Table S4.

Post hoc analyses of agent and patient identification performance within their corresponding ROIs when generalizing to each verb context and new pairwise noun discriminations

Content	Patient in patient ROI	Agent in anterior agent ROI	Agent in posterior agent ROI
Generalization by verb context
Verb
“Chase”	2.12, P = 0.022	2.04, P = 0.026	1.59, P = 0.063
“Block”	1.67, P = 0.054	2.37, P = 0.013	2.16, P = 0.021
“Bump”	3.03, P = 0.003	1.78, P = 0.044	2.37, P = 0.013
“Approach”	2.15, P = 0.021	2.61, P = 0.008	3.19, P = 0.002*
“Scratch”	4.31, P = 0.0001*	2.77, P = 0.005	1.87, P = 0.037
Generalization by noun pair
Noun pair discrimination
Man/Girl	1.85, P = 0.039	2.68, P = 0.0065	3.21, P = 0.0019*
Man/Dog	1.23, P = 0.115	1.47, P = 0.077	2.10, P = 0.023
Man/Cat	2.90, P = 0.004	2.22, P = 0.018	2.02, P = 0.027
Girl/Dog	1.20, P = 0.121	1.08, P = 0.145	2.62, P = 0.0075
Girl/Cat	3.53, P = 0.0009*	3.2, P = 0.0019*	1.56, P = 0.066
Dog/Cat	1.71, P = 0.05	1.20, P = 0.121	1.06, P = 0.15

Values are t statistics, with one-tailed P values against chance. Our main goal is to qualitatively describe the pattern of results. However, asterisks (*) mark accuracies that withstand Bonferroni correction for multiple comparisons.