Slashdot Mirror
Neuroscientists At MIT Developing DNI
coolphysco1010 wrote to discuss the possible development of a direct neural interface, ala 'The Matrix', that could eventually allow for instant object recognition. From the article: "Now, neuroscientists in the McGovern Institute at MIT have been able to decipher a part of the code involved in recognizing visual objects. Practically speaking, computer algorithms used in artificial vision systems might benefit from mimicking these newly uncovered codes ... In a fraction of a second, visual input about an object runs from the retina through increasingly higher levels of the visual stream, continuously reformatting the information until it reaches the highest purely visual level, the inferotemporal (IT) cortex. The IT cortex identifies and categorizes the object and sends that information to other brain regions."
wohoo, alter my IT neurons to think my girl friend looks like buffy :)
Now were you R'ingTFA, or were you looking at the woman in the red dress?
I read an article many years ago about them doing this to live human patients. Via a fiber cable brain wet-ware implant, a blind man was able to discern colors and rudimentary objects. He did have a short seizure during the interview; however, once the subject got passed that he immediately requested that the researchers continue.
. htm
Unfortunately this was so long ago I cannot remember the magazine or relocate the article. But googling artificial vision shows a few parts of history and HOWSTUFFWORKS has a full set of details
http://health.howstuffworks.com/artificial-vision
i am not at MIT, but I can tell you this aint about to happen any time soon.
i am working on optical neuron-computer interfaces, and this is probably the most efficient and direct route for reading neurons. I know of researchers who can also stimulate neurons to fires via light, so in principle, we could build a complete neuroptical computers tomorrow... if neurons were not complete bastards to work with.
you see, they just dont like to stay place. where i research, they often build tiny fences to keep them in place, but even then, they go shooting theyre axons anywhere they feel, with no concern for the feelings of the researcher.
we also grow neurons on microchip surfaces, which allows for high speed and high resolution stimulation and reading of single neuron activity, but in two dimensions, which is excellent for retina etc.
but the neuron-chip or old fashioned neuron-electrode are hard to place, and optical reading of neurons still has bugs to sort out (id guess from 4-10 years more basic research). whenever you see these cool brainscan pics with MRI etc, remember theyre resolution is on the order of millimeters, and thats a lot of complexity lost.
http://www.biochem.mpg.de/mnphys/ has a nice review of the problems involved, if you like hardcore solidstate chemistry, silicon physics, and neurobiology
This reminded me of the research by Quian Quiroga et al in which they performed single-neuron recordings from MTL (upstream of IT, if I recall correctly) in humans. In that study they found neurons which would respond selectively to particular objects, such as Jennifer Aniston, Halle Berry, and the Sydney Opera House. Here's the abstract:
R. Quian Quiroga, L. Reddy, G. Kreiman, C. Koch & I. Fried Invariant visual representation by single neurons in the humanbrain. Nature (2005) 435, 1102-1107
It takes a fraction of a second to recognize a person or an object even when seen under strikingly different conditions. How such a robust, high-level representation is achieved by neurons in the human brain is still unclear. In monkeys, neurons in the upper stages of the ventral visual pathway respond to complex images such as faces and objects and show some degree of invariance to metric properties such as the stimulus size, position and viewing angle. We have previously shown that neurons in the human medial temporal lobe (MTL) fire selectively to images of faces, animals, objects or scenes. Here we report on a remarkable subset of MTL neurons that are selectively activated by strikingly different pictures of given individuals, landmarks or objects and in some cases even by letter strings with their names. These results suggest an invariant, sparse and explicit code, which might be important in the transformation of complex visual percepts into long-term and more abstract memories.
Kudos to MIT!
Every visual neuroscientist, ever, has been working on "deciphering part of the code involved in recognizing visual objects." Poggio and DiCarlo's contribution is mostly that they were able to record from a large number of neurons simultaneously in the inferotemporal cortex (IT). It's a logical (but interesting, to be sure) progression of work that has been done for decades in IT -- most of that work done elsewhere.
Neural prosthetics and DNI are the bullshit that people trot out to make neuroscience interesting to the public. It's worth pointing out that neither of the actual named scientists in this work raise the possibility, and in fact, other than the abstract, there's nothing that even hints at the idea. These guys aren't working on a DNI. They're doing basic science. Years, decades down the road, some engineers might take the work that built on Poggio and DiCarlo's work and turn it into a DNI. Or at least, we can so hope.
Name a university, and I can guarantee that the odds are that they'll have some basic science research underway with as much potential for the betterment of society as this stuff. So when you say "kudos to MIT" like this, remember that you're praising their PR department, not their scientists.