The influence of consonance–dissonance contrasts on perceived pleasantness of concluding tonic chords in short chord sequences

The present study

This study aims to examine whether and how the contrast between dissonance and the following consonance induces pleasantness. We employed chord progressions consisting of five chords that all ended on the tonic, and we manipulated the level of dissonance of the penultimate chord by using chords with a low or high level of roughness and cultural familiarity. A chord progression is a small but dynamic musical component where the contrast between dissonance and consonance may induce pleasantness. It is a suitable component to use for the purposes of this study because even a two-chord sequence induces an emotional response (Smit et al., 2020) and creates tension and release (Bigand & Parncutt, 1999; Krumhansl, 1996; Lehne et al., 2013).

We hypothesized that the contrast between the penultimate and final chords will contribute to pleasantness as follows: the pleasantness rating for the final chord will be higher when the cultural familiarity of the penultimate chord is low and its roughness is high. A chord with low cultural familiarity is perceived as being more dissonant (e.g., Lahdelma & Eerola, 2020), as is a chord with high roughness (e.g., Harrison & Pearce, 2020). The low cultural familiarity and high roughness of the penultimate chord will create a greater contrast between the dissonance of the penultimate chord and the consonance of the final chord which, on the basis of theorizing by Meyer (1956) and Huron (2006), will lead to a higher pleasantness rating. Also, this effect may be more pronounced for musically trained listeners than for less trained listeners.

Method

Stimuli

Sequences with five chords consisting of four notes were used as stimuli (Figure 1). All sequences consisted of the same chords apart from the penultimate chord (I–vi–IV–[manipulated chord]–I). To see the contrast effect of dissonance on the pleasantness of the following chord, the dissonance of the penultimate chord was manipulated. One of eight different chords with a low or high level of roughness and cultural familiarity was used each time for the penultimate chord: low roughness and low familiarity (LR–LF); low roughness and high familiarity (LR–HF); high roughness and low familiarity (HR–LH); high roughness and high familiarity (HR–HF). Familiarity was calculated based on the frequency with which each interval and chord appears in 739 popular songs collected in the Billboard corpus (Burgoyne, 2012), following Harrison and Pearce’s (2020) model. Roughness was calculated using a model by Hutchinson and Knopoff (1978). Cultural familiarity and roughness were used because cultural familiarity is a significant contributing factor to C/D perception (Eerola & Lahdelma, 2021), and roughness is influential in affective responses to chords (Armitage et al., 2021; Lahdelma et al., 2022). We did not include harmonicity for manipulation because harmonicity seems largely to overlap with cultural familiarity (Eerola & Lahdelma, 2021; Lahdelma et al., 2022).

OPEN IN VIEWER

Figure 1.

Eight Variations of the Chord Sequences in C Major Key.

The chord sequences were created using MuseScore 4 using the piano timbre, with a constant volume level and the tempo set at 80 bpm. The duration of each sequence was 8 s. The chord sequences with eight different manipulated chords were all played in seven keys whose tonic tones are within the range of three semitones below and above C (the keys of A major, B flat major, B major, C major, C sharp major, D major, and E flat major), yielding a total of 56 trials.

Results

The inter-rater reliability for each type of stimulus was calculated using Cronbach’s alpha, and all types of stimuli had very high reliability: LR–LF (α = .915), LR–HF (α = .905), HR–LF (α = .947), and HR–HF (α = .937). A linear mixed model with two fixed factors (Roughness and Familiarity, both with two levels: low and high) and three random effects (Participant, Musicianship, and Transposition) was applied to the ratings, using lmer function from lme4 package (Bates et al., 2015) in R (R Core Team, 2022). The mean ratings of the final chords for each of the eight penultimate chords are shown in Figure 2. For fixed effects, the analysis showed significant effects of both Roughness, F(1,2578.9) = 9.15, p < .001, ε²_p = .004, and Familiarity, F(1,2578.9) = 204.0, p < .001, ε²_p = .07, indicating that the final chords were rated more pleasant when Roughness was lower, and when Familiarity was higher. There was also an interaction between them, F(1,2578.9) = 40.8, p < .001, ε²_p = .02. For random effects, Transposition and Musicianship failed to show significant effects (p = .334 and p = 1.0, respectively), and the participant slopes indicated a strong effect (p < .001, all random effects tested with likelihood ratio tests).

OPEN IN VIEWER

Figure 2.

Mean Pleasantness Ratings of the Final Chords for Each Sequence With Different Penultimate Chord.

A post hoc pairwise comparison with Bonferroni correction was conducted to see whether each penultimate chord gave rise to different pleasantness ratings for the final chord. The majority (19 out of 28 pairs) were significantly different from each other (p < .001), while some adjacent pairs were not statistically different from each other. These are: Diminished triad and Diminished with a major seventh; Diminished triad and Minor-major seventh; Diminished triad and Minor triad; Diminished with a major seventh and Minor-major seventh; Dominant seventh and Major triad with a major second; Dominant seventh and Major triad with the root doubled on the octave; Dominant seventh and Power chord with a major sixth; Major triad with a major second and Power chord with a major sixth; and Minor triad with major seventh and Minor triad.

Discussion

Our experiment tested whether the dissonance of the penultimate chord increases the pleasantness of the following consonant chord, based on the idea of the contrast effect (e.g., Beebe-Center, 1932; Cantor & Zillmann, 1973; Parker et al., 2008; Schellenberg et al., 2012), which states that if there is a large contrast between dissonance and a following consonance then this will contribute to positive aesthetic responses (Bartók, 2004; Huron, 2006; Meyer, 1956). This is, to our knowledge, the first study to systematically examine the applicability of the contrast effect to C/D between chords. The results demonstrate that, although both the Roughness and Familiarity of the penultimate chord did indeed influence pleasantness ratings for the final consonant chord, they did so in the opposite way from how we had anticipated. High Roughness and low Familiarity of the penultimate chord led to lower, rather than higher, pleasantness ratings for the final chord. Our results did not reveal the contrast effect, unlike previous studies (Parker et al., 2008; Schellenberg et al., 2012; Smit et al., 2022), although in the case of these studies the contrast was between different affective characteristics of music and chords rather than the contrast between dissonance and consonance. The findings of this study instead demonstrate the influence of dissonance on the pleasantness of the following chord and reaffirm a well-documented association between dissonance and unpleasantness in line with previous research (e.g., Lahdelma & Eerola, 2020).

There are several possible explanations for this finding. One such explanation lies in the tonal relationship between the penultimate chord and the final tonic. According to Bharucha and Krumhansl (1983), when two chords are more closely related, the second chord is processed more quickly and as more consonant, and vice versa. Krumhansl (1979) states that non-diatonic tones are less closely related to the tones that make the triad, so non-diatonic tones in the dissonant penultimate chords we employed may have made these penultimate chords and the final, tonic chord less closely related. This may have resulted in the final chord being perceived as more dissonant and less pleasant. A second possible explanation is that the dissonance in the penultimate chord may have violated listeners’ tonal expectations, as listeners prefer the authentic cadence V-I (e.g., Tillmann & Marmel, 2013). The violation of expectation often results in negative responses (Huron, 2006), so in our case, the final chord is perceived as less pleasant.

Regarding the effect of Familiarity and Roughness, it is interesting that Familiarity seems to have a stronger effect on pleasantness ratings than Roughness. This can be seen from the larger effect size for Familiarity than Roughness (ε²_p = .07 and ε²_p = .004, respectively) and that HR–HF chords led to higher pleasantness ratings compared with HR–LF chords (see Figure 2). Familiarity’s stronger effect is in line with the findings of Eerola and Lahdelma (2021) that cultural familiarity is the dominant predictor of C/D perception.

We included participants’ level of musical sophistication in the analysis as a random effect. However, there was no significant difference between the ratings of listeners with different levels of musical training, even though previous studies report that musicians are more sensitive to the C/D of single chords (Arthurs et al., 2018; Weiss et al., 2020). This may be because the differences between the two levels (low and high) for both Roughness and Familiarity in the penultimate chords were very clear for any listener familiar with Western tonal music, regardless of their musical training, thus not finely graded enough to reveal individual differences. Also, regardless of their musical training, our participants may have had similar levels of cultural familiarity with the harmonic progression used, since it is very common in popular music (e.g., Temperley, 2018). This may explain the lack of significant difference between the pleasantness ratings of musicians and non-musicians.

Conclusion

In sum, this study found that the dissonance of the penultimate chord led to the following consonance being perceived as more unpleasant, demonstrating dissonance’s well-established association with unpleasantness (e.g., Lahdelma & Eerola, 2024) and in this case, its lasting effect on the following chord. Hence, we did not observe the contrast effect between dissonance and consonance on the pleasantness of consonant chords, contrary to theorizing by Bartók (2004), Meyer (1956), and Huron (2006). This is not to say, however, that there is no contrast effect inside a piece of music. We employed the five-chord sequences as stimuli in order to observe the effect of dissonance while strictly controlling other variables such as melody, voice leading, or rhythm. However, dissonance manipulation with such strict controls may have resulted in unnatural-sounding harmonic progressions. Strict controls also limited our ability to explore the potential effects of these other variables. Also, as in previous studies on the perception of chord progressions (e.g., Bigand et al., 1999; Jansen & Povel, 2004; Kuusi et al., 2021), the current experimental design did not allow for a clear distinction between the pleasantness of the final chord and that of the entire chord sequence, even though the participants were specifically instructed to rate the pleasantness of the final chord rather than the whole sequence. Future research could benefit from a more elaborate experimental design to control other variables by using longer chord progressions or real pieces of music for higher ecological validity, and by collecting continuous ratings in order to shed further light on the contrast effect that underlies our aesthetic response to music.