Cyberbrain Taskforce

edited February 2015 in Everything else
Let's build a Cyberbrain.

Recently on irc, there was a discussion going on about scanning of neurons and their connections. This works well in 2 dimensions at the moment (like a "slice" taken from a brain). The question that popped up in my mind was:"if neurons transmit information with 120m/s ... would simulating the same network on electronic-based hardware with a transmission speed 3mio times faster than that effectively turn reality into super-slow motion ?"
This lead to the idea of scanning an entire network of neurons, including the way they are connected, building a simulation model from that and running that model on a transistor based system. This effectively would be, building a cyberbrain.

The big steps to accomplish:
1. Scanning of the neural system.
2. Process the scanned images to map out the connections between neurons
3. Building a model that simulates the original body in a virtual world.

Since this is not something that can be done by a simple person, i wanted to survey if we can combine our knowledge, available tools and experiences in order to build a Cyberbrain Taskforce.

The list of skills needed would be (probably not completed):
-biology/neurology ( to help preventing damage of neurons, and to help with the simulation feedback)
-microscopy ( sample preparation, microscopic systems available, working mechanisms, limits ...)
-mechanical engineering and process automation ( i can contribute a bit here)
-lasers ( especially excimer might be good for blasting off thin layers from samples)
-image processing
-dynamic simulation of mechanical systems, building mathematical models for the virtual body ( i can also help here)

so long story short: the plan is to kill a living creature (a roundworm for starters, 5000 synapses sounds reasonable to simulate ), slicing it into thousands of thin pieces (or grinding it off layer by layer), reconstructing the whole neural network into a simulation, and letting the creature live in a virtual space and see how it behaves so the model can be improved. upon success, repeat with a more complex organism.

besides the awesomeness factor of transforming a real world organism into a virtual one and the usualy transhumanistic stuff like immortality etc. there are a couple of great questions that arise. like the preception of time by altering the clock-speed of the virtual environment. or what happens if you meet a copy of yourself. or how neural growth can be modeled which allows learning and "not being stuck in your brain".

many questions and challenges. i summon you, internet! 

ps: i'd vote for running the roundworm simulation on the server with a live-stream of his simulated environment so you can interact with it. cyberpet so to say.

[edit:] very related


  • I read the book Connectome recently. I strongly recommend it if you're interested in this.

    There are academic neuroscientists working on this already.
    - Kenneth Hayworth
    - EyeWire
    Depending on how serious you want to get, you should probably contact them at some point.

    And there are some free neuron modeling programs available online, NEURON and GENESIS.
  • Your summons have been answered. Are you familiar with the work of Randal Koene?

    substrate independent minds. kinda related.

    I probably wouldn't be helpful....unless you need some brains. I'm damned supportive though, and can try to recruit in whatever area we need. I do know that my friend Jimmy and I had discussed this at length a while back and he had devised a way to reprogram a virus to create carbon fibers within a neuron. Then bacteria would be used to replace the myelin sheath with silicon. After that you could peel the nerves away one by one. Not sure if that is useful.
  • I have no idea if this skill would be of any help, but I can program decent artificial neural networks, and I'm learning more about that as we speak (well, type).
  • so.. summarized, many people are working on it, there's still no DIY sollution, and no one has a good solution to pull of the actual microscopy thing.

    microscopy aside, there seems to be a number of useful projects going on so, scanning and reconstructing the network is exactly what's lacking.
  • I have access to a lot of 3T fMRI time. There is some mention of MR as a technique on the openconnectome site...

    I don't think it'll be directly useful for the roundworm project (as such) based on the voxel size we typically work with... but maybe DTI could be used as a macro QA (sanity check) process for any eventual microscopy results?

    In any case, I am strongly interested and will volunteer where possible. I have a basic familiarity with common image processing software used in functional imaging (SPM) and access to a lot more expertise through my labs.
Sign In or Register to comment.