eli5 why brain memory is imperfect but computer memory is near perfect


eli5 why brain memory is imperfect but computer memory is near perfect

In: 0

Brain memory works with neurons. Basically a huge spiderweb of “weighted” connections. The more you recite/strengthen something, the more weight you give to a certain pathway, making that whole chain of connections stronger. That’s also why association and understanding with memory is so powerful. You can find connections to those chains via related things. It’s more a puddle of gradients.
Computer memory uses binary bits with only the most extreme weights of 1 and 0. Either on or off. So you can store 10010110 as a finite determined block of 8 on and off states. These are very easy to manipulate, move, and maintain since the margin of error for telling if a light bulb is on or off is almost nonexistent.

Brains optimize themselves all the time. Connections that you use a lot are reinforced, connection that you don’t use a lot will slowly diminish and be removed.

That’s important for efficiency, it makes the first thing that comes to your mind more likely to be relevant. Brains work by connections/associations, unlike a computer you can’t search the whole memory for a specific thing, you have to reach it through things it’s connected to.

Computer memory is made by having a physical thing which represents something. Human brains remember things through association.

Binary is how we store information on computers. This is because its very clear and easy to represent binary. Binary has two states 1 and 0. So to physically represent something in binary you only need something that has two states. Like a light being on or off, or a hole being in a piece of paper. Nowadays we store information on special electrical components which can hold charge for a very long time without being powered, but the princable is the same. The only way computer memory becomes imperfect is if that physical representation of the thing changes. In this way it’s sort of like writing something down on a piece of paper in a different language. Unless something ruins the paper or changes what’s on it, the papers ‘memory’ is perfect.

Where as human brains remember things through association. Neurons link and connect in complex patterns and your brain is just good and know what neurons normally fire together. So when we experience something certain neurons fire to represent that experience. When we recall something those same neurons fire recreating the experience. But obviously there is a difference between remembering something and experiencing it, this is because the exact same neurons don’t fire. Without reminders over time less and less of the same neurons fire and the sensation we recreate is different from what the experience first was.

Computer memory is more analogous to a pencil and paper than to a human brain.

If you write something on a piece of paper, then the paper’s “memory” of what was written is perfect. Barring physical destruction of the paper, or using an eraser to get rid of the writing, the words will persist in the same form forever. This is why we can read things that were written thousands of years ago – as long as we still know the alphabet and the language.

So instead of thinking of a computer’s memory as somehow being like a brain, try to think of it as being billions and billions of sheets of paper, each of which will either stay with the value it has forever (or until the computer is switched off, at least), or else the computer metaphorically erases the contents of that sheet of paper and writes something else on it.

Computer memory uses a different address/transistors on the memory chip to store each piece of data, so storing something new doesn’t interfere with what’s already there.

Brain memory doesn’t work that way – there isn’t a separate neuron (or group of neurons) for each memory. Instead, groups of neurons are used to collectively store multiple memories and new memories can therefore interfere with old ones.

The computer-brain analogy isn’t very helpful – they are really very different. A computer has separate components for processing and memory, but in our brain both functions are combined – we don’t really have separate “memory” as such, but rather just a processor (our cortex) that is modified by experience to reflect “memories”, which is why it works the way that it does.

If you wanted to have a *slightly* better computer analogy of how brain memory works, then something like a computer hash table might give a better idea… there isn’t a separate location for each thing to be stored, but because most hash codes are different, there usually aren’t any clashes. This is comparable to how our brains use groups of neurons and sparse/embedded representations to store things.