[ExI] Digital Consciousness

[Ben Zaiboc]

From: Gordon <gts_2000 at yahoo.com> wrote:

>I am myself a programmer, by the way (C++). I can make computers do interesting things that appear conscious, but I don't suppose I could ever write a program that would make a computer actually conscious. That would require magic, like the magic that made Pinocchio come alive. What would the code look like? How do I write a "become conscious" function?


Aha, a question about something real and useful! (rare in these kinds of discussion).

It seems likely that consciousness and self-awareness are tied in with the various models we make of the world, and of agents that act in the world.  We have at least one (probably several) models that represent ourselves, so I suspect that if you could write some classes that implement such models, and make them interact with each other and with inputs from other areas of the brain - sensory maps of all kinds, memories, associations, language areas, attention-directing networks, etc. - you might be on the right track.

I expect that taking an incremental approach, using comparative evolutionary neuroanatomy as a guide, would be a good idea.  Pick a simple reactive creature with a nervous system to model, see what comes next in the evolution of complex nervous systems, figure out how that works, etc., bearing in mind how brains seem to evolve - duplicating functional modules, then specialising them - and the fitness environment they evolved in (what kind of problems drove the development of the next level of complexity?).  I wouldn't be surprised if at about the level where social interactions start to become crucially important, some kind of self-awareness starts to emerge.

I doubt if any system without some way of creating internal models of agents can be said to be conscious in any sense, but once the models start, then interesting things happen, and when the models start to include self-models, then it gets /really/ interesting.

Ben Zaiboc