As in conventional functional connectivity-based analyses at lower resolutions, ‘connectivity’ is estimated based on the temporal similarity of functional time series. This method is prone to overestimating the connectivity strength if multiple areas are affected by the same source of erroneous signal fluctuations. One prominent cause of unwanted signal fluctuation can be respiration induced signal changes, which can introduce biases of false positive connectivity between neurally unconnected brain areas. In layer-fMRI analyses, this can have a higher effect in the superficial layers with a larger vascular density compared to deeper layers with reduced vascular density. Another source of unwanted signal fluctuations, when estimating the connectivity of two brain areas of interest, is a common input from a third unknown area. In this case, the common input would induce the same fluctuations in both ROIs and make them look more connected. Figs. 4–7 depict a number of potential approaches to minimize the effect of these unwanted sources of shared variance. One approach would be to restrict connectivity interpretations to differential analyses. Unwanted sources of fluctuations can be removed, by orthogonalizing the
