This thesis presents an original methodology, as detailed below, applied to the testing of an existing computational model of human rhythm perception. Since the computational model instantiates neural resonance theory (Large and Snyder,2009), the thesis also tests that theory. Neural resonance theory is a key target for testing since, as contrasted with many other theories of human rhythm perception, it has relatively strong physiological plausibility. Rather than simply matching overt features of human rhythm perception, neural resonance theory shows how these features might plausibly emerge from low-level properties of interacting neurons. The thesis tests the theory using several distinct research methods. The model stood up well to each family of tests, subject to limitations that are analysed in detail. Firstly, the responses of the model to several types of polyrhythmic stimuli were compared with existing empirical data on human responses regarding beat identification to the same stimuli, at a variety of tempi. Polyrhythmic stimuli closely resemble real life complex rhythmical stimuli such as music, and were used for the first time to test the model. It was found that the set of categories of response predicted by the model matched human behaviour. Secondly, the model was systematically analysed by exploring the degree of dependence of its behaviour on the values of its parameters (sensitivity analysis). The behaviour of the model was found to retain consistency in the face of systematic numerical manipulation of its parameters. Thirdly, the behaviour of the model was compared to that of related models. In particular, the focal computational model, which balances physiological plausibility with mathematical convenience, was compared with other models that relate more directly to brain physiology. In each case, all relevant behaviours were found to be closely in line. Lastly, the outputs of the model under polyrhythmic stimuli were analysed to make new testable predictions about previously unobserved human behaviour regarding the time it takes for people to perceive beat in polyrhythms. These predictions led to the design and conduction of new human experimental studies. It was found that the model had successfully predicted previously unobserved aspects of human behaviour, more specifically it predicted the timescale within which people start to perceive beat in a given polyrhythmic stimulus.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:618268 |
| Date | January 2014 |
| Creators | Angelis, Vassilis |
| Publisher | Open University |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://oro.open.ac.uk/40519/ |
Page generated in 0.0022 seconds