Improving ecological validity in research on social cognition