Participants listened passively to stimuli in the Reversed Speech condition. The Dabrafenib clinical trial task was explained verbally by the experimenter before the start of the functional data acquisition

to ensure participants understood it and could overtly produce a small set of target stimuli. A short practice was given to the participants inside the scanner immediately before the start of data acquisition. During this practice they heard five stimuli for the Speech condition followed by five stimuli for the Reversed Speech condition. Participants were instructed not to overtly produce the target word because speaking produced head movements during scanning. They were asked instead to “think of the word inside their heads” and keep as still as possible. The practice stimuli were not used again during the functional data acquisition. If the participants were happy to proceed with the task, functional data were acquired. The Speech and Reversed Speech conditions and a

baseline condition during which learn more no stimuli occurred were presented in 30-s blocks and repeated four times each in a fixed pseudorandom order so that no condition was presented consecutively. Each 30-s block of the Speech and Reversed Speech conditions comprised six stimuli presented one every 5 s. The T1-weighted structural brain images were analysed with an ‘optimised’ ADAMTS5 voxel-based morphometry (VBM)-style protocol (Good et al., 2001) within FMRIB’s Software Library (FSL v4.1, www.fmrib.ox.ac.uk/fsl). The skull was stripped from this image using the Brain Extraction Tool (Smith, 2002) and the brain images were segmented to form images representing partial volume estimates of each tissue class (i.e. how much of the signal in each voxel was grey or white matter or cerebrospinal fluid) (Zhang, Brady, & Smith, 2001). The total volume of grey matter was calculated from these images (by multiplying

the average voxel value by the total number of voxels). These images were also used in the functional analyses below as voxel-dependent covariates. For the VBM-style analyses of structure, the 32 images of grey matter were non-linearly registered to the MNI-152 grey matter template using FMRIB’s Nonlinear Registration Tool (FNIRT) (Andersson et al., 2007a and Andersson et al., 2007b). Each image was flipped across the midline to create a mirror image and the 64 images were averaged to create a left–right symmetric study-specific grey matter template. The 32 original images of grey matter were then non-linearly transformed to this new template.