For this analysis, we trained the classifier with data from one location (indicated by the label on the x-axis), and we tested the classifier with data from either the same location or from a different location. While in general training and testing at the same location led to slightly better results, the difference was small (as can be seen by comparing the left most three bars on each plot), and even this small difference disappeared when training with data from all locations (right two bars), showing that it is easy to learn a position invariant representation from data in IT. The results in this plot all comes from 7 object stimulus set (from monkey 2).