On the determinants of the conjunction fallacy: confirmation versus probability