Abstract: Some years ago I suggested that consciousness pays its way in the functional economy of the brain by unlocking the savings hidden in the mutual dependencies among target selection, action selection and motivational ranking through multi-objective constraint satisfaction among them (Merker 2007, p. 70). This would place consciousness at a late stage in the brain's operations, suggesting a subcortical implementation of key mechanisms of consciousness in sites of global convergence in midbrain and diencephalon. No doubt our cortical machinery is the source of much of our conscious contents, but that does not mean that the cortex also must be the site where those contents become conscious. Recent information-theoretic analyses of the probabilistic data format of cortical operations point to the utility of collapsing cortical probability density distributions to estimate form in extracortical locations (Ma et al. 2006). I propose that this essential step in neural operations is implemented in a subcortical "global best estimate buffer" whose contents - alone among neural activities - are conscious. They are so not by virtue of anything being "added" to them in order to "make them conscious," but as a direct consequence of the format they must adhere to in order to provide a global best estimate within the narrow time constraints of inter-saccadic intervals. That format directly matches the global format of our phenomenal experience, which in its sensory aspects is that of naive realism.
Ma, W.J., Beck, J.M., Latham, P.E. & Pouget, A. 2006. Bayesian inference with probabilistic population codes. Nature Neuroscience, 9, 1432-1438.http://psych.stanford.edu/~jlm/pdfs/Ma%20et%20al%20with%20figs.pdf
Merker, B. 2007. Consciousness without a cerebral cortex: A challenge for neuroscience and medicine. Target article, peer commentary and author--response. Behavioral and Brain Sciences, 30, 63: 134. http://eprint.ncl.ac.uk/file_store/production/87645/626A59CF-2311-4DA6-BFBD-2D4AACC68DD3.pdf