It’s been a while since I’ve written anything, but a post on reddit about memory got me thinking, so I put down some thoughts about how memory works, and memristors!
* * *
I think the reason you have nobody answering is that there is no definitive answer to your question. Memory and consciousness are the big unknowns in philosophy and neuroscience, endlessly debatable and almost impossible to prove. It’s all just theories, which are little more than mental ‘what ifs’.. However, I’ll take a stab and tell you my theory.
So our brains, the most complex object that we know of in existence, are just neural networks of neural networks of neural networks. Think about how we learn, from babies to adults. I imagine that a baby’s brain is a mass of unconnected and pre-mapped out neural networks. Some of these networks will have been formed as a result of our DNA, things like controlling your breathing and heart-rate to the medulla, the thalamus and hypothalamus, probably pre-wired to route information to and from different areas of your brain. The cerebellum is probably hard-wired to control your limbs, but all of these things are connected via pathways to the cerebrum, which I’m betting is mostly unmapped.
So you have a baby, he’s born, he’s basically raw-unmapped sensations, stimulation coming in from all over the place and probably no way to process it properly. All this baby knows is that things are hot, things are cold, he’s hungry, he’s tired, he’s warm, he’s safe, he’s scared, just these raw emotions that are built in for survival. Once he starts growing, he starts learning. He sees that if he moves his hand, he can see his fingers, and if he wiggles his fingers, they move. You start getting these feedback networks getting burnt into the cerebrum, this cause and effect portions of the brain – if he cries, there’s comfort, he see’s his mom, she tickles him, little networks getting formed in the cerebrum. As the child grows up, more and more actions and stimulus are getting defined in the cerebrum. Walking, running, making sound, tying the sounds together to make speech, networks building upon networks. As the child is learning, he’s starting to chain these networks together. Associations between networks. Saying something that’s unintentionally funny to all the adults in the room, and watching them burst into laughter. Important events are getting written into the cerebrum, things that are different are getting saved for future use.
Now as he grow up, and his consciousness is forming and solidifying, he starts to see associations with stuff that happened in the past. He see’s a child saying something funny and everybody laughing, and that get’s hashed back to the time when he was a child doing the same thing. Whatever part of his brain that has evolved to be his consciousness rolls back over those networked connections that were saved in his cerebrum, and it’s played out again in his internal movie theatre guy. What exactly the nature of that internal movie theatre guy is, that’s another question and much more philosophical / spiritual.
Anyways, that’s my interpretation of how memory works. Whenever your consciousness is rolling back over the network that formed a different or important or rewarding point in your life. Most of your day to day stuff seems to be ignored, because it’s not eventful, and it doesn’t change the who of you are.
As for memristors, yes, I’m pretty sure that if they actually get developed, we’re going to see a whole new level of processing. They really do seem to be made for neural networks, however keep in mind the physiology of a computer neural network and a wetware neural network are very different – with our neurons, you have tens if not hundreds of incoming connections to the dendrites of a neuron, which is then in turn connected to hundreds of other neurons. Memristors are single input single output, but I bet that you could possibly emulate the same behaviour using a bunch in parallel or something. Anyways, that’s my two cents.