The Learning Trajectory Of Musical Memory: From Schematic Processing Of Novel Melodies To Robust Musical Memory Representations
| dc.contributor.author | Agres, Kathleen | en_US |
| dc.contributor.chair | Field, David James | en_US |
| dc.contributor.committeeMember | Spivey, Michael James | en_US |
| dc.contributor.committeeMember | Goldstein, Michael H. | en_US |
| dc.contributor.committeeMember | Zevin, Jason | en_US |
| dc.contributor.committeeMember | Pizarro, David A. | en_US |
| dc.date.accessioned | 2013-09-05T15:26:14Z | |
| dc.date.available | 2018-01-29T07:00:36Z | |
| dc.date.issued | 2013-01-28 | en_US |
| dc.description.abstract | This dissertation utilizes a multi-method approach to investigate the processes underlying musical learning and memory. Particular emphasis is placed on schematic processing, musical structure, temporal aspects of learning, statistics-based predictive models, efficiency, and the role of musical expertise. We employed a set of behavioral change detection studies with musician and nonmusician participants to test what is encoded into gist memory upon hearing unfamiliar melodies varying in musical structure. These studies demonstrate that listeners abstract a schematic representation of the melody that includes tonally and metrically salient tones. In well-structured music, change detection performance improves when a musical event does not conform to the listener's schematic expectations. Musical expertise is also shown to benefit change detection, especially when the melodies conform to the conventions of Western tonal music. In a study examining learning over a period of increasing musical exposure, we used an information theoretic approach to capture how the statistical properties of music influence listeners' musical memory. This work highlights how patterns and predictability can facilitate musical learning over time. In further investigation of what underlies this learning process, a series of neural network studies revealed that a compressed representation arose in the internal structure of a computational network as tonal and stylistic information were learned over time. Population sparsity of the SRN's hidden layer strongly predicted the sophistication of the network's musical output as rated by human listeners. Electroencephalography (EEG) methods were utilized to investigate the neural correlates of musical learning and memory, and to further explore the notion of increasing efficiency over the time-course of learning. These experiments suggest that the listener's implicit internal model of musical expectation is gradually developed and made increasingly accurate with repeated exposure to initially unfamiliar music. Both the computational and EEG experiments illustrate how efficiency accompanies successful learning over time. These findings, as well as those from the change detection and information theory studies, provide evidence that schemata are formed as the probabilities of forthcoming music are gradually learned with increasing experience. Schematic expectations dynamically guide perception and influence memory, and generally allow for more efficient musical processing. | en_US |
| dc.identifier.other | bibid: 8267375 | |
| dc.identifier.uri | https://hdl.handle.net/1813/33890 | |
| dc.language.iso | en_US | en_US |
| dc.subject | musical learning and memory | en_US |
| dc.subject | schematic processing | en_US |
| dc.subject | change detection | en_US |
| dc.subject | neural networks | en_US |
| dc.subject | EEG | en_US |
| dc.subject | efficiency | en_US |
| dc.title | The Learning Trajectory Of Musical Memory: From Schematic Processing Of Novel Melodies To Robust Musical Memory Representations | en_US |
| dc.type | dissertation or thesis | en_US |
| thesis.degree.discipline | Psychology | |
| thesis.degree.grantor | Cornell University | en_US |
| thesis.degree.level | Doctor of Philosophy | |
| thesis.degree.name | Ph. D., Psychology |
Files
Original bundle
1 - 1 of 1