Modeling rhythm perception and temporal adaptation: top-down influences on a gradually decaying oscillator

Abstract

Adapting to a dynamically-changing auditory environment is a challenging task that our brains are able to achieve most of the time seemingly without much effort. Dynamic attending theory posits that this ability is governed by rhythmic entrainment, namely, synchronization of internal oscillators to the regularities in sound signals. Here, we investigated the properties of these oscillators, based on their behavior prior to, during and after entraining to rhythmic stimuli. We fitted a linearized oscillator model and several variants to empirical datasets, obtained from a duration discrimination paradigm that involved a wide range of stimulus rates. Two sessions of the experiment, to which the models were fitted separately, differed in their requirement for temporal adaptation: trial-to-trial changes in stimulus rate were maximal in one session, and minimal in the other. We first compared models that assumed either complete, gradual, or no decay towards the preferred rate after cessation of a stimulus rhythm, either in a silent gap between the stimulus sequence and to-be-judged comparison interval, between consecutive trials of the experiment, or both. Then, we obtained parameter estimates from the best-fitting model and compared them across session types. Results revealed that the internal oscillators gradually decay towards their preferred rate, on similar timescales within and between trials. Critically, the oscillators’ behavior is mainly determined by their preferred rate and can be modulated by task demands. The findings are in line with theoretical predictions and neuroscientific literature on oscillatory mechanisms underlying rhythm perception and temporal adaptation.