Re:A New Theory of How Consciousness Evolved
by simonh on Tuesday June 07, 2016 @08:40AM
(#11854031)
I think the problem of androids, sometimes referred to as zombies, is based on a fallacy because I don't think it's possible for non-conscious systems to exhibit the behaviours of conscious systems.
I do believe that the experience of consciousness is based in a model of consciousness in the brain. This also leads to our ability to form mental models of other beings, systems and processes which enable us to predict or form theories about their behaviour. It's this existence of a model of our mental world in the brain and our brain's interrogation and interaction with it that leads to concious experience. I don't know the details of how that works and I'm no Philosopher so maybe I'm over simplifying, but it seems to me that something along these lines is the most plausible theory.
Given this, the existence of such models in the processing system of a correspondent should be testable. You should be able to test the ability of the correspondent to form and generate models of itself, beings, processes and systems and interrogate and mutate those models. One example of such systems are games. My favourite ways to test a chat bot Turing Test style is to try and teach it a game or to introspect on itself and it's understanding and expectations of others. A correspondent that cannot form and process such models cannot be conscious, but if it can then if such models lead to consciousness, by definition it must be concious.
I do believe that the experience of consciousness is based in a model of consciousness in the brain. This also leads to our ability to form mental models of other beings, systems and processes which enable us to predict or form theories about their behaviour. It's this existence of a model of our mental world in the brain and our brain's interrogation and interaction with it that leads to concious experience. I don't know the details of how that works and I'm no Philosopher so maybe I'm over simplifying, but it seems to me that something along these lines is the most plausible theory.
Given this, the existence of such models in the processing system of a correspondent should be testable. You should be able to test the ability of the correspondent to form and generate models of itself, beings, processes and systems and interrogate and mutate those models. One example of such systems are games. My favourite ways to test a chat bot Turing Test style is to try and teach it a game or to introspect on itself and it's understanding and expectations of others. A correspondent that cannot form and process such models cannot be conscious, but if it can then if such models lead to consciousness, by definition it must be concious.