Stimuli were presented using a ProPixx LED projector (VPixx Technologies, Saint-Bruno, CA) and were rear-projected onto a textured screen in the bore of the MRI scanner, which could be viewed on the screen via a mirror attached to the head coil. To monitor participants’ fixation stability, we used an eye tracker to record a video of the eye, allowing us to monitor eye movements in real time throughout the experiment (https://www.crsltd.com/tools-for-functional-imaging/mr-safe-eye-tracking-for-fmri/livetrack-fmri/ ).
PNG images of faces were processed in MATLAB (Mathworks). First, all images were converted to grayscale with flattened histograms, to control grayscale ranges across images. Following this, the mean luminance value of each image was set to 705 cdm−2 so that they would be undetectable in regions of the retina that were adapted to bright light of ~15,000 cdm−2. All images were re-scaled to take up 50% of a 400 × 400-pixel size stimulus. Finally, all images were placed on a dark grey background matching mean luminance of the images. To scramble the faces, each image was divided into a 20 × 20 square grid and each square was randomly shuffled through ±90 or 180. Finally, we applied a slight blurring to the intact and the scrambled images with a Gaussian filter (SD = 1px) to soften the lines created when scrambled.
PNG images of faces were processed in MATLAB (Mathworks). First, all images were converted to grayscale with flattened histograms, to control grayscale ranges across images. Following this, the mean luminance value of each image was set to 705 cdm−2 so that they would be undetectable in regions of the retina that were adapted to bright light of ~15,000 cdm−2. All images were re-scaled to take up 50% of a 400 × 400-pixel size stimulus. Finally, all images were placed on a dark grey background matching mean luminance of the images. To scramble the faces, each image was divided into a 20 × 20 square grid and each square was randomly shuffled through ±90 or 180. Finally, we applied a slight blurring to the intact and the scrambled images with a Gaussian filter (SD = 1px) to soften the lines created when scrambled.