Colloquium - Wayne Wu, Carnegie Mellon
Abstract: Historically, the empirical case for unconscious vision in neuropsychological patients relies on a discrepancy between objective measures, say accurate performance in detection tasks, and introspective reports, say the subject’s denial that they see the relevant stimulus. The problem is that the empirical argument works only if the measures are reliable, yet we have no way of assessing the reliability of introspective report. In this talk, I treat introspection of visual consciousness as a phenomenon that can be understood empirically and for which we can give a plausible information processing model. The model draws on fairly uncontroversial theses about attention and the structure of action. From this, I derive a condition for introspective unreliability and show that two standard neuropsychological cases purported to demonstrate unconscious vision, (a) the visual agnosia patient D.F. and (b) blindsight, involve unreliable introspection. Accordingly, the argument for unconscious vision that draws on these cases is unsound. I draw some general conclusions about the use of introspection to establish the presence of conscious states and alternative measures that might replace it.