Understanding the nature of consciousness is one of the grand outstanding scientific challenges of the 21st century. Two features of consciousness stand out: the integrated and differentiated quality of conscious experience and its delayed appearance relative to the actions of the agent leading to speculations that consciousness is not necessarily the cause of actions and thoughts. These two properties create the paradox that evolution has rendered solutions to the challenge of survival that includes epiphenomenal processes. The Distributed Adaptive Control of Consciousness in Humans and Machines (DAC-CHM) project aims at resolving this paradox by using a multi-disciplinary approach to show the functional role of consciousness in adaptive behaviour and to identify its underlying psychological principles and neuronal mechanisms by constructing a neuromorphic robot based real-time conscious architecture (DACC). The embodied and situated DACC model will serve the purpose to overcome the so-called, hard problem of consciousness by showing that a third person perspective on first-person states can be achieved through a synthetic methodology deploying consciousness machines. We can call this methodology quale parsing.
The core hypothesis underlying the DAC-CHM project is that consciousness is a transient integrated sequential memory system that provides a unitary and normative description of the interaction between the agent and its (social) environment serving monitoring and valuation. The need for this sequential normative unitary virtualized task description is that it optimizes real-time performance in an increasingly more complex world by virtue of the presence of other agents. Indeed, the shift from surviving in a physical world to one that is dominated by intentional agents requires radically different control architectures that are capable of, for instance, Theory of Mind (ToM) and action prediction. At the neuronal control level, this requires combining parallel, distributed model predictive control loops to assure real-time operation with a second level of control that assures unitary intentional monitoring and valuation through sequential integrated representations of the self and its task performance in a social world. More specifically, conscious experience is driving credit assignment and planning beyond the immediately given sensory information of the physical world into the realm of hypothesized states of the social world. The DAC-CHM project will realize an existence proof of the hypothesis that this latter process is served by conscious experience. DAC-CHM advances a comprehensive framework progressing beyond the state of the art and will be realized using system-level models of an embodied and situated conscious architecture, detailed computational studies of its underlying neuronal substrate focusing on the thalamo-cortical system, empirical validation with a humanoid robot and psychophysical experimentation. The DAC-CHM project directly addresses one of the main outstanding questions in science: the function and genesis of consciousness and will advance our understanding of mind and brain, provide a new generation of robots with advanced social competence and contribute to radically new neurorehabilitation technologies.
SPECS_lab: Director Paul Verschure
PRINCIPAL INVESTIGATOR: Paul Verschure
KEYWORDS: Distributed Adaptive Control, Consciousness, Social Interaction, Robot, Declarative Memory, Computational Neuroscience, Memory