please, leave your brain where it is
A while ago, I took a look into an idea of the Technological Singularity that looks towards a future in which the human mind will reside as pure information within a complex network of computers. Since the first post on the subject, I’ve been getting a slow but steady stream of feedback from proponents who say that with enough time, we can’t discount something like this from happening or that I haven’t given a valid reason why a future technology for transferring human minds to machines won’t work. According to them, if you have a perfect, one to one transfer of the information from the brain to a computer, the human and his consciousness should all be there in an electronic format like a huge collection of files. All you’d need are the right tools for the job…
The concept of machine aided immortality is one of those ideas that’s just too good to let go. I may be tempted to believe it myself as an avid reader of science fiction since childhood. But it just so happens that technology is my area of expertise and I’ve worked with computers too long not to hear alarm bells ringing when I picture mad scientists trying to replicate a brain in machine form. When Ray Kurzweil talks about melding minds with microchips, he focuses on the idea that our minds have electrical impulses like computers and our thoughts can be read with proper equipment. However, the issue is a lot more nuanced than that. Just because we can design a probe that can read electrical activity in our nervous systems and activate preprogrammed functions, doesn’t mean that we took a step towards replacing neurons with silicon.
Let’s say you wanted to truly download a human mind into a mechanical vessel of some sort. To make it work, your machine would have to work the same exact way as a human brain. That’s not going to happen with just a few futuristic updates. You’d need to rethink how it works from the bottom up, starting with memory and how it uses electrical activity for its basic functions.
Electrical pulses in digital equipment like computers come from transistors switched on and off as electricity from a power supply flows to them. Data generated by myriads of these pulses is then recorded to a disk in a computer’s hard drive and can be retrieved when needed with the use an index which keeps track of where on the disk the data is actually recorded. So if you needed to find a file on your computer, the device finds all the places on the hard disk where the file’s data lives and presents it in the correct form via an operating system. A human brain is very different. In the brain, those pulses come from chemical reactions between sodium and potassium. There’s no hard disk where data is stored. Instead, clusters of neurons store the data temporarily and the more you reuse the information, the stronger those connections get and the better you remember it.
Another major difference is that computers have limited amounts of data storage because hard disks can fit only so much. But the brain doesn’t have a limit on the amount of information it can store. Forgetting seems to be more of a function of not using the information or being unable to effectively retain it rather than our limit on potential knowledge. Estimates of what capacity the brain has are completely meaningless because they try to apply a the constraints of an electronic system designed to have X amount of storage to a biological network that evolved over millions of years and records memories in much more sophisticated and dynamic ways.
So if the Singularity proponents can overlook so basic and so important, what are we to make of the idea that human brains can actually be downloaded or merge with machinery? Well, again, there’s a lot of excitement over the fact that our nervous system generate electrical activity and carry information with pulses of ions. In a computer network, you can get the same pulses to get you the same data on a different machine. But keep in mind that there’s no hard disk to which the data gets recorded in the brain and the electrical activity is used to fire neurotransmitters rather than record files. The brain is the storage, the RAM and the operating system so trying to transfer over the signals neurons send to each other into a computer isn’t going to give you any of the information stored in the person’s mind, not to mention that all you’d be doing in interrupting pulses which will start again as soon as your futile effort is over.
The neurons in our brains aren’t just transistors that send ions to each other. They’re what make us who and what we are. Human minds are a product of chemistry and organic connections, swayed by neurotransmitters and hormones, prone to emotions that were passed on from our earliest ancestors as survival mechanisms. When someone like Ray Kurzweil talks about abandoning our bodies, he’s taking his very circumstantial and drastically incomplete knowledge of both computers and human brains, and applies a hefty dose of what can only be described as a technological New Ageism. His idea is based on a typical religious model which sees our bodies as nothing more than vessels for our souls. It’s that soul, that concept of the transcendent human essence, Kurzweil wants to extract and put into a microprocessor to gain a sort of immortality. And in this case, he thinks he found the human soul in the everyday electrical activity of the brain.
…applies a hefty dose of what can only be described as a technological New Ageism. His idea is based on a typical religious model which sees our bodies as nothing more than vessels for our souls.
Man, don’t ever tell those guys that, especially folks like Michael Anissimov.
You’ll get flamed worse than the poor women at the Salem Witch Trials!
Excellent post. We are our bodies. Not just the thoughts in our brains.
All those details basically boil down to the question of whether a brain is (just) a Turing machine or not. If it is, then the Church-Turing thesis means we can overlook the implementation differences, and run a brain on any UTM. If not, then not…
I disagree with a number of your points.
“But the brain doesnt have a limit on the amount of information it can store.” That’s a heavy heavy claim to make sir, and one that a generous amount of people don’t agree with, I’m wondering how you come to this conclusion?
“Forgetting seems to be more of a function of not using the information or being unable to effectively retain it rather than our limit on potential knowledge.” Again, you seem to be speaking like this is 100% correct and everyone agrees. Why do I remember lying in ym crib and getting my diaper changed? It’s not something I use all the time, which makes your claim not correct.
“Estimates of what capacity the brain has are completely meaningless because they try to apply a the constraints of an electronic system designed to have X amount of storage to a biological network that evolved over millions of years and records memories in much more sophisticated and dynamic ways.” I don’t see how you get this either. a SSD is far far more complex than say the old punchcards, but the amount of data they store is measured in the same way. the unit of measurement is a man made concept.
But a challenge to you sir. Say I design a perfect human neuron replacement. something that takes the exact same input,a nd gives the same exact output, and I replace only 1 neuron in your head with this mechanical one…are you still a person then? According to you it seems liek no. But if you are, say I do 5…10…15….etc, what’s the threshold?
No sir, I find your arguments very lacking, and don’t disprove the possiblity of a technological singularity at all.
Jason,
This is actually just one subset of Ray Kurzweils rendition of an idea which covers a lot of ground, so I would be more surprised if you thought I gave a thorough rebuttal to the entire concept in just one, 850 word post.
To clear up the confusion over what’s meant by the brain having no limits on the amount of information it can store, I added a link to a thorough overview of the Atkinson-Shiffrin Model of memory which is pretty much the default model studied in psychology classes. Basically, there’s no evidence that the brain ever actually runs out of room to store data like a data volume would over an entire human lifespan. So if we really wanted to nitpick, we could say that there is no practical limit to human memory.
When you’re comparing SSD and punchcards, you’re comparing the same concept of storing data with very different implementations. The technology has changed but what it fundamentally does hasn’t, which is why computer memory has always been measured the same way. Bits and bytes have not changed since the invention of digital technology.
Finally, your idea of replacing neurons in a human brain is really a Ship of Theseus Paradox, a philosophical problem that has nothing to do with literally downloading a human mind to a machine. I’m not sure where you saw anything about that in the original post.
I must say, the idea of downloading a personality from a meat body into a mech body sounds ridiculously difficult what with the issue of transmigration. Getting software to go from one computer to another is one thing, but getting “YOU” from something you recognize as yourself to a foreign body may well be impossible … except through death – which, all things considered might suck to the extreme.
The idea of having the capacity to create a machine that could hold an entire personality, though, I think, is going to happen with the advent of quantum computation. Combine quantum computers with neural technologies and it just might put humanity over the threshold.
Very nice post and I share your thoughts exactly. I liked your point about the fact that the brain might not really have a memory capacity in the same way that a computer does. In fact, the whole idea of downloading the brain seems to require a dualist stance (the mind and the body are separate and separatable) which seems rather religious to me. I’ve got a blog committed to pointing out all of the ridiculous claims that Ray Kurzweil and other Singulitarians subscribe to. If you want, please check it out at: http://antisingularity.wordpress.com/
Very interesting topic, but i am a bit unsure on the information that the brain stores, i don’t think there is a limit to this, but then again this depends on individuals….
Leakage?? There must be a point where something must collapse to make way for the new knowledge.
My reply goes out to all the people who replied that said this idea was ridiculous. Just because you cant figure out how to make this happen, does not make it impossible. It is possible, and technological efforts have already proved it. Do your homework bitches!
I’m afraid I don’t agree with very much of your post. I agree that the lack of a model of the brain is a bigger obstacle to uploading than any hardware shortage is. But given enough time and effort, that can also be fixed. Here’s how:
Nanotechnology is likely to make it possible to attach a robot from a swarm to every single neuron in a subject’s brain, and thereby collect detailed data about how it interacts both with other neurons and with neurotransmitters and hormones. (True, it may be difficult to observe the neurons on such a small scale without affecting them in the process, but it isn’t proven impossible.) Once this data can be analyzed, a computable model of that person’s brain can be built, tested and refined.
“I’m afraid I don’t agree with very much of your post.”
No need to fear. I generally don’t bite and even when I do, you can relax knowing that I had all my shots.
“Nanotechnology is likely to make it possible to attach a robot from a swarm to every single neuron in a subject’s brain…”
That would mean about 100 billion machines about the size of a neuron, or a neuron’s soma at best. Your skull would have to double in size to hold that much machinery. And there could also be the potential for this swarm of nanobots in your head blocking or interfering with normal operations by getting between dendrites.
“and thereby collect detailed data about how it interacts [...] with other neurons”
Or we could just develop more and more sophisticated fMRI to see what neurons are firing, when and how. No need to literally cram our heads full of robots.
“Once this data can be analyzed, a computable model of that person’s brain can be built, tested and refined.”
Sure, but then what? It would just be a model of a brain in a computer used for medical illustrations or visualization rather than a full blown human. To see more about how the human brain can be simulated by computers and what will happen, check out my debate with the Singularity Institute’s Michael Vassar on the subject.
Gfish: the gradual neuron replacement scenario is a thought experiment which intuitively shows why the functionalist-materialistic view of mind is correct. If you really truly believe mind uploading is not possible, answer this:
Imagine a future technology that allows individual neurons to be replaced with nanotech equivalents. The nanotech neurons are functionally equivalent – they connect to the same synapses of the original neuron, and they perform the same functional roles. Replace one neuron with a nanotech equivalent, and nothing changes – the rest of the brain doesn’t notice. Replaced one by one, your entire brain could be replaced with nanotech, but you would have the same information content, you would think the same thoughts, etc.
Now imagine if these nanotech devices allowed you to interface with computing systems and transfer all of their internal state, and so on. Your body could die, but the nanotech devices could continue functioning, even be removed from the skull, reform your brain in some other device, and connect you to a virtual reality environment – ala the matrix.
If functionalism is wrong, one of the following must be true:
1. at some point during the neuron replacement you lose consciousness (as stated this is impossible because the successful neuron replacement wouldn’t change brain function – so for this to be wrong you must choose a worthless definition of consciousness)
2. or you become a different, new conscious person (again impossible unless you use a worthless definition of consciousness and identity)
3. or the gradual neuron replacement is somehow not allowed by the laws of physics (not true from current theory)
All 3 are untenable, and so functionalism is correct. Thus uploading is possible.
Apologies for the necropost, but I thought I had something to add.
gfish: you seem to be using a different concept of identity than most people do intuitively. Correct me if I’m wrong, but you’re implying that a perfect emulation of the brain, whereby the brain is self-aware, sapient and can’t tell the difference, would not be equivalent to the original? I think that your saying “the human and his consciousness should all be there in an electronic format like a huge collection of files” is a gross oversimplification of the issue. A potential straw man, really. I wouldn’t swear to it, though; I’m neither a transhumanist nor a neurobiologist, but a computer scientist.
In essence, I see mind uploads as possible through emulation of the neural net, with special attention paid to the chemical in neurochemical. It would require ridiculously fast hardware and is pretty much out of the realm of current computer architecture, but theoretically possible. More likely, the abilities of the higher brain functions would be layered on top of an abstraction layer that hides the idea of disk IO from the “brain” when it attempts to recall a memory, or peripheral control for moving a limb.
Jake Carnell: You’ve missed at least one possibility: what if functionally equivalent nanotech has the same problem in that you can’t pull the state from them? Or even if you could, the state of each individual neuron-alike combined doesn’t add up to the entirety of a mind? This doesn’t necessarily require dualism; there are quite a lot of chemicals hanging around outside of neurons and synapses in the brain.
“[are] you’re implying that a perfect emulation of the brain, whereby the brain is self-aware, sapient and can’t tell the difference, would not be equivalent to the original?”
Not at all. I was talking about the idea of transferring consciousness via uploading. Now, if you have a system that can run the exact same way as a brain and produce sapience and self-awareness, then yes, it would have to be an equivalent to the original since it does the same set of functions. But could you take the thoughts out of someone’s mind and put them into a computer that runs the same way as a human brain? Probably not.
“I think that your saying ‘the human and his consciousness should all be there in an electronic format like a huge collection of files’ is a gross oversimplification of the issue.”
Then you might be happy to know that I’m satirizing the sentiments of Ray Kurzweil and critiquing that sort of approach to the concept of uploading as someone who’s aware of some of the neurological issues involved and as a systems analyst.