Infra-slow scale-free dynamics modulate the connection of neural and behavioral variability during attention.

The activities of the human brain vary across different timescales, exhibiting scale-free dynamics. Previous research has highlighted the psychological and physiological significance of brain dynamical fluctuations across the Delta to Gamma bands. However, there has been less focus on infra-slow scale-free dynamics, e.g. power law exponent (PLE), and neural variability, e.g. standard deviation (SD), and sample entropy (SE), in mediating brain-behavior connection during attention. In this study, we recruited 49 participants and recorded functional magnetic resonance imaging (fMRI) resting-state and task data during a sustained attention task paradigm to investigate how the three measures-SD, SE, and PLE-modulate the dynamics of behavioral performance. Our findings demonstrate the following: (i) PLE, SD, and SE exhibit differential topographic distribution with a hierarchical structure from sensory to associative networks, during their rest-task modulation. (ii) PLE, SD, and SE show different topographic extensions from visual cortex to default-mode network in their relationship with behavioral variability. (iii) The relationship between SD and SE is mediated by PLE in the empirical data, which (iv) is further confirmed in simulation. Collectively, our results highlight the topographically- and dynamically-layered mechanisms of distinct neurodynamical features during attention processing: scale-free dynamics modulate neural and behavioral variability.