Neural theories of visual consciousness frequently focus on the question of what is needed for a visual stimulus to enter consciousness. A common notion is that representations and processes in sensory regions of the brain can operate outside of conscious perception, and that some “extra property of processing” has to come on top in order to let these representations enter conscious experience (e.g., Dehaene & Naccache 2001). This extra processing property can range from neural synchronization of neurons encoding the stimulus (Engel & Singer 2001), recurrent and feedback processing (Lamme 2006, this collection; Pascual-Leone & Walsh 2001; Singer this collection), to participation in a global coherent process, known as neuronal workspace theories (Baars 2002; Dehaene & Naccache 2001). Discussion of the neural correlates of consciousness (NCC) has often focused on this extra ingredient needed to bring a stimulus representation into consciousness. However, a related, but somewhat different question has often been neglected: Which neural representations (can) precisely participate in encoding the various dimensions of conscious experience? For this it is not enough to establish a correlation between conscious perception and neural signals. That would yield a far too large set of candidate brain regions, including, say, signal patterns in the retina that also correlate with conscious perception. Instead, it would be desirable to identify which neural representations most closely encode specific contents of consciousness and can be used to explain dimensions of conscious perception under as many different conditions as possible, and down to the level of single contents. This article will focus on how to identify such core neural correlates of contents of consciousness (NCCCs; Chalmers 2000; Block 2007; Koch 2004).
It is desirable that studies of visual awareness take NCCCs into account because specific theories of visual awareness make specific predictions regarding the encoding and distribution of sensory information (e.g., Dehaene & Naccache 2001; see also Baars 2002). In the following, I will first outline the more standard techniques for identifying NCCCs, along with their shortcomings. The next step proposes to use multivariate decoding techniques (reviewed e.g., in Haynes & Rees 2006) as a tool to identify NCCCs. Decoding can serve as an empirical technique that can establish which brain regions bear most information about specific contents of visual experience. This is an important first step towards establishing a more rigid mapping between visual phenomenal states and content-encoding brain signals. Then, several examples will be presented where multivariate decoding of visual experiences can help inform specific questions regarding NCCCs, such as whether information in V1 participates in visual awareness, whether imagery and perception share the same underlying neural codes, or whether the prefrontal cortex contains any dynamic NCCCs for coding specific dimensions of conscious experience.