We conducted further confirmatory analyses to ensure that the hierarchical regression was robust. Specifically, we subsampled the data in order to reverse the direction of eye movement differences across the conditions. In the original data set, there are more saccades in the Attention-High conditions than the Attention-Low conditions. In order to reverse the direction of this effect on a participant-wise basis, we sorted the trials within each condition according to the number of GW3965 saccades that occurred on that trial. In each Attention-High condition, we took all

scores below the 60th percentile. In each Attention-Low condition, we took all scores above the 40th percentile. As shown in Figure 3A, in the subsampled data, the number of saccades was much larger in the Attention-Low conditions than the Attention-High conditions (F(1,29) = 148.97, p < 0.001). In fact, CH5424802 clinical trial the absolute value of the difference between conditions was much larger in the subsampled data than in the original data. As in the original data, the main effect of Memory was significant

(F(1,29) = 4.44, p < 0.05) and the interaction was not significant (F(1,29) = 2.47, p = 0.13). The subsampled data were then subjected to the same analysis as the original data set. If the hierarchical regression is robust, the subsampled data should lead to similar conclusions: the effects of eye movements only have already been satisfactorily modeled, so any further classification of the data on the basis of eye movements should have no effect. Alternatively, if the activation presented in Figure 2 reflects the effects of eye movements, there should be a substantial reversal of these effects when the sub-sampled data are subjected to the same analysis. The same basic pattern of activation

seen in the main analysis (Figure 2) is also seen in the subsampled data (Figure 3). Although there is an expected slight reduction in the overall magnitude and extent of activation, which results from a reduction in power, the peak activations in parietal cortex are still clearly apparent. Time courses from the subsampled data (Figures 3C and 3D) closely resemble those obtained from the original data set. Similar conclusions were obtained when using the number of saccades between pictures as the measure of interest (Figure S4). There is a hint of residual effects of eye movements in early visual cortex (Figure 3, cool colors). Critically, however, activation of the dorsal attention network persisted despite these modest residual effects. These confirmatory analyses indicate that the hierarchical regression was robust and that the findings reported in Figure 2 cannot be attributed to the effects of eye movements. To identify regions associated with the retrieval of specific perceptual detail, we identified regions showing a significant main effect of Memory.