3 ms·

You cannot copy "all" of the information stored in the human body. Briefly, this happens because the physical limitations of information storage require cooling

by scythe 10y ago

You cannot copy "all" of the information stored in the human body. Briefly, this happens because the physical limitations of information storage require cooling (Landauer's principle) which requires kT×log(2) energy per bit, or about 2.5 zJ at room temperature. The brain's memory capacity is about 2.5 * 10^15 bytes (https://www.scientificamerican.com/article/what-is-the-memory-capacity/ https://www.scientificamerican.com/article/what-is-the-memor...), but the amount of information which can be retrieved is much smaller than the amount of information required to fully describe the system -- in the same way that knowing the contents of my hard disk will not help you recreate my laptop. The brain weighs about 1300 grams, of which maybe 5%, or 65 grams, is protein, composed of amino acids which weigh about 0.25 zeptograms, so contains about 2.6 * 10^23 amino acids. If we assume that we need a kilobit of information per amino acid, which seems like a lowball, it would require about 650 kilojoules of energy at the absolute thermodynamic minimum (vastly more efficient than real computers) just to store the information in the human brain. The electrical power required to simulate its dynamics would be tectonic, and that's assuming perfect efficiency. Therefore, no computer that is composed of components which interact with the electromagnetic field and which operates at a temperature comparable to Earth temperatures will ever simulate a human mind to complete accuracy. It will always be an approximation.

cyan_atrus 10y ago
this is a really cool comment -- thank you! -- do you have any more reading you can link to?
- selestify 10y ago
  Same, would love to read more about this!
Razengan 10y ago
Copying a disk does in no way require simulating each atom of the disk, though? • We can already clone mammalian bodies. • I'll assume that "growing" an adult human body, with a "blank" or "default" brain, isn't too inconceivable. • Your real, natural brain can access the information within itself. • At the moment of "copying", we figure some way to biologically link the two brains together, and write some biological software to pull data from the source brain to the destination brain.
- JetSpiegel 10y ago
  While the information stored in the disk is stored in the "analog" world (magnetic potential of parts of the platters), there's an infinite amount of states that lead to the same digital disk contents. You can't take the digital representation and reconstruct the analog state that leads to it, since by doing the analog-to-digital conversion there was information loss. And growing a body with a "blank" brain, what does that mean? Your conscience is absolutely intertwined with your body. If you take my mind and put it in Usain Bolt's body, that Frankenthing won't run 100m in <10s.
- scythe 10y ago
  The disk is designed to be copyable. The brain is not. See e.g. this evolved circuit: https://www.damninteresting.com/on-the-origin-of-circuits/ https://www.damninteresting.com/on-the-origin-of-circuits/ The brain is, as far as we can tell, more like that circuit than a disk. Information is encoded within the structure of the processor itself, rather than being one of 2^N easily defined conformations of a memory device. As such, there is no way to copy the memories of brain A into brain B -- it simply isn't a thing. It would be like saying: copy all of the stains on this shirt, but the stains are integrated into the fibers and are unique to the stitching. In particular, this statement is false: >Your real, natural brain can access the information within itself. It cannot, for example, I cannot fully describe how to walk, given a pair of human legs and a butt. My brain can only encode information into language in the particular ways in which it is "designed" to do so. There is no way at all for the brain to report, e.g., the number of synapses within it, even if the firings of some neurons are recorded electronically. I say "design" in this sense because the structure of my brain is contingent on its fitness-for-purpose in a way that affected its formation, via biological evolution, similar to the way that the structure of an intentionally designed object is contingent on its fitness-for-purpose, which affected its formation as it was constructed by the entity which made it.
  - Razengan 10y ago
    Well then. If copying's out of the question (for now), then I propose Plan B: A single brain, in a secure location, remotely controlling multiple bodies (one at a time, but able to switch between them.) • Identify all the points of contact – the I/O pins – between the brain and body. • Identify the medium and format of communication between brain/body/senses. I think most of this has already been achieved. • Grow a body, a shell, without a brain. • Take your brain out and keep it alive in a vat. • Put wireless transceivers at each I/O point on the controlling brain and the shell body/bodies. Will anything like this be remotely (<pun) possible? You'd have to get used to lag but in the case of a fatal accident/violence* and most diseases your brain would be safe and able to switch to a new body.. I expect something like that would be very valuable to soldiers and workers in hazardous environments, like space.
HamSession 10y ago
It is good to note that Lauder's principal is a statistical argument for degrees of freedom in a system. The Landauer principal on a statistical averaging of this effect, meaning you will sometimes be able to store a little more information. I think you have your calculations off for the amount of energy required to store a human brain. Assume 8.6 x 10^10 neurons (more accurate number as its simple density measurement) each holding a 64-bit weight which gives us 5.5 x 10^12 bits. This multiplied by the lower bound e.g. k(1)ln2 or 9.56 x 10^-24 J, gives us 5.3 x 10^-12 J which is lower than the energy required to type a letter on this keyboard. Now even if we increase this by body temperature say 37 C we get 310.15 K or about 1.7 x 10^-8 J. This is of course a lower bound on the problem but even if you pump up the number of neurons or weight value you still get less than a joule of energy required.
- scythe 10y ago
  >each holding 64 bits There is no way in hell you're going to store the entire physical state of a neuron in anything near 64 bits. The question is not "information stored in the brain". The question is "information required to recreate the brain". In fact a single neuron has many more than 64 synapses, each of which itself has a highly intricate structure. All of the information describing this structure must be retained intact in order for the image brain to be equivalent to the substrate.

by scythe 10y ago

cyan_atrus 10y ago

this is a really cool comment -- thank you! -- do you have any more reading you can link to?

selestify 10y ago
Same, would love to read more about this!

Razengan 10y ago

Copying a disk does in no way require simulating each atom of the disk, though? • We can already clone mammalian bodies. • I'll assume that "growing" an adult human body, with a "blank" or "default" brain, isn't too inconceivable. • Your real, natural brain can access the information within itself. • At the moment of "copying", we figure some way to biologically link the two brains together, and write some biological software to pull data from the source brain to the destination brain.

JetSpiegel 10y ago
While the information stored in the disk is stored in the "analog" world (magnetic potential of parts of the platters), there's an infinite amount of states that lead to the same digital disk contents. You can't take the digital representation and reconstruct the analog state that leads to it, since by doing the analog-to-digital conversion there was information loss. And growing a body with a "blank" brain, what does that mean? Your conscience is absolutely intertwined with your body. If you take my mind and put it in Usain Bolt's body, that Frankenthing won't run 100m in <10s.
scythe 10y ago
The disk is designed to be copyable. The brain is not. See e.g. this evolved circuit: https://www.damninteresting.com/on-the-origin-of-circuits/ https://www.damninteresting.com/on-the-origin-of-circuits/ The brain is, as far as we can tell, more like that circuit than a disk. Information is encoded within the structure of the processor itself, rather than being one of 2^N easily defined conformations of a memory device. As such, there is no way to copy the memories of brain A into brain B -- it simply isn't a thing. It would be like saying: copy all of the stains on this shirt, but the stains are integrated into the fibers and are unique to the stitching. In particular, this statement is false: >Your real, natural brain can access the information within itself. It cannot, for example, I cannot fully describe how to walk, given a pair of human legs and a butt. My brain can only encode information into language in the particular ways in which it is "designed" to do so. There is no way at all for the brain to report, e.g., the number of synapses within it, even if the firings of some neurons are recorded electronically. I say "design" in this sense because the structure of my brain is contingent on its fitness-for-purpose in a way that affected its formation, via biological evolution, similar to the way that the structure of an intentionally designed object is contingent on its fitness-for-purpose, which affected its formation as it was constructed by the entity which made it.
- Razengan 10y ago
  Well then. If copying's out of the question (for now), then I propose Plan B: A single brain, in a secure location, remotely controlling multiple bodies (one at a time, but able to switch between them.) • Identify all the points of contact – the I/O pins – between the brain and body. • Identify the medium and format of communication between brain/body/senses. I think most of this has already been achieved. • Grow a body, a shell, without a brain. • Take your brain out and keep it alive in a vat. • Put wireless transceivers at each I/O point on the controlling brain and the shell body/bodies. Will anything like this be remotely (<pun) possible? You'd have to get used to lag but in the case of a fatal accident/violence* and most diseases your brain would be safe and able to switch to a new body.. I expect something like that would be very valuable to soldiers and workers in hazardous environments, like space.

HamSession 10y ago

It is good to note that Lauder's principal is a statistical argument for degrees of freedom in a system. The Landauer principal on a statistical averaging of this effect, meaning you will sometimes be able to store a little more information. I think you have your calculations off for the amount of energy required to store a human brain. Assume 8.6 x 10^10 neurons (more accurate number as its simple density measurement) each holding a 64-bit weight which gives us 5.5 x 10^12 bits. This multiplied by the lower bound e.g. k(1)ln2 or 9.56 x 10^-24 J, gives us 5.3 x 10^-12 J which is lower than the energy required to type a letter on this keyboard. Now even if we increase this by body temperature say 37 C we get 310.15 K or about 1.7 x 10^-8 J. This is of course a lower bound on the problem but even if you pump up the number of neurons or weight value you still get less than a joule of energy required.

scythe 10y ago
>each holding 64 bits There is no way in hell you're going to store the entire physical state of a neuron in anything near 64 bits. The question is not "information stored in the brain". The question is "information required to recreate the brain". In fact a single neuron has many more than 64 synapses, each of which itself has a highly intricate structure. All of the information describing this structure must be retained intact in order for the image brain to be equivalent to the substrate.