A wide variety of hypotheses have been put forward as to the depth of non-conscious processing. According to some, it is confined to certain circuits: subcortical regions, sensory areas, the dorsal visual pathway, etc. Research in psychology and brain imaging, however, has rapidly refuted the view that high-level cortical processing is necessarily conscious. A subliminal stimulus readily activates the primary visual cortex, and even ventral visual regions, where it leads to subliminal priming effects of a high degree of invariance. In the 2007 lecture on reading, we pointed out, for example, that a subliminal word such as 'radio' can prime the subsequent processing of the same word 'RADIO' written in a different font. Numerous similar experiments involving numbers, faces, images of objects or tools leave little doubt that the whole of visual processing is likely to take place outside the conscious mind.
A major controversy, summarized by Lionel Naccache in his book Le nouvel inconscient, has focused on the next stage in word processing: access to the lexicon and a semantic representation. Several classic experiments by Tony Marcel (as early as 1974) seemed to show significant non-conscious semantic effects. However, these results were strongly criticized in a now classic article (Holender, 1986). Anthony Greenwald's team responded with a methodologically sophisticated approach that seemed to demonstrate the possibility of an unconscious semantic classification of words (Greenwald, Draine & Abrams, 1996)... until the author himself, following further experiments, demonstrated its strictly visual and not semantic origin.
