Abstract
Studying task modulations of brain connectivity using functional magnetic resonance imaging (fMRI) is critical to understand brain functions that support cognitive and affective processes. Existing methods such as psychophysiological interaction (PPI) and dynamic causal modeling (DCM) usually implicitly assume that the connectivity patterns are stable over a block-designed task with identical stimuli. However, this assumption lacks empirical verification on high-temporal resolution fMRI data with reliable data-driven analysis methods. The present study performed a detailed examination of dynamic changes of functional connectivity (FC) in a simple block-designed visual checkerboard experiment with a sub-second sampling rate (TR = 0.645 s) by estimating time-varying correlation coefficient (TVCC) between BOLD responses of different brain regions. We observed reliable task-related FC changes (i.e., FCs were transiently decreased after task onset and went back to the baseline afterward) among several visual regions of the bilateral middle occipital gyrus (MOG) and the bilateral fusiform gyrus (FuG). Importantly, only the FCs between higher visual regions (MOG) and lower visual regions (FuG) exhibited such dynamic patterns. The results suggested that simply assuming a sustained FC during a task block may be insufficient to capture distinct task-related FC changes. The investigation of FC dynamics in tasks could improve our understanding of condition shifts and the coordination between different activated brain regions.
Challenging the Static Connectivity Assumption
In functional magnetic resonance imaging (fMRI) research, "block designs"—where a participant performs a task for a set period followed by a rest period—are a gold standard. Most common analysis methods operate on the assumption that the functional connectivity (FC) between brain regions is sustained and stable throughout that task block. However, this assumption has rarely been tested due to the low temporal resolution of typical fMRI scans. This study utilizes high-speed fMRI to determine if brain "wiring" is truly static during a task or if it exhibits rapid, stimulus-driven dynamics that standard models might miss.
Methods: High-Temporal Resolution and TVCC
To capture these elusive dynamics, researchers employed a multiband EPI sequence with a sub-second repetition time (TR = 0.645s), providing enough data points within a single 20-second task block for meaningful time-series analysis. They used a Time-Varying Correlation Coefficient (TVCC) with a sliding window approach to estimate how the correlation between regions changed second-by-second.
Crucially, they regressed out the main BOLD (blood-oxygen-level-dependent) response triggered by the task to ensure the measured connectivity reflected the synchronization of underlying neural fluctuations rather than just a shared reaction to the stimulus. The study focused on four activated visual regions and regions in the Default Mode Network (DMN) as controls.
The "Transient Drop" in Connectivity
The most striking finding was that FC between hierarchical visual regions (MOG and FuG) was not sustained. Instead, it decreased significantly shortly after the task began and then gradually returned to baseline levels while the task was still ongoing. This "transient drop" was specifically observed in connections between different levels of the visual hierarchy. In contrast, connections between bilateral "twin" regions (e.g., Left MOG to Right MOG) and DMN control regions remained relatively stable.
Dissociation Between Activation and Connectivity
A key insight from the results is the dissociation between signal intensity and inter-regional communication. While the BOLD activation remained high and steady throughout the task block, the connectivity was dynamic. This suggests that signal intensity and inter-regional coordination are governed by different neural mechanisms.
Implications for Brain Function Models
The authors suggest these dynamics might be explained by "predictive coding." When a simple, repetitive stimulus like a checkerboard appears, the brain initially communicates intensely across hierarchies. Once higher-level regions successfully "predict" the stimulus, the need for constant communication drops, leading to the observed decrease in functional connectivity.
This research challenges the practice of assuming sustained FC during task blocks and suggests that investigating FC dynamics can improve our understanding of how different brain regions coordinate during condition shifts. Future work should explore if more complex stimuli produce different connectivity signatures and use simultaneous EEG-fMRI to confirm the neurophysiological basis of these BOLD dynamics.
Summary updated