UC Merced

Timing is an essential component of human actions, and is the foundation of any sort of sequential behavior, from picking up a glass to playing an instrument or dancing. Because of this, our understanding of how we represent time in the brain (i.e., the human timing system) critically relies on basic research on simple behaviors. Perception of temporal regularities is central to a wide range of basic actions, but also underpins abilities unique to humans such as the creation of complex musical scores. This dissertation is an in-depth examination of endogenously and exogenously guided timing behavior, and how context is a critical component of understanding rhythmic entrainment in humans. We previously validated “gold standard” functional magnetic resonance imaging (fMRI) findings on action-based timing behavior using functional near infrared spectroscopy (fNIRS) (Rahimpour et al., 2020). In particular, we observed significant hemodynamic responses in cortical areas in direct relation to the complexity of the behavior being performed. To do so, we probed multiple levels of contextual influence on action-based timing behavior and patterns of cortical activation as measured using fNIRS. Our findings highlighted several distinct, context-dependent parameters of specific timing behaviors. Here we further interrogate human timing abilities by introducing variations of our original experimental design, observing that subtle contextual variations have a significant impact on the degree of rhythmic entrainment given the presence/absence of metronomic input. We used electroencephalogram (EEG) to further validate our fNIRS findings, demonstrating that single trial neurobiological activity can be used to predict whether behavior is exogenously or endogenously guided. We also found that patterns of neural activity correspond to differential use of the internal timing system, and that specific differences in neural activity correlate with accuracy of action-based timing behavior. These findings emerged from our use of a novel deep learning approach to extract person-specific, neural-based features as predictors of behavioral performance. Finally, we examined whether fNIRS and EEG produced similar localization information, finding that the influence of training factors on cortical localization must be accounted for to make such comparisons.