Positive versus Negative. A cognitive perspective on wording effects for contrastive questions in attitude surveys