Cognitive modeling of human music information processing

要将一系列的音调感知为一个“旋律”，有两个感知过程是必不可少的：一个是评估韵律组织的过程，另一个是评估序列的音调组织的过程。前者在声调序列上赋予了连贯的时间结构，因此，人们可以感知到“节拍”和“节拍”等心理时间单位。类似地，致力于音调组织的处理将连贯性分配给序列中各个音符的音高之间的关系，从而赋予序列一种“调性”的感觉。致力于韵律和音调关系的知觉过程被认为是由独立的模块组成的。然而，毫无疑问，在处理的某个阶段，这两个处理模块的输出在某种程度上是综合的。本研究旨在阐明韵律组织和声调组织这两种知觉加工之间的相互关系。我们对声调组织和韵律组织所涉及的过程进行了认知建模，假设它们是独立的模块。我们构建了计算机模型，这些模型能够预测给定音调序列的“基调”分配和“基调”的程度，只使用组成音符的音高作为数据。通过将模拟研究的结果与实验对象的记录反应进行比较，我们的模型被发现可以预测大约80%的听众的关键解释。这一结果优于我们的实施方案所获得的任何其他现有模型的拟合结果。我们还构建了一个计算机模型，该模型能够仅使用组成音符的音调长度来解释给定音调序列的节拍和节拍。

To perceive a sequence of tones as a "melody, " two perceptual processes are essential ; a process for evaluating the metrical organization and a process for evaluating the tonal organization of the sequence. The former assigns a coherent temporal structure on the tone sequence and, as a result, such psychological time units as "beat" and "meter" are perceived. Similarly, processing dedicated to the tonal organization assigns coherence to the relationship between the pitches of the individual notes in the sequence thus conferring a sense of "tonality" upon the sequence. The perceptual processes devoted to the metrical and tonal relationships are considered to comprise independent modules. Nevertheless, there is no doubt that the outputs of the two processing modules are integrated, in some way, at some stage of processing. The present study aimed at clarifying the inter-relationship between the two kinds of perceptual processing, i.e., the metrical organization and the tonal organization. We engaged in cognitive modeling of processes involved in the tonal organization and the metrical organization, assuming that they are independent modules. We constructed computer models that were capable of predicting the assignment of "key" of a given tone sequence and the degree of the "keyness, " using only pitches of the constituent notes as data. Our model was found to predict approximately 80% of listeners' key interpretations in a comparison between the results of our simulation study and the recorded responses from the subjects of our experiment. This result is superior to the fits obtained by our implementing programs of any other existing models. We also constructed a computer model that was capable of interpreting the beat and meter of a given tone sequence using only tone-lengths of the constituent notes.