October 23, 2004

• Time to fire all the pilots. "Who needs you? A dead rat could do your job!"
• Un-freaking-believable!
• Anyone remember Cordwainer Smith? Wish he were alive to read this.
• OK, so we figure a way to implant a rat brain in an automobile then we can keep the cell phone jammed in our ear 100% of the time.
• sTep 2: dissEMBoDIed raTt NuEronz laern HOw 2 posTT 2 MunkIFilTer!!!! SoooN th wORLd WiLL be OUWRZ!!!!
• o puppy-brained man forgo your fore-doomed howling! somewhere a cat-brained girl with deep green gaze is soft-paw prowling
• cool- it's interesting that he can keep them alive long enough to learn how to control that feedback loop between the MEA and the aircraft simulator. I wonder how many cells are interacting with the MEA electrodes though (they show a standard 64 element array in the image, vs 25000 neurons plated) and which electrode/s they choose. I'll have to read the journal articles.
• I once upset a helicopter pilot by suggesting that he could be replaced by a dragonfly's brain. Mind you, in his case, the conversational skills would have been better too...
• Fantastic again, beeswacky!
• Thanks for the kind words, chimaera.
• "And as the data comes in, it slowly modifies the (neural) network so over time, the network gradually learns to fly the aircraft." I'm trying to understand something. Polychrome, does the network 'think' its a plane? This is a goofy question but I don't know the right terminology.
• Ah, Bees. *sighs softly
• *heart lurches and thumps*
• PAtB- haven't had a chance to read anything else about it yet, so I'm going out on a limb here but..... what I think is probably happening is that the neurons are, over time, being trained to act as a series of feedback circuits. The way I suspect it works is that a deviation from level flight, a pitch or a yaw, is translated into an electrical impulse which is delivered to one (or more) of the electrodes. This will stimulate the neurons in the local area, resulting in a series of action potentials (the nerve cells firing) which will, in turn, stimulate other neurons around them. Unchecked, this would result in a whole series of uncontrolled electrical impulses. But some of the cells in the culture will be inhibitory, and so gradually the cascade of electrical activity will dampen out. Or it would, if the electrode wasn't still active because the plane was out of control. So the next bit is to hook up another electrode/s to the simulator's controls, completing a _possible_ feedback loop. The trick is to keep the cells alive for long enough so that they discover that if this electrode fires, they can make it stop by tickling that electrode over there with electricity. Yes, the neurons will rearrange their electrical connections, given enough time, into a feedback ciruit that will do that. They don't know anything, in any real sense, you see, but just doing what nerve cells do, which is build and destroy electrical connections to keep themselves in as electrically stable a configuration as possible, will result in a feedback loop which can keep a plane steady.
• The trick is to keep the cells alive for long enough so that they discover that if this electrode fires, they can make it stop by tickling that electrode over there with electricity. There. That's the sentence I needed right there. I didn't really think the little suckers 'thought', per se, but I didn't know how else to put it. Thank you, polychrome I'M tickled at the thought of electrodes being tickled by rat brain cells.
• This is old news. Didn't we elect one of these president? I kid, I kid.