I agree that the "pure" variant of "once it has enough connections, it spontaneously wakes up!" is utterly a misunderstanding based on folk biology. I think, though, that the variant "something very complex that we train to emulate most of the activities of consciousness spontaneously develops consciousness" is a somewhat different case. If you put "simply having enough circuits = consciousness" on one end of the spectrum -- a giant, empty Neumann-architecture with blank RAM, you simply flip the switch and it starts talking in Voiceover Of God -- and on the other end of the spectrum you put "The Lifecycle of Software Objects" -- machine consciousness as a painstakingly developed, heartrendingly fragile one-on-one project -- it seems to me most extropian/singularitarian stories are somewhere in the middle.
I think that these guys are excited about the frisson of "emergent consciousness" as an *unintended* consequence, not a planned project, and that therefore there's some aspect of "it wakes up!" But this doesn't have to be wholly naive folk biology. It might be that various elements and capacities of consciousness are developed separately -- even painstakingly as per "Lifecycle" -- and the "surprise" is only when they at some point accidentally combine to make a whole greater than the sum of their parts.
I note that Vinge's original paper, which you and I traditionally like to cite, does not involve any "enough connections and suddenly it wakes up!" element, but rather leaves plenty of room for the "development of superhuman intelligence" to be an arduous and purposeful process. Indeed there need not be any dramatic "takeover" by the intelligent computer, it seems rather to be implied that the same economic forces which incentivized developing the computer of intelligence N lead to asking it to develop the computer of intelligence N+1.
(I also note that while, in that paper, Vinge leaves us the range 2005-2030 for the Singularity to occur, at one point within it he places a more casual offhand bet of next year: "And for all my rampant technological optimism, sometimes I think I'd be more comfortable if I were regarding these transcendental events from one thousand years remove ... instead of twenty." That's 2013! I really hope someone commissions Vinge to write a followup paper in 2030...)
I think the "folk biology" argument you're making is accurate, but I think it's insufficient to just take vague swipes at "extropians" without citing which stories you mean. Because if you're just talking about random internet posters in brain-freezing forums, that's a weaker argument than if you really find this folk biology among the thought leaders: in Kurzweill's essays or in the stories of Vinge, Egan, Stross, Rucker, Reed, et al. My own impression is that you're more likely to find "enough circuits and it wakes up!" in Golden Age sf writers, and that the writers in the Singularity genre generally are talking about neural nets, further breakthroughs in strong AI, programmatic development of machine consciousness, etc. (Maybe the new Kessel & Kelly antho would be a good place for us to look...)
I will allow that there are two stories I can think of where I'm guilty of explicitly referencing Neumann architectures to describe posthuman consciousness -- "True Names" and "The House Beyond Your Sky" -- though in my defense I will say that in the first case I was trying to be funny, and in the second case to explicitly mark the Neumannesque language as being technological metaphors employed in the same way as the technologies "clay hut" and "parakeet cage"...