June 11, 2008
The Secular Rapture
Via The Valve, the IEEE has a special issue on The Singularity. Ray Kurzweil and Neil Gershenfeld discuss 
"Two Paths to the Singularity" can be found here. The views of some prominent computer scientists, biologists and all-around smart people such as T.J Rogers, Gordon Moore and Steven Pinker can be found here. Pinker:
WHO HE IS Professor of psychology at Harvard; previously taught in the department of Brain and Cognitive Sciences at MIT, with much of his research addressing language development. Writes best sellers about the way the brain works, like The Blank Slate (2002) and The Stuff of Thought (2007).
SINGULARITY WILL OCCUR Never, ever
MACHINE CONSCIOUSNESS WILL OCCUR “In one sense—information routing—they already have. In the other sense—first-person experience—we'll never know.”
MOORE'S LAW WILL CONTINUE FOR 10 more years
THOUGHTS “There is not the slightest reason to believe in a coming singularity. The fact that you can visualize a future in your imagination is not evidence that it is likely or even possible. Look at domed cities, jet-pack commuting, underwater cities, mile-high buildings, and nuclear-powered automobiles—all staples of futuristic fantasies when I was a child that have never arrived. Sheer processing power is not a pixie dust that magically solves all your problems.”
Posted by Robin Varghese at 11:11 AM | Permalink
Comments
Well, for all his intellectual shortcomings, Pinker does have his moments. Of course, when you're in a room full of mental midgets with a mutual delusion, you don't have to be terribly brilliant to point out that the emperor has no clothes. (Excuse the mixed metaphor, please.)
Kurzweil and Gershenfeld are snake-oil salesmen. I find it unlikely that even they believe the nonsense that they spout. And it's embarrassing that anyone but 13-year-old boys with visions of flying cars takes any of this fantasy seriously.
Posted by: Picador | Jun 11, 2008 11:52:11 AM
If Pinker is right that Moore's law won't continue, then it's plausible he'll be right that there won't be any sort of singularity. But if the "sheer processing power" needed to simulate a human brain does become available, Pinker doesn't offer any argument as to what fundamental obstacles could prevent scientists from mapping and simulating a human brain at a sufficiently detailed level that its behavior would be just like that of the organic brain the simulation was based on (the idea known as mind uploading). If the reductionist view of brain function is correct, then scientists would not have to understand much about the large-scale functioning of the brain to do this, they'd just need to be able to accurately simulate interactions between individual neurons, and to faithfully reproduce the arrangement of all the neurons of a real brain in the simulation.
The same IEEE issue also has an interesting article on current efforts to do something just like this with a fruit fly brain--see here.
Posted by: Jesse M. | Jun 11, 2008 1:10:44 PM
Thank you, Jesse M. -- long time, no see.
Posted by: Elatia Harris | Jun 11, 2008 1:53:58 PM
Hi Jesse,
Pinker doesn't offer any argument as to what fundamental obstacles could prevent scientists from mapping and simulating a human brain at a sufficiently detailed level that its behavior would be just like that of the organic brain the simulation was based on (the idea known as mind uploading).
Does Pinker deny this, here? I take his point to be that no matter how much a machine behaves like a conscious being, we won't have access to its experience and thus won't be able to verify that it is conscious.
If the reductionist view of brain function is correct, then scientists would not have to understand much about the large-scale functioning of the brain to do this, they'd just need to be able to accurately simulate interactions between individual neurons, and to faithfully reproduce the arrangement of all the neurons of a real brain in the simulation.
The $1,000,000 question is: does it matter what sort of stuff is doing the simulating? After all, the only perfect brain-simulator is a brain. Information is going to be lost on any digital or virtual replication. Furthermore, we have no conclusive arguments to rule out the possibility that neurons themselves may be necessary for consciousness, that there is something to the stuff that makes them up which is essential.
As I understand it, the cutting-edge AI is not aimed at replicating or simulating any present mind but in growing one, on the reasonable assumption that all known minds start from almost nothing and grow according to exceedingly complex biological and environmental interactions. If I recall correctly, I think this is one of Pinker's main motivations, here.
Posted by: Nick Smyth | Jun 11, 2008 2:37:38 PM
I'm interested in the cognitive leap (of faith?) that jumps from simulating brains to actually re-creating them in some other medium (hardware or software)--and, further--that imagines that these manufactured brains won't need human bodies in order to have human experiences.
Picador is exactly right to call this a fantasy. Even if the singularity is a technical possibility, which I find no reason to believe, why would we want to trade our world of senses for a simulation? (theoretically we could gain extended life spans or even immortality--if you call that living).
Jesse, you seem to be a singularity-booster. What's in it for you?
Posted by: Chris Schoen | Jun 11, 2008 3:59:44 PM
Nick Smyth wrote:
Does Pinker deny this, here? I take his point to be that no matter how much a machine behaves like a conscious being, we won't have access to its experience and thus won't be able to verify that it is conscious.
True, but then you also have no way to verify that other humans are conscious either. This is a philosophical issue, and not really relevant to the question of whether some sort of "singularity" scenario will come about. (And as philosophy goes, I think David Chalmers' dancing/fading qualia argument makes a good case for believing that a being which behaved just like a human would have the same sort of inner experience too).
The $1,000,000 question is: does it matter what sort of stuff is doing the simulating?
As long as reductionism is correct--i.e. the behavior of any complex system can in principle be boiled down to interactions between its parts which follow lawlike rules--and as long as it's true that the rules governing these interactions can be approximated arbitrarily well by an algorithm of sufficient complexity, then it should be possible in principle to approximate the behavior of the system as a whole arbitrarily well in a computer simulation (the computing power needed is another question). I think it would be very surprising to most scientists if this failed to be true for any physical system, whether the brain, the earth's atmosphere, the interior of a star, the motions of a fluid...in every scientific field it's generally assumed that the behavior of all systems can in principle be explained in reductionist terms, and all the most fundamental physical laws describing the interactions of the most basic entities like particles are thought to be computable ones.
And as a practical matter, if you just want to simulate the high-level behavior of some complex system, you can pretty much always do it using some coarse-grained picture of the "parts" making it up...you don't need to simulate every quark in the atmosphere to get a simulation which reproduces the sort of weather and climate phenomena that interest us. In the case of the brain, we'd be interested in reproducing the same patterns of electrochemical impulses in neurons that are found in real brains, so we'd need simulated neurons whose "output" in response to "inputs" was close to that of real neurons...as long as you have that you don't need to simulate every movement of every particle in the neuron, the simulation just needs to be faithful enough to reproduce this high-level behavior of neurons.
P.S. Hi Elatia--yeah, been a while since I posted any comments huh? I've been taking a bunch of trips this summer, but I still always manage to find time to read over the latest posts here!
Posted by: Jesse M. | Jun 11, 2008 4:23:00 PM
Although I disagree with Jesse's premise, I don't see his post as one of a 'singularity-booster.' Maybe a singularity advocate... He's just posing an argument.
Along the same logical thread, however, it is important to note that the neurons composing 'the brain' are connected pretty intimately to the rest of the nervous system, which is connected to the body, etc. So specifying the 'boundary conditions' to model the interactions of the brain with the rest of the body can't really be done in the reductionist mode, which doesn't even work that well in (say) fluid mechanics.
More to the point, there is any reductionist model of a neuron that would fit the bill. The reference linked in Jesse's post refers to the fruit-fly work as a "20-year experiment." Their approach is highly speculative... At this point, certainly, the neuron is not a known quantity.
On this basis, I'd refute Jesse's suggestion that 'just need[ing] to be able to accurately simulate interactions between individual neurons' is a well-defined idea.
At a gut level, I am kind of with Picador on this one. The arguments supporting this stuff seems so weak as to invite scorn. Though I assure you from direct experience, there are lots of folks much older than 13 that take this stuff seriously and will also allocate (and spend!) our tax dollars on such fantasies.
Posted by: sifta | Jun 11, 2008 4:35:41 PM
Sorry, first sentence in paragraph 3 of my previous post should read:
"... there is not any reductionist model ..."
Posted by: sifta | Jun 11, 2008 4:42:20 PM
True, but then you also have no way to verify that other humans are conscious either.
No, but I do have inductive evidence for it. I am certain that I am conscious, and I know that my consciousness is somehow produced by my brain. I know that other human beings have brains. An AI machine of the sort we're considering, on the other hand, has no brain, in fact, it doesn't even have a physical representation of it, but merely a virtual one.
In short, I'm willing to bite this bullet. Of course I don't know for sure if you're conscious. I can, however, be fairly certain of it.
I would definitely dispute your claims about scientists and the supposed orthodoxy that all sciences reduce to physics. Any rational practicing scientist must realize that not only is physics seriously incomplete, but that it is leading us into realms where the notions of physical "matter" and standard causal laws are seeming positively quaint. Furthermore, microphysics isn't the only area of difficulty: The ontological problems of supervenience, holism and emergence extend from quantum mechanics to condensed-matter physics and thermodynamics.
I'd also note that reductionism is largely unpopular in modern philosophy of mind, even if scientists are "surprised" by this. As I'm sure you're aware, the problems of Qualia, Intentionality and Multiple Realizability form a powerful set of arguments against the thesis. So, I hope we can agree that Pinker isn't totally nuts to claim that processing power isn't going to solve this problem.
Posted by: Nick Smyth | Jun 11, 2008 5:18:42 PM
Chris Schoen wrote:
I'm interested in the cognitive leap (of faith?) that jumps from simulating brains to actually re-creating them in some other medium (hardware or software)
What is the distinction you are making between "simulating" a brain and "re-creating" it? Obviously a simulated brain is physically distinct from a biological brain, but if it is behaviorally the same, then you can have a social relationship with it, it can come up with ideas and art and so forth, just like with a real brain. So the impact on society would be pretty significant, especially if the simulation could run much faster than a biological brain (and the speed of signals in real brains is fantastically slow compared to the speed of signals in a computer).
--and, further--that imagines that these manufactured brains won't need human bodies in order to have human experiences.
Of course a simulated brain would require inputs like those from a real body or it would probably go crazy or become comatose, like a person locked in a sensory deprivation tank forever. But if you can accurately simulate the behavior of neurons in a brain, you could presumably simulate sensory neurons too, so you could give it a simulated body which would feel just like a real one, or a robotic body to interact with the real world (or even a real biological body without a brain). That's the ideal, thought I could see it happening that the first uploading projects wouldn't have all that much concern for the comfort of the upload, and would stick them with clumsy bodies that give them enough sensory information so they don't go crazy but which the uploads themselves would find much less than ideal. But if uploading is possible at all, I imagine over time the population of uploads would grow and there would be more push for bodies (simulated or robotic) that could reproduce the same richness of sensations as biological ones.
Picador is exactly right to call this a fantasy. Even if the singularity is a technical possibility, which I find no reason to believe, why would we want to trade our world of senses for a simulation? (theoretically we could gain extended life spans or even immortality--if you call that living).
Again, having a simulated brain does not imply you must live in a simulated environment, a simulated brain can receive sensory information from a physical body just like a biological brain does. The main advantage of spending time in a simulated environment would be speed--if the simulated brain was running much faster than a biological brain, it would presumably prefer a body and environment that moved at "normal" speed from its subjective point of view, rather than in glacially slow motion. And if there were communities of upload researchers willing to spend a few years or decades of subjective time doing research in a realistic simulated world while only days or minutes or seconds passed in realtime, I think this would likely lead to a "singularity" scenario where the rate of progress would become amazingly fast to those of us in the "real world" (also, one of the subjects that uploads might spend years of subjective time researching would be attempting to enhance various mental abilities by tinkering with their own simulated brains, adding new simulated neurons to different regions and so forth).
The fact that uploads could make copies of themselves would also have a lot of interesting implications. If you have a few uploads that are good at a particular job and willing to do it for a minimal fee (just the price of enough computing power to provide comfortable simulated accommodations, say), and these helpful individuals are willing to make arbitrarily many copies of themselves, then anyone who needs this job done could request a fresh copy, and with a large community of uploads with many different skills it's not hard to see how this might lead to a post-scarcity society where most goods and services would become essentially free for everyone (self-copying uploads could also provide the "brains" for self-replicating mining and manufacturing facilities in space, which could hugely increase the resources available to produce goods while removing environmental problems associated with mining and manufacturing on Earth...the fact that uploads wouldn't need food, water or shelter might also reduce the environmental effects of civilization).
Jesse, you seem to be a singularity-booster. What's in it for you?
My interest is mainly in the effects this sort of scenario would have on the world, not on any pressing personal desire to leave behind my body or become immortal. A huge acceleration of the rate of progress on scientific and technical problems, the possible creation of an eco-friendly post-scarcity society, the idea that uploads might in short order be able to bootstrap their intelligence to beyond-human levels...all these possibilities are fascinating to me, and they all begin to seem fairly plausible (though by no means certain) if mind uploading is achievable.
Posted by: Jesse M. | Jun 11, 2008 5:51:44 PM
Nick Smyth wrote:
No, but I do have inductive evidence for it. I am certain that I am conscious, and I know that my consciousness is somehow produced by my brain. I know that other human beings have brains. An AI machine of the sort we're considering, on the other hand, has no brain, in fact, it doesn't even have a physical representation of it, but merely a virtual one.
And what is special about the category "human brains", as opposed to some broader category like "vertebrate brains" or "computing machines", or some more narrow category like "male brains" or "brains weighing over 1400 grams"? It seems rather arbitrary to accept the argument "I am conscious and I have a human brain, therefore I can be confident others with human brains are conscious too, but cannot be confident that cat brains or simulated brains are conscious", but to reject the argument "I am conscious and I have a brain weighing over 1400 grams, therefore I can be confident others with brains weighing over 1400 grams are conscious too, but cannot be confident that brains weighing under 1400 grams are conscious" or the argument "I am conscious and I have a computing machine which produces intelligent behavior (my brain), therefore I can be confident that others with computing machines that produce intelligent behavior are conscious too."
Also, did you read Chalmers' dancing qualia/fading qualia argument which I linked to? I think this is a pretty strong argument against the idea that there could be systems which were behaviorally indistinguishable from human brains but which did not have the same sort of inner experiences.
I would definitely dispute your claims about scientists and the supposed orthodoxy that all sciences reduce to physics. Any rational practicing scientist must realize that not only is physics seriously incomplete, but that it is leading us into realms where the notions of physical "matter" and standard causal laws are seeming positively quaint.
I am not sure what you mean by "physical matter" or "standard causal laws". All that's important is that the behavior of more complicated entitities can be reduced to combinations of more basic entities which interact according to well-defined mathematical laws; these entities and laws can be totally abstract, like quantum wavefunctions guided by the Schroedinger equation (no need for them to correspond to any commonsense idea of what 'physical matter' is supposed to look like), and the laws can also include a probabilistic element rather than being totally deterministic; but this basic outline is still reductionist, and it is still an "orthodoxy" which most scientists would take as a working assumption.
Furthermore, microphysics isn't the only area of difficulty: The ontological problems of supervenience, holism and emergence extend from quantum mechanics to condensed-matter physics and thermodynamics.
Quantum mechanics is reductionist in the sense I describe above, and it is certainly assumed by almost all physicists that if you had the complete initial conditions for the starting state of a system's wavefunction, then all the higher-level behavior seen in thermodynamics or condensed matter physics would naturally arise in a computer simulation which evolved the wavefunction according to the basic quantum rules.
I'd also note that reductionism is largely unpopular in modern philosophy of mind, even if scientists are "surprised" by this. As I'm sure you're aware, the problems of Qualia, Intentionality and Multiple Realizability form a powerful set of arguments against the thesis. So, I hope we can agree that Pinker isn't totally nuts to claim that processing power isn't going to solve this problem.
You are conflating physical reductionism with reductionism in a much broader philosophical sense here. For example, David Chalmers (who Pinker is apparently a fan of, and I am too) is definitely a physical reductionist in the sense of believing that all facts about physical events (like whether a particular neuron fires at a particular time) can be explained in terms of fundamental physical laws, but he also believes there are nonphysical facts about qualia which cannot be explained in this way. If we are just interested in the question of whether we will be able to create uploads which behave like humans, and on the implications this would have for our predictions about the future, then it is only physical reductionism which is relevant to this question.
Posted by: Jesse M. | Jun 11, 2008 6:20:50 PM
Looks like a bad italics tag in that second-to-last comment is putting everything afterwards in italics, my apologies...I wish comments were editable! I'll try putting some starting and ending italics tags here and see if it fixes the problem.
Posted by: Jesse M. | Jun 11, 2008 6:23:27 PM
The distinction I intend between simulation and replication is that the former only approximates its subject, while the latter at least purports to reproduce it in full.
You suggest it's a distinction without a difference if we can create a digital brain that is "behaviorally the same" as a biological human brain. That's a big if. The whole purpose of modeling is to be able to simplify that which we intend to simulate; otherwise there would be no point. Your argument presumes that we can succesfully isolate the important functions and morphology of the neural brain, and scrap the rest. I'd wish you luck, but your vision of an accelerated simulated world is so abhorrent to me that even my supreme lack of confidence in reductionism of mind does not afford me enough generosity to do so.
One thing I'm not sure you've considered in your post-scarcity paradise: the species of simulated human you are advocating will have very little interest in the "actual" world we know and love, except perhaps academically. We lowly flesh humanoids need water, food, sunlight, and some measure of beauty (even if that is only a loose signifier of the overall health of the biosphere). Uploads will need only a reliable power source, and will most likely have a much different sense of beauty than we do, and very likely one at odds with our own, if the needs of existing machines are any guide.
But mostly I think the AI dreams are just a massive waste of money and other resources to provide a hobby for a bunch of people who like computers more than they like people and other living things.
Posted by: Chris Schoen | Jun 11, 2008 6:41:11 PM
Chris Schoen wrote:
You suggest it's a distinction without a difference if we can create a digital brain that is "behaviorally the same" as a biological human brain. That's a big if. The whole purpose of modeling is to be able to simplify that which we intend to simulate; otherwise there would be no point. Your argument presumes that we can succesfully isolate the important functions and morphology of the neural brain, and scrap the rest.
Well, all outward behavior such as speech acts are determined by discrete electrochemical impulses traveling from the brain to motor neurons to muscles, so if you can accurately simulate the way each neuron's output of electrochemical impulses depends on whatever factors influence this (the primary influence probably being incoming electrochemical impulses from other neurons, along with some other things like blood chemistry), then you're simulating everything relevant to behavior.
One thing I'm not sure you've considered in your post-scarcity paradise: the species of simulated human you are advocating will have very little interest in the "actual" world we know and love, except perhaps academically. We lowly flesh humanoids need water, food, sunlight, and some measure of beauty (even if that is only a loose signifier of the overall health of the biosphere). Uploads will need only a reliable power source, and will most likely have a much different sense of beauty than we do, and very likely one at odds with our own, if the needs of existing machines are any guide.
I think your professed dislike of the future I'm imagining is distorting your thinking here, making you think of uploads as some alien "other" and avoiding putting yourself in the shoes of an upload and realistically imagining how they would think and feel. Remember, the definition of successful uploading would be a being psychologically indistinguishable from the flesh-and-blood brain the upload was based on, someone who would have a lifetime of memories of being a biological human and would still feel like the same person--if you woke up one day to find you had been uploaded, but mentally you felt like the same person and you felt all the same physical sensations and desires in your simulated body as before, do you really think the mere intellectual knowledge that your body was not "real" would cause you to lose all empathy for biological humans, become uninterested in any world outside a simulation, and experience a total change in your standards of beauty? This doesn't strike me as at all likely--the only way the changes you suggest would happen would be if mind uploading was wildly unsuccessful and produced beings who were still intelligent but whose minds bore very little resemblance to those of the brains they were based on.
But mostly I think the AI dreams are just a massive waste of money and other resources to provide a hobby for a bunch of people who like computers more than they like people and other living things.
That's kind of a cheap attack, and here you seem to be turning real present-day humans into caricatured "others" so you can dismiss their opinions--have you actually known any people who work on AI personally (like, not on the internet), and if so would you say they fit this mean-spirited image? I had a friend in college who was into AI, and she wasn't at all lacking in empathy or feeling for her fellow humans.
Posted by: Jesse M. | Jun 11, 2008 8:00:57 PM
Well, despite the fact that there is a vast gulf between artificial intelligences and "uploaded" intelligences...
Jesse. Presumably, you would prefer that you were uploaded while your mental faculties were at their peak. What would you do with your still healthy meat self? After all, he would still think of himself as you, unchanged, even if the new you did not. Would killing him be murder? Or could this turn into a "my body" privacy right a la Roe v Wade? What if "it" has second thoughts after having gone through the upload? Tough luck?
At what point are the rights of artifical humans going to supercede the rights of meat humans?
Posted by: Carlos | Jun 12, 2008 10:30:11 AM
McLuhan is right on this also--
It is the medium.
I'm with Pinker on this one.
Moore is already in the headlights, and about to be run down, as, along with capitalism and other systems with physical restraints, expansion into a finite systems eventually has consequences on unlimited growth.
Posted by: Dave Ranning | Jun 12, 2008 11:38:15 AM
Carlos, the most likely near-term method for mind uploading would be destructive--one would have to freeze a brain and slice it into ultrathin sections which can each be mapped individually and then put back together in a computer model. This is what they're attempting to do with the fruit fly brain project I mentioned earlier. But if a nondestructive method of scanning the brain at the synaptic level was found, I have a hard time believing anyone would seriously suggest the biological person should be murdered afterwards (I suppose he could commit suicide if he had a strong desire that there only be one copy of him in the world, and preferred that it be the upload, but I can't imagine many people would really feel this way). Likewise, I have a hard time believing the rights of uploads would ever "supercede" those of biological ones--why would they? If you were uploaded, and continued to have the exact same memories and personality, would you suddenly begin to look down on your non-uploaded friends and relatives, or think you were superior to them?
Posted by: http://www.jessemazer.com | Jun 12, 2008 11:38:16 AM
Without the physical body, a human being is not human. The body and all its cells are as important and as intelligent as the brain. Just how did the brain get so bossy?
Posted by: Jared | Jun 12, 2008 12:03:19 PM
Jesse,
It's not every day you get Carlos, Dave Ranning, Jared, Picador, Nick, and myself all on the same side of an argument. You deserve some praise for that.
The question is not so much what the first generation of "uploads" might think or feel, but their descendents, the ones who have no "real" memories of being biological and thereby of being connected to the biosphere as a whole. Why should they care about "life," except to the extend that it doesn't get in the way of their access to electricity?
As Jared wrote, you are exhibiting a remarkable faith in the ability of a human "mind" to exist outside a human body:
Whence this conviction that the self resides exclusively in the brain, rather than in the whole body? It all sounds suspiciously dualistic.
Posted by: Chris Schoen | Jun 12, 2008 12:24:16 PM
Chris,
I agree. The brain is a part of the body; the body is not merely an extension of the brain. Though the brain would often like to play abstract games with itself, too much of this is unhealthy and even unnatural. As I have mentioned before, I am very impressed by the intelligence of life at the cellular level. The body is wiser than the mind.
Posted by: Jared | Jun 12, 2008 12:43:06 PM
Chris Schoen:
The question is not so much what the first generation of "uploads" might think or feel, but their descendents
How would they have descendants, exactly? Are you imagining simulated embryology as well as simulated copies of mature biological brains? If so, you'd need to be simulating entire bodies at the cellular level, which means these simulated human bodies would have all the same sensations and desires as regular biological humans. Unless you're imagining we can create bodies and minds which are functional but which are not created by slavishly copying real human biology--but this would have nothing to do with "uploading", it would mean the creation of alien A.I. intelligences de novo. Personally I doubt we're smart enough to create entirely novel beings with the same complexity as humans in this way, I think the only route to humanlike A.I. is likely to lie in precisely copying actual human biology in a simulation.
Whence this conviction that the self resides exclusively in the brain, rather than in the whole body? It all sounds suspiciously dualistic.
People can lose limbs and organs and it doesn't change their personality or intelligence, whereas relatively small amounts of brain damage can seriously change these things. If you want to define the "self" as something other than a human's personality, intelligence, thoughts and feelings, then I suppose you could say that a quadruple amputee has lost a significant part of their "self" even if these mental/behavioral qualities haven't changed a bit. But this looks like just a matter of definition to me--you're free to say an upload does not have the same "self" as the biological human it was based on, but all my arguments about the possible consequences for society of mind uploading (including the arguments about uploads continuing to share the same values as biological humans) were just based on them having the same mental characteristics.
Posted by: Jesse M. | Jun 12, 2008 1:31:22 PM
Well personally, were I to destroy my meat self to continue in some fashion as software, I would need to have the ante upped a bit.
I'd be faster, have much wider spectrum sensor arrays (naturally), with vastly richer access to the infostream, and I'd be able to communicate far more data intensively with other similarly "evolved" entities than would be possible with meat people. Perhaps I would still love poetry, but I might find trillion word poems vastly more delightful both to compose and to appreciate. Not for me the decontextualized smells of food and pheremone, but the formerly off limits delights of superheated chlorine gas or mid atlantic ridge vent emissions might more than make up for it.
So while my meat relatives might feel disinclined to invite me to Thanksgiving, or to set me up with a cousin's gorgeous roommate, I'd be busy anyway uploading myself directly up to GLAST III for a direct look at the spectacular Gamma Ray displays you squishy types can't even see without remapping everything to visible light which, as any purist would tell you, completely ruins the effect. I'd write you a million line poem in iambic googlameter to describe what we feel when we see such wonders, but you'd never read it.
We used to have so much in common. People change.
Posted by: Carlos | Jun 12, 2008 5:05:05 PM
Carlos, why do you think we'd have any easier of a time figuring out how to give you those abilities as an upload than we'd have giving your regular biological brain those abilities through some kind of brain/computer interface? One of the main points of uploading is that it doesn't require you to actually understand how the brain actually does its magic, just to slavishly copy it, whereas altering a brain to give it amazing new mental abilities like "vastly richer access to the infostream" or dashing off million line poems would require quite a lot of this understanding (and if we had that understanding, we could presumably give biological brains the same abilities with the right interface). As for seeing outside the visible spectrum, you can basically do that now by looking at cameras with a display that translates invisible frequencies into a false-color image...but I doubt you'd want to spend all your time looking at galaxies rather than "normal" human-scale things, so I imagine your upload wouldn't want to either unless the parts of his brain associated with motivation and desire had been messed around with a whole lot.
Now, I think it might be that uploads could eventually figure out how to alter their brains in ways that they found useful rather than just crazy-making, through a long process of trial-and-error (unlike with surgical alterations to a biological brain, trial-and-error experiments for uploads would be pretty safe since they could just erase any new simulated neurons they added to their brain if they weren't working out). And if years of subjective time for uploads translated to much less time in the outside world, this sort of thing might lead to really different kinds of intelligence emerging fairly quickly for the rest of us, which is one reason I think a 'singularity' scenario is at least plausible. But again, I think any successful techniques they discovered for changing their simulated human brains could be translated into techniques that could allow non-uploaded people to gain the same abilities if they wished. In either case I guess there's always the danger that modified minds would treat unaltered humans badly, but I don't see why uploads would be more likely to behave badly than people who still had biological bodies but had altered their mental abilities through a mind-machine interface. And the trend in evolution is for bigger-brained creatures to be more social, more capable of empathy, so I'm optimistic that if beyond-human intelligences ever arise they won't turn out to be raging sociopaths.
But I don't mean to come off like a total advocate for uploading or the singularity, it's true that there are a lot of ways these things could potentially lead to disaster. The original point of this discussion was not whether or not these things would be good but whether or not they are plausible as scenarios for the future. Those who find mind uploading an unpleasant possibility might also find it unpleasant to think that over millions of years of natural evolution humans will likely evolve into quite different animals (possibly with greater mental abilities of their own), but I don't think they'd deny that this is a likely future scenario if humans don't go extinct before then.
Posted by: Jesse M. | Jun 12, 2008 6:05:10 PM
And what is special about the category "human brains", as opposed to some broader category like "vertebrate brains" or "computing machines",
Nothing in my argument relies on the premise that human brains are special. "Brain" is pretty close to a biological natural-kind category, as close as you're going to get in biology. It picks out a certain kind of entity that is as rigidly definable as anything in macro-biology can be.
The equivocation of "cat brain" and "simulated brain" in your argument is just not valid at all. There is a fundamental, radical difference between a brain and a simulation of a brain, and it is this difference that provides us with strong inductive evidence that animals and other human beings are conscious (evidence that we necessarily will not have for a simulation).
Chalmers' article is very good, and does indeed provide a strong argument for accepting that qualia will exist in a functional duplicate, on the premise that functionalism is true. But the possibility I'm considering here (and in good faith I must admit that I have no settled position whatsoever) is that the actual physical composition of brains is necessary for consciousness, something that functionalism denies right off the bat. I don't think this possibility is refuted by an argument that assumes it is false before the refutation begins.
In any case, I don't really doubt your claim that the behaviour of an organism can be simulated... it seems clear that this will obviously be done some day. Thanks for clarifying reductionism and its relation to this issue.
Posted by: Nick Smyth | Jun 12, 2008 6:11:47 PM
Jesse,
The Matrix Trilogy is far from perfect, but one thing it gets right is the inherent conflict of interest between digital minds (should they ever arise) and biological ones.
We humans are only now learning that we can only satisfy many of our appetites at the expense of a degraded, if not decimated, biosphere. Like the intelligent species we've always proclaimed ourselves to be, we very sensibly coming to the realization that we need to live harmoniously with other life on the planet, not just for our survival but for our quality of life, too.
We cannot, however, assume that "uploads" would share these interests with us in the long run. (My references to "descendents" notwithstanding, it doesn't matter whether uploads can reproduce or just live indefinitely; either way, memories of carnal life will fade). As Carlos wrote: people change. Especially when you disencarnate them.
The sad irony will be that this conflict, if it arises, will have been entirely of our own manufacture. We have an impossibly lush and rich relationship with the world right now, as mortal primates. Devoting the enormous resources to try to simulate this in ways that can never fully succeed seems to me as criminally ridiculous as Star Wars missile defense, or war in Iraq. Do you really suppose that digital "people" who have no need and no aptitude for food or sex will be just the same as they were when they had biological bodies? You write that Carlos's upload would be interested in the same things he is now "unless the parts of his brain associated with motivation and desire had been messed around with a whole lot." Wouldn't that be a cruel fate? To never see art or hear music in the same way you once did? To never touch loved ones in the same way? To never again eat, in the way we understand it?
I realize that in theory simulations of these experiences could be built in, but as in the Matrix, what would be the point? Unless one was to be deluded that one was still a meat human. But then, why not just be a meat human?
You seem to want to have it both ways, on the one hand saying that uploads would have all sorts of advantages we don't currently enjoy, and on the other saying they'd be just the same people as we are now. Those are not compatible propositions. If there are advantages to being an upload, they arise from the upload's being different from a biological human. I don't see how you can account for any kind of firewall around the upload's ethical precepts. Why shouldn't he/she/it be able to reason an ethical stance from it's own contingencies? Why be shackled to the nostagia of a biological past? Why, in other words, should we assume a culture of uploads would maintain common cause with lifeforms they have little need for?
Posted by: Chris Schoen | Jun 13, 2008 2:02:13 PM
Chris Schoen wrote:
The Matrix Trilogy is far from perfect, but one thing it gets right is the inherent conflict of interest between digital minds (should they ever arise) and biological ones.
Do you think the actions of Neo and his fellow humans would be less plausible if they discovered they were actually uploads living in a simulated world rather than real biological brains whose only experience (until they were unplugged) was of a simulated world? The only difference here would be the intellectual knowledge that their brains were digital as opposed to biological, in both cases their entire lives have been spent interacting with a simulated world that feels totally physical to them. Do you think this intellectual knowledge would radically change their priorities and motives?
We humans are only now learning that we can only satisfy many of our appetites at the expense of a degraded, if not decimated, biosphere. Like the intelligent species we've always proclaimed ourselves to be, we very sensibly coming to the realization that we need to live harmoniously with other life on the planet, not just for our survival but for our quality of life, too.
We cannot, however, assume that "uploads" would share these interests with us in the long run.
Why not? Do you think biological humans who spend their lives in artificial environments, like city-dwellers or hypothetical future space colonists, are likely to be uninterested as a group in preserving the biosphere? If not, why would uploads be any different?
And on the subject of appreciation for nature, I imagine that if we were capable of simulating brains at a cellular level, it would just be a matter of additional research to simulate embryology, the development of multicellular organisms from the information contained in a single fertilized egg...if so, one could then begin to upload all manner of organisms, not just humans, and make the simulated environments for uploads less sterile. Just as spending time in parks can help city-dwellers appreciate nature, so this sort of thing could help maintain the uploads' appreciation of the need to preserve natural environments outside their simulated worlds (and again, uploads could spend time in the real natural world just like we do, by connecting their simulated brains to physical bodies).
(My references to "descendents" notwithstanding, it doesn't matter whether uploads can reproduce or just live indefinitely; either way, memories of carnal life will fade).
Do you imagine that uploads would discard the experience of virtual or robotic bodies, and choose to exist as disembodied minds without any bodily sensations? This is a little like imagining a cult of flesh-and-blood humans who would voluntarily spend their lives in isolation tanks, I can't imagine such an option ever holding much appeal. And if you agree that uploads would continue to at least have virtual bodies which would feel just like real ones (much like the virtual bodies that people had in the Matrix), then again, I can't see how the mere intellectual knowledge that these carnal experiences are simulated rather than real is going to radically change anyone's perspective, when from a gut-level perspective they would feel just like our own experiences. We know intellectually that many of our own intuitions about solid objects are a type of "illusion", that all objects are mostly empty space and the particles making them up are guided by highly counterintuitive and abstract mathematical rules; does this intellectual knowledge cause anyone to devalue physical pleasures like touching and eating and sex? If not, why would the mere intellectual knowledge that the seemingly solid objects around you were "really" just bits in a computer cause any greater shift in priorities and values?
As Carlos wrote: people change. Especially when you disencarnate them.
But uploads would only be disencarnated in the abstract intellectual sense that they know their bodies are not "real", not in the sense of immediate felt experience from moment to moment. Again, if you suddenly found out that your own life up until now had been a Matrix-like simulation, would this cause you to think it was unimportant to preserve whatever real natural world existed outside the simulation? Please think about this and tell me how you think you would really feel in such a scenario.
Do you really suppose that digital "people" who have no need and no aptitude for food or sex will be just the same as they were when they had biological bodies?
Intellectually they might know that they no longer "really" needed these things, but their brains would work exactly the same, so of course they would have exactly the same psychological desires for them as before. You really don't seem to have thought this out very well, if there could be some scenario where a society of fully human minds would plausibly decide to give up such experiences, you need to lay out the details instead of just vaguely imagining that because they are "really" computer programs naturally all these desires will fade away (which is a bit like arguing that once we realize we are 'really' just clouds of quantum particles our desires for food and sex should naturally fade...it seems like a gigantic category error).
You write that Carlos's upload would be interested in the same things he is now "unless the parts of his brain associated with motivation and desire had been messed around with a whole lot." Wouldn't that be a cruel fate? To never see art or hear music in the same way you once did? To never touch loved ones in the same way? To never again eat, in the way we understand it?
I don't consider it a "cruel fate" to know that the real physical truth about the nature of solid objects and sound waves and so forth bears almost no resemblance to my experience of them through eating and seeing and hearing, so I don't think it'd be a cruel fate to know that these objects and sights and sounds were ultimately created by bits in a computer either. It's the experiences themselves that are important to me, not dry intellectual knowledge about the ultimate nature of what gives rise to them. And meaning in life always has to do with the effects one's actions have on other minds--if I experienced my own consciousness as an upload (i.e. I knew firsthand that uploads were not philosophical 'zombies' lacking qualia), then I would trust that other uploads were real minds too, so my actions would still have the same meaning even if I considered the physical elements of my environment to be less "real".
I realize that in theory simulations of these experiences could be built in, but as in the Matrix, what would be the point?
"Could be"? How could an upload possibly function without the sensory experience of a body? How long do you think a normal biological brain could get along with no sensory inputs before going crazy or comatose? Again, much of your argument here sounds like a poorly thought-out category error.
Unless one was to be deluded that one was still a meat human. But then, why not just be a meat human?
Since you know intellectually your desire for sex is "really" based on the evolutionary need for organisms to pass on their genes, would you consider it "deluded" for an infertile individual to continue to desire sex?
You seem to want to have it both ways, on the one hand saying that uploads would have all sorts of advantages we don't currently enjoy, and on the other saying they'd be just the same people as we are now. Those are not compatible propositions.
Your formulation is overly vague. What I am saying is that they'd be psychologically the same, but that they'd have various sorts of advantages which have nothing to do with any change in psychology, like being able to think much faster than us (which would of course not be experienced as thinking any faster from their perspective, but just as seeing the external world moving in slow motion), or being able to make duplicates of themselves (of course each duplicate would feel like the same person as before the duplication, having all the same memories up until the moment of the split).
If there are advantages to being an upload, they arise from the upload's being different from a biological human.
But not psychologically different.
I don't see how you can account for any kind of firewall around the upload's ethical precepts. Why shouldn't he/she/it be able to reason an ethical stance from it's own contingencies?
I suppose some might reason new intellectual stances based on mere intellectual knowledge that they were a different form of life (even if they felt exactly the same), just as some real humans might reason new intellectual stances based on the intellectual knowledge that they are "really" just animals or just swirls of interacting quantum particles. I don't see any reason to see why this sort of thing would be more likely for uploads than for us though.
Why, in other words, should we assume a culture of uploads would maintain common cause with lifeforms they have little need for?
Do you think your appreciation for the biosphere is primarily based on the intellectual knowledge that you "need" it? If something like nanotechnology allowed you to get all the needed food, air, water etc. from inorganic raw materials, would you say to hell with nature? If we were able to build self-sustaining space colonies which didn't need fresh supplies from Earth, do you think the inhabitants would be likely to collectively decide their biological heritage on Earth wasn't worth preserving? Do we preserve elements of our historical heritage, like the pyramids or the Mona Lisa, because we have any non-psychological "need" for them? These aren't rhetorical questions, I really would like to hear your answers.
Posted by: Jesse M. | Jun 13, 2008 3:31:32 PM
I really don't think your insistence that there would be no psychological difference between meatmen and metalmen is supportable. If you want to mandate that all metalmen must have psuedo-endochrine systems to make them feel fear, love, lust, belonging, protectiveness, sorrow, joy, etc., it would not be long before they hacked those limitations away. Pure thought, perfect simulated manifestations of unfading memory would make for a different way of looking at the world. Consider how you look fondly back at your already dim and fading memory of a wonderful experience. You long for as good a recollection as possible, but your system will not support it. Theirs would. You can't mandate strength of character, or the willingness to risk unending mediocre experiences when endless variations of perfect memories are so near to hand. How many uploads would just close themselves in eternal reverie (appearing to the rest of us as if they were in a coma)? And what would be the psychological effect of that?
From a metaphysical standpoint, these people would be able to design their experience of life in ways that would bear no necessary resemblance to the existence experienced by anyone else. Imaginary friends, imaginary trips up the Nile with an amiably chatty Roger Bacon, imaginary roast Phoenix (they cook themselves!), saving the imaginary world from an imaginary comet by discovering a new imaginary particle beam weapon, and then saving an appropriately grateful 20ish Sophia Loren from an imaginary dragon in time to catch the opening curtain of your new opera? Would that life be less worth living than my nasty brutish and short one? Would it be more worth living?
Posted by: Carlos | Jun 13, 2008 5:17:42 PM
Carlos wrote:
I really don't think your insistence that there would be no psychological difference between meatmen and metalmen is supportable. If you want to mandate that all metalmen must have psuedo-endochrine systems to make them feel fear, love, lust, belonging, protectiveness, sorrow, joy, etc., it would not be long before they hacked those limitations away.
What makes you so sure that there would be any feasible way to mess with the emotional and motivational aspects of the brain this way and produce something that was non-crazy and non-comatose? I think it's unlikely that rationality in humans is something that can be separated from emotion and bodily feelings--read Damasio's book Descartes' Error to see a good case that they are inseparable. And even if we do ever gain sufficient understanding of brain function to radically alter it in the way you suggest and produce a non-crazy being, there is no special danger from uploads in this regard, regular flesh-and-blood humans might equally well modify their brains in such ways.
Consider how you look fondly back at your already dim and fading memory of a wonderful experience. You long for as good a recollection as possible, but your system will not support it. Theirs would.
How? I suppose an upload could preserve a copy of their brainstate at an earlier time, but I don't see how this would help them with recollection unless they could somehow merge their later brain with the copy from an earlier time, no easy task. Also, there are a few people in the real world who seem to have near-perfect recall of every moment in their lives due to an unusual neurological condition called Hyperthymesia, but they don't spend all their time reliving pleasant past experiences, and in the cases I've read about they actually consider it something of a curse (see here for example).
How many uploads would just close themselves in eternal reverie (appearing to the rest of us as if they were in a coma)? And what would be the psychological effect of that?
Well, those that did would at least present no danger to the rest of us...this would be sad, but not that different from cases like people who spend all their time in an opiate-induced haze. And this is another case where the danger isn't particular to uploads...what if normal humans choose to get brain implants which constantly stimulate areas of their brain that give them pleasure, for example? If the technology exists, is there any reason biological humans would be less likely to do this than uploads?
From a metaphysical standpoint, these people would be able to design their experience of life in ways that would bear no necessary resemblance to the existence experienced by anyone else. Imaginary friends, imaginary trips up the Nile with an amiably chatty Roger Bacon, imaginary roast Phoenix (they cook themselves!), saving the imaginary world from an imaginary comet by discovering a new imaginary particle beam weapon, and then saving an appropriately grateful 20ish Sophia Loren from an imaginary dragon in time to catch the opening curtain of your new opera?
I don't believe a fantasy world could have really convincing characters that passed the Turing test (like your chatty Roger Bacon) unless the characters were themselves independently intelligent uploads or A.I.s, in which case they would have goals and desires of their own beyond the person's control. I suppose generous uploaded actors might volunteer to make duplicates of themselves to play roles in the fantasy worlds of anyone who wanted them. Still, as I said to Chris Schoen above, I think almost all humans find meaning in their lives through relationships to other individuals or to society...I think only a pathological few would want to spend their lives entirely in the company of actors who were just playing a role to entertain them, or in the company of simpler videogame characters who spout canned dialogue but don't have any real intelligence of their own. I think most of us would want actual friends, actual romance, actual family relationships, etc., not just a permanent charade.
Posted by: Jesse M. | Jun 13, 2008 6:51:05 PM
My reference to the Matrix was meant to discuss the role of the computer "agents" (Smith etc.) and their antagonistic attitude toward human life. As Carlos notes, once artifical intelligence is free to pursue pure thought, the "real" world just can't compete.
But to answer your question: what makes the Matrix an interesting drama is that Neo, Morpheus, Trinity et al actually have "real" incarnated selves outside the matrix. If they "discovered" they were really uploads, the number of meaningful dramatic choices available to them would be quite small.
Because city dwellers are still biological humans. The biosphere is our natural habitat; we emerged from it and are supported by it throughout our lives, even if we live in high rises. I've been arguing that our aesthetic values are strongly linked to our biological nature. We like to surround ourselves with environments that suit us--temperate climates, fragrant smells, sociability and human touch, clean air, water and food, etc. These would be sheerly academic to a simulated mind. If we're going to give these uploads bodies, which persumably they can modify themselves, then we cannot assume these bodies will mimic our own. At any rate, they won't need to. We are wedded to our bodies, but the uploads will not be, by any force more powerful than whim.
As for simulated environments, I think this is where your argument really jumps the rails. It's already a stretch to say that we're going to someday be able to describe the functional brain in enough detail as to model it digitally. But to imagine that we can do the same with anything we might care to examine or experience, from numerous attitudes and perspectives? The sun would burn out before we even got started.
I'll have to respond to the rest of your commment this weekend. Must catch a train.
Posted by: Chris Schoen | Jun 13, 2008 6:57:52 PM
Chris Schoen wrote:
My reference to the Matrix was meant to discuss the role of the computer "agents" (Smith etc.) and their antagonistic attitude toward human life. As Carlos notes, once artifical intelligence is free to pursue pure thought, the "real" world just can't compete.
And as I said in my latest response to Carlos, I think the sci-fi notion of beings of "pure thought" not relying on emotion and intuitions drawn from embodied experience is highly implausible--again, I recommend Damasio's Descartes' Error for more on this subject, perhaps along with the writings of people like Francisco Varela on enactivism and embodied cognition in cognitive science and AI research.
But to answer your question: what makes the Matrix an interesting drama is that Neo, Morpheus, Trinity et al actually have "real" incarnated selves outside the matrix. If they "discovered" they were really uploads, the number of meaningful dramatic choices available to them would be quite small.
Why? An upload would be perfectly able to interact with the physical world just like we do--by connecting their brains to some kind of body (either android or bioengineered) with its own sensory and motor pathways.
Because city dwellers are still biological humans. The biosphere is our natural habitat; we emerged from it and are supported by it throughout our lives, even if we live in high rises. I've been arguing that our aesthetic values are strongly linked to our biological nature. We like to surround ourselves with environments that suit us--temperate climates, fragrant smells, sociability and human touch, clean air, water and food, etc. These would be sheerly academic to a simulated mind.
You keep saying things like this, but you never acknowledge that although an upload would not actually have a biological need for things like food or water, they would still have precisely the same desires for them, since the idea of uploading is to create a simulated brain which is a precise copy of a real human brain. By assumption, if you took a group of uploads and placed them in a simulated world, and then took a group of biological humans and hooked up their sensory and motor pathways to the same simulated world "Matrix" style, and you didn't tell any of these individuals whether they had biological bodies outside the simulation or whether they were uploads, then there would be no collective differences in the behavior, feelings, or desires of the two groups.
Are you unwilling to grant this premise? If so, then you are basically rejecting the whole idea of "mind uploading", whose point is to produce simulated brains which function precisely like the biological brains they were made from, and if you reject this premise then any further discussion of how "uploads" would behave is pointless since you aren't willing to grant their existence even for the sake of argument. And if you are willing to grant the premise that uploads would behave exactly the same as biological brains fed inputs from simulated bodies in a simulated Matrix-style world, then your argument would seem to be that the mere intellectual knowledge that they were uploads would cause them to behave radically differently and suppress all these "natural" desires ingrained in them through years of life experience (and millions of years of evolution shaping their brains), which also implies that if you lied to the real biological humans and convinced them that they were uploads, they too would suddenly suppress all their sensual desires and try to become beings of "pure thought" or something. This doesn't seem remotely consistent with my sense of human psychology.
If we're going to give these uploads bodies, which persumably they can modify themselves, then we cannot assume these bodies will mimic our own. At any rate, they won't need to. We are wedded to our bodies, but the uploads will not be, by any force more powerful than whim.
But our brains, through both experience and evolutionary hardwiring, are only able to make sense of inputs from, and control the muscles of, our own human bodies. Learning to use a different kind of body or different kinds of sensory organs would require changes in the brain...given neural plasticity it might be that through a long training process an upload could learn to make sense of a different type of body (much like the monkey here who learned to roughly control a robotic arm with its brain), but I'd imagine there'd be limits to how many bodily changes you could make and still have an unmodified brain adapt to it. Perhaps through a long process of trial and error uploads could learn to modify their brains so that they could adapt to more radically different bodies, but I'm convinced all intelligence would have to be embodied in one way or another, and I don't really see why having different sorts of bodies available would make a being less likely to care about things like the natural environment, any more than we would be uninterested in the natural environment of an alien planet just because our bodies didn't evolve to suit that planet.
As for simulated environments, I think this is where your argument really jumps the rails. It's already a stretch to say that we're going to someday be able to describe the functional brain in enough detail as to model it digitally. But to imagine that we can do the same with anything we might care to examine or experience, from numerous attitudes and perspectives? The sun would burn out before we even got started.
No one said anything about exactly replicating every aspect of real-world environments, just replicating enough so that uploads can be reasonably comfortable. With enough computing power to simulate entire brains at the synaptic level, I don't think it'd be hard to simulate all sorts of physical objects with different textures and consistencies (as well as simulating things like the sound waves that arbitrary collisions or vibrations of simulated objects would produce), and naturally for any given simulated object, this sort of physical modeling program could show the object from arbitrary "attitudes and perspectives". The first uploads might find themselves in sparse environments, but over time individuals familiar with the language of the modeling program (both uploads and non-uploads) could constantly add to the library of available stuff.
Posted by: Jesse M. | Jun 13, 2008 8:12:03 PM
(Where but here...?)
Thanks for a fascinating conversation to follow, Jesse M., Chris and Carlos. And Nick Smyth, who seems to have dropped out. As someone who tends already to ponder whether there is world outside mind, I truly appreciate it.
I would just love to be uploaded, mainly because of "all those moments," as someone famously said. But I deeply suspect that all those moments have supreme value to us because they arise mysteriously and are gone forever, memory being kind of a card trick. To be a really good upload that would feel itself intensely human, there would have to be an allowance made for how memory, over time, transforms our moments.
Rene Magritte told a story of a terrible fire that consumed his childhood home when he was well under 10, killing -- I think -- his mother. More vividly than any other thing on that awful night, Magritte remembered a police officer, who wrapped him in the distinctive white kepi the Belgian police wore in those years, and spirited him away from the conflagration. This rescue gave him an enduring sense, he said, of being "the pathetic center of a tragedy." (I hope I'm not misquoting -- the line means a lot to me.) This precise sense of himself was one of the wellsprings of his art, and so, when he learned as an adult that there had been no police officer present that night, that he had found safety in the garden, he felt quite adrift. He could see and smell and feel the white cape -- it was bedrock. And yet.
An upload worth its salt would encompass the processes in this story, would have an inner reach, a psychic reality which new information would from time to time necessitate its questioning, with these consequences: it would wonder about itself, experiencing deep doubt; it could be shaken to its very foundations; it would know the dread, and the certainty, of having staked much on the counter-factual, and would see what a wild thing it was therefore; and, it would painfully perceive the moral dimension of all this.
It's easy to envision an upload as self-editing, indeed self-correcting in many ways. It's not easy to understand how one could correct for all the incorrect things that happen in the mind, and simulate the ingeniously flawed lens through which the mind gathers information about itself over time. That's an awful lot of misfiring for a upload to take on. Maybe you have to be an organism after all, if one of the central dilemmas of self-hood can't make it to the upload.
To return to _Blade Runner_, wasn't Rachael's best moment, her most human, when she realized she must be a Droid, that her memories were but a program? She hadn't thought of that before, having had no previous occasion for ontological insecurity.
Posted by: Elatia Harris | Jun 13, 2008 11:41:53 PM
Interesting, Elatia, as always. But could a mere copy of a mind, which formerly struggled to separate dream and memory even as both faded over time, do this? Now held in a matrix that does not have the "shortcoming," how would it manage the process of deciding how to degrade or muddle its information capture and data storage?
Would you choose a container that deliberately leaked? Where wires "randomly" shorted out by design? Run everything through complex algorithms that introduced rounding errors in times of stress? Or would you forgo the things that make us human that are limitations of the flesh in order to "improve" on last years model?
Perhaps the perfect upload process Jesse envisions brings with it the subconscious engine that handles all this extraction and mapping of associations from the kalidoscopic gestalt of memory. Perhaps also, the obvious differences in people's memories have something to do with what condition the containing structures are kept in. Nutrition and patterns of use seem to play a role in maintaining mental abilities and memory retention. Presumably, metalmen would have optimizing subroutines to keep things properly configured so things we see as symptomatic of senility would not occur. It's possible, though, that those kinds of degradations are always happening and are a normal part of how we sort and store what our subroutines determine we are most going to need – and what we are going to have to thin out to make room. Would an upload have the exact upper limit of storage capacity as a meathead? If not the thinning process needn't take place but then the associative process reasoning employs wouldn't have just what's available to draw from but the whole dataset (properly prioritized, of course). It would be ironic if that process ended up being slower than the snap judgments we are capable of making because we can't pull everything out on demand.
That's another thing. No neuroscientist I, but aren't things like Schitzophrenia currently thought to be at least partially due to a glitch in the information processing structures in the brain? Presumably, an upload would find itself cured any genetic flaws contributing to the problem, which might be a good thing. But how much of creativity and the evolution of new memes is driven by mutation of the known by faulty meat structures corrupting the data?
Posted by: Carlos | Jun 14, 2008 10:42:49 AM
Too much perfection of the mind can be a painful "imperfection" in itself. But in this case, pain is irrelevant, I suppose. What I do not understand is the utility or the desirability of preserving the essence of our minds - something like the grin of the Cheshire cat without the cat? Also, while it will certainly be more "utilitarian" to upload a mind that is at its keenest in terms of IQ, memory and analytical ability, it may be more "desirable" to preserve another which is at its "wisest" despite some rustiness in other parts of the machinery. Who decides?
Fascinating discussion. Besides everything else I learnt, Carlos' apt "meatman" versus "metalman" distinction best encapsulates the central point.
Posted by: Ruchira | Jun 14, 2008 12:07:07 PM
Elatia and Ruchira--now it's a party!
Elatia's discussion of Magritte and Ruchira's mention of the "essense" of mind both touch on something I was thinking last night on this.
If I understand Jesse (and Wikipedia) correctly, the possibility of uploading minds depends on the ability to essentially take a high-definition 3-D picture of the brain that captures all relevant neural activity. OK. So let's say for the sake of argument this technology becomes available. The first question is going to be, how do we stop the action long enough to map it? Neurons don't stand still, at least not while we're alive. Mapping a living brain is going to be a little like mapping the water molecules of a river, and I hope it doesn't sound too neo-Luddite of me to say this sounds near-impossible. (Jesse said something earlier about mapping the brain after death, at least when the technology is still young. But then all we have is the map of a dead brain. If we "are" our brain states, any mind generated from this map should also be dead after uploading, no?)
The next problem I foresee is that our brain states are extremely contingent. That is to say, we are always experiencing something different. I would imagine that my brain state at this moment intimately reflects the fact that I am indoors, sitting in front of a laptop, bare feet resting on the base of the chair, the hum of the cat water fountain in the background, and on and on. If we mapped my brain right now, and uploaded my mind, would I not be living this (admittedly pleasant) moment for all of eternity--or at least until the hard drive fails?
What if I had a hangover, or was grieving the loss of a family member, or had just fallen in love. Wouldn't these experiences be dramatically reflected in my brain state? So the question becomes, how do we get to the "me" that transcends all of these state-specific experiences? That's what we want to upload, not the phenomenal noise of an everyday moment. But where is that me? Where in the map? How do we separate out the neural data that relates to the context of the moment, to leave behind the essential me, that likes, say, Magritte paintings and Sean Young movies? Is this even a meaningful question?
I think it is not. I think that the mind that upload-enthusiasts have in mind is actually a fiction--a story we tell ourselves about ourselves so that our experiences have a certain unity. The seach for this mind will, I think, be about as fruitful as the search for the soul was a couple centuries back.
Carlos's point about the relationship between error and creativity is also a good one. In present day computing, errors have no creative power (this is a large part of Jaron Lanier's argument against AI: human-made software is brittle and inflexible. Humourless, we might even say. Or at least, far too literal). If we treat meatmen as robots ("human behavior is explicable as a function of neural activity") then our metalmen wil be robotic too.
It's a nice fantasy to think that a digital Shakespeare or Emily Dickinson, running on a server somehere in Paramus, NJ, would continue to write verse of the same breathtaking immediacy, but before we get too attached to it I think we need to remember that every map, every simulation, has by definition a quality of approximation; of incompleteness. Jesse wants to claim that this level of approximation would be either indetectable to us, or so slight as not to matter. Does anything about the history of science and technology support this idea? Anybody have an example of a simulation they'd gladly trade for the real thing? (Another one of Lanier's critiques of AI is that the Turing Machine can work two ways: by the computer getting "smarter," or by humanity getting stupider. All that matters is that we can't tell the difference.)
I continue to be unimpressed by the idea that replacing mother's milk with formula and calling it "the same" is an example of human progress. What great problem facing mankind are we solving here?
Posted by: Chris Schoen | Jun 14, 2008 1:38:12 PM
Jesse,
I hinted at a response to you on these points earlier, but I want to be explicit since we seem to be going back and forth on this.
I don't acknowledge it because I have no reason to believe it. Most of these desires arise, at least in part, in our bodies. Hunger and thirst, to be sure, and sex drive too. Do you suppose a being with no biological cells to hydrate is still going to get thirsty just because it has an idea in its mind that it used to like to drink water?
Hopefully by now it's clear that I am.
You are moving the goalposts here. If we are talking about simulated beings in a simulated world, then their subjective exprience might very well be just like ours. I think this is an impossible scenario in reality, but as a thought experiment, I agree that we would expect life in the matrix to be "just like" life in the natural world.
When we talk about uploads, though, we are presumably talking about simulated minds interacting with the natural world, somehow. And the resulting biological disconnection would be unavoidable.
Are you saying that if we all moved to Jupiter, we wouldn't want to do a little terraforming to make it more comfortable? Not that I'm advocating such a move, but I personally don't have any interest in an ammonia-based atmosphere. I'd get headaches.
But why would anyone want to live in the simulated world, when we can live in this one where everything we investigate more closely turns out actually presents to us a rich existence, with an infinite panoply of properties, instead of a dialog box that tells us the sleekness of a dog's fur or the feel of the teeth of a comb against our scalp have not been rendered in this release?
What's the upside? There are a lot of things we can devote our time, money and energy to in this world, and lots of serious challenges facing us. Why this?
Posted by: Chris Schoen | Jun 14, 2008 2:22:37 PM
Good points, Chris. That upload running on a server amusingly located in Paramus, NJ, would be, as we can now envision it, too much like a player piano, even if Rudolf Serkin were the template. Or like the sourdough loaf without leaven that failed, also, as matzoh. We might have some Uncanny Valley problems with it, too, problems that would make Eliza look cute and freckle-faced by contrast.
Or, is that not the upload "as we envision it" but simply as I do? I can only see what would be missing -- error in its vastest sense, error that makes strivers of us and endows us with insight. This is at the top of my list of qualities that would be missing, if only because we don't understand them well enough to translate them to a program. We aren't so intelligent that we can make an upload that accurately reflects the entirety of the systems we rely on -- a dilemma at the heart of the human condition if ever there were one. But maybe I just fail to fathom parallelism.
Decades ago ago, AI first came to my attention through the work of Danny Hillis. It struck me at the time that Hillis & Co were visionaries of representation, maybe not so different from the artists of earlier eras who dissected the human body to better represent it, and were conventionally thought to be doing a nefarious thing thereby, seeking to understand what was not their proper material and getting in the way of the eventual Resurrection of the Body. The heroic impulse -- to understand, the better to recreate, to summon something from nothing -- seemed to me then to be passing from the arts, relocating to visionary technology. Flash forward 20+ years, and I'm not so sure that's inaccurate. Certainly the most interesting minds of our time are not thinking about painting and how to do it.
The forbidden aspect of representation is well documented in its consequences for art. But what about for philosophy of mind? We know not to make a golem but we do it anyway, possibly because it is forbidden, and because we are curious: we must know, and do. The Muslim ceramic artists working in the 19th century on Lord Leighton's Arab Hall in London were asked to represent a hunt in a huge tile mural. That sat badly with them, but they did it. Except. Looking closely at the myriad hunters on the gleaming tiles, one sees slits painted across their jugular veins; they have been represented, as Lord Leighton desired, but they have not been represented as alive. Would the upload under discussion be similarly afflicted? Or does the question even obtain?
Posted by: Elatia Harris | Jun 14, 2008 3:36:45 PM
I don't know, Elatia. Sourdough matzo sounds good to me.
Posted by: Chris Schoen | Jun 14, 2008 10:02:52 PM
We got from The Singularity to wild yeast??? The conclusion is obvious -- our uploads couldn't have done it.
Posted by: Elatia Harris | Jun 14, 2008 10:27:52 PM
Chris Schoen wrote:
I don't acknowledge it because I have no reason to believe it. Most of these desires arise, at least in part, in our bodies. Hunger and thirst, to be sure, and sex drive too. Do you suppose a being with no biological cells to hydrate is still going to get thirsty just because it has an idea in its mind that it used to like to drink water?
The causes may arise in the body (although there are psychological causes as well, like people's desire to eat when hungry, and especially in the example of sex), but they are only experienced subjectively as desires because the body sends certain kinds of signals to the brain. If you had a simulated body which sent indistinguishable signals to the brain (whether the brain is a simulated one or a real one receiving counterfeit input as in the Matrix movies), do you disagree that the brain would have just the same sorts of subjective experiences and desires?
As I said, my argument is that uploads would find life pretty intolerable and perhaps even go insane without continuing to receive a stream of sensations similar to the ones they once received from their biological body. I suppose in the case of desires which mostly only arise periodically from very specific types of signals from the body to the brain, like hunger and thirst, some uploads might choose simulated bodies which never sent such signals (though I doubt most would, eating is a pleasurable experience for a typical human). Some might also choose to shut off their sex drive, though I don't see why this would be any more popular than it is for biological humans to voluntarily suppress their own sex drives through drugs or castration. But desires which have a large psychological component in the brain itself rather than mostly only being triggered when the body sends specific types of signals to the brain, like the desire for visual beauty or music, or interpersonal pleasures relating to touching and being touched, would almost certainly continue in an unaltered brain connected to a body that still provided the same basic sort of visual, audible, and tactile sensations.
Are you unwilling to grant this premise? If so, then you are basically rejecting the whole idea of "mind uploading"
Hopefully by now it's clear that I am.
Well, see my comments above. And perhaps it would help to ask you about another thought-experiment, the philosophers' beloved "brain in a vat" (i.e. a biological brain removed from its body but kept alive in isolation, with inputs being sent through its sensory pathways). Would you reject the premise that a biological brain in a vat which was being fed inputs from a simulation which were indistinguishable from those it had once been fed from its body would continue to think the same way and have the same desires? If you accept this premise, and if you grant the premise that an uploaded brain would behave just like a biological brain in response to the same inputs, then this naturally leads to the conclusion that an uploaded brain, too, would continue to think the same way and have the same desires if fed realistic inputs from a simulated body. If you are unwilling to accept this conclusion (even for the sake of argument), then it would help if you point out which step in this chain of reasoning is the first one you can't accept.
which also implies that if you lied to the real biological humans and convinced them that they were uploads, they too would suddenly suppress all their sensual desires and try to become beings of "pure thought" or something. This doesn't seem remotely consistent with my sense of human psychology.
You are moving the goalposts here. If we are talking about simulated beings in a simulated world, then their subjective exprience might very well be just like ours. I think this is an impossible scenario in reality, but as a thought experiment, I agree that we would expect life in the matrix to be "just like" life in the natural world.
When we talk about uploads, though, we are presumably talking about simulated minds interacting with the natural world, somehow. And the resulting biological disconnection would be unavoidable.
I don't understand at all why you say this...first of all, why must we necessarily assume that uploads wouldn't spend all their time in a simulated environment? They could still interact with the outside world in the same way I can interact with people in places I'm not physically present, by exchanging news and messages and videos and so forth. This could still be enough to bring about rapid singularity-like change in society, since uploads could spend years of subjective time working on problems in a simulated world while much less time would pass in the real world, and then they could communicate their results to us.
But as for interacting more directly with the real world, if we are able to create simulated bodies whose simulated sensory nerves send realistic inputs to the uploaded brains, then it seems to me it would be a fairly short step to create realistic humanlike robot bodies with visual/tactile/audio sensors which would translate the information from these sensors into a pattern of simulated nerve impulses to feed to the uploaded brain (or even bioengineered living bodies with the normal sensory systems but no brains of their own),.
Are you saying that if we all moved to Jupiter, we wouldn't want to do a little terraforming to make it more comfortable? Not that I'm advocating such a move, but I personally don't have any interest in an ammonia-based atmosphere. I'd get headaches.
Unlike us, an upload would have no "natural" external environmental, since they'd either carry their simulated world with them or create an artificial body tailored to their environment. Anyway, I think you're missing the point of the analogy, which was meant to address the question of whether uploads would be interested in the biosphere and think it worth preserving even if it wasn't their "natural home" in the same way it is ours. I think if we found thriving alien ecosystems in the atmosphere of Jupiter, those who care about preserving ecosystems on Earth would care about preserving these alien ecosystems too.
But why would anyone want to live in the simulated world, when we can live in this one where everything we investigate more closely turns out actually presents to us a rich existence, with an infinite panoply of properties, instead of a dialog box that tells us the sleekness of a dog's fur or the feel of the teeth of a comb against our scalp have not been rendered in this release?
If the simulation is modeling the detailed physical behavior of any object, then general-purpose rules can determine how it will interact with a simulated body and how its sensory nerves will react, you don't need to fill in the details of each sensation on a case-by-case basis, any more than you need to tell a 3D modeling program how a given object looks from every angle on a case-by-case basis. But as for your question about what the advantages of uploading would be, again, there are at least two potentially revolutionary things uploads could do that we couldn't--think at greatly faster speeds, and create as many duplicates of those with particular skills as needed. Tinkering with their own simulated brains in order to try to increase different mental abilities would also be much easier and risk-free for uploads (as I said before, they could easily erase any new neurons they had added as an experiment if they weren't working out), so this could be one path to superintelligence. On a more personal level, many might see the virtual immortality of life as an upload as being worth it. Even if not many people would intentionally discard a still-functioning body to upload themselves, if it were possible to upload a mind after the body had died I think plenty of people would put this in their wills.
As I said before, I am not really trying to be a pure advocate of uploading or say it would be completely beneficial with no downsides, but I think the technology is fairly plausible if you're a reductionist about brain function, and I think if it can be done it very likely will be done, so it's interesting to think about the sorts of effects it would have on society and the world.
Posted by: Jesse M. | Jun 14, 2008 11:48:15 PM
And to respond to Chris Schoen's earlier post:
Mapping a living brain is going to be a little like mapping the water molecules of a river, and I hope it doesn't sound too neo-Luddite of me to say this sounds near-impossible. (Jesse said something earlier about mapping the brain after death, at least when the technology is still young. But then all we have is the map of a dead brain. If we "are" our brain states, any mind generated from this map should also be dead after uploading, no?)
Generally in reductionism the idea is that you can boil down the behavior of some complex system into the interactions of a bunch of simpler units whose behavior can be predicted using general rules, given a snapshot of their state at any given moment and given the way other units interact with them. If you know the physical shape of a neuron, the density of different neurotransmitter molecules at its synapses, and other similar information about its structure at a given moment, the hope is that you could have a general model of neurons which you could feed this information into and it would predict pretty accurately how that neuron would respond to inputs from other neurons. I suppose it's possible that neurons are so idiosyncratic in their responses that this sort of thing could never be done regardless of how much research we do, but in that case you could still try to pick even smaller and simpler units--it's a pretty safe bet you could predict the behavior of individual molecules given a snapshot of their position and shape at a particular moment, for example.
The next problem I foresee is that our brain states are extremely contingent. That is to say, we are always experiencing something different. I would imagine that my brain state at this moment intimately reflects the fact that I am indoors, sitting in front of a laptop, bare feet resting on the base of the chair, the hum of the cat water fountain in the background, and on and on.
Your brain activity (i.e. the pattern of neural pulses in your brain) would reflect this, but the actual arrangement of synaptic connections between neurons isn't changing from moment to moment. Presumably the more stable long-term elements of your mind, like memories and personality and skills, are encoded somehow in more long-term structure like synaptic connections. Waking up as an upload would then be a bit like waking up from anaesthesia, with the transient brain activity that allows for moment-to-moment perceptions "rebooting" but still having continuity with your past self thanks to the persistance of this underlying structure.
Carlos's point about the relationship between error and creativity is also a good one. In present day computing, errors have no creative power (this is a large part of Jaron Lanier's argument against AI: human-made software is brittle and inflexible. Humourless, we might even say. Or at least, far too literal). If we treat meatmen as robots ("human behavior is explicable as a function of neural activity") then our metalmen wil be robotic too.
Again I think there is a category error here. Even if the computer doing the simulation is making no error in its simulation of each individual neurons, whatever high-level errors real brains make thanks to the arrangement of their real neurons, the simulated brain should make the same sort of errors since its simulated neurons are arranged the same and behave the same. Douglas Hofstadter has a good discussion of high-level vs. low-level errors on p. 575 of Godel Escher Bach (you can read this section online here), and even today we see that neural networks (where the nodes are very simple caricatures of how actual neurons behave) make plenty of errors. I suspect if you gained some familiarity with neural nets you'd be less likely to make these sorts of category errors involving ascribing characteristics of computers to the high-level behavior of uploads; if you're interested, a good book on the subject can be found here.
Posted by: Jesse M. | Jun 15, 2008 1:38:26 AM
Ah, Lord Leighton. Thanks for the smelling salts Elatia. I’m not sure about “most interesting minds” however. Aren’t we talking about different skill sets, different talents? The achievements of a Ferdinand Porsche or a Mario Andretti don’t invalidate or replace the marathon runner’s achievement in victory. Similar impulses of accurate representation might be at work with neuroscience and Western art, but very different terrains no? The critical spotlight has unfortunately waned for painters working within the older parameters, but a heroic impulse still seems to be there, perhaps more than ever given the hostility of academic theorists of art.
Sorry for the subject shift folks, the tangibility of art objects got to me.
Posted by: Jesse (not M) Fleshy | Jun 15, 2008 8:29:12 AM
Jesse Fleshy, if 3QD gets around 500,000 unique visits monthly, then I guess the odds favored two visitors inside 24 hours giving a fig about Victorian painting. But we've only been cued by thinking about the tangible, right?
Your point is taken, but I was actually more concerned with the impulse than with the terrain, as you beautifully put the distinction. That we don't live in a heroic age for painting does nothing to invalidate -- and as you say, possibly much to play up -- the heroic efforts of some painters working now. Real painting will always involve inquiry and risk, even at the purely private level. The drama of abstraction that drove 20th century art was one in which there were heroic actors, seeking, as Brancusi said in reference to his _Bird in Space_, the truth of a thing, not the truth about it.
The argument whether all art is representation, because we are the ones who are making the art and we cannot so easily escape ourselves, is not the one to pursue here, but it does speak to the human condition -- which, if we could indeed be uploads, would turn out to have been a highly romanticized notion. Whether uploads are, themselves, a romantic vision in that the very idea speaks to a longing for personal immortality, and a vision of the adaptation and survival of the human race into an unimaginably far future outside the dross and decay of flesh, outside a biosphere teeming with threats -- well, that's a question we should be asking, while there is still enough imagination left to improve life on Earth, not just improve on it.
Posted by: Elatia Harris | Jun 15, 2008 12:12:25 PM
Minor correction, in my second-to-last post I wrote:
The causes may arise in the body (although there are psychological causes as well, like people's desire to eat when hungry
That should have been "people's desire to eat when not hungry", since my point was about desires that aren't just passively triggered by signals from the body...I, for example, will eat chocolate just about any time the opportunity presents itself...
Posted by: Jesse M. | Jun 15, 2008 2:12:50 PM
Hello, good people (and how reasuring to know that you are all [I hope] still flesh and blood). I can't pretend to have read all this thread, but I'd like to say that I agree with Jesse that a simulated person inside a simulated world would end up being nothing more than a comatose cybernaut, and that if someone (or something) one day switches the power off on all those Raptured geeks in their perfect imitation universes, then no one will miss them. This reminds me of the story about the reporter who asked (a much younger) Woody Allen: "Would you like to keep on living forever through your work?"
Allen replied: "No, I'd rather go on living in my apartment."
Posted by: aguy109 | Jun 15, 2008 6:09:11 PM
aguy109:
but I'd like to say that I agree with Jesse that a simulated person inside a simulated world would end up being nothing more than a comatose cybernaut
What I said is that without a steady stream of inputs from a body (simulated or physical) that resembled the stream of inputs biological brains get from biological bodies, the simulated brains would probably go crazy or comatose. With those inputs, I think they'd behave in the same way normal humans do.
and that if someone (or something) one day switches the power off on all those Raptured geeks in their perfect imitation universes, then no one will miss them.
I gave a few arguments as to why uploads would be unlikely to stop interacting with the real world or spend all their time in simulated fantasy worlds. Here was one of them:
I don't believe a fantasy world could have really convincing characters that passed the Turing test (like your chatty Roger Bacon) unless the characters were themselves independently intelligent uploads or A.I.s, in which case they would have goals and desires of their own beyond the person's control. I suppose generous uploaded actors might volunteer to make duplicates of themselves to play roles in the fantasy worlds of anyone who wanted them. Still, as I said to Chris Schoen above, I think almost all humans find meaning in their lives through relationships to other individuals or to society...I think only a pathological few would want to spend their lives entirely in the company of actors who were just playing a role to entertain them, or in the company of simpler videogame characters who spout canned dialogue but don't have any real intelligence of their own. I think most of us would want actual friends, actual romance, actual family relationships, etc., not just a permanent charade.
And here was another:
must we necessarily assume that uploads wouldn't spend all their time in a simulated environment? They could still interact with the outside world in the same way I can interact with people in places I'm not physically present, by exchanging news and messages and videos and so forth. This could still be enough to bring about rapid singularity-like change in society, since uploads could spend years of subjective time working on problems in a simulated world while much less time would pass in the real world, and then they could communicate their results to us.
But as for interacting more directly with the real world, if we are able to create simulated bodies whose simulated sensory nerves send realistic inputs to the uploaded brains, then it seems to me it would be a fairly short step to create realistic humanlike robot bodies with visual/tactile/audio sensors which would translate the information from these sensors into a pattern of simulated nerve impulses to feed to the uploaded brain (or even bioengineered living bodies with the normal sensory systems but no brains of their own).
Posted by: Jesse M. | Jun 15, 2008 8:56:44 PM
Jesse M, you are totally tripping me out.
If if if, then yes. If the signals were "indistinguishable," then, yes, they'd be "the same," by the power of tautology alone. It's that first "if" I'm skeptical of. You are using technology as a sort of get out of jail free card here, but as I asked before, what simulation has our species developed up till now that would be preferable to (or at least indifferently interchangeable with) the real thing?
(Flight simulation, for example, is a great way to reduce the risk of accident for novice pilots. But at some point these pilots take off the training wheels and fly actual planes. The simulators aren't ends in themselves, and no one would want them to be).
Considering the possibility of mind uploading as a thought exercise has a pretty limited utility, it seems to me. All things being equal, if I could travel to distant galaxies in an instant, or remember my past lives like Tiresias, speak and understand any language I wanted, or travel back in time to talk to Rembrandt or Mencius, I might be tempted. But all things aren't equal. Even if any of these things were made possible through technological innovation (which I doubt), there would be a monumental cost to any of them. For me, at least, this cost would chill my enthusiasm right down to the point of indifference and beyond.
So, I prefer not to discuss this as though it were just a reverie, but as something taken seriously by high profile theorists such as Minsky, Kurzweil, and Chalmers. I fail to see any inherent problem posed by the limitations of the "speed" at which humans currently think. What matters isn't how fast we get answers to our questions, but that we get the right answers, and I don't see how Moore's law helps us with that at all. If uploading is possible, it will radically transform both our physical and cultural worlds in ways that will not be easy to predict or control. Meanwhile we'll be diverting trillions of dollars* and the research time of many thousands of intelligent and well trained minds away from other projects that might be innocuous at worst, and truly positive at best.
This type of reductionism is clearly false. There is nothing, famously, about oxygen or hydrogen atoms that is predictive of the properties of water. We can "reverse engineer" water into hydrogen and oxygen, but it is far from evident we could ever predict water, given only the behavior and properties of oxygen and hydrogen. Reductionism is, in this way, intractably a posteriori.
This, I think, is the main problem with functionalism. We define functional neurons in terms of some empirical neural "behavior." You earlier used speech as an example of a behavior that we can observe, and then trace back analogously to certain electro-physiological patterns in the brain. But without the insight of subjectivity (and, often, even with this insight), there is a vast wealth of neural "behavior" that we cannot capture just by seeing lips move and hearing phonemes uttered.
For example, we might overhear someone say "I don't care what we have for supper, honey; you decide." To say that this string of sounds is a single, isolatable behavior, is to overlook any number of things the speaker is thinking in the course of this statement: the speaker could have a number of opinions that might be at odds with the sentiment expressed; the speaker could be completely distracted and not thinking about dinner at all; or could be fantisizing a future conversation; or acting in a film or play.
Granted, the observer could probably make good guesses about these things, based on tone of voice, personal history, and other factors. But the fact remains that only a small fraction of true behavioral neural function is available to either subjective or objective analysis. We delude ourselves into thinking that our minds and our consciousness are the same thing, but it doesn't logically follow.
Think of all the things we are unaware of, and could never identify with conscious recall, but which we would instantly notice if they were changed. (The concept of health is a good example of this; we know it best in its absence.) There's a huge Unknown Unknowns problem here that casts serious doubts on your claims that we could replicate a digital sensorium that would be indistinguishable from the biological one.
This echoes sifta's comments above about boundary conditions. It is not really possible to say where the brain "ends" and the CNS begins. They are part of a single system.
This seems overly dualistic to me, suggesting active stimulae inducing responses from a passive neural structure. I don't want to get all blank slate on you, but it seems reasonable to consider that the synaptic connections between neurons are relatively plastic. It's not clear to me that we can be so easily separated from our moment to moment psychological existence.
That's probably enough for one day.
*If the Blue Brain supercomputer were online today it would use $3 billion in electricity every year. Future generations of computers will be more efficient, to be sure, but that's just one virtual mind. Our old fashioned meat-brains, meanwhile can get by on a few handfuls of grains a day.
Posted by: Chris Schoen | Jun 16, 2008 5:44:10 PM
I read Fantastic Voyage, The Age of Spiritual Machines and The Singularity is Near, and they changed my life. I even found some of his lectures on Itunes and I find myself impatiently awaiting his next book.
Recently read another incredible book that I can't recommend highly enough, especially to all of you who also love Ray Kurzweil's work. The book is ""My Stroke of Insight"" by Dr. Jill Bolte Taylor. I had heard Dr Taylor's talk on the TED dot com site and I have to say, it changed my world. It's spreading virally all over the internet and the book is now a NYTimes Bestseller, so I'm not the only one, but it is the most amazing talk, and the most impactful book I've read in years. (Dr T also was named to Time Magazine's 100 Most Influential People and Oprah had her on her Soul Series last month and I hear they're making a movie about her story so you may already have heard of her)
If you haven't heard Dr Taylor's TEDTalk, that's an absolute must. The book is more and deeper and better, but start with the video (it's 18 minutes). Basically, her story is that she was a 37 yr old Harvard brain scientist who had a massive stroke in the left hemisphere of her brain. Because of her knowledge of how the brain works, and thanks to her amazingly loving and kind mother, she eventually fully recovered (and that part of the book detailing how she did it is inspirational).
There's a lot of learning and magic in the book, but the reason I so highly recommend My Stroke of Insight to this discussion, is because we have powerfully intelligent left brains that are rational, logical, sequential and grounded in detail and time, and then we have our kinesthetic right brains, where we experience intuition and peace and euphoria. Now that Kurzweil has got us taking all those vitamins and living our best ""Fantastic Voyage"" , the absolute necessity is that we read My Stroke of Insight and learn from Dr Taylor how to achieve balance between our right and left brains. Enjoy!
Posted by: Patience | Jun 16, 2008 11:42:13 PM
If you had a simulated body which sent indistinguishable signals to the brain (whether the brain is a simulated one or a real one receiving counterfeit input as in the Matrix movies), do you disagree that the brain would have just the same sorts of subjective experiences and desires?
If if if, then yes. If the signals were "indistinguishable," then, yes, they'd be "the same," by the power of tautology alone. It's that first "if" I'm skeptical of.
C'mon, give me a little credit here, if I offer an argument that seems obviously tautological to you, consider the possibility that you may have misunderstood! Obviously I didn't mean to assume from the start that the signals would be subjectively indistinguishable from the signals from a real body from the perspective of the brain receiving them, since by definition that would mean the brain would "have just the same sorts of subjective experiences and desires". Rather, I meant that from a third-person physiological point of view they'd be just like the signals from a body--if, for example, you studied the way that light hitting the retina was translated into nerve impulses in the optic nerve in the greatest detail, then you'd see that the simulated light was translated into simulated optic nerve signals going into the brain in just the same way.
It isn't obvious that you'd say that even if the simulated input fed to a brain-in-a-vat was perfect from a physiological point of view, the brain would have the same experiences. You seemed to suggest earlier that you might disagree that the way the body creates sensations is just a matter of the discrete measurable signals (like nerve impulses or chemicals in the bloodstream) that it feeds to the brain; some of your comments suggested some sort of more holistic notion of consciousness originating throughout the body in some non-reductionist way, like your statement "As Jared wrote, you are exhibiting a remarkable faith in the ability of a human "mind" to exist outside a human body ... Whence this conviction that the self resides exclusively in the brain, rather than in the whole body? It all sounds suspiciously dualistic", which was in response to Jared's "Without the physical body, a human being is not human. The body and all its cells are as important and as intelligent as the brain. Just how did the brain get so bossy?" So it would help if you'd tell me whether you agree in principle that the brain-in-a-vat could in principle still have experiences just like those of an embodied brain, if the right pattern of nerve impulses and other discrete signals was fed into it (leaving aside the technical difficulty of simulating such a pattern).
You are using technology as a sort of get out of jail free card here, but as I asked before, what simulation has our species developed up till now that would be preferable to (or at least indifferently interchangeable with) the real thing?
(Flight simulation, for example, is a great way to reduce the risk of accident for novice pilots. But at some point these pilots take off the training wheels and fly actual planes. The simulators aren't ends in themselves, and no one would want them to be).
"Interchangeable with the real thing" is kind of vague...simulations are typically designed to mimic some specified aspects of a physical system, not every possible aspect (which would require simulating every last subatomic particle that makes it up). If you looked at data from a simulation of planetary orbits vs. data on the orbits of real planets, they'd probably be indistinguishable, for example. Perhaps you're talking primarily about simulations of human-scale things which are indistinguishable from reality by the human senses? If so, obviously we don't have the computing power to do physical simulations which render 3D objects at the smallest level of detail humans can perceive, with physical properties like flexibility factored in so they change shape realistically in response to applied forces, and with the simulation running in realtime. We also don't have any sort of interface that can hook into the human senses of touch and smell and taste, nor do we have 3D visual displays that can show the full range of color and brightness the eye can perceive. But if we had the computing power to simulate an entire human brain at the synaptic level (something that is vastly beyond the abilities of current supercomputers even if we had the ability to map brains and the knowledge of how individual neurons function), it seems pretty certain to me that this sort of detailed newtonian physics simulation of everyday objects, which doesn't need to be in nearly the same level of microscopic detail as the brain, would be relatively easy in comparison. That's not to say that an upload's simulated environment would be indistinguishable from the real world, because even if you can simulate all the sensations associated with touching and seeing any single object, creating detailed models of even a fraction of all the objects we can encounter in reality would be hugely time-consuming and might not be a priority for the earliest mind uploading projects. The first mind uploads might find that their simulated environment was limited to something like a very sterile-looking hotel room, but the experience of that room might at least be close to indistinguishable from staying in a real version of the same room. With more time and a larger community of uploads interested in living comfortably, the range of objects could continue to grow, and if biologists came to have a detailed understanding of the molecular signals that cells use to coordinate embryonic growth in multicellular organisms, the simulated worlds could eventually come to contain a large variety of cellular-level simulations of living, reproducing organisms other than humans, which would help in making things seem less sterile.
But let's assume that the first uploads do live in a pretty sterile, boring simulated environment with much less variety of objects lying around than in the real world (they'd still be able to look at video feeds from the real world of course). Is this much more terrible than the type of sacrifice that biological humans on long-term space missions would have to make? In your comments you often talk as though the idea of uploading is somehow to replace our boring ol' reality with a more exciting simulation, an idea which you then scoff at. But I have always emphasized that the appeal of uploading has nothing to do with wanting to escape the world of everyday experience, which I imagine early uploads would probably sorely miss; instead, it's about specific things one could do as an upload that one couldn't do in the real world, like think at a greatly accelerated rate, make duplicates of oneself, tinker with one's own neurology without fear of permanent damage, and avoid death. Surely there are a lot of people who would accept being uploaded either because they could do things of great benefit to humanity, or if the only alternative was being dead (as I said before, my prediction would be that if mind uploading technology was available, most people who did so would opt to have their minds uploaded after the point of natural physical death, at least at first when the community of uploads was small and simulated environments were lacking in much variety).
So, I prefer not to discuss this as though it were just a reverie, but as something taken seriously by high profile theorists such as Minsky, Kurzweil, and Chalmers. I fail to see any inherent problem posed by the limitations of the "speed" at which humans currently think.
It's not a matter of an inherent problem, but an inherent advantage in thinking faster. You don't think there could be great advantages to humanity if technical problems which might otherwise take a hundred years to solve could become available in a much shorter time? Of course there could be great dangers to such rapid change too, but I think it's unrealistic to imagine that if it was technologically feasible to create such accelerated minds, the groups responsible for funding research (governments, corporations) would not bother to give it a try. As I've said a few times, my purpose in these discussions is not mainly about advocacy, it's also to discuss the likelihood of mind uploading and the kinds of consequences that would flow from it, because I'm interested in scenarios for what the future may be like.
If uploading is possible, it will radically transform both our physical and cultural worlds in ways that will not be easy to predict or control. Meanwhile we'll be diverting trillions of dollars* and the research time of many thousands of intelligent and well trained minds away from other projects that might be innocuous at worst, and truly positive at best.
OK, so if you had absolute control of the funding of research, you might intentionally avoid funneling any money into mind uploading research. But considering things from the perspective of a futurist who is trying to predict what the world of 100 years from now will likely look like, not from the perspective of someone trying to imagine their own ideal future, don't you think it's fair to say that if mind uploading becomes technically feasible through increases in computing power and brain-mapping technology, then someone's probably going to do it?
Generally in reductionism the idea is that you can boil down the behavior of some complex system into the interactions of a bunch of simpler units whose behavior can be predicted using general rules, given a snapshot of their state at any given moment and given the way other units interact with them.
This type of reductionism is clearly false. There is nothing, famously, about oxygen or hydrogen atoms that is predictive of the properties of water. We can "reverse engineer" water into hydrogen and oxygen, but it is far from evident we could ever predict water, given only the behavior and properties of oxygen and hydrogen. Reductionism is, in this way, intractably a posteriori.
I think most physicists would say there is nothing fundamental that would prevent you from predicting the behavior of water molecules from basic quantum-mechanical laws, just that the computations get more complicated the more interacting particles you have (the properties of individual hydrogen atoms can be predicted accurately using quantum physics, but adding more protons and electrons as in the higher elements makes things more complicated). In fact, I found this article which seems to indicate that the best modern computers are up to the task of simulating water purely from quantum principles, saying:
Again, even if you are dubious about reductionism, I don't think you can plausibly deny that it's the mainstream point of view among scientists.
This, I think, is the main problem with functionalism. We define functional neurons in terms of some empirical neural "behavior." You earlier used speech as an example of a behavior that we can observe, and then trace back analogously to certain electro-physiological patterns in the brain. But without the insight of subjectivity (and, often, even with this insight), there is a vast wealth of neural "behavior" that we cannot capture just by seeing lips move and hearing phonemes uttered.
But I never said that observing speech would be enough to deduce the patterns in the brain, just that speech was caused by the output of pulses from the brain to the motor neurons that control muscles. The output of individual neurons would be expected to be a lot simpler to predict based on their electrochemical inputs, and experimenters can take also try to break down all the different changes in neurons (like changes in the density of neurotransmitters at the synapses) which influence this input/output relationship for an individual neuron. Again, that's the idea of reductionism; if some large complicated entity like the brain is hard to understand, break it down into simpler entities which behave in more predictable ways.
For example, we might overhear someone say "I don't care what we have for supper, honey; you decide." To say that this string of sounds is a single, isolatable behavior, is to overlook any number of things the speaker is thinking in the course of this statement: the speaker could have a number of opinions that might be at odds with the sentiment expressed; the speaker could be completely distracted and not thinking about dinner at all; or could be fantisizing a future conversation; or acting in a film or play.
Granted, the observer could probably make good guesses about these things, based on tone of voice, personal history, and other factors. But the fact remains that only a small fraction of true behavioral neural function is available to either subjective or objective analysis. We delude ourselves into thinking that our minds and our consciousness are the same thing, but it doesn't logically follow.
Again, I nowhere said or implied that you could deduce all the important details of what was going on in the brain just by looking at motor outputs like speech. The point of mind uploading is to take apart the brain and map its structure out in detail, not to figure out its structure from external clues like speech.
Your brain activity (i.e. the pattern of neural pulses in your brain) would reflect this, but the actual arrangement of synaptic connections between neurons isn't changing from moment to moment.
This seems overly dualistic to me, suggesting active stimulae inducing responses from a passive neural structure. I don't want to get all blank slate on you, but it seems reasonable to consider that the synaptic connections between neurons are relatively plastic. It's not clear to me that we can be so easily separated from our moment to moment psychological existence.
No, I'm not suggesting the neural structure is passive, in fact neuroscientists think learning happens based on long-term changes in the strength of connections between synapses (again, this conversation might be easier if you read up on the basics of how simulated neural networks work). But changes are happening on different time scales. Your immediate ability to respond to changing sensory information from second to second is thought to have to do with changes in which clusters of neurons are firing in synchrony, while long term learning over days or years is thought to be based on more long term changes in synaptic connections, and I think on some intermediate time scale changes in the levels of neurotransmitter molecules at each synapse would also influence cognition. The point is that even if the second-to-second pattern of neural firing events is disrupted as in anaesthesia, it can start up again and you'll still have all your long-term memories and personality since longer-term structures have not been disrupted. The simulated neurons of an upload would have to change connection strengths over the long term just like real neurons (these strengths are thought to change according to some sort of Hebbian rule), otherwise the uploads would have no ability to form new long-term memories.
Posted by: Jesse M. | Jun 16, 2008 11:51:11 PM
Jesse,
In principle, no. I don't think these signals would be separable from the means of their transmission.
I would agree. I think that upload theory takes as given that consciousness is the "specific aspect" of brain activity that could be simulated by encoding that activity into algorithms, but I'm not sure why that would be.
Here I have to return to Jaron Lanier's comment that the Turing Test can be considered two ways: either the machines are getting smarter or the humans are getting stupider. I certainly *hope* that hotel wouldn't be indistinguishable from an actual (sterile) hotel room experienced by a biological human, because this would imply that our perceptual and conceptual abilities had really atrophied.
We could spend the rest of this day brainstorming all the things are person in a hotel room might see, touch, smell, taste, hear, or otherwise be aware of, and we wouldn't even scratch the surface. And this doesn't even consider the physiological possibilities (what if we observed the room standing on our head--would we feel the blood rush to our heads, or have to make constant micro-corrections to our balance as our legs and feet slowly waved above us?
I think talk of persuasively simulating human experience enormously underestimates the wide range and deep complexity of our sensorium. Regardless of processing power, the properties of the simulated environment would have to be developed by the simulation programmer; a theoretically endless task.
I don't. What's the hurry? There is no prize for cleverness at the end of the line; just heat death. And processing power is not the same as wisdom. I think accelerated thinking poses a great advantage to the technological-industrial complex. To humanity, not so much.
Yes, it's likely, but that doesn't mean we should be fatalistic about it. If everything that's technically feasible is considered inevitable, then there isn't much argument left against, say, a cobalt bomb.
Yes, but we already know the principles of water (at a macro level) from personal experience. This is what I mean by a posterior explanations. Given water, and a chemical understanding of its molecular structure, we can see how it is derived from oxygen and hydrogen. But that is not the same as saying that, having never experienced water, we could predict it from the properties of elemental hydrogen and oxygen.
What I think you're still missing is that if you reduce the subjectively experienced activity that is associated with a speech act (complex thoughts and feelings) to the objectively observable data of mouth and jaw musculature and production of auditory phonemes, and then use only that data to correlate to specific neural patterns perceivable in the brain, then all you have demonstrated is a correlate between physical speech activity with certain "functional" neurons. You haven't demonstrated a correlate to thought or consciousness.
You write that "speech [is] caused by the output of pulses from the brain to the motor neurons that control muscles." That's misleading, in that the causal chain is highly indirect at best. Speech comes with a bundle of thoughts, feelings, motivations, and other "hidden" conceptual activity. If any given utterance is a result of specific neural activity, all the associated cognition must either (a) also be a result of that same activity, in which case our simulation is missing a lot of important data, or (b) be caused by some other neural activity, in which case we have nothing to map to.
I don't see how a detailed correlation of thought and neural activity could ever be established. We can't possibly report the full content of our internal dialog in real time, and even if we could this would not reveal any thoughts and feelings that were not fully conscious, which very likely outnumber our conscious mentation by several magnitudes.
For a map to be meaningful it has to be representational (that's practically redundant). If we are going to map the brain and hope to capture everything relevant to "mind" then we need to be clear about our signifiers and signifieds. How else to do this but by associating brain function with "external clues"?
The concept of a "functional" brain implies that we can isolate the various functions of brain tissue that we are interested in. If we can't, then what value is our map to be? Our uploads are going to be very disappointed when they wake up.
Posted by: Chris Schoen | Jun 17, 2008 5:55:00 PM
So it would help if you'd tell me whether you agree in principle that the brain-in-a-vat could in principle still have experiences just like those of an embodied brain, if the right pattern of nerve impulses and other discrete signals was fed into it (leaving aside the technical difficulty of simulating such a pattern).
In principle, no. I don't think these signals would be separable from the means of their transmission.
We are talking about a brain in a vat...although the body is only simulated, in order to feed information into the brain, you'd need an interface that would translate inputs to simulated sensory organs into real patterns of electrochemical impulses to feed into the brain. Are you saying that even if you sent the right pattern of real physical nerve impulses in through the brain's optic nerves, the brain would somehow be able to "know" whether or not the optic nerves were receiving impulses from a real retina taking in light or an interface which was stimulating different optic nerves based on the simulated light hitting a simulated retina? This would suggest some sort of mysterious psychic connection between the brain and the body, instead of the brain becoming aware of what was going on in the body through a pattern of discrete, localizable signals like nerve impulses. It would also go against the universal principle in physics that a system's behavior can only be affected by events on the system's boundary, that a system cannot have any nonlocal information about events outside it except to the extent that they influence events on the boundary.
The first mind uploads might find that their simulated environment was limited to something like a very sterile-looking hotel room, but the experience of that room might at least be close to indistinguishable from staying in a real version of the same room.
Here I have to return to Jaron Lanier's comment that the Turing Test can be considered two ways: either the machines are getting smarter or the humans are getting stupider. I certainly *hope* that hotel wouldn't be indistinguishable from an actual (sterile) hotel room experienced by a biological human, because this would imply that our perceptual and conceptual abilities had really atrophied.
Why do you think this exactly? Of course a modern simulation of a room wouldn't be indistinguishable from a real room, but that's because we lack the computing power to simulate all the objects in the room at the same level of fine-grained detail that human perception can resolve. If we could simulate the pattern of bumps in an object's surface down to something like the micrometer level, simulate variations in the way different parts of the surface scatter light at that scale, simulate characteristics like flexibility so the object would respond realistically to applied forces, and respond with its own forces on the simulated flesh of the upload (which would determine how the simulated nerves responsible for tactile sensations would fire), what do you think would be the visual or tactile giveaway that this wasn't real? A realistic Newtonian physics simulation of everyday objects at the micrometer level, rendered in realtime, isn't possible today because we don't have the computing power, but if we did I don't think writing the software would actually be all that much of a challenge, as I said it's basically just some Newtonian physics and optics. Even if we had the computers and the software, as non-uploads ourselves we might lack the right sort of interface to feed realistic tactile data into our bodies, but at least at a visual level I think we could produce images of a simulated hotel room which were indistinguishable from images of a real one, even if we were free to move the camera around and had something like a simulated stick that could be used to poke things to move them around and see how their shape changes without getting any tactile feedback. Again, the main difficulty here is the amount of computing power that would be needed, and beyond that a some more research into the fine-grained details of the way particular objects (say, carpet fibers) bend and twist in response to applied forces would be necessary, but it would be an extension of modern (much more coarse-grained) simulations of physical objects and substances like the ones you can see here and here and here.
We could spend the rest of this day brainstorming all the things are person in a hotel room might see, touch, smell, taste, hear, or otherwise be aware of
But you don't have to design all these sensations a case-by-case basis, you just need a general set of Newtonian laws governing how objects reflect light and change shape and create vibrations when they move (both in other objects and in the air), and a sufficiently detailed description of the properties of each object in the room (very small details of surface texture, light scattering, and responses to physical stress). If you want to make it an immersive environment for an upload, of course you'd also need to simulate the physics of their body at the same level of detail--the same rules that tell you how other objects deform in response to stress could tell you how skin, fat, muscle etc. deform when any other object in the room is pressed against them, and then you'd need to know the way that sensory nerves in the skin and muscles respond to such deformations. But compared to simulating an entire human brain, I'm pretty sure this would be a much simpler task that would require much less microscopic detail.
You don't think there could be great advantages to humanity if technical problems which might otherwise take a hundred years to solve could become available in a much shorter time?
I don't. What's the hurry?
There are all sorts of problems where the longer it takes us to figure out solutions, the more we irrevocably lose--environmental problems are a big one, finding ways to save human lives is another.
I think most physicists would say there is nothing fundamental that would prevent you from predicting the behavior of water molecules from basic quantum-mechanical laws, just that the computations get more complicated the more interacting particles you have.
Yes, but we already know the principles of water (at a macro level) from personal experience. This is what I mean by a posterior explanations. Given water, and a chemical understanding of its molecular structure, we can see how it is derived from oxygen and hydrogen. But that is not the same as saying that, having never experienced water, we could predict it from the properties of elemental hydrogen and oxygen.
The article seems fairly clear that they were able to simulate the behavior of water molecules using only the fundamental physical laws which are thought to govern all particles at the quantum level. I'm pretty sure no a posteriori knowledge about anything specific to water was used at all, and these fundamental quantum laws were not found by studying water. It seems to me you are really failing to understand something basic about reductionism if you don't see that the whole point is that specific properties of complex structures like water molecules are understood to be uniquely derivable solely from universal laws which govern the parts they're made of (in this case the protons, neutrons and electrons that make up both hydrogen and oxygen atoms). This is very much the mainstream belief among scientists in every field.
But I never said that observing speech would be enough to deduce the patterns in the brain, just that speech was caused by the output of pulses from the brain to the motor neurons that control muscles. The output of individual neurons would be expected to be a lot simpler to predict based on their electrochemical inputs, and experimenters can take also try to break down all the different changes in neurons (like changes in the density of neurotransmitters at the synapses) which influence this input/output relationship for an individual neuron.
What I think you're still missing is that if you reduce the subjectively experienced activity that is associated with a speech act (complex thoughts and feelings) to the objectively observable data of mouth and jaw musculature and production of auditory phonemes,
I never did anything like this. The subjectively experience presumably has to do with all sorts of brain activity that you can't deduce from the speech act itself; my only point was that the output from the motor neurons is determined by the activity of the neurons in the brain, so if you simulate the entire brain accurately, presumably the motor neuron activity will be accurate to and thus the upload will be able to carry on a normal intelligent conversation.
and then use only that data to correlate to specific neural patterns perceivable in the brain
What do you mean by using speech data to "correlate to" patterns of activity in the brain? Again, if you're suggesting that we could somehow deduce the brain activity from the speech acts themselves, that is definitely not what I was saying.
then all you have demonstrated is a correlate between physical speech activity with certain "functional" neurons. You haven't demonstrated a correlate to thought or consciousness.
Again, subjectivity is a philosophical matter--you have no way of knowing whether other people are really conscious either, but you take the fact that they use language in a way that is similar to the way you use language to express your own subjective states as evidence that they have similar subjective states of their own, so it would be reasonable to do the same for an upload who behaved like a biological human.
You write that "speech [is] caused by the output of pulses from the brain to the motor neurons that control muscles." That's misleading, in that the causal chain is highly indirect at best. Speech comes with a bundle of thoughts, feelings, motivations, and other "hidden" conceptual activity.
This sounds a little like Cartesian dualism, where some nonphysical mental entities like the "soul" are able to influence the behavior of physical things like muscles. In the normal scientific picture of the universe, physical events have physical causes (possibly including a random element). I suppose one could use "thoughts" and "feelings" as a shorthand for certain high-level patterns of physical activity in the brain, rather than using them to refer to subjective qualia, but then we'd get back to reductionism and the fact that the behavior of high-level physical entities like the brain or water is thought to be reducible to the interactions of the smaller physical entities that make it up; high-level entities are not thought to have any independent causal powers of their own.
If your entire argument against mind uploading depends on the idea that the physical world in not "causally closed" (physical events having physical causes), or that reductionism is not an accurate picture of the way causation works in the physical world, then there is probably nowhere further this discussion can go; just as I wouldn't know how to change the mind of an "intelligent design" advocate who has a strong conviction that their must be something more to the origin of living forms than natural selection operating on random variations, I also wouldn't know how to change the mind of someone with a strong conviction of the wrongness of the reductionist paradigm which has been so successful throughout the sciences. But the record is pretty clear that non-reductionist alternative ideas like vitalism and intelligent design have consistently failed to produce any useful research, and suffer from the "god of the gaps" syndrome where anything that currently isn't understood is ascribed to mysterious non-reductionist forces, and when it turns out that a reductionist physical explanation for that phenomenon is found, the non-reductionists simply point to some new phenomenon. I'm reminded of this section of Kenneth Miller's book Finding Darwin's God:
I hope you would at least be willing to acknowledge that if the reductionist picture of physical causation was correct, then the idea of simulating a brain in a computer would become at least somewhat plausible, even if you don't in fact believe in that reductionist picture (but I hope you would also be willing to acknowledge that rejecting this picture puts you out of the scientific mainstream).
If any given utterance is a result of specific neural activity, all the associated cognition must either (a) also be a result of that same activity, in which case our simulation is missing a lot of important data, or (b) be caused by some other neural activity, in which case we have nothing to map to.
I don't see how a detailed correlation of thought and neural activity could ever be established. We can't possibly report the full content of our internal dialog in real time, and even if we could this would not reveal any thoughts and feelings that were not fully conscious, which very likely outnumber our conscious mentation by several magnitudes.
In the reductionist picture, any high-level activity of the brain emerges entirely from the local interactions of the parts that make it up. So if we had found rules that let us accurately predict how any individual neuron behaves in response to local physical influences at its boundary (like incoming electrochemical impulses from neighboring neurons), then if you accurately map every neuron in the brain and simulate each one's interactions with its neighbors using these rules, all the appropriate high-level patterns would just emerge from this simulation "naturally" without our having to have any independent understanding of these high-level patterns whatsoever. Similarly, in the water simulation project as I understood it, there was no need for any high-level knowledge of how water molecules should behave, instead the researchers just took the general quantum rules governing the behavior and interactions of the basic units making up atoms such as electrons, and the appropriate water molecule behavior emerged naturally from this simulation.
For a map to be meaningful it has to be representational (that's practically redundant). If we are going to map the brain and hope to capture everything relevant to "mind" then we need to be clear about our signifiers and signifieds.
Not if you're a physical reductionist. High-level properties of the mind like "signifieds" are understood to just emerge from the collective interactions of large numbers of neurons (or molecules or whatever the simplest unit you want to look at), you wouldn't have to plug them into a simulation separately, just like you don't have to plug in high-level properties of water into a simulation which accurately simulates all the electrons and other basic particles that make up water molecules.
Posted by: Jesse M. | Jun 17, 2008 11:35:51 PM
Post a comment