But you've got some damned serious brainpower backing the alternative position, and I really don't think that could happen if the T-test was so pathetic that a group of freshman college students could rip it apart.
It really depends what you're talking about. I do think the Turing test is pretty pathetic when taken out of context, and a group of college freshmen certainly could tear it apart. Basically, Turing's test says that humans are like black boxes with input and output. If you have another black box that gives the same output to a given input, you have to admit that black box #2 displays consciousness and all that because you do that for people all the time. It's hard to debate the truth of this because it's almost more of a statement defining what consciousness and sentience are as opposed to saying anything about them. However, if you take the "strong AI" position, you would argue that replicating human behavior means that you have replicated the behavior of the human brain. The counter-argument to this is that the two black boxes could have two vastly different things going on inside them that give indistinguishable answers. Strong AI is largely what Searle's argument is directed against, but he takes it too far. I'm a computational neuroscience grad student, so basically I think strong AI is so obviously incorrect that it's laughable (and I think there are very few today who would disagree with that). However, if you take this argument down to a lower scale and say that you have a bunch of black boxes that are each the input-output equivalent of a single neuron, and you hook those together exactly the same way that real neurons are hooked together (there is no experimental technology that allows you to do this, BTW), you could say that you've replicated human consciousness and intellect.
Research and even a few things that made it to mass-consumption have shown that rapidly flashing or rythmically scanning a "safe" (read: low- power) light can be dangerous.
It's true that flashing lights can cause seizures, but those have to be flashing extremely slowly relative to video displays (something like 4Hz). This is a different situation than scanning, which is what goes on in a CRT. If the refresh rate is high enough, the photoreceptor cells in your retina can't tell that it's different from a continuous light source. LEDs are certainly fast enough to implement insane refresh rates--the limitation lies in the scanning apparatus itself (doing the faster of the scan axes). Doing it mechanically with a vibrating mirror (like it sounds like they're doing) can be difficult because at high speeds the mirror will want to oscillate sinusoidally, which can be corrected in various ways but would be a pain in the ass. I'm not sure what the limitation is on how fast they can be used without that type of correction, but there are more exotic things like acoustooptic deflectors you can use that don't have these limitations.
Someone correct me if I'm wrong, but I think it's supposed to be partially in reference to an interview with Timothy McVeigh, when he called the kids he blew up collateral damage.
That article raises a couple of valid points and several misleading half-truths. I should say that this is not my specialty, but it is something I know a bit about. First of all, no one claims that pseudogenes are entirely non-functional, only that they don't code for functional proteins. They can contain regulatory sequences for other genes that are still functional. The "evolutionary" argument that they should have been taken out of the DNA is misleading because it's really only true for single-celled organisms. There are three reasons (that I can think of) this is false for mammals. First, kinetics. It would take a lot longer for a gene to be excised than it would to be pseudogenized. Just because it hasn't happened yet doesn't mean it never will. Second, remember that pseudogenes are not necessarily useless, they just don't code for functional proteins. They can still contain regulatory elements absolutely critical to the survival of the organism. The third and most important reason is that it is not evolutionarily advantageous for organisms like humans to have genetically unstable genomes where things like excision of random chunks of DNA happen frequently. It is for some organisms--HIV is a good example. I forget the exact numbers, but something like half of HIV viruses floating around in the blood are actually infectious, and the reason why is because the rate of mutation is so high. However, one virus can have thousands of offspring in hours, so the high rate of mutation doesn't prevent it from replicating and is actually advantageous to the virus (and disadvantageous to HIV patients). Viruses use their genome extremely efficiently, and you would never find pseudogenes there. However, humans replicate slowly, and if our genome is unstable it leads to cancer and death, so the DNA is actually not used very efficiently. Like a hard disk that's 10% full and never gets defragmented, there will be clusters around containing deleted files, or pseudogenes. The tradeoff is that it's very reliable. Anyone who understands enough molecular biology to write useful articles on things like pseudogenes knows these things.
In any case, this is all irrelevant because mammalian olfactory receptor genes have certain special traits that make them exempt from these arguments. Think about some random genes. They've got some homology to other things, there might be two or three different genes for fairly similar enzymes or whatever that have maybe 50% sequence identity. In mammalian OR genes, you have about 1000 of them that are practically identical, with a couple of hypervariable regions that code for the binding pocket of the odor molecule. In addition, they all have no introns, meaning that each of them is one contiguous piece of DNA. In the article it says that you can't rule out translation to useful proteins by sequence information alone. There are certain far-fetched cases in which this could be true, but this is clearly not the case for OR genes for the reasons discussed above. If there are premature stop codons or frameshift mutations, that gene will not make a functional protein, and we can tell that it isn't "meant" to be that way because we basically know what the gene "should" look like since there are nearly 1000 of them that are nearly identical. To be fair, there are people out there trying to make all kinds of crazy arguments based on weak data from cross-species sequence homologies or pseudogenes or whatever, but this particular example is pretty much unassailable on those grounds.
As you correctly pointed out this is not macro-evolution. No information was gained.
Well, in a sense this is as close to direct evidence of macroevolution as one could ever get. It's vestigial DNA from when we were a dramatically different species. I guess I should rephrase that and say that if you believe in microevolution (which is directly observable), and it's easy to see how pseudogenization of that type would occur by microevolutionary means, this is evidence of microevolution occurring on a macroevolutionary time scale. So I should change my earlier claim and say that this is evidence for macroevolution.
As to your reference as to why we have so many pseudonized genes, are geneticists truely certain those genes are pseudonized? There is still alot of work to be done in the field, infact we have barely scratched the surface.
Yes, they are pseudogenes in that they do not code for functional proteins. That we have barely scratched the surface is a common misconception--at this level of analysis, the questions are fully understood in directly experimentally demonstrable ways that no one can argue. Comparing the problem of computationally finding new genes (as in the # of genes in the genome) to finding new members of a large gene family with high sequence homology is like comparing a needle in a haystack to a needle in a box of toothpicks. As far as the number of genes in the genome goes, we still don't know how many there are, and people in the biology community weren't really that surprised by the (preliminary) results, contrary to what the press latched onto. No offense or anything, but you're obviously not a molecular biologist, and your questioning the methods of the experiments and molecular biology in general is analogous to saying that the speed of light could never be measured since it's just so unimaginably fast, timing circuitry and lasers still aren't really understood, and it wasn't that long ago that physicists thought that light propagated through ether with infinite velocity. There's no point questioning the data. Your closing statement is very interesting, but I don' understand how you reached your conclusion. Nonrandom pseudogenization argues for a highly SELECTIVE process. Two good possibilities would be intelligent design or evolution (but not nonselective genetic mutation, which could cause these changes on a microevolutionary time scale). The question is, why would an intelligent designer create nonfunctional genes? The only answer I can think of is to create the false impression that evolution had taken place. The alternative explanation, of course, is that evolution actually did take place.
There is no single thing that proves evolution, but the more biology you study, the more interesting facts pop up that simply have no plausible explanation except evolution. Look for my post regarding pseudogene formation in human olfactory receptor genes (grep for lukesl) and see if you can find an alternative explanation. You'll probably find it interesting because it's strong quantitative bioinformatics-based evidence for descent of humans from a non-human organism. Taking it for what it is, one tiny piece in a big puzzle, evolution provides a beautifully clear explanation of the data in a way that no other explanation I'm aware of can. Right now I'm in med school and grad school for neurobiology, and things like that pop up every week or so. As far as I know, I've never met a biologist who even thinks that evolution requires further proof. We put it up there with "cigarette smoking causes cancer," another difficult thing to prove, yet the preponderance of evidence is simply overwhelming. In our society it's hard to imagine people who don't believe that cigarette smoking causes cancer. For most biologists it's equally hard to imagine people who don't believe in evolution.
One other thing, though--they do not teach real biology in high school any more than they teach real physics in high school. If you haven't taken any college-level molecular or cellular bio courses that come after organic chem and biochem, you don't know what biology is (or at least molecular bio--for me, they're synonymous, but I know others probably disagree). Don't take this the wrong way, but when you say "But I for one reserve the right to doubt an idea like evolution, that if true would completely invalidate my world-view, without more evidence than we currently have," if all you're working with is high-school bio, no one's ever shown you the real evidence.
There's a paper in the latest issue of Nature Neuroscience from Stuart Firestein's group at Columbia that provides some really interesting evidence from a very different angle. Some background: mice are animals that rely heavily on olfaction, or their sense of smell. Over half their brain is dedicated to it. In "lower" mammals (or however you want to look at it), the sense of smell is also very important (dogs, cats, etc.) For humans, however, smell is not as important. We don't smell predators coming or track prey by scent; we use vision (and a huge portion of our brain is dedicated to it).
Anyway, In this article they do a rigorous analysis of the data on olfactory receptor (OR) genes from the recently acquired mouse genome compared to the data from the human genome project. I forget the exact numbers, but mice have about 1000 OR genes. Humans have about the same number, but something like 75% of ours are pseudogenized. Basically, this means they've been converted to pseudogenes, or sequences in the genome that obviously used to be functional genes but have mutated to a nonfunctional state. This much was known before. In this paper, however, they use techniques based on similarity of sequences to group the mouse OR genes into families and subfamilies. Then they group the human OR genes into the same families. To sum up what they found, if you were to take a random group of say six mouse OR genes, there will be five or six human genes that are the human counterparts of those mouse genes (over 90% of human OR genes have a mouse gene that's over 95% identical at the protein sequence level, and 77% have a mouse gene that's over 99% identical, so reliable identification is not a big issue). However, within that group of five or six human genes, all but one of them has been converted to a pseudogene. They find this over and over again. There's only one functional gene in each group. Each group, BTW, can be thought of as sensing when a certain class of feature is present on a molecule. In an analogy to vision, it would be like if mice could see different shades of six types of red, but we could see only shades of one.
Okay, here's an evolutionary explanation. A long time ago humans were monkey-like animals. Before that, dog-like animals, before that mouse-like animals, etc. Whatever animal we used to be, it was heavily dependent on a highly-developed sense of smell for survival (hence an entire 2% of our genome being dedicated to it--think about that). However, as we progressed evolutionarily, having an exquisitely sensitive and precise sense of smell became less and less important, but smelling things in general was still necessary. The genes mutated and mutated, but if the last member of a family became pseudogenized, that would compromise our ability to smell molecules with a certain class of molecular features (by analogy, if we couldn't see red at all), and those pre-humans would die. As a result, we're left with the HIGHLY nonrandom distribution of working genes. I want to point out that while this could be written off as "microevolution," consider two things: #1- all humans on earth will turn out to have >98% identical OR genes. #2- I say this because that nonrandom of a pattern with that many genes involved would take a LONG time to evolve, or at least a lot longer than humans have been on different continents. Almost certainly longer than we have been Homo sapiens.
Can anyone come up with a non-evolutionary explanation that explains 1) why so many of the genes are pseudogenized, 2) why the selection of which genes are pseudogenized is so highly nonrandom and optimized for the real world? I'm not asking for a critique of my evolutionary explanation, as the reasoning as I've presented it is not intended to be bulletproof. However, I do think that the underlying model is correct and I don't think there's any better explanation.
Re:Interesting press release
on
Think And Click
·
· Score: 1
Personally, I think the project has a low probability of success. A neural prosthetic device should be interfaced with as peripheral part of the nervous system as possible. This group has chosen to use as abstract a part of the nervous system as possible. But maybe they'll prove me wrong.
I think there's some truth in that, but the fact that they used posterior parietal cortex instead of primary motor cortex or the spinal cord makes the work that much more interesting for actually understanding brain function. I mean, if you really wanted a neural prosthesis to work, you could stick electrodes in every muscle and control whatever you want based on small twitches that were too small to actually move the body around. The interesting thing about this is that there's been lots of work done, by Andersen and others (including a lab I used to work in), on the role of posterior parietal cortex in figuring out the important aspects of sensory input (e.g. what to look at in a given visual scene). Since there is a role in motor planning, it's an interesting place to look at how sensory input comes in and is analyzed and an appropriate motor response generated. In essence, that's almost all of what the brain really does.
Well, it is and it isn't. It isn't because the society for neuroscience meeting where this was presented was in November. However, comparing this to the "cyberlink interface" or other EEG methods, which use electrodes stuck on the scalp, is not really valid. Basically, EEG electrodes record synchronous neural activity in the cortex (or synchronous input from the thalamus), and there's a limited amount of useful information (still a lot) that could be recorded through scalp electrodes. Single-cell recording like in the monkey is the way to go, except even in humans a similar device (quadruplegics moving a mouse pointer around a screen or signaling yes/no) was implemented several years ago (http://www.ncbi.nlm.nih.gov:80/entrez/query.fcgi? cmd=Retrieve&db=PubMed&list_uids=10896186&dopt=Abs tract) In monkeys they've even made a robot arm controlled by electrodes stuck in the brain, and they were able to put the robot arm in one city and have the monkey in another city controlling the arm over the internet. I can't remember who did that, but some of that work was also presented at the same conference that this stuff was at. The difference is that I'm pretty sure both the robot arm and the human mouse-moving experiments were done with electrodes in the precentral gyrus, also known as primary motor cortex, which is much closer to controlling actual movement. This part of the brain was mapped out in the 50's by Penfield, the neurosurgeon who found that giving patients small shocks in different parts of this area would cause different parts of the body to move, and that the map between brain anatomy and body part was the same from person to person. According to the article, however, this monkey study used electrodes in posterior parietal cortex, which could represent somewhat higher-order activity (e.g. movement planning, not actual execution). So in a sense it's that much more interesting. However, the interesting thing is that we've been doing single cell recording from different brain regions for decades, but what's really enabled this type of technology is fast computers to analyze the data and translate it into moving the cursor or whatever in real time.
Slashdot Mirror
But you've got some damned serious brainpower backing the alternative position, and I really don't think that could happen if the T-test was so pathetic that a group of freshman college students could rip it apart.
It really depends what you're talking about. I do think the Turing test is pretty pathetic when taken out of context, and a group of college freshmen certainly could tear it apart. Basically, Turing's test says that humans are like black boxes with input and output. If you have another black box that gives the same output to a given input, you have to admit that black box #2 displays consciousness and all that because you do that for people all the time. It's hard to debate the truth of this because it's almost more of a statement defining what consciousness and sentience are as opposed to saying anything about them. However, if you take the "strong AI" position, you would argue that replicating human behavior means that you have replicated the behavior of the human brain. The counter-argument to this is that the two black boxes could have two vastly different things going on inside them that give indistinguishable answers. Strong AI is largely what Searle's argument is directed against, but he takes it too far. I'm a computational neuroscience grad student, so basically I think strong AI is so obviously incorrect that it's laughable (and I think there are very few today who would disagree with that). However, if you take this argument down to a lower scale and say that you have a bunch of black boxes that are each the input-output equivalent of a single neuron, and you hook those together exactly the same way that real neurons are hooked together (there is no experimental technology that allows you to do this, BTW), you could say that you've replicated human consciousness and intellect.
Research and even a few things that made it to mass-consumption have shown that rapidly flashing or rythmically scanning a "safe" (read: low- power) light can be dangerous.
It's true that flashing lights can cause seizures, but those have to be flashing extremely slowly relative to video displays (something like 4Hz). This is a different situation than scanning, which is what goes on in a CRT. If the refresh rate is high enough, the photoreceptor cells in your retina can't tell that it's different from a continuous light source. LEDs are certainly fast enough to implement insane refresh rates--the limitation lies in the scanning apparatus itself (doing the faster of the scan axes). Doing it mechanically with a vibrating mirror (like it sounds like they're doing) can be difficult because at high speeds the mirror will want to oscillate sinusoidally, which can be corrected in various ways but would be a pain in the ass. I'm not sure what the limitation is on how fast they can be used without that type of correction, but there are more exotic things like acoustooptic deflectors you can use that don't have these limitations.
Two turntables and a microphone refers to Beck's penis. Three LEDs and a mirror sounds like something out of medical genetics textbook.
Someone correct me if I'm wrong, but I think it's supposed to be partially in reference to an interview with Timothy McVeigh, when he called the kids he blew up collateral damage.
That article raises a couple of valid points and several misleading half-truths. I should say that this is not my specialty, but it is something I know a bit about. First of all, no one claims that pseudogenes are entirely non-functional, only that they don't code for functional proteins. They can contain regulatory sequences for other genes that are still functional. The "evolutionary" argument that they should have been taken out of the DNA is misleading because it's really only true for single-celled organisms. There are three reasons (that I can think of) this is false for mammals. First, kinetics. It would take a lot longer for a gene to be excised than it would to be pseudogenized. Just because it hasn't happened yet doesn't mean it never will. Second, remember that pseudogenes are not necessarily useless, they just don't code for functional proteins. They can still contain regulatory elements absolutely critical to the survival of the organism. The third and most important reason is that it is not evolutionarily advantageous for organisms like humans to have genetically unstable genomes where things like excision of random chunks of DNA happen frequently. It is for some organisms--HIV is a good example. I forget the exact numbers, but something like half of HIV viruses floating around in the blood are actually infectious, and the reason why is because the rate of mutation is so high. However, one virus can have thousands of offspring in hours, so the high rate of mutation doesn't prevent it from replicating and is actually advantageous to the virus (and disadvantageous to HIV patients). Viruses use their genome extremely efficiently, and you would never find pseudogenes there. However, humans replicate slowly, and if our genome is unstable it leads to cancer and death, so the DNA is actually not used very efficiently. Like a hard disk that's 10% full and never gets defragmented, there will be clusters around containing deleted files, or pseudogenes. The tradeoff is that it's very reliable. Anyone who understands enough molecular biology to write useful articles on things like pseudogenes knows these things.
In any case, this is all irrelevant because mammalian olfactory receptor genes have certain special traits that make them exempt from these arguments. Think about some random genes. They've got some homology to other things, there might be two or three different genes for fairly similar enzymes or whatever that have maybe 50% sequence identity. In mammalian OR genes, you have about 1000 of them that are practically identical, with a couple of hypervariable regions that code for the binding pocket of the odor molecule. In addition, they all have no introns, meaning that each of them is one contiguous piece of DNA. In the article it says that you can't rule out translation to useful proteins by sequence information alone. There are certain far-fetched cases in which this could be true, but this is clearly not the case for OR genes for the reasons discussed above. If there are premature stop codons or frameshift mutations, that gene will not make a functional protein, and we can tell that it isn't "meant" to be that way because we basically know what the gene "should" look like since there are nearly 1000 of them that are nearly identical. To be fair, there are people out there trying to make all kinds of crazy arguments based on weak data from cross-species sequence homologies or pseudogenes or whatever, but this particular example is pretty much unassailable on those grounds.
As you correctly pointed out this is not macro-evolution. No information was gained.
Well, in a sense this is as close to direct evidence of macroevolution as one could ever get. It's vestigial DNA from when we were a dramatically different species. I guess I should rephrase that and say that if you believe in microevolution (which is directly observable), and it's easy to see how pseudogenization of that type would occur by microevolutionary means, this is evidence of microevolution occurring on a macroevolutionary time scale. So I should change my earlier claim and say that this is evidence for macroevolution.
As to your reference as to why we have so many pseudonized genes, are geneticists truely certain those genes are pseudonized? There is still alot of work to be done in the field, infact we have barely scratched the surface.
Yes, they are pseudogenes in that they do not code for functional proteins. That we have barely scratched the surface is a common misconception--at this level of analysis, the questions are fully understood in directly experimentally demonstrable ways that no one can argue. Comparing the problem of computationally finding new genes (as in the # of genes in the genome) to finding new members of a large gene family with high sequence homology is like comparing a needle in a haystack to a needle in a box of toothpicks. As far as the number of genes in the genome goes, we still don't know how many there are, and people in the biology community weren't really that surprised by the (preliminary) results, contrary to what the press latched onto. No offense or anything, but you're obviously not a molecular biologist, and your questioning the methods of the experiments and molecular biology in general is analogous to saying that the speed of light could never be measured since it's just so unimaginably fast, timing circuitry and lasers still aren't really understood, and it wasn't that long ago that physicists thought that light propagated through ether with infinite velocity. There's no point questioning the data. Your closing statement is very interesting, but I don' understand how you reached your conclusion. Nonrandom pseudogenization argues for a highly SELECTIVE process. Two good possibilities would be intelligent design or evolution (but not nonselective genetic mutation, which could cause these changes on a microevolutionary time scale). The question is, why would an intelligent designer create nonfunctional genes? The only answer I can think of is to create the false impression that evolution had taken place. The alternative explanation, of course, is that evolution actually did take place.
There is no single thing that proves evolution, but the more biology you study, the more interesting facts pop up that simply have no plausible explanation except evolution. Look for my post regarding pseudogene formation in human olfactory receptor genes (grep for lukesl) and see if you can find an alternative explanation. You'll probably find it interesting because it's strong quantitative bioinformatics-based evidence for descent of humans from a non-human organism. Taking it for what it is, one tiny piece in a big puzzle, evolution provides a beautifully clear explanation of the data in a way that no other explanation I'm aware of can. Right now I'm in med school and grad school for neurobiology, and things like that pop up every week or so. As far as I know, I've never met a biologist who even thinks that evolution requires further proof. We put it up there with "cigarette smoking causes cancer," another difficult thing to prove, yet the preponderance of evidence is simply overwhelming. In our society it's hard to imagine people who don't believe that cigarette smoking causes cancer. For most biologists it's equally hard to imagine people who don't believe in evolution.
One other thing, though--they do not teach real biology in high school any more than they teach real physics in high school. If you haven't taken any college-level molecular or cellular bio courses that come after organic chem and biochem, you don't know what biology is (or at least molecular bio--for me, they're synonymous, but I know others probably disagree). Don't take this the wrong way, but when you say "But I for one reserve the right to doubt an idea like evolution, that if true would completely invalidate my world-view, without more evidence than we currently have," if all you're working with is high-school bio, no one's ever shown you the real evidence.
There's a paper in the latest issue of Nature Neuroscience from Stuart Firestein's group at Columbia that provides some really interesting evidence from a very different angle. Some background: mice are animals that rely heavily on olfaction, or their sense of smell. Over half their brain is dedicated to it. In "lower" mammals (or however you want to look at it), the sense of smell is also very important (dogs, cats, etc.) For humans, however, smell is not as important. We don't smell predators coming or track prey by scent; we use vision (and a huge portion of our brain is dedicated to it).
Anyway, In this article they do a rigorous analysis of the data on olfactory receptor (OR) genes from the recently acquired mouse genome compared to the data from the human genome project. I forget the exact numbers, but mice have about 1000 OR genes. Humans have about the same number, but something like 75% of ours are pseudogenized. Basically, this means they've been converted to pseudogenes, or sequences in the genome that obviously used to be functional genes but have mutated to a nonfunctional state. This much was known before. In this paper, however, they use techniques based on similarity of sequences to group the mouse OR genes into families and subfamilies. Then they group the human OR genes into the same families. To sum up what they found, if you were to take a random group of say six mouse OR genes, there will be five or six human genes that are the human counterparts of those mouse genes (over 90% of human OR genes have a mouse gene that's over 95% identical at the protein sequence level, and 77% have a mouse gene that's over 99% identical, so reliable identification is not a big issue). However, within that group of five or six human genes, all but one of them has been converted to a pseudogene. They find this over and over again. There's only one functional gene in each group. Each group, BTW, can be thought of as sensing when a certain class of feature is present on a molecule. In an analogy to vision, it would be like if mice could see different shades of six types of red, but we could see only shades of one.
Okay, here's an evolutionary explanation. A long time ago humans were monkey-like animals. Before that, dog-like animals, before that mouse-like animals, etc. Whatever animal we used to be, it was heavily dependent on a highly-developed sense of smell for survival (hence an entire 2% of our genome being dedicated to it--think about that). However, as we progressed evolutionarily, having an exquisitely sensitive and precise sense of smell became less and less important, but smelling things in general was still necessary. The genes mutated and mutated, but if the last member of a family became pseudogenized, that would compromise our ability to smell molecules with a certain class of molecular features (by analogy, if we couldn't see red at all), and those pre-humans would die. As a result, we're left with the HIGHLY nonrandom distribution of working genes. I want to point out that while this could be written off as "microevolution," consider two things: #1- all humans on earth will turn out to have >98% identical OR genes. #2- I say this because that nonrandom of a pattern with that many genes involved would take a LONG time to evolve, or at least a lot longer than humans have been on different continents. Almost certainly longer than we have been Homo sapiens.
Can anyone come up with a non-evolutionary explanation that explains 1) why so many of the genes are pseudogenized, 2) why the selection of which genes are pseudogenized is so highly nonrandom and optimized for the real world? I'm not asking for a critique of my evolutionary explanation, as the reasoning as I've presented it is not intended to be bulletproof. However, I do think that the underlying model is correct and I don't think there's any better explanation.
Personally, I think the project has a low probability of success. A neural prosthetic device should be interfaced with as peripheral part of the nervous system as possible. This group has chosen to use as abstract a part of the nervous system as possible. But maybe they'll prove me wrong.
I think there's some truth in that, but the fact that they used posterior parietal cortex instead of primary motor cortex or the spinal cord makes the work that much more interesting for actually understanding brain function. I mean, if you really wanted a neural prosthesis to work, you could stick electrodes in every muscle and control whatever you want based on small twitches that were too small to actually move the body around. The interesting thing about this is that there's been lots of work done, by Andersen and others (including a lab I used to work in), on the role of posterior parietal cortex in figuring out the important aspects of sensory input (e.g. what to look at in a given visual scene). Since there is a role in motor planning, it's an interesting place to look at how sensory input comes in and is analyzed and an appropriate motor response generated. In essence, that's almost all of what the brain really does.
This is really nothing new.
? cmd=Retrieve&db=PubMed&list_uids=10896186&dopt=Abs tract) In monkeys they've even made a robot arm controlled by electrodes stuck in the brain, and they were able to put the robot arm in one city and have the monkey in another city controlling the arm over the internet. I can't remember who did that, but some of that work was also presented at the same conference that this stuff was at. The difference is that I'm pretty sure both the robot arm and the human mouse-moving experiments were done with electrodes in the precentral gyrus, also known as primary motor cortex, which is much closer to controlling actual movement. This part of the brain was mapped out in the 50's by Penfield, the neurosurgeon who found that giving patients small shocks in different parts of this area would cause different parts of the body to move, and that the map between brain anatomy and body part was the same from person to person. According to the article, however, this monkey study used electrodes in posterior parietal cortex, which could represent somewhat higher-order activity (e.g. movement planning, not actual execution). So in a sense it's that much more interesting. However, the interesting thing is that we've been doing single cell recording from different brain regions for decades, but what's really enabled this type of technology is fast computers to analyze the data and translate it into moving the cursor or whatever in real time.
Well, it is and it isn't. It isn't because the society for neuroscience meeting where this was presented was in November. However, comparing this to the "cyberlink interface" or other EEG methods, which use electrodes stuck on the scalp, is not really valid. Basically, EEG electrodes record synchronous neural activity in the cortex (or synchronous input from the thalamus), and there's a limited amount of useful information (still a lot) that could be recorded through scalp electrodes. Single-cell recording like in the monkey is the way to go, except even in humans a similar device (quadruplegics moving a mouse pointer around a screen or signaling yes/no) was implemented several years ago (http://www.ncbi.nlm.nih.gov:80/entrez/query.fcgi