Temporal Signal‐to‐Noise Ratio (tSNR)

tSNR values were calculated for each voxel separately by dividing the voxel's mean signal by its standard deviation across the scan [Murphy et al., 2007]. This stage was performed on the functional images after motion correction, slice‐timing correction and normalization, but before smoothing, nuisance covariates regression and filtering (which reduce voxels' mean signal to zero). A two‐tailed paired‐sample t‐test was performed to compare between average tSNR values of high‐intensity voxels and of low‐intensity voxels across subjects. In addition, for each brain atlas tSNR values were averaged across all voxels included in each region (as determined by the specific brain atlas) to compute a regional tSNR value. Correlation was computed between the tSNR value of each atlas region and the number of low‐intensity voxels it contains. Signal variance across time was also computed for each voxel, and its average in low‐intensity and high‐intensity voxels was compared across subjects using a two‐tailed paired‐sample t‐test.