 
Summary: ExperienceDependent Representation of Visual Categories in Parietal Cortex
David J. Freedman and John A. Assad
Supplementary Methods
Computation of BCD, WCD and categorytuningindex values. The strength of neuronal
categorytuning was estimated by a categorytuning index that was computed from two values: (1)
the average withincategory difference (WCD) in firing rates between pairs ofdirections in the
same category, and (2) the average betweencategory difference (BCD) in firing rates between
pairs of directions in different categories. Our goal was to have equal numbersof paired
comparisons forthe BCD and WCD calculations, and to only include comparisons that were
represented in both BCD and WCD calculations. For these reasons,both BCD and WCD values
were determined by computing the averageabsolute firingrate differences for 18 pairs of
directions, including comparisons across 30°, 60°, 90°, 120°, and 150° intervals. 180° was
excluded fromthe BCD calculation because therewere no corresponding pairs of directions for the
WCD calculation that were 180° apart and in the same category (although, including the 180° BCD
interval gave qualitatively similar results). The 18 pairs used for the WCD calculations were
identical in structure to the 18 pairs used forthe BCD, except that they were all rotated 90° with
respect to the category boundary. In other words, we used the 18 withincategory pairs of
directions (nine in each category) for the WCD calculation that, when rotated 90°,crossed the
category boundary. For example, the 30° BCD value was computed across the two pairs of
directions that were 30° apart and crossed the category boundary (i.e. 120°  150° and 300°  330°;
