Scientific requirements for an engineered model of consciousness

Abstract

The building of a non-natural conscious system requires more than the design of physical or virtual machines with intuitively conceived abilities, philosophically elucidated architecture or hardware homologous to an animal’s brain. Human society might one day treat a type of robot or computing system as an artificial person. Yet that would not answer scientific questions about the machine’s consciousness or otherwise. Indeed, empirical tests for consciousness are impossible because no such entity is denoted within the theoretical structure of the science of mind, i.e. psychology. However, contemporary experimental psychology can identify if a specific mental process is conscious in particular circumstances, by theory-based interpretation of the overt performance of human beings. Thus, if we are to build a conscious machine, the artificial systems must be used as a test-bed for theory developed from the existing science that distinguishes conscious from non-conscious causation in natural systems. Only such a rich and realistic account of hypothetical processes accounting for observed input/output relationships can establish whether or not an engineered system is a model of consciousness. It follows that any research project on machine consciousness needs a programme of psychological experiments on the demonstration systems and that the programme should be designed to deliver a fully detailed scientific theory of the type of artificial mind being developed – a Psychology of that Machine.