An integrated information theory of consciousness

The Annual Human Mind and Brain Lecture

Over the past decades, studies have investigated the neural correlates of consciousness with increasing precision. However, why experience is generated by the cortex and not the cerebellum, why it fades during certain stages of sleep and returns in others, or why some cortical areas endow experience with colors and others with sound, remains unexplained. Moreover, key questions remain unanswered. For example, how much consciousness is there when only a few brain 'islands' remain active? How much during sleepwalking or psychomotor seizures? Are newborns conscious, and to what extent?
Are animals conscious, how much, and in which way? Can a conscious machine be built? To address such questions, empirical observations need to be complemented by a principled theoretical approach. The information integration theory (IIT) has several related aims: to characterize, starting from phenomenology, what consciousness is and how each experience is structured; to account for several neurobiological observations about its neural substrate; and to develop measures of consciousness that can be applied, at least in principles, to humans, animals, and machines.