ABSTRACT
Empirically, the central characteristic of unconscious processing is the observation that an agent’s behaviour is influenced by knowledge of which it remains unaware. Connectionist models have provided genuine insights into how knowledge can influence processing without access—a hallmark of unconscious processing—and of how change can accrue as a result of mere information processing—a hallmark of the phenomena of implicit learning. An important consequence of the fact that long-term knowledge in connectionist networks accrues in connection weights as a mandatory consequence of information processing is that connectionist models capture, without any further assumptions, two of the most important characteristics of implicit learning, namely the fact that learning is incidental and mandatory, and the fact that the resulting knowledge is difficult to express. Numerous theories of consciousness have been proposed over the last twenty years. Global Workspace Theory (GWT) is the most consensual account of the functional characteristics of consciousness.
