I have argued that early vision is cognitively impenetrable because its processes do not operate over cognitive contents. Recently it has been argued that pre-cueing guided by cognitively driven attention affects early vision rendering it cognitively penetrated. Since the signatures of these effects are found in early vision, early vision is directly affected by cognition since its processes use cognitive information. Here, I defend the cognitive impenetrability. First, I define early vision and cognitive penetrability. I argue that a set of perceptual processes is cognitively penetrated only if cognition undermines the epistemic role of these processes in grounding empirical beliefs. Second, I discuss the problems cognitive penetrability causes for the epistemic role of perception and relate them to the impact of cognitive penetrability on the sensitivity of perception to the data. Third, I examine the epistemic role of early vision and argue that the cognitive effects underpinning pre-cueing do not undermine this role and, thus, do not render early vision cognitively penetrable. In addition, they do not entail that early vision uses cognitive information; they are indirect effects similar to the shifts of overt or covert attention.