4.3.1 Data pre-processing

The data were individually corrected for head-motion using MCFLIRT and spatially smoothed using a Gaussian kernel of FWHM 5mm. Mean-based intensity normalisation of all volumes by the same factor was applied, followed by high-pass temporal filtering (see above) was performed. The individual data sets were registered into MNI space using FLIRT [Jenkinson and Smith, 2001] while keeping the data at the functional resoultion in order to decrease computational load. The final 3-way data $\mbox{\protect\boldmath$X$}$ was of size $122\times 12839\times 5$. Based on the estimated sample covariance matrix of the matrix $\mbox{\protect\boldmath$X$}_{I\times JK}$, the Laplace approximation to the model estimated a 12-dimensional signal sub-space. The data for each session was projected onto the space spanned by the first 12 Eigenvectors and spatially normalised by the voxel-wise variance estimate from the residuals of the projection.