This article presents a critical review and analysis of key studies that have been done in science education and other areas on the effects and effectiveness of using diagrams, graphs, photographs, illustrations, and concept maps as "adjunct visual aids" in the learning of scientific-technical content. It also summarizes and reviews those studies that have students draw diagrams, graphs, maps, and charts to express their understandings of the concepts and relationships that are present in the text they read or/and empirical data provided (i.e., student-generated "adjunct visual productions"). In general, the research and theory on instructional aids is fragmented and somewhat unsystematic with several flaws and a number of key uncontrolled variables, which actually suppress and mask effects in the studies that have been done. The findings of these studies are compared to relevant literature and empirical research and findings in the areas of cognitive psychology, computer science, neuroscience, and artificial intelligence that help to clarify many of the inconsistencies, contradictions, and lack of effects found for visual (e.g., diagrams and graphs) instructional aids in the science education literature currently and in the past 20 years. A model and a set of criteria and goals for improving research in this area is then described, as visuals are a first step in the process of learning formal (scientific) models, which are most often visually represented. Understanding how students learn formal models is one the outstanding research challenges in the next 20 years, both within and outside of science education.