Fig. 3
From: The “Narratives” fMRI dataset for evaluating models of naturalistic language comprehension
Intersubject correlation (ISC) in early auditory cortex (EAC). (a) ISC in left and right EAC across all subjects and tasks (i.e. stories). Each tick mark represents the leave-one-out ISC (x-axis) for a left-out subject in either left- (purple) or right- (orange) hemisphere EAC across tasks (y-axis). Violin plots illustrate the distribution of leave-one-out ISC across all left-out subjects for each task. The circular markers indicate the mean ISC across left-out subjects for each task and hemisphere (error bars indicate the 95% bootstrap confidence interval of the mean). At bottom, ISC for left and right EAC is summarized across all tasks. (b) Lagged ISC captures temporal synchronization in left EAC. Leave-one-out ISC (y-axis) was computed for left EAC at 61 lags ranging from -30 to + 30 TRs (−45 to 45 seconds; x-axis). In the upper plot, each line represents the mean ISC across lags for each task. At bottom, all lagged ISCs are visualized in a single plot where each left-out subject corresponds to a semi-transparent gray line; the red line reflects the average lagged ISC across all subjects and tasks. See the plot_isc.py script in the code/ directory for details.
