LONGECITY

The melding of flesh and silicon is taking shape as scientist at the Georgia Inst. of Technology develope a technique to give "life" to "thousands of rat neuron cells."
--BJK



Rat-Brained Robot

By David Cameron December 18, 2002

Rat neuron cells on silicon are the brains behind a new robot—a breakthrough that may lead to better computer chips.

Steve Potter’s brand new robot would probably never make it to the second round of Battlebots. The size of a coffee mug, the cylindrical robot slides across a round meter-sized playpen on an apparently chaotic path. But this robot is a thinker, not a fighter, and it does its thinking with a network of neurons—culled from rat embryos—that resides a few feet away on an electrode-activated silicon chip.

Article

I'm one of Potter's grad students. Let me know if you have any questions.

What a fun place to work!



Do you have an estimate on the progress of increasing the number electrodes from 60. Are you doing better now?

Do you see more problems with a larger system with more electrodes, or will it eventually take over and mend and grow on its own?

Here's a follow up paper from gatech.

Neurons from embryonic rat cortex spontaneously connect and form intricate patterns of electrical activity when grown in culture. We and others studying neuronal cultures call stimulation-induced changes in this activity "synaptic plasticity." Learning is usually defined in behavioral terms, as a change in behavior as a resulting from experience.
Link: http://www.neuro.gat...orts/Animat.doc

In response to Bruce's questions:

[?] We're going to stick with the 60 electrode chip for now until we get the kinks worked out. We could easily add more electrodes (there is a 128 model already commercially available) but we still don't understand all the signal coming out of 60 channels. We are currently working on things such as:

[>] increasing our signal to noise ratio
[>] understanding how to recognize patterns in ~10,000 neurons
[>] figuring out what good stimulus parameters are for this size network
[>] building our visual recording system (fluorescent neurons that change color when they fire!)

We just got a kick ass microscope system installed. I'm building an environment (temp, humidity, CO2) chamber for it so we can do long term stimulation with electrical and visual recording simultaneously. This will help us answer alot of the questions we need to attack now.

[?] Our biggest problem with more electrodes is sorting out all the data in a reasonable amount of time. Right now, we stimulate and record from the neurons for a limited period of time (~2 hrs). Then we spend weeks offline analyzing all the data to try and figure out what it all means.

We need better math models and software for crunching all of it. I would like to apply the work that has been done in the communications industry, in emergent systems math, and in nonlinear dynamics (chaos math) to building some of the models and software. My fantasy is to learn how to send signals to (and program) networks of neurons and to be able to decode their output in real time. This is what will be needed for a robust neural implant that you can both send and receive from.

Best Stuff,
Peter [B)]

This is the lab and equipment where Peter and crew are working at GA Tech(Correction: This lab is Steve Potter's Lab at Caltech) They had to make a camera to film the neurons growing in vitro. http://www.neuro.gat...tter/HSCCD.html

Here is a link to a couple of Timelaps Movies of the neuronal cells in action. Peter answered many questions about his work during the Sun Jan 19, 2003 ImmInst Chat



Living rat hippocampal neurons sending processes out from wells of a Neuroprobe.


http://www.neuro.gat...er/2phPics.html

Hi Bruce and Co,

Acutally, that lab pic above is Steve Potter's Lab at Caltech, which is still up and running too. I'll post some pictures of the new lab soon.

Peter

Ahh, thanks.... I added a correction.

this resarch was also discussed during the

[>] Imminst chat event January 19th, 2003