by HarryStottle » Wed Apr 01, 2009 12:16 pm
First, just in case anyone thinks I'm claiming credit for "grok" let me clarify that we owe that one to Robert Heinlein though I can't remember which of his novels it came from, possibly Starship Troopers...
Second, by happy coincidence, this months Cafe Sci, which took place less than 40 hours ago as I write, was focussed on a presentation by , who, as you can see from that link, is researching deep inside the brain. I had a reasonably lengthy chat with him after the event and spent some time on the essence of the question we're discussing here.
In short, although he doesn't believe there are yet any good digital analogues for the neuron, he (and others in his field) do not perceive any material or logical barriers to creating them. We simply don't yet understand the real thing well enough yet to create an accurate simulation, but that's just a matter of time.
More interesting to me (because I'd always taken that "matter of time" for granted) was what he had to say about what I've always seen as the more intractable problem; viz the neurotransmitters.
It's always seemed to me to be the harder problem. Yes we can imagine how we'll simulate information storage and transfer between our digital neurons. That's the easy bit because it aint that far from what we're already doing with computers today. But how do we deal the "emotions" and the "modulators" of information transmission?
Unlike information storage and transmission, emotions are chemically mediated. Serotonin or Dopamine or whatever floods into the brain and changes the way information is being stored or transferred. How would we simulate that? Indeed WHY would we simulate that?
The "Why" question is the easiest. For me, at least, unless we can authentically mirror our emotions as well as our memories, then "it aint me" will remain true. What we'd have would be a super sophisticated version of a telephone answering service; capable of answering any question thrown at it as "I" would have, but completely lacking the "motivation" that makes me "Me".
The reason I'd always thought the "How" would be the difficult one was because I was under the impression that these "modulators" flooded the brain in a different way to the "information transfer" transmitters. And that this would not be easy to mimic in any "natural" way. I am rather pleased to learn that I was wrong.
There are essentially two different types of neurotransmitters (assuming I've absorbed the Prof's information correctly). One type is released at the synapse only and provides the "information transfer functionality. The other I had imagined being released only from key parts of the brain like the Amygdala and flooding a whole area (which would make it very difficult to simulate accurately). But it doesn't happen like that.
Instead, these too are released by individual neurons but NOT at the synapse. They diffuse more generally from the body of the neuron. THAT is much more straightforward for simulation and it has significantly increased my confidence that we will indeed be able to mirror "feelings" just as easily as basic memory. The combination provide "motivation" and the core of "self".
So, as of now, I'm even more confident that our digital future is plausible if not yet inevitable. There are, of course, still some amazingly difficult technical barriers to overcome but none look like showstoppers. Frankly the only "showstopper" looks to be the ongoing war between us and the Authoritarians. THAT could prevent our digital future which is why "sidelining" the authoritarians has got to be, in my view, our main priority...