Journal: bioRxiv
Article Title: Parallel and dynamic attention allocation during natural reading
doi: 10.1101/2025.05.27.656336
Figure Lengend Snippet: Each trial began with a fixation-cross at the screen centre, followed by a “starting box” on the left. Fixating on the starting box would trigger the sentence onset. Participants ( n = 42) were instructed to read 188 sentences silently and then gaze at the “ending box” at the screen bottom to trigger the sentence offset. Randomly, 25% of trails included a simple comprehension question requiring a button response. Each sentence contained one or two target words, either of low or high lexical frequency. The target words were unpredictable from the prior context and all sentences were plausible. RIFT was applied by continuously flickering rectangle patches underneath the target and post-target words at frequencies f t and f p respectively (60 and 65 Hz sine waves, balanced across participants). A Gaussian mask was applied over the patch to smooth the sharp luminance changes around the edges to reduce their visibility across saccades. Two discs were positioned at the bottom corners of the screen, with their luminance oscillating at sine waves of f t and f p separately throughout the entire trial. These discs were covered by two photodiodes, and, during sentence presentation, their luminance changes mirrored those of the patches beneath the flickering words. Eye tracker and MEG data were acquired simultaneously. ITI, inter-trial interval.
Article Snippet: The MEG data analysis was conducted using MATLAB R2020a (Mathworks Inc, USA), incorporating the FieldTrip toolbox (version 20200220; Oostenveld et al., 2011), the FLUX MEG analysis pipeline , and custom-made scripts.
Techniques:
