| 其他摘要 | Musical tension is an affective state related to conflict, dissonance, instability, or uncertainty. It originates from future-directed prediction to generate a desire for resolution, and bridges between musical structure and emotions. Predictive theories of musical tension claim that in the process of music listening, listeners continuously predict forthcoming music events on the bases of their prior knowledge and the current context for the better understanding of the meaning of music. Previous empirical studies have explored how predictions affect musical tension, and mainly concerned about prediction errors. However, as assumed by musical prediction theories, musical tension is affected by prediction build-up processes, predictive outcomes, and prediction updating processes, as well as by preceding context and high-level pitches. Therefore, how predictive processes affect musical tension? How different levels of predictions interact with musical tension? The dissertation explored these questions through two parts of studies (four experiments).
Stuty 1 manipulated tonality with a phrase, observing the relationships between predictive processes and musical tension. Experiment 1 using behavior and EEG explored how predictive processes affect tension in induction stage by manipulating contextual predictability and congruency. The major results were that at the anticipatory stage, compared with high-prediction condition, in low-prediction condition tension curve fluctuated faster and unstable, and a larger NS in ERP response was elicited; at the integration stage, compared with congruent condition, in the incongruent condition the behavioral rating of tension was higher regardless of the predictability of the final chord; a right negativity and P600 were elicited, and the amplitude of P600 was modulated by the predictability of the final chord. These results show that the predictability of preceding context affects tension, and further weight prediction errors to mediate musical tension. Experiment 2 further investigated how prediction updating affect tension by manipulating contextual certainty and congruency. The major results were that for the incongruent conditions, tension curves fluctuated faster, and a larger P600 was evoked in the certain condition compared with the uncertain condition. For the congruent conditions, there was no contextual effect on behavioral ratings and tension curves, but a larger P600 was elicited in the uncertain condition. These results show that prediction updating elicited by prediction errors was mediated by contextual certainty, and further affect tension.
Study 2 took into account the hierarchy of music, and explored how different hierarchical levels of predictions affect tension from the perspective of phrase and period (experiment 3), tonal and melodic perspectives (experiment 4) respectively. The major results were that different levels of predictions affect musical tension respectively, and more importantly, interactive effects were found in the experiment 3 and experiment 4. Specifically, the results of experiment 3 showed that for the period congruent condition, the amplitude of P600 was higher in the phrase incongruent condition than the congruent condition. For the period incongruent condition, the amplitude of P600 was smaller in the phrase incongruent condition than the congruent condition; meanwhile, tension curves fluctuated faster, and increased beta oscillations were elicited in the phrase incongruent condition compared with the congruent condition. For the phrase congruent condition, tension ratings and the amplitude of P600 were higher in the period incongruent condition than the congruent condition; meanwhile, decreased beta oscillations were found in the period incongruent compared with the congruent condition. For the phrase incongruent condition, all Through the above series of experiments, this paper investigated the effects of predictions on musical tension in pitch hierarchical structures, and obtains the following conclusions. Firstly, in the tension induction processes, the predictability of preceding context build-up processes affect tension, and further weight prediction errors to mediate musical tension; secondly, in the tension resolution stage, prediction updating elicited by prediction errors was mediated by contextual certainty, and further affect tension; thirdly, the different levels of predictions affect musical tension, and the local effect plays a more superiority role. these period effects disappeared, but tension curves of the period incongruent condition fluctuated faster than the congruent condition. The results of experiment 4 showed that the effect size of tension score induced by the melodic contour and tonal incongruent condition is lower than that induced by the melodic contour congruent but tonal incongruent condition. We also found that the melodic effects of tension scores and the NS disappeared under tonal incongruent conditions. These data showed that different levels of predictions affect musical tension, and global effects could mediate local effects, but local effects could block global effects, suggesting that local effects might plays more superiority role in tension experiences. |
修改评论