Aligning Robot Representations With Humans