[neilep (OP)]

If 'memory' is something intrinsically stored chemically as engrams in the brain...is it possible that either they could be transplanted into someone else or replicated into someone elses brain ?

[DoctorBeaver]

I've not looked at neuro-psychology in any depth, but as far as I'm aware we still don't know the mechanism behind memory. However, the physicalist view is that memories are nothing more than a network of connected neurons controlled by electro-chemical stimuli. If that is the case, and we can find a way of establishing exactly what the connections are that give a person certain memories, then I suppose it's theoretically possible that the pattern could be imported into another brain or a computer.

One big problem I can foresee is that memories are not clear cut. Remembering 1 thing can trigger the memory of something else. We may catch a whiff of scent that reminds us of a person and then other memories of that person come flooding back. There must be a connection between memory and olfactory processing. The same is true of all our senses.

Sufferers from Alzheimer's, or those with brain damage, can have some memories impaired or lost whilst others stay firmly in the mind. That would imply that memories are scattered throughout various parts of the brain. Therefore, it may be that we would have to ascertain every connection from every braincell in a person's brain before we could guarantee that we had captured a certain memory.

We have billions of braincells, billions of interconnections. And the connections are not like binary notation in a computer; either 1 or 0, yes or no. There are varying degrees of interconnection akin to fuzzy logic. At present we simply don't have the computing power to mimic an entire brain. I don't know if quantum computers could provide the requisite power, but from what little I know I don't think so.

[neilep (OP)]

Thank you Dr Eth.

I quite agree that at this time it is a matter of fantasy !..but I can see it being a real possibility.

I don't believe it is prerequisite that memories must be associated with  olfactory processing or any senses for that matter. Though, clearly there are associations which are definite and true.  But I do believe that it's possible to recollect without the need for a sensation.

It doesn't matter if a memory is clear cut or not....If so..then it's non clear cut nature will also be transferred

Certainly emotions play a key role too and in most cases , it's the memory  that triggers the emotion...though, granted, the whole process can be initially stimulated by a sense.

I see ourselves as physical beings...The answer is in the brain somewhere and though it may be the most least understood and perhaps the most complex thing in the Universe..I think one day we'll unravel it's true nature.

I am sure you are right that quantum computers will be an aide....perhaps...that is the true nature of our brains...in that they themselves are quantum computers.

[another_someone]

If indeed, as is suspected, that the brain is connected as a neural network, then it is indeed the case that memory is scattered, like a hologram, across the whole brain.

If that is the case, then I suspect it will be a matter of all or nothing (i.e. you copy an entirety of a persons memory, or none of it; but you cannot selectively copy just a few of them - and in copying the entirety of a persons memory, you will have to supplant the existing memories, for they would interfere with the new memories).

[DoctorBeaver]

Neil - I didn't say that memories must be associated with sensory processing; I merely used that as an example of the interconnectedness of memories.

It is suspected, but not yet proven, that any given neuron can be utilised in more than 1 memory. The fact that 1 memory can trigger another indicates, in my opinion, that this may be the case.

And, yes, you're right that emotions come into it too. Remembering a good time can make us feel happy, memories of a bad time make us sad. That, in itself, raises the question of what else is involved in memory. The memory triggers a physiological response - certain chemicals are released that cause physical reactions. In order to effectively transfer the memory, we would have to cause the same physiological response in the recipient. Without that, the recipient would experience the memory in a robot-like fashion - totally devoid of emotion. That may be what's required for accelerated learning, but it's no use for transferring someone's life experiences into a new host.

[Bored chemist]

It's a whole lot easier to ask someone what they remember.

[DoctorBeaver]

It's a whole lot easier to ask someone what they remember.

That depends how drunk they were.

[coglanglab]

It is suspected, but not yet proven, that any given neuron can be utilised in more than 1 memory. The fact that 1 memory can trigger another indicates, in my opinion, that this may be the case.

It depends on how, exactly, memories are stored. Yes, current neuroscience models use neural networks, in which case each neuron must be involved in multiple memories, because there aren't that many neurons relative to the number of memories we have. Plus, that's just how neural networks work.

However, neural networks are provably finite state computing machines. That is, they aren't Turing machines. This also means that they aren't capable of many or most human cognitive activities. So the neural network model, at least as currently developed, is probably wrong.

The debate right now is between people who say, as I just did, "The human brain can do many things that require non-finite state computing. Since neural networks can't accomplish those tasks, there must be more to the brain than neural networks," and people who say, "We know of know possible neural mechanism to allow for symbolic processing such as in a Turing machine. So the brain must be composed of neural networks."

That's neither here nor there with regards to the original question, but I think it's an interesting point.

----
Please take a moment for science at http://coglanglab.org [nofollow]

[another_someone]

However, neural networks are provably finite state computing machines. That is, they aren't Turing machines. This also means that they aren't capable of many or most human cognitive activities. So the neural network model, at least as currently developed, is probably wrong.

I half understand what you mean by that, but am not totally sure if it is correct - or maybe only half correct.

Any CPU is only a finite state machine, and a Turing machine can only be such if it has some memory.  A Turing machine itself is just a finite state machine with memory.

We know that the brain has, at a low level, memory (i.e. it will change structure, and will alter its sensitivity to signals).  This is essentially the programming phase of a neural network - the difference is that most neural networks we build function in two distinct phases, a programming phase and an execution phase, and within the execution phase it acts as a finite state machine.  An animal brain on the other hand has both programming and execution functions happening continually, and so it is not in the kind of fixed configuration that would legitimately regard it as a true finite state machine (the neural system of an insect might be more so, but this is not the case with more complex brains).