Is it theoretically possible to imitate the human brain on a computer?

Question

Our brain consists of billions of neurons, which mainly work with all incoming data from our senses, process our consciousness, emotions and creativity, as well as our hormonal system, etc.

So, I'm completely new to this topic, but every neuron does not have a fixed function? For example: if a strength signal x comes in, if the last signal was x ms ago, redirect it.

From what I learned in biology about our nerve system, which includes our brain, because both are composed of simple neurons, it seems to me that our brain is one big, complex computer .

Perhaps it is so difficult that things like intelligence and cognition become possible?

How are the most complicated things about a neuron to a large extent the chemical aspects of generating an electric single, saving its life and ultimately segmenting itself, should it be pretty easy to imitate some on a computer, or? You don’t have to worry about keeping the virtual neuron alive, or?

• If you can emulate a neuron on one on a computer that shouldn't be too complicated, can you theoretically emulate more than 1000 billion of them, recreating intelligence, cognition, and possibly even creativity?

In my question I leave the following aspects:

• The speed of our current (super) computers.
• Actually write a program for emulating neurons

I don’t know much about this topic, please tell me if I have something wrong :)

(My secret goal: make a copy of my brain and save it to a 10 millionth hard drive with a hard drive and run it in the future)

Answer 1

A neural circuit can be built with multiple transistors. Let them say that on average about a dozen transistors. (See http://diwww.epfl.ch/lami/team/vschaik/eap/neurons.html for an example.)

The brain chain will require 100 billion of these neurons (more or less).

This is 1.2 trillion transistors.

The Itanium Quad-Core has 2 billion transistors.

You will need a server rack with 600 quad-core processors the size of a brain. Consider buying servers worth $ 15 million. You will need power management and cooling plus real estate to support this mess.

One major problem in brain modeling is scale. The actual brain dissipates only a few watts. Power consumption is 3 squares per day. A pint of gin. Maintenance - 8 hours of downtime. The property is a 42-foot sailboat (22 tons of volume measured by ships) and a place where you can drop the hook.

A server cage with 600 quad-core processors uses a lot more power, cooling and maintenance. This will require two full-time people to support this brain-sized server farm.

It seems easier to just teach two people what you know and skip equipment investments.

Answer 2

Imitation of a neuron is possible and, therefore, theoretical modeling of the brain is possible.

Two things that always haunt me as a problem are entry and exit.

We have a lot of nerve endings that provide input to the brain. Without them, the brain is useless. How can we simulate something complex, like a human brain, not only simulating the entire human body!?!

Conclusion, as soon as the brain "cope" with all the inputs it receives, what does this mean? How could you say that the “copy” of your brain was actually you, without re-linking it to a real human body that could speak and speak to you?

All in all, a fascinating subject !!!!

Answer 3

Roger Penrose presents the argument that human consciousness is not algorithmic and, therefore, cannot be modeled by the usual computer type of Turing computer. If so, you can forget about creating a brain using a computer ...

Answer 4

The main problem with simulating neural networks (and the human brain is a neural network) is that they operate continuously, while digital computers operate in cycles. Thus, in a neural network, different neurons function independently of each other, and on a computer you only simulate discrete states of the system.

The fact that adequate modeling of real neural networks is very problematic at the moment, and we are very far from it.

Answer 5

Researchers are much smarter than most, therefore, see Blue Brain from IBM and others .

The Blue Brain project is the first comprehensive attempt to reconstruct a mammalian brain in order to understand brain function and dysfunction through detailed modeling.

Theoretically, the brain can be modeled using a computer (because software and hard / wetware are compatible or mutually expressed). The question is not theoretical, since computer science is on, but philosophical:

Can we model a (chaotic) way of brain development. Is the brain a powerful equipment or an environment that shapes the development and emerging properties of this equipment when it learns

Even smarter:

If I, with 100% accuracy, simulated my own brain, then I started the simulation. And this brain had my memories (since I had my physical form of the brain) ... is it me? If not, what do I have, what is wrong?

I think that if we ever imitate the brain, we should work on a logical system based on biological principles with better applications than the brain itself.

We all have a brain, and we all have access to it, tremendous power already;)

Answer 6

I do not think that they are remotely close enough to understand the human brain, even to start thinking about replicating it.

Scientists would think that we are almost there, but with respect to the brain we are not much further than Dr. Frankenstein.

Answer 7

What is your goal? Do you want a program that can make smart decisions or a program that provides a realistic model of how the human brain works? Artificial intelligence can be approached from the point of view of psychology, where the goal is to simulate the brain and thereby better understand how people think, or from the point of view of mathematics, optimization theory, decision theory, information theory and computer science, in this case the goal is to create a program capable of making intelligent decisions using computational efficiency. The latter, I would say, has been largely resolved, although success has not yet been achieved. When it comes to realistic brain modeling, I think we have only recently been able to simulate a cat's brain semi-realistic; when it comes to people, at the moment it would not be very appropriate.

Answer 8

Yes, the Blue Brain Project is approaching, and I believe that Moore Law has a $ 1,000 computer that gets there by 2049.

The main problem is that our brains are based mainly on the control of the human body, which means that our understanding and production in the language, the basis of our high-level reasoning and recognition of semantic objects, depends heavily on its potential and practical results for the larynx, language and facial muscles. In addition, our reward systems are tied to signals that indicate nutrition and social approval, which are not goals that we usually want in brain AI.

Exact modeling of the human brain will be useful in studying the effects of drugs and other chemicals, but I think the next steps will be to isolate the ways that will allow us to do what is difficult for computers (for example, the visual system, fusiform gyrus, face recognition) and the development of new or modifications of known structures to represent concepts.

Answer 9

In short: yes, we can certainly reproduce artificial brains, but no, maybe it won’t be with our modern computers (Turing machines), because we just don’t know enough about the brain to know if we need new computers ( super-Turing or biologically engineered brains), or if current computers (with plenty of power / storage space) are enough to mimic the whole brain.

---

Long

Disclaimer: I work in the field of neuroscience research, and I am interested in both the neurobiological side and the computational (artificial intelligence).

Most answers suggest, as a true OP postulate, that simulating neurons is enough to preserve the state of the entire brain and, thus, simulate the whole brain.

This is not true.

The brain is more than just neurons.

Firstly, there is a connection, synapses that are of paramount importance, perhaps even more than neurons.

Secondly, there are glial cells, such as astrocytes and oligodendrocytes, which also have their own connectivity and communication system.

Thirdly, neurons are heterogeneous, which means that there is not only one template model of a neuron that we could scale to the required number to simulate the brain, we also need to identify several types of neurons and place them in the right places. In addition, types can be continuous, so you can actually have neurons that are halfway between three different types ...

Fourth, we know little about the rules for processing and managing brain information. Of course, we found that the cerebellum works pretty much like an artificial neural network using stochastic gradient descent, and that the dopaminergic system works like TD learning, but then we have no idea about the rest of the brain, even memory is inaccessible (although we assume that this is something close to the Hopfield network, but there is no exact model yet).

Fifth, there are many other examples from current research in the field of neurobiology and computational neuroscience that show the complexity of the dynamics of brain objects and networks, which this list can go on and on.

So, in the end, your question cannot be answered, because we just don’t know enough about the brain to find out if our modern computers (Turing machines) are enough to reproduce the complexity of biological brains to bring about a full range of cognitive functions.

However, the field of biology is getting closer and closer to the field of computer science, as you can see with biologically engineered viruses and cells that are programmed in the same way as you design a computer program, and genetic treatments that basically process a living system based on its template is a "class" (genome). Therefore, I dare say that when we know quite well about the architecture and dynamics of the brain, reproduction in silicone will not be a problem: if our current computers cannot reproduce the brain due to theoretical limitations, we will develop new computers. And if only biological systems can reproduce the brain, we can program the artificial biological brain (we can already 3D-print functional bubbles, skin and veins and hearts, etc.).

Therefore, I would dare to say (even if it may be inconsistent, here is my own statement) that yes, artificial brains will undoubtedly be possible someday , but will it be like a Turing machine a computer, a Turing super-simulator or a biologically designed the brain remains to be seen depending on our progress in knowledge of brain mechanisms.

Answer 10

A word of caution. Current brain modeling projects are working on a human brain model. Your idea of ​​keeping your mind on your hard drive crazy is: if you need a copy of your mind, you will need two things. First, another “empty” brain. Secondly, develop a method that allows you to fully transfer all the information contained in your brain: up to the quantum states of each atom in it.

Good luck with that :)

EDIT: The dog ate part of my text.