Figure 7.

Percentage signal change for trained and untrained object classes in trained voxels. Trained voxels were selected based on a significant post-training preference for trained objects over untrained objects. Data are shown for different sets of runs: post-training selection runs (the post-training runs used to select the voxels), post-training test runs (the other runs in the post-training scan session), and pretraining runs (all runs in the first scan session). The error bars show the SEM across subjects of the difference in the response between trained and untrained objects.