We aim to find the limits of subliminal information processing and clarify the function and brain architecture underlying conscious processing in adults, infants, patients and non-human primates. (1) We will design experimental tests applicable to non-verbal organisms that can reveal behavioral and cerebral signatures of conscious processing. Those tests, respectively called rule extraction and central collision , each comprise an automatic sensory component and a central component thought to require conscious access. (2) Using these tests, we will identify brain signatures of non-conscious and conscious processing using neuroimaging techniques (ERPs, MEG, fMRI, intracranial recordings) in normal human adults. To validate our approach, we will manipulate stimulus perceptibility (masking), attention (distraction by another task) and vigilance (sleep and anesthesia). (3) We will then extend the approach to brain-damaged adults with coma, persistent vegetative state or minimal consciousness, in order to detect residual processing and to obtain predictors of recovery. We will design computer systems to extract signatures of conscious processing in real time. (4) We will also examine when these signatures first appear in human infancy. (5) Finally, we will measure fMRI activation in monkeys during the same tests, thus allowing for a direct comparison of monkey and human signatures of conscious processing. We will study the effects of anesthesia on the loss of these signatures, and the potential beneficial impact of thalamic stimulation on their restoration. This research will clarify the brain mechanisms of conscious processing, illuminate their ontogeny and phylogeny, and pave the way to clinical intervention studies in patients with impaired consciousness.