"This dissertation tested two related hypotheses about the mental representation of rhythms in the sub-second to second time range, namely, a preferred tempo hypothesis and an input noise hypothesis. The preferred tempo hypothesis proposed that the inherent period of an internal oscillatory timekeeper biases the perceptual representation of rhythms, such that perceived rhythm tempo is systematically distorted towards the period of the oscillator. The oscillator period was hypothesized to be about 500 to 600 ms, based on past results showing that the perception and reproduction of sequence tempi are systematically biased towards this tempo. The input noise hypothesis proposed that increased variability in the initial encoding of a rhythm leads to greater bias towards preferred tempo in the perceptual representation of the rhythm. To test these hypotheses, three experiments used a novel approach to the study of rhythms, which was a serial reproduction (SR) paradigm. In the SR paradigm, participants are exposed to a stimulus and then reproduce it from memory, with their reproduction serving as the stimulus for the next participant in the series. This process is iterated across a series of participants, forming a chain of reproductions in which each participant's reproduction influences the content being transmitted. Typically, the final reproduction in a chain incorporates the accumulation of systematic distortion across reproductions in the chain, thereby magnifying the bias introduced at each iteration. Bias in serial reproductions results from the adjustment of noisy input representations towards a value consistent with prior experience during perceptual representation. Greater noise in the stimulus input leads to greater bias, and thus faster chain drift. The experiments herein varied the seed tempo of rhythms initiating SR chains, asked a series of participants to reproduce the timing of iterated rhythms by tapping, and then compared the final reproduced tempo to the seed tempo of the chain. In line with the preferred tempo hypothesis, across experiments, when participants passively observed the stimulus rhythms before reproducing them, chains initiated at tempi ranging from 150 to 1709 ms converged to participants' preferred tempo. In line with the input noise hypothesis, when participants synchronized finger taps with stimulus rhythms during exposure, reproductions were less biased towards preferred tempo, leading chains to converge to preferred tempo more slowly or not at all. A similar pattern of results was obtained regardless of whether SR chains consisted of reproductions by different participants or by the same participant. Also consistent with the input noise hypothesis, when participants were exposed to and reproduced visual rhythms instead of auditory rhythms, reproductions tended to be more biased towards preferred tempo, leading chains to converge more quickly for visual than auditory rhythms. Synchronizing with visual rhythms slowed convergence of SR chains to preferred tempo, but to a lesser degree than for auditory rhythms. Overall, by testing these hypotheses in the domain of rhythm using the method of serial reproduction, the results of this dissertation advance basic research on tempo biases in rhythm perception and production. More broadly, the findings demonstrate the feasibility of using the serial reproduction paradigm in a novel domain and lay the groundwork for the future application of this approach to probe underlying biases in the mental representation of rhythm."--Pages ii-iii.
Read
