|
|
|
|
|
|
by sandinmytea
2768 days ago
|
|
|
For anyone trying to generalize Mr. Jayne's theory from an article or set of comments - I suggest that you definitely do not. Really, don't do that with anyone's theories, but instead READ THE BOOK. The way that Jaynes "carves up consciousness" as Daniel Dennet puts it in one review, is very different than the way that a reflexive intellectual instinct would choose to. The taxonomy that Jaynes decides upon is distinctive from our mental reflexes, and not at all the same as what nearly everyone begins with. I feel that this "Bicameral" consciousness discussion keeps reappearing on Hacker News in part because it is crucial for us to create a proper distinction between these terms - AI is rapidly going to become conscious in exactly the way Julian Jaynes suggests we became "conscious" and that consciousness is very much different than what we think it is (or must be) for AI to become generally aware of itself. I suggest that anyone who is working towards AGI MUST discover the distinctions that Julian Jaynes made. That does not mean that his theory regarding our consciousness is correct, but merely that his ways of describing "self-awareness" or whatever other term could be applied is crucial to getting a grip on what the true mechanism of "the self" is in the truest engineering sense. That is to say, that if the "self" is the licensed self-aware driver of a vehicle and the engine and transmission is the remainder of cognition, than the order of development was actually from "self-driving car" into human driven vehicles - metaphorically speaking. Basically, up until recently, we just DID, and in attempting to describe WHY we were forced to invent metaphor until collectively we were able to create the language necessary to actually BE self aware. Without the language, we actually were NOT self aware. We were self-driving cars with no human drivers. The human driver only arrived after there were ways of describing "him." |
|
|
Self-awarenes was worked out with "I know that I don't know" and again, but in positive terms with "I think therefore I am". These are the most basic forms of reflection, a y-combinator differentiating endlessly. With literarry writing, this reflection could be stretched out over generations, yielding a national identity, but oral transmission of religious teachings had been practiced before, just with a higher rate of mutation. Such verbalization is important for stability, and similarity is important for discoverability, but if verbalization hadn't happened, the perhaps because it goes without saying.
PS: A single threaded cpu doesn't need to spell out distributed computing primitives, either. It still may use signals, in many cases, even if implemented completely in software.
PPS: maybe there's an interesting parallel from natural language to the progression from imperative language over object oriented or declarative, to agent based programming. Bicameral Petri Nets ...