Testing hypotheses about social targets: The effects of instructions on asymmetric strategies