On ecological inference from laboratory experiments conducted under optimum conditions