View Single Post
  #37  
Old 04-14-2016, 08:03 AM
joema joema is offline
Guest
 
Join Date: Jul 2006
Location: Nashville, TN
Posts: 455
Quote:
Originally Posted by Senegoid View Post
I don't think this follows at all. If you can duplicate an ant over there, down to the molecular/atomic level, and if all the innate behavior is encoded therein, then one can expect the new instance to have all the same behavior. There is no need at all for the inventors of the duplicating machine to have any idea how behavioral encoding works for this to happen. Just copy the ant, verbatim.

The idea is similar to copying an encrypted data file: You can make a bit-for-bit copy of a file that you can't interpret (because it's all encrypted gobbledygook), and the copy will be exactly as decryptable as the original (by someone who knows the key).
While this is correct, it assumes a magical perfect copy mechanism. If some advanced alien civilization gave you a black box which did that, of *course* you don't need to understand how biological function works.

But to develop that on our own in would seem to require this understanding. Otherwise what scanning approach would you use? Cellular? Synaptic? Molecular? Atomic? Is the behavior encoded statically or is it in the ongoing dynamic spatio-temporal firing pattern among neurons?

Consider a much easier replication scenario than the human brain. A functioning Intel Xeon E7-8890 v3 (18 cores, 5.6 billion transistors) is sent back in time so an electronics researcher from 1958 could examine it and try to replicate it.

He could try to examine the signals with an oscilloscope, but his equipment could not resolve the gigahertz-rate bus signals.

He could de-cap it and look at it the die with an electron microscope, but this would not reveal the underlying operational principles. He could make a photomicrograph of the silicon die, but this would reveal nothing about the hidden functional complexity. It would not reveal the microcode, or anything about the immense complexity of the instruction decoding -- prefetching, dependency checking, branch prediction, hyperthreading, register scoreboarding. He could make a lithographic copy of that die and reproduce it to the degree 1958 technology allowed, but it would not function. It would be a child's crayon drawing of a machine, and would have about equal success of working.

If his boss asked what direction should they pursue to replicate it, he would likely say he doesn't know until he understands more about the internal function.