Clarifications on uploading

PZ has written a reply to my post on brain uploading. I think he gets some things definitely wrong, but in this post, I just want to make a few clarifications that I should have made in my original post:

Uploading will require significant advances in the technologies we use to study the brain. PZ’s post mentions the nematode worm, an animal that many neuroscientists study because it has such a simple nervous system. I almost mentioned the nematode worm myself in my post to illustrate this point.

I had a professor in undergrad, Tony Stretton, who had spent decades studying the nematode worm. He once told us that when they started studying it, they thought it was so simple it would be easy to completely work out how its nervous system works. They were wrong, it turned out to be much harder than that.

Based on that, I think it’s safe to say that just doing more of what we’ve been doing in neuroscience will not bring us complete, accurate computer models of the human brain in even two centuries. We’ll need better technologies for studying the brain. But that will take time, which makes me very skeptical about predictions that uploading will be here in a couple decades. (I was probably too understated about that last point in my original post.)

I don’t think uploading will necessarily be the technology that radically changes the world. As I said in my previous post, I focused on uploading because I think in some ways it’s more straightforward than the issue of AI written from scratch (de novo AI, it’s been called).

Still, it may be that de novo AI is a much better approach to building powerful AI than brain uploading is, and consequently de novo AI will be the big disruptive change that will happen first. That view is compatible with the view that brain uploading would become possible given a couple centuries of uninterrupted technological progress.

I’m apprehensive about the effects uploading and other future technologies may have on the world. On balance, technology has so far been a plus for humanity. So I don’t want to be too pessimistic about uploading and other possible future technologies. But I wouldn’t be too sure the effects will be good either.

For example, Robin Hanson has argued that uploading would mean the return of slavery and a return to virtually everybody living a subsistence existence. Hanson has also tried to argue that wouldn’t be such a bad thing, but (no surprise) others are dubious. Or: is there a risk that a copy clan (as described in  Muehlhauser and Salamon’s paper) could take over the world?


Slavery abolition and animal rights: the biggest problem
Notes on Robert Fogel’s Without Consent or Contract
How selfish are voters?
No scientific evidence for that
  • Denisha Pitsenbarger

    Warning ! This message is for the webmaster , i see you have a great blog . But beleave it or not this site wont get you any cash i learned it the hard way . You Probably heard of this “Money is on the list” (not in BLogging) . I know Blogging is not bad if you are a great writer but wont it be better if you make 10x , 20x or maybe 50x more ! i cant give you all info here , if you are interrested join me here NOBODYSDEAD.COM feel free to contact me ;)

    • mikespeir

      Oh, man! I want money? Can I have some money?


    • Chris Hallquist

      I was going to disemvowel this comment, but I then I realized that would obscure the bad spelling, punctuation, and capitalization. The comment is funnier intact.

  • SAWells

    None of your new points touch on the major original problem. From your original: “The version of the uploading idea: take a preserved dead brain, slice it into very thin slices, scan the slices, and build a computer simulation of the entire brain.”

    This would only work if you could “scan” to the molecular level. We can’t do that. Quantum limits are such that we’ll probably never be able to do that. Are you going to count individual low-copy-number proteins one at a time with an electron microscope? We can’t slice to that level of “very thin”. And you’re scanning a _dead brain_. If you want a computer simulation of a dead brain doing nothing, why bother scanning anything? Uploading would require scanning a _live brain_ at molecular levels of detail in nanoseconds or faster. Good luck with that!

    • Hunt

      That’s one of the least convincing arguments from PZ’s post. Why are you sure you need a molecular record of the brain to reconstruct its operation? –particularly considering that we know that variations in brain chemistry from hour to hour don’t fundamentally change brain operation or identity. An individual synaptic record, along with certain parameters from each synapse, may be more than enough to record brain operation well enough to simulate it and record individual identity, and that might be doable with even an imperfectly preserved brain. Now, there are a lot of “maybe’s” and “mights” in all that, but the important thing is that PZ is in no greater position of certainty than I am, or anyone with a sanguine view of uploading is. That’s why it remains an interesting idea and one not deserving such off-base ridicule.

      • wanderfound

        Yes, a lot of this debate is on subjects where there is a great deal of uncertainty.

        However, I think you’ll find that the vast majority of people who regularly do hands-on work with brains (such as PZ and myself) are of the opinion that the people most enthusiastic about uploading have dramatically underestimated the complexity of the brain.

        Will uploading be possible someday? Unlikely, but maybe. Is this going to happen during the lifetime of anyone alive today? Absolutely not.

        • Chris Hallquist

          “The people most enthusiastic about uploading have dramatically underestimated the complexity of the brain.” I agree with this, for what it’s worth. But mainly because you have Kurzweil predicting uploading in a couple decades and claiming the brain isn’t that complex.

          • wanderfound

            Even if we grant a less-daft-than-Kurzweil trajectory of development, the issue of timing is still fundamental.

            You’re concerned about the impact of this hypothetical technology on the world, and appear to desire that we formulate our ethical and/or political response to this possibility in advance. But history would suggest that when responses of this sort are formed before having a proper understanding of the nature of the situation involved they tend to get things spectacularly wrong.

            If uploading ever arrives, the precise details of how the technology operates are likely to be highly influential in shaping the outcomes. It’s also highly likely that the date of reaching this technology is going to be so distant from the present that the shape of human society will have already changed to a degree as to make any current predictions of the social impact nonsensical.

        • Hunt

          You know, you really can’t say. Science doesn’t proceed in a way that can be predicted by linear extrapolation or “dead reckoning.” You might recognize this from Kurzweil’s infamous reference to exponential advance, but even more than that, these things simply can’t be predicted with certainty. “Probably” you are correct, but there will always be the small but significant chance that you’re the analog of the informed and well meaning physicists in the 50′s who said that travelling to the moon was impossible. There were those in 1955 who just. could. not. imagine a man walking on the moon, but there you go…

      • SAWells

        Your “some parameters” involve the state of a whole bunch of membrane proteins, internal cell signalling proteins, the expression state of nuclear genes… yes, you do need detail at the molecular level, even if some of them are pretty big molecules.

        This: “variations in brain chemistry from hour to hour don’t fundamentally change brain operation or identity” is a whole hairy problem. Consider the difference between your brain when you’re sober, and your brain with the addition of a small concentration of ethanol. Consider the difference between your brain when you’re hungry and when you’re full. Molecules matter.

        You don’t get to make the most blithely optimistic assumptions. This is biology. Everything is always messier than you think in biology.

        • Hunt

          Yes, but the molecules don’t matter enough to change who I am, what I know, or what I can do, which is my point. Within the entire state space of possible molecular configuration, “I” am preserved to a recognizable extent, which means that I am not mapped to any single molecular configuration. That alone is enough to counter PZ’s requirement. Just what kind of higher level abstraction is necessary to record identity remains the question.

          • Dunc

            A hundred micrograms of LSD should be enough to disabuse you of that notion…

          • sawells

            “Yes, but the molecules don’t matter enough to change who I am, what I know, or what I can do, which is my point.”

            Your body devotes considerable effort to not letting stuff get into your brain- blood-brain barrier – precisely because the molecules absolutely matter.

            Hi, this is ethanol. Ethanol, meet a guy who doesn’t believe the molecules matter. You guys have fun. An error in “scanning” of no greater magnitude than mistaking your water content for water+0.1%ethanol would be enough to make the difference between a model of you sober and a model of you permanently pissed out of your skull. Would you accept the latter as an accurate upload of your mind? I’m thinking not.

          • Hunt

            What I meant is that specific molecular configurations don’t determine the general character of cognitive function, mental states, identity, etc. I drink a cup of coffee, I’m still me, think pretty much the same thoughts, etc. More than that, moment to moment my brain changes precise chemical status. It seems obvious that this directly refutes PZ’s point about having to record every protein, epi-genetic state, etc. He’s operating at the wrong level of abstraction. To PZ, emulating an operating system means simulating electrons and transistors of a computer running it directly.

          • jamesskaar

            taking a standard human template, the brain already is, up to some level, just a template, and every human(most animals in our basic shape) has nearly the same template. hardware is hardware, throw a hard drive from one machine, into a similarly configured machine, it’ll be the same operating system, unchanged, even then, your operating system is based on ultra common software. well, it may need to have drivers for the different hardware installed, but most new operating systems have drivers already there, just needing activation. the parts that are YOU, are the things that are put in the private documents directory, the common parts can be replicated with barely a yawn from the person making them, it’s just a vat of goo with a programmed shape.

            knowing where the molecules are, is not that important, knowing the levels in each part may be. our brains are a large cluster of simple computers, they just need boot code, some specialised interfaces on a set of them and a saved state for the ones that deviate from template. the more primitive parts of the brain, balance, vision, interface for voluntary muscle control, all the same, there’s parts in us that aren’t too different from the same ones in a crocodile…

  • Brad

    For me to be interested in uploading, I would need to experience continuity of consciousness. Don’t care to be the guy in the tank in the prestige.

    • Chris Hallquist

      Fair. Though as I’m trying to get people to realize, what you’re interested in personally will have only a very small effect on how the technology impacts society.

      • Brad

        Certainly. And even without that crucial continuity, there’s still a pseudo immortality, and newborns could have their brains uploaded and then something neurosciencey to put the upload back into the meatbag with a implanted chip, resulting in continuitous immortality as standard procedure for the people born after the technology has proliferated.

    • Hunt

      Hmmm, every time you fall asleep must be somewhat traumatic for you.

      • Brad

        I’ve acclimated to that particular anxiety. I am however, interested in the development of wakefulness drugs like modafinil, but that’s more resenting of roughly a third of life being sleep than worrying that somebody else wakes up in the morning when I go to bed.

        • Hunt

          You seem to be interested in waking up tomorrow morning, yet not reawakening after “falling asleep” on your deathbed. I see a contradiction.

          And if you thought about it this way; that you *are* a different person every morning you wake up, who correctly remembers what it was like to be you the night before, how would that affect your consideration of mind uploading?