Fig. 4. Decoded viewing behavior explains network-wide brain activity.
GLM group results for the contrast ‘far versus short eye movements’ during visual search. We plot the F-statistic of this contrast superimposed on a template surface (fsaverage) for gaze labels obtained with camera-based eye tracking (left) as well as for three DeepMReye cross-validation schemes. For the within-participants scheme, all participants of a dataset were included with different partitions in model training and testing. For the across-participants scheme, different participants were included during model training and testing. For the across-datasets scheme, different datasets (and hence also different participants) were included during model training and testing; FWE, family-wise error. Source data are provided.