Where to Look Next? Combining Static and Dynamic Proto-objects in a TVA-based Model of Visual Attention