Firstly I'm not 100% sure if Neurocryptology actually exists in any specified form since a search on google only came up with two references.

The first reference refered to a translation of something someone wrote from French to English which would have been wrongly interpreted if taken as was.

http://ase.tufts.edu/cogstud/papers/intrptxt.htm

However this reference doesn't really define neurocryptology and might itself be wrong in how it's refered.

For the actual post however, I'm going to suggest that the definition of the wording is about to be defined.

In a normal working human brain there is the capacity to function with both short-term and long-term memory, however in some people the short-term memory is impaired which in turn can impair long-term memory.

To deal with the misfunction of loss of using memory correctly, people that suffer from short-term memory loss can actually increase their processing capacity. As every time they come across something they have previously learnt and suppose to 'be trained in', they might attempt to generate an answer to fill the missing gap.

The gap itself might not be completely vacant, it might actually have some residual memory left which generates a filter. This filter I term as a "HASH", my terming is coming from the Cryptographic vocabulary in which it means it's not the input or the output but a Vector(multiple) of Factors.

My suggestion also is that this can be extended on with say for instance, "String Theory".

"String theory" is where a problem exists that needs a solution, so you can see the input and output, however the HASH hasn't been arrived at because the overall processing of the data hasn't been decided. This means that many different variations of theory can be created to fill the HASHES location, and some might even LOOK correct, however only the correct HASH will actually connect the overall string theory.

In for instance this Forum, there are alot of people that end up creating theories about subjects, where they have perceived their version of the HASH, however their HASH (and even my own on occasion) can be completely false and generates a Pseudoscience.

The reason I'm posting this is it is just an interesting possibility in how to gain answers for intelligent systems and even possibly adoptive by people not just machines.

Took me a while to understand. Or at least I think I do with regard to hashes.

A HASH function in cryptography gives out a sort of small fingerprint of a given source which is used for checking authenticity.

Are you saying: "If short term memory was poor and hence long term imprints were less prominant then you would end up with a smaller-than-usual memory, that can later be recompiled when the memories are drawn upon" ?

Which I agree would obviously require more processing power as suggested.

However, HASHs do not 'look' like the original data at all (The idea is that if you change a single character anywhere in a text the HASH will be a completely different number), where as a less prominant portion would at least resemble the original and hence be of little crytographic usage. I'm probably missing the point though!

I'm not sure I see how you would solve a problem without first having a hash, implying the prob has already been solved.

Well in the world of Cryptology if someone attempts to Crack a password, they are usually attempting to simulate the HASH.

They might not know the actual password, however there can be many different letter combinations that cause the same value HASH to be created.

This is pretty much where the idea came from, since multiple different input's can generate the same HASH value, although only one of them is the actual password.

This is pretty much where the idea came from, since multiple different input's can generate the same HASH value
Highly unlikely...Please Read Collision and Coupling Laws of hashing...

bye!

It's not as unlikely as you think, Zion.

It's a pity I can't remember the name of a particular puzzle I use to do as a kid. The Puzzle itself was a Rectangle that was cut into various shapes a bit like what you get in Tetris, and the pieces to begin with are inside a rectangular frame. When you start the puzzle you make all the pieces fall out of the frame and scramble them about, you've then got to put the pieces back into the frame.

The puzzle consisted of a finite amount of pieces but had more than one way of arranging the pieces to eventually create a rectangle within the frame.
[Rectangle=Oblong]

A mathematical example would be to look at how many different ways there are to write a sum that equals "9", Only one sum will be correct and others will only create a Representation that meets the overall answer.

I'm not suggesting my example of a HASH is correct in my theorum, however it does explain that sometimes 2+2 does always equal 4 depending on what sum was used to generate the answer 4.

Stryder, funnily enough I have that game - except with mine there are two or three completely different ways to solve. I fluked it once.

In MySQL (the version I use, anyway) password hashes are 64 bit I think. Anyway it's a string of chars about 14 long, which is (probably) longer than most passwords. Other texts can obviously generate the same HASH key, but they may not be practical (i.e. lengthly).

Hashes are one way functions, but of course can be reversed with arbitrary constants thrown in were there are unknowns to give a password that would work. There are many obsticles if you wanted to do that btw, like not all binary combinations are valid chars ($,!,*,., Return, Spacebar etc...).

Stryder,
You didnt read my sentence about Collision Rules.I would presume to dont know about Hash Function Requirements.Let me throw some tid-bits of knowledge for you...

Okay.Hash function produces a fixed value of any variable block.say x is that block.Then H(x) = h,here H(x) is the hash function and h is the hash value.While designing a Hash function you have following requirements :

H can be applied to block of any size.
H produces a Fixed length output.
For any given block x,it is computationally infeasible(yes it is Infeasible) to find x such that H(x) = h.This is sometimes called one-way property.
For any given block x,it is computationally impossible to to find y such that y!=x(not equal to in C),with H(y)=H(x).This is sometimes called weak collision resistance.
It is computationally infeasible to find any pair x,y such that H(x)=H(y).This is called Strong Collision Resistance

I hope the strong collision resistance explains my highly unlikely expression.The above matter was from Ref-Book,during 6th Semester in Computer Engg.Subject was Crypto-Graphy and Netwrok Security(William Stallings).If you have any queries we can discuss here off-course...

~~Zion
bye!

Well Zion, I know HASH in Cryptology is different from my reference, however it's the nearest thing to explain the overall theory that people can suggest Facsimiles that look correct (pseudosciences) but still come to the conclusion that people look for. They might have had all the same inputs but juggled the ordered around the wrong way (Compute the variables) to get their answer.

For any given block x,it is computationally impossible to to find y such that y!=x(not equal to in C),with H(y)=H(x).This is sometimes called weak collision resistance.

Surely if the output is smaller than the input, there will be multiple y's - which could be determined by reversing the hash algorithm and randomly picking the necessary unknowns?

You dont get it Blue_uk.We are talking about x & y as two input blocks.if "Y" block has a same hash value as x block then y should be equal to x block.where's the point of multiple y's...?

bye!

PS: Did you know of birthday attacks?


bye!

Stryder,explain your post once again please.(original one i mean).I have difficulty in grasping the idea of yours.

thanks.
bye!

You dont get it Blue_uk.We are talking about x & y as two input blocks.if "Y" block has a same hash value as x block then y should be equal to x block.where's the point of multiple y's...?

bye!
I do get it.

If the input block is greater than the size of the output block (number of bits, not value) then there must different inputs giving the same h. Not necessarily for all h, of course, but when the input is very much larger than the output, a greater proportion of h's will have multiple sources.

E.g.
input 10 bits, hash output 2 bits, you only have to encode 4 different texts before you are guarenteed to get a duplicate h.

The point of a multiple y is to have a password you can use effectively even if it's not the one chosen by the victim.

(For the benefit of others, password checkers only compare the post-hashed passwords.)

Stryder,explain your post once again please.(original one i mean).I have difficulty in grasping the idea of yours.
Seconded, I'm not sure I understand what neurocrypt~ is.

in MD5 perhaps.But SHA? RIPMED?...prove it.

bye!

That exceeds my knowledge, but what I said above applies to any function.

What I was suggesting was, a person with a short-term memory problem potentially increases their capacity to compute. The reason for this is the redundant short-term capacity that is failing is capable of being used for processing.

I'm also suggesting that because their memory doesn't function properly that they have to re-compute an answer for a value that should already be stored in the memory.

When I tried pulling "String Theory" into it, I was trying to suggest that such people that have to re-compute these great vectors, could on occasionally come up with the answers to problems through problem solving purely based on how their brains condition is.

[I hope thats better, since it's not late yet and it was early back when I did the first post, if you catch my drift.
Btw Zion, It's great having you about as it makes me simplify it a great deal and make it clearer.]

stryder search "neural networks" and "cryptography" I don't know if thats what your looking for but there is a group in i believe israel that is highly interested in this implication...of course sthere are those who have their doubts, because of a neural nets stabiliity