Differential patterns of preference for modified linear perspective in children and adults