Denaturalizing Information Visualization

This dissertation investigates how, behind the recent proliferation of information visualization technologies, a number of opaque epistemic conventions structure distinct practices of data analysis, sensemaking, argumentation, and persuasion. The dissertation's primary claim is that contemporary infovis practices and tools naturalize specific representational forms and modes of interaction (e.g. ocularcentric objectivity) at the expense of others. It examines this naturalization process through a series of four interconnected case studies that consider, in turn, the translation of 2D images and graphics into tangible 3D objects; the development of two multimodal visualization projects; scale experiments in a virtual reality-based visualization environment; and an empirical study of 14 blind and visually impaired citizens interacting with tactile data objects. By surfacing specific naturalizations and attempting to denaturalize them, these four case studies illuminate a host of critical issues (e.g. a long-simmering tension between truth and aesthetic concerns) raised by the appearance of new interactive approaches to visualization. In doing so, they also demonstrate alternative visualization methods that might be deployed in service of a more critical infovis practice. As a result, the dissertation contributes to an emerging body of critical infovis research that spans domains ranging from HCI to critical data studies. Visualization media has become the de facto currency of data interpretation, and a rich and nuanced understanding of the epistemology of information visualization is absolutely necessary. Discussions around the challenge of multimodal representation, the tendency for new media to produce epistemic conflict, and the benefits of critical data literacy, all of which appear in the dissertation, will help colour in this rich and nuanced picture of visualization.Ph.D