Say that a model's interpration assigns true to A and false to B. Then the valuation function assigns false to "not A", true to "not B", false to "A implies B", false to "A and B", and true to "A or B".