Slashdot Mirror
Will You Ever Be Able To Upload Your Brain? (nytimes.com)
An anonymous reader points out this piece in the Times by professor of neuroscience at Columbia and co-director of the Center for Theoretical Neuroscience Kenneth Miller, about what it would take to upload a human brain. "Much of the current hope of reconstructing a functioning brain rests on connectomics: the ambition to construct a complete wiring diagram, or 'connectome,' of all the synaptic connections between neurons in the mammalian brain. Unfortunately connectomics, while an important part of basic research, falls far short of the goal of reconstructing a mind, in two ways. First, we are far from constructing a connectome. The current best achievement was determining the connections in a tiny piece of brain tissue containing 1,700 synapses; the human brain has more than a hundred billion times that number of synapses. While progress is swift, no one has any realistic estimate of how long it will take to arrive at brain-size connectomes. (My wild guess: centuries.)"
If an elderly but distinguished scientist says that something is possible, he is almost certainly right; but if he says that it is impossible, he is very probably wrong.
- Arthur C Clarke
Locality of self.
The problem with almost all "uploading" schemes is that it creates a copy of your brain structure, so it's a copy of you, rather than you. Externally, there might be no apparent difference to an outside observer, but internally, you're kind of dead, if that 1 cubic foot of meat space is no longer functional.
The only hope of an upload of the actual "you" would be an incremental replacement of brain structure, such that you lived in both meat-you and electronic-you at the same time, until the electronic-you completely replaced the meat-you, without a loss of continuity of consciousness.
Otherwise, you're just building pod people. Which could be useful, if you wanted to embed one of them in a starship (or more likely, a tank or other weapon of war), or if you wanted to make a lot of duplicate copies of a particular mind, and didn't care about their locality of self, either.
Connectome will be done not in centuries but a decade or less, really that's problem to be solved by automation and computing
However, the 2nd reason, left out of the quote but in the article, has to do with the function rather than physical configuration of synapses and neurons. We don't understand that well at all. And that is probably where the "mind" is.
[ - ]
Because #Concise
Get thee glass eyes, and, like a scurvy politician, seem to see things thou dost not.--King Lear
The primary problem with this recurrent geek fantasy is that at best it's not really a copy; it's an emulation on different hardware. And that means a different added layer of possible breakdowns, bugs, glitches, etc. "All abstractions are leaky", per Joel Spolsky I think. Will the person feel hungry, thirsty, sleepy, horny, too cold/hot, react the same way to their favorite booze/weed/drugs, etc.? Probably not. Will there be outages due to power, networking, input/output devices? Likely so. And it's really hard to pretend that in the face of those radically changed experiences of the world that it's the same person.
This thought experiment serves as a pretty good case study that the Western attempt to cast a hard distinction between mind and body is not really tenable. You are your body, and your body is you.
We know where leadership by an anti-intellectual "strongman" who scapegoats minorities and likes boisterous rallies goes
The entire concept of uploading/duplicating is based on a deterministic view of the universe - one without quantum mechanics.
This viewpoint is false. Not only is quantum mechanics part of the universe, but the specific reactions involved in the brain require quantum mechanics.
As such, the concept of a physical copy or uploading is nonsensical. It can not be done. The best we can do is make a poor copy - one that will NOT react the way the real you would.
What?
The specific activities involved in the computer you used to type your message require quantum mechanics. Perhaps that explains whey the poor copy that appears on my screen seems somehow incomplete or off-base.
It's possible that quantum activities in the brain make the processes of consciousness somehow non-classical and incapable of replication, but not only is the jury still out on that, I'm not even sure we've finished arraigning the suspects.
In Mind Children, Moravec described a fascinating scenario. A probe equipped with molecular-scale surgical tools, encloses a few brain cells and simulates them in software while you lie on a table. You have a switch in your hand; as you press it, you flip back and forth between the simulation and the working cells; when you can't tell the difference, the cells are removed. The probe continues to work its way through your brain until no real cells are left. You have been slowly, gradually uploaded into software. This is you, your continual awareness, not a copy of you that takes your place after you've died.
Suppose you have a stroke, and it damages a small section of your brain.
The (cerebral cortex surface) brain is made up of a repeating pattern of cortical columns, which is a structure that connects vertically among it's 6 layers, but not laterally beyond the column boundary. There are connections out the top to the higher order layers in the brain, and connections into the bottom from lower layers, but it's an independent function(*).
As far as anyone can tell, the cerebral cortex is composed of a repeating array of these columns.
Suppose you have a synthetic "plug" that can take the place of a number of cortical columns. You remove the damaged part of the brain and replace it with the synthetic plug.
The plug contains processing units which then learn from the existing connections. The human helps to train the connections by giving feedback: as the plug tries out the connections and actions, the human can tell whether the output is right or wrong, and act accordingly.
For example, if the plug was within the speech centers, the human would have to relearn that part of speech which was damaged, but he would have all the rest of his experiences and knowledge as a basis. His environment and other humans (family, friends) would also help support the learning process.
Eventually, the plug would learn the correct responses to any of the inputs, and it would be a replacement for the damaged part.
Now suppose you have another stroke, and it damages another part of the brain.
Continue the process to its logical conclusion, and you migrate the essence of the person from the biological into the synthetic. This is possible because the information in the brain is not stored in one place, but distributed over many areas. If you lose one area, the information can still be reconstructed from information in other areas.
I can well imagine when the technology gets advanced enough, that rich people might be able to get "stroke plugs" implanted, and over time completely replace the biological portions of their brain.
Is this not a sufficient definition for uploading?
(*) Yes, a glossy, simplistic description.
The entire concept of uploading/duplicating is based on a deterministic view of the universe - one without quantum mechanics.
We don't understand quantum mechanics, and we also don't understand how the brain works ...however, that doesn't mean that the brain is quantum mechanical.
Two things that have similar characteristics sometimes turn out to be quite different, and relying on "we don't understand this" as the similar characteristic that makes two things equivalent is dubious at best.
A deterministic model may be a sufficient emulation even if not a perfect emulation.
After all, a lot of people take drugs, caffeine, alcohol, get smashed in the head in football or a swimming accident, get diabetes, and still are usually more or less themselves. The brain is designed to handle a degree of "noise" and damage, and this could very well include the "noise" of an imperfect model of itself. How much is "good enough", we don't yet know.
Table-ized A.I.
Dear Future,
Please DON'T extract and emulate the trolls.
Thanks
-Present
Table-ized A.I.
> The specific activities involved in the computer you used to type your message require quantum mechanics.
So is basic chemistry, looked at closely enough. The idea that something cannot be created or functionally replicated because it's quantum mechanical is, I'm afraid, a nonsensical one.
Whether the complex interaction of state and process between a brain and its senses, between physical layout of neurons and ongoing biochemical interatctions, can be replicated to an electromechanical system seems unlikely in the extreme. Complex analog interactions are difficult to model precisely, much less replicate to the kind of essentially "digital" structure of modern computer systems.
Alan Turing said in 1936 that it's impossible to construct an algorithm that generally solves the halting problem.
So who's wrong: Clarke or Turing?
We know where leadership by an anti-intellectual "strongman" who scapegoats minorities and likes boisterous rallies goes
If the scale required is your only argument you have made a very common error regarding the speed of change in exponential processes.
What can we do now?
What is the rate of technology doubling, D?
How many times, X, do we need to do it to get to the required magnitude?
It will take D*X years where 2^X = one hundred billion
And that is without anything radically new being discovered in that time period, so 20 to 30 years is actually possible.
Imagine what a large scale 3D quantum computing array would be capable of. We have just seen silicon based quantum logic fabrication developed and we already have 3D silicon based memory arrays.
If we can reduce the number of synaptic connections in the average human brain while we are working on improving the technology, we ought to get the two to meet much sooner than the few centuries that TFS predicts.
Have gnu, will travel.
Second, we get distracted halfway through a small list.
At the bottom of the
OK, so we're mostly software geeks here who have a vague idea how the underlying digital hardware works. It's not surprising that we think of 'uploading' a mind into our limited area of expertise. But why?
Is there something wrong with biology and existing brains? We can grow brains. We are learning the first steps toward interfacing with them. Let's do what we can with real brains while adventurous explorers probe the distant frontier of digital brains.
...omphaloskepsis often...
Isn't the job of the nerves in the brain supposed to be to communicate?
Shouldn't we just have to play the role of a nerve, and just 'ask' the brain nerve to tell us its contents, and those of its close neighbors?
I mean,there's parasites that do this to an extent, such as toxoplasma gondii, seems odd that we haven't created an interface to work with nerves and just get them to communicate to us, as nerves logically have to do, in order to act like minds.
Even if the process is slow, we should be able to do it at lots of locations simultaneously, so long as it's non-destructive communications. Sure, we'd be reinforcing connections by doing the queries, but so long as it was even-handed, it would be *nothing* compared to acts like dreaming or most of regular life.
Worst case, even if we couldn't recreate a living landscape of a mind completely right away, we could at least save the long-term memories, and have something better than the complete destruction of being that happens with death now.
Even if it would be embarrassing by conventional standards, I'd actually like the idea of my complete memory set continuing after I'd dead, rather than the feeble methods we currently use to leave something of ourselves. Add a query system to it, could be very odd, but really neat too - real life information ghosts.
Far better than nothing, for my preferences at least.
Ryan Fenton
I recently read that a fairly large swath of top AI researchers were polled about when we may be likely to see human- and superhuman level artificial intelligence. The median was around 2060. It seems to me that once computers are perhaps millions of times smarter than we are, seemingly insurmountable problems such as this one will be rapidly solved. When that happens I question whether humanity will even remain biological as there are clearly disadvantages to this format.
On finding the answer the partial would signal the originating consciousness that it had completed and was ready.
At death, you consciousness was available for restoration to either reality or a simulated environment. Which didn't help if your body ended up in some inaccessible place.
My ism, it's full of beliefs.