Gravar-mail: Reply to Henrich et al.: Behavioral variation needs to be quantified at multiple levels