Embodied cognition summary — LessWrong