This paper argues that naive realism—the view that veridical perceptions are essentially relational—should be understood as a transcendental project rather than an ordinary theory to be tested by cost-benefit analysis. Drawing on Strawson's account of reactive attitudes, the author proposes that naive realism occupies a special foundational role in explaining how perceptual experience is possible. This transcendental status would make naive realism, in a certain sense, immune to falsification by empirical evidence or competing theories. The paper develops a modest version of this transcendental naive realism.
Laypeople hold conflicting beliefs about vision, simultaneously endorsing both Direct Realist and Indirect Realist conceptions, according to three studies using the newly developed Direct/Indirect Realist Belief Inventory (DIRBI). These conflicting beliefs are not merely superficial agreement but reflect genuine beliefs anchored in implicit knowledge structures: experiential event knowledge about vision and an implicit model of endogenous attention. The findings challenge the common philosophical assumption that there is a single, coherent common-sense conception of vision that can serve as an epistemic default in debates about perception.