Time-Perception Network and Default Mode Network Are Associated with Temporal Prediction in a Periodic Motion Task

Updating internal models to predict future events relies on both frontal and parietal brain regions involved in uncertainty-driven updating and a separate network for temporal attention. Using fMRI, this study examined how continuous manipulation of temporal prediction engages these networks. Participants viewed periodic (simple harmonic oscillation) and non-periodic (variable acceleration) motion patterns. Non-periodic motion activated the exogenous temporal orienting network, including ventral premotor and inferior parietal cortices, cerebellum, presupplementary motor area, and motion-sensitive area MT+, with a right-hemisphere bias suggesting explicit timing. Periodic motion activated default-mode network midline areas (left DMPFC, ACC, bilateral PCC/PC), indicating the DMN may process contextually expected information and validate prospective internal models. Findings show continuous temporal prediction engages both temporal expectation representations and task-independent internal model updating.

Study at a glance

Abstract

The updating of prospective internal models is necessary to accurately predict future observations. Uncertainty-driven internal model updating has been studied using a variety of perceptual paradigms, and have revealed engagement of frontal and parietal areas. In a distinct literature, studies on temporal expectations have also characterized a time-perception network, which relies on temporal orienting of attention. However, the updating of prospective internal models is highly dependent on temporal attention, since temporal attention must be reoriented according to the current environmental demands. In this study, we used functional magnetic resonance imaging (fMRI) to evaluate to what extend the continuous manipulation of temporal prediction would recruit update-related areas and the time-perception network areas. We developed an exogenous temporal task that combines rhythm cueing and time-to-contact principles to generate implicit temporal expectation. Two patterns of motion were created: periodic (simple harmonic oscillation) and non-periodic (harmonic oscillation with variable acceleration). We found that non-periodic motion engaged the exogenous temporal orienting network, which includes the ventral premotor and inferior parietal cortices, and the cerebellum, as well as the presupplementary motor area, which has previously been implicated in internal model updating, and the motion-sensitive area MT+. Interestingly, we found a right-hemisphere preponderance suggesting the engagement of explicit timing mechanisms. We also show that the periodic motion condition, when compared to the non-periodic motion, activated a particular subset of the default-mode network (DMN) midline areas, including the left dorsomedial prefrontal cortex (DMPFC), anterior cingulate cortex (ACC), and bilateral posterior cingulate cortex/precuneus (PCC/PC). It suggests that the DMN plays a role in processing contextually expected information and supports recent evidence that the DMN may reflect the validation of prospective internal models and predictive control. Taken together, our findings suggest that continuous manipulation of temporal predictions engages representations of temporal prediction as well as task-independent updating of internal models.