Geoscience visualizations are commonplace; they appear in television news programs, classroom lectures, conference presentations, and internet hypermedia. But to what degree do individuals who view such visualizations actually learn from them, and if so, why? As visualizations become more commonplace in school, laboratory, and entertainment settings, there has been a concurrent interest in considering the effectiveness of such presentations. How can we build effective collaborations that address pedagogical questions in the earth sciences while also informing theories about the cognitive processes that underlie visualization experiences? In this chapter, we contend that only through directed, collaborative projects between earth scientists and cognitive scientists will significant advances in visualization research take place. We describe two specific models of such collaboration, the advisory model and the reciprocal model, and argue that a reciprocal model presents a more effective framework for addressing important questions about the nature of visualization experiences. Such a model will inform both the design of effective visualizations for teaching complex geoscience topics, as well as provide insight into the processes that underlie learning from visualizations.