Modeling the global neuronal workspace with communicating timed automata

Overview & Synopsis

The Global Neuronal Workspace Theory (GNW-T) -one of the most empirically supported theories of consciousness -describes conscious processing as a global broadcast of information from specialized brain modules to many others via a network called the global neuronal workspace. Despite a growing body of computational models of the GNW-T, none use the formalism of communicating automata, which offers modular compositionality, visual representations, scalability, and, in some cases, verification. We address this gap by introducing a new variant of systems of communicating timed automata for building neurocognitive theories grounded in the GNW-T, in which unconscious and conscious communication are modeled as distinct synchronization mechanisms. We prove that the untimed version of this model is both simulated by and simulates Petri nets. We instantiate this architecture in a formal model covering certain key elements of the theory: unconscious processing in a visual module, attention, preconscious stimuli, attentional binding, and verbal report.