
44
,
144
156
.
Wilson, B.S., & Dorman, M. F. (
2008
).
Cochlear implants: A remarkable past and a brilliant future
.
Hearing Research
,
242
,
3
12
.
Wouters, J., Mcdermott, J. H., & Francart, T. (
2015
).
Sound coding in cochlear implants: From electric pulses to hearing
.
IEEE Signal Processing Magazine
,
32
,
67
80
.
Zarlino, G. (
1558
)
Le Istituzioni harmoniche
.
Venezia
.
Zeng, F.-G., Tang, Q., & Lu, T. (
2014
).
Abnormal pitch perception produced by cochlear implant stimulation
.
PLoS One
,
9
,
e88662
.