Typically, T2*-weighted GE-EPI is sensitive to the blood-oxygenation-level-dependent (BOLD) effect, which is coupled to gray matter (GM) neuronal activity in myriad complex ways [1], and bears little relation to signals in white matter (WM). Yet WM signals yield functional characteristics including low frequency fluctuations and anisotropic temporal correlations [2] that are orientation-dependent with respect to the direction of B0 [3], which suggested a possible hypothesis that local functional connectivity anisotropy (LFCA) could be related to electric current transmitting along WM fiber pathways. We tested this hypothesis in principle with phantom studies and controlled variables that may confound the interpretation, such as correlation bias between through-plane and in-plane voxel-pairs, motion via Lorentz forces, and electric current intensity dependence. This research addressed these confounds and the sensitivity of LFCA method in determining the mechanism that gives rise to B0 orientation-dependent anisotropic local temporal correlations in WM resting state fMRI time series. T2*-weighted fMRI (TE/TR 31/2000 ms) were recorded from a spherical distilled water-filled phantom with either 1 mA or .1 mA DC fluctuating at approximately .05 Hz and oriented either parallel or perpendicular to B0. Data were preprocessed according to typical resting state fMRI protocol. Temporal correlations were computed in a neighborhood radius of 2 voxels, and an isotropic tensor model was fitted to the local functional connectivity matrix. Motion correction parameters were calculated using an industry standard realignment algorithm. Detrended motion parameters were regressed out of the phantom data and anisotropy and Eigen values from the LFCA tensor were compared before and after. In addition, phantom data were recorded with the readout gradient in the axial (Bz), sagittal (Bx) or coronal (By) planes, and with the electric current oriented parallel to either Bz or Bx. Both current intensities produced paradigm-correlated motion in the phantom when the current was oriented perpendicular to B0, but not when oriented parallel to B0 (figure 1). The 1 mA generated detectable motion in 5 of 6 directions (roll, pitch, yaw, dS & dL), with displacement of approximately 0.02 mm rotation, and .03 mm of translational movement. The 0.1 mA also generated detectable rotational motion in the roll and the yaw (approximately .02 degrees). These results clearly show the effect of the Lorentz force. For both current intensities, the motion was significantly negatively correlated with the raw (detrended) signal from a sample voxel, and all positive correlations were non-significant. This shows that when the current is ‘active’ the fMRI signal drops out and global displacement of the volume is detected. After correction of the motion, however, the relation between the paradigm time course and fMRI signal became positive, and anisotropy (and 1st Eigen values) remained highly significant (figure 2). In fact, the 1st Eigen value was larger for the 0.1 mA after correction of the global motion caused the Lorentz force. The temporal correlations between the nearest voxel-pairs are indeed stronger for in-plane directions than between slices, which biased eigenvector alignment. Importantly, such bias, along with LFCA, was not observed when the electric current was oriented parallel to B0, which replicated our previous in vivo findings. Finally, the weaker (0.1 mA) current produced a weaker global movement, however, it accounted for as much of the signal fluctuation as the stronger current. Local anisotropic temporal correlations in a phantom are related to electric current after controlling for several confounds. Fluctuating electric current caused detectable global motion of the phantom, which contributed to but cannot fully account for local correlation anisotropy, nor the corresponding Eigen values and their eigenvector alignment. The B0 orientation-dependence of local temporal correlations cannot be explained by the slice selection orientation of the scan and thus, can be ruled out as a possible cause of the absence of LFCA in the B0 direction. Despite causing a weak signal change, fMRI is sensitive to detect the fluctuation associated with electric current. Together, these results support the hypothesis that orientation-dependent anisotropic local temporal correlations in vivo are related to neuroelectric current in the WM.Fluctuating electric current is a most likely cause for the correlated signal changes through neighboring voxels that appear as orientation-dependent local functional connectivity anisotropy in white matter withT2*-weighted fMRI.