Theoretical Question - Memories Question 3

[dimumurray 0]

MEMORIES- I am a computer scientist so I really have no medical background but does anybody really know how our memories work? How they are stored in our minds. The amazing part is that they are the only record we have of our past. I read somewhere that there is not a cell in your body that was there 7 years ago. Your body constantly makes new cells and gets rid of the old ones, but if that is true what links us to ourselves. Are our memories the only thing that makes myself me? Well sure I guess there is DNA but the only way we can see that is our appearance, which changes anyway. What if you somehow found out that what you remember was not what you actually did. What if your memories were a lie?

 

So this brings me to my ultimate question for this thread....

What if you were trapped in the same time lapse for your entire life but you didn't know it. What if your life is only a few hours long and after those hours are up then your mind was restored how it was two hours ago and then you relive the same two hours again without realizing it. And everything you remember from before those two hours never really happened, they are just false memories... I know its a stretch but I find myself thinking this a lot especially when I am really looking forward to something. I think "what if I never get to that point in time? What if I am constantly reliving the hours before it and never experience it?

I too, am a computer scientist. I don't know how far along you are in your studies, but in time you'll find -- with or without a medical background -- that ours is one of the few disciplines that draws upon and in turn influences all the other branches of science. We create abstractions of the world around us. Many of these abstractions find themselves applied to problems altogether outside of their domains potentially providing a more holistic view of science in general.

 

But I'm moving away from the focus of this topic.

 

Like I said before we create abstractions of the world around us. Of relevance to this topic are our attempts to model how the human brain functions. It is but one of many promising approaches to develop truly intelligent systems. The holy grail of AI research.

 

One such model are neural networks. There are special cells, called neurons, that make up the brain and are thought to be responsible for memory and learning. Each neuron shares several connections with its neighboring siblings. Some connections act as inputs streaming in from other neurons. Other connections ferry outputs to yet other neurons.

 

Here's a sample scenario. Imagine several neurons firing sending a multitude of tiny electrical charges along the connections (synapses) into a receiving neuron. The length and thickness of the synapses determines the speed/rate and strength of the charge and therefore its net influence on the receiving neuron. The net charge of all the incoming inputs may or may not cause the neuron to fire. Each neuron has its own threshold which when breached causes it to "fire", that is, releasing a charge of its own. The strength of this charge is also variable depending on the neuron. Now imagine billions of these little neurons, interconnected and firing away; that is your brain in action. The brain's neural network also exhibits a kind of structure. That structure gradually changes as new knowledge and learning is acquired over time.

 

Even though neurons go though the typical life cycle of any other cell the impact of cell death and replacement (except in the case of massive brain damage) on structure is minimal. For the systems that scientist use to model neural networks, artificial neural networks, seek to maintain an optimal structure for a given problem domain.

 

In term of computer science the neuron is modeled as a system of inter-connected nodes with a number of weighted inputs and outputs with special bias values that help shape the outcome. The inputs are calculated via a weighted sum and if this sum exceeds the node's threshold output is produced.

This digital analog of the brain can be 'trained' on target data and then let loose upon more random input.

 

Now my thinking is this. If an artificial network can be trained to act on data in a specific way, then I do not think it too far fetched that sometime in the future we will be able to scramble the structure of a person' neurons. First art imitates life and then, coming full circle, life imitating art. There are obviously some scary negative and amoral consequences of such a technology. It would give the term 'behavioral modification' new meaning.

 

But hopefully that's a long way off (fingers crossed)...

 

Here's a link to some slides I made on the topic of Neural Networks while taking an AI course in college:

Neural__Networks.ppt

 

It should elucidate some of the precepts of the technology and its origins.

Edited August 5, 2008 by dimumurray (see edit history)