Posting in Science
It seems the sci-fi industry has done it again. Predictions made in novels like Johnny Mnemonic and Neuromancer back in the 1980s of neural implants ...
It seems the sci-fi industry has done it again. Predictions made in novels like Johnny Mnemonic and Neuromancer back in the 1980s of neural implants linking our brains to machines have become a reality.
Back then it seemed unthinkable that we'd ever have megabytes stashed in our brain as Keanu Reeves' character Johnny Mnemonic did in the movie based on William Gibson's novel. Or that The Matrix character Neo could have martial arts abilities uploaded to his brain, making famous the line, "I know Kung Fu." (Why Keanu Reeves became the poster boy of sci-fi movies, I'll never know.) But today we have macaque monkeys that can control a robotic arm with thoughts alone. We have paraplegics given the ability to control computer cursors and wheelchairs with their brain waves. Of course this is about the brain controlling a device. But what about the other direction where we might have a device amplifying the brain? While the cochlear implant might be the best known device of this sort, scientists have been working on brain implants with the goal to enhance memory. This sort of breakthrough could lead to building a neural prosthesis to help stroke victims or those with Alzheimer's. Or at the extreme, think uploading Kung Fu talent into our brains.
Decade-long work led by Theodore Berger at University of Southern California, in collaboration with teams from Wake Forest University, has provided a big step in the direction of artificial working memory. Their study is finally published today in the Journal of Neural Engineering. A microchip implanted into a rat's brain can take on the role of the hippocampus—the area responsible for long-term memories—encoding memory brain wave patterns and then sending that same electrical pattern of signals through the brain. Back in 2008, Berger told Scientific American, that if the brain patterns for the sentence, "See Spot Run," or even an entire book could be deciphered, then we might make uploading instructions to the brain a reality. “The kinds of examples [the U.S. Department of Defense] likes to typically use are coded information for flying an F-15,” Berger is quoted in the article as saying.
In this current study the scientists had rats learn a task, pressing one of two levers to receive a sip of water. Scientists inserted a microchip into the rat's brain, with wires threaded into their hippocampus. Here the chip recorded electrical patterns from two specific areas labeled CA1 and CA3 that work together to learn and store the new information of which lever to press to get water. Scientists then shut down CA1 with a drug. And built an artificial hippocampal part that could duplicate such electrical patterns between CA1 and CA3, and inserted it into the rat's brain. With this artificial part, rats whose CA1 had been pharmacologically blocked, could still encode long-term memories. And in those rats who had normally functioning CA1, the new implant extended the length of time a memory could be held.
The next step is to test the device in monkeys, and then in humans. Of course at this early stage a breakthrough like this brings up more questions than solutions. Memory is hugely complex, based on our individual experiences and perceptions. If we have the electrical pattern for the phrase, See Spot Run, mentioned above, would this mean the same thing for you as it does for me? How would such a device work within context? As writer Gary Stix asked in the Scientific American article, "Would "See Spot Run" be misinterpreted as laundry mishap instead of a trotting dog?" Or as the science journalist John Horgan once put it, you might hear your wedding song, but I hear a stale pop tune.
We are provided with the same structural blueprint for our brains, but its circuitry is built from experience and genetics, and this is a tapestry unique to each of us. Something that many scientists feel we'll never be able to fully crack and decode, let alone insert into it an experiential memory.
[Via press release]
Jun 17, 2011
The experiment's implant duplicated a specific ELV signal (to replace one they negated with a drug). To call that the "Successful Implanting of an Artificial Memory System" is comical hyperbole. It's not even a 'system' per se; simply a low power electro-shock to one (pre-selected) hippocampal location. At least map the hippocampus before you call this science-fair demonstration an 'Artificial Memory System'. Arbitrarily blocking one neuro-synaptic path chemically, then activating it via externally-applied extreme low voltage electricity (exactly what was reported) is NOT an "Artificial Memory System'. False alarm, people----nothing to see here. Not yet, at least.....
Is anyone working on doing this the other way around? Rather than stuffing all this into someone's brain, how about figuring out a way that a normal person can 'plug in' to an external 'brain', and maybe do it remotely. I don't need to carry around with me the details of how to fly a jet plane, I just need it when I am flying a jet plane (or speaking certain languages, or etc.etc.). I would become really 'smart' when I was plugged in, and go back to 'normal' when not plugged in. And no one is 'messing with' my memories!
The description of the experiment was to train the animal for a specific task and then block part of the hippocampus and use the artificial memory for the same animal. It would be interesting to know if they were able to put memory recorded from one trained animal and put it into an untrained animal. This may be good for people with memory problems like loss of short term memory or the inability to form new memories as well as some stroke victims.
If a computer program maps our brain waves then in theory it would be possible to create a program which could decipher what we are thinking, which could lead to advances in many areas from interrogation to advertising. I suppose this may depend on the individuals willingness to participate or maybe not if brain waves are generic in nature.
(like your username...) but memory is substantially different, machine to animal. In machines (so far) the memory must be complete in every detail, or those details are lost. Animals encode the memory using more or less a "least energy" strategy. That is, the biased nerve connections actually remember the most (subjectively) important details, but require many, many "accessory" channels, stored as trace adjustments through the network; using those channels to "refocus" the details during the process of recall. In other words, the "facts" we remember are only the tip of an iceberg that is rebuilt during recall. If recall is put off for too long, or if other memories are "too" similar (and therefore use the same networks), the quality of the memory is degraded, as it can no longer be reliably rebuilt when needed. The benefit of this tech would be a form of memory that would "never" degrade, that would be as sharp as it ever was, for as long as the chip reliably kept the values intact. If it used the same strategy as organic networks, however, storing ONLY the most important parts, then the memory could over time become "other"-ized; that is, it might come to seem as though it did not belong in your own network, as its accessory fractions in the larger, organic network, might degrade completely away. And I'd expect the same thing as well, the first time new (downloaded) memories were used; as at that point, the extended network is not yet built (the use of the core memories is what (re-)builds the extended memory network). In this conception, the actual use of the memories would be what kept them feeling current -- same as organic memories behave.
- Why not? Memory is memory. There are computer programmers who understand how the mind works better than a lot of brain specialists, precisely because of their familiarity with memory issues in computers. It's pretty much the same stuff, just wired a bit differently.