Remote photoplethysmography enables the extraction of cardiovascular metrics using standard digital cameras. To combat signal degradation, contemporary methods typically merge multiple facial regions of interest or analyze the entire face to improve the signal-to-noise ratio. While effective for robust heart rate estimation, this spatial integration assumes that cardiovascular signals across the face are temporally synchronous. This assumption ignores potential delays caused by physiological factors like Pulse Transit Time and hardware artifacts like rolling shutter effects. This study investigates the validity of this assumption by quantifying inter-ROI temporal lags using a public 30 FPS dataset and a custom 60 FPS dataset. Signals were extracted from the forehead, chin, and both cheeks. Cross-correlation analysis combined with parabolic interpolation was applied to resolve sub-frame temporal offsets. The results reveal a consistent spatial delay pattern where the forehead leads the lower facial regions by a median of approximately 16 to 23 ms. Conversely, horizontally aligned regions like the left and right cheeks remain synchronized. These findings demonstrate that merging facial regions without temporal alignment introduces destructive interference that blurs pulse morphology. To ensure the temporal fidelity required for advanced diagnostics like heart rate variability, pipelines must abandon global face averaging in favor of spatially aware fusion strategies.