Have I mentioned I just got back from the 2007 meeting of the Society for Neuroscience, held in sunny San Diego, California? Yes, in my last two posts? Okay, well, I still haven’t given my roundup of just went on there and what conclusions one might take away from it.
So, let me do that now…
First off, since I’ve gotten the question from friends, a note as to why I’m suddenly paying so much attention to neuroscience. In short, it’s caught my interest. The longer answer: Of all the fields of human endeavor, neuroscience strikes me as the most likely to have the biggest impact on how we understand and experience the world in my lifetime. After centuries of speculation and philosophy, we have just in the last few decades begun to solve the problems of how to look inside the living human brain, how to understand what we’re seeing, and how to use that understanding to alter the brain and allow the brain to alter the world. The brain has a language, we’ve just begun to translate it, and what we learn (and what we learn to do) is going to turn major parts of the world upside down.
So, with that, let me try to find some decent way to take you through things. First off, this is a BIG conference. Some 30,000+ folks come out for it — scientists, students, exhibitors, a small gaggle of press. A team of reporters couldn’t keep up with one-tenth of the presentations, posters, symposia, minisymposia, workshops, meetings, socials, and satellite events; so I can only give you the snapshot of what I saw.
#1 — Jeff Hawkins, inventor of the Palm Treo and founder of the Redwood Neuroscience Institute and Numenta
The first big talk I made it out to on Saturday, one of the featured lectures, was Jeff Hawkins on the topic, “Why Can’t a Computer Be More Like a Brain?” Good question. Despite all outward appearances, it turns out computers are very, very stupid. The example Hawkins used was this: Give a computer a bunch of sets of pictures of cats and dogs, then ask it which pictures are of which animals; no computer in existence can pass this test. (My old Apple IIc would just crawl under a table and hide; my MacBook Pro would, too; as would Big Blue.) The point here is that even the simplest task, something that can be accomplished by a two-year-old human, is virtually impossible to accomplish with a computer at this time.
Why?
Well, we, as humans, have a tremendous amount of very complex information in our brains about cats and dogs. Computers don’t. Unless, of course, we teach it to them. The question, though, is how to design a computer that can learn. Hawkins looks at the question as one of how the human brain distributes information. His model is Hierarchical Temporal Memory (explained somewhat on the Numenta Web site). The idea, essentially, is that the brain stores information hierarchically, with many pictures of dogs and cats stored together under the headings “dogs” and “cats.” Each category is a subset of some larger category (animals, say, or furry animals), and each category has various subsets (labs, beagles, puppies, pugs). Each node on the hierarchy represents an algorithm that allows the brain to process what it’s looking at.
Using this concept, Hawkins and his lab have created a platform that can be taught to look at rudimentary drawings (of things like mugs, helicopters, dogs, etc.) and classify them. After a significant amount of “training” with test images, they’ve had decent success in getting a computer to recognize what it’s looking at. The drawings are very crude, but after training the computer is correctly classifying unique images it’s never been trained on (it’s seen a lot of drawings of mugs, for instance, but it hasn’t seen this one) — so that seems pretty promising.
A computer that can learn is sort of the holy grail (Hawkins quoted Bill Gates, who said a computer that can learn would be worth 10 Microsofts). And, of course, once we have such a thing working, it doesn’t have to be limited by the human senses. The concept could be applied to things like weather or seismology — imagine a computer brain whose “senses” were every weather station in the world or every seismometer.
Of course, this talk wasn’t purely, or even mostly, neuroscience. But without the insights we’re gaining into the brain, this progress wouldn’t be being made.
[Another talk by Hawkins on the same subject is here on Google Video. Examples of the drawings the computers are looking at appear around 30 minutes in.]
#2 — Press panel on the normal aging brain
On Sunday, there was a panel for the press on the normal aging brain — that is, a look at what happens to the brain as we age in the absence of disease. It’s truly depressing to look at exactly how one’s brain will decline, even if one dodges various bullets in terms of disease. Nonetheless, a couple bullet points from the presentation:
* Old people really do have a worse sense of direction — Scott Moffat, PhD, at Wayne State University in Detroit, presented a study in which old people and young people navigated a virtual environment. Both old people and young people did a pretty decent job at remembering a series of landmarks in the virtual world. But old people did much worse at remembering which way to go at those landmarks. Also, old people did a worse job of filtering out useless information, remembering meaningless landmarks (not associated with a choice point) that younger people correctly ignored.
* Risk factors for stroke, like high cholesterol and a physically inactive lifestyle, may be linked to an increased risk of Alzheimer’s.
* Physical fitness is closely related to maintaining cognitive ability — We’ve known this about cardiovascular health for a while. But research presented by Claudia Völcker-Rehage of Jacobs University in Bremen, Germany, indicates that balance training (such as Tai Chi) may also be important.
# 3 —Andy Grove, former CEO of Intel
Also on Sunday, Andy Grove gave a talk about, “Translating Neuroscience: Can Systems Engineering and Lessons from High-Tech Take Us Beyond the R01 Culture?” Grove, who had prostate cancer and now has Parkinson’s, has come to take an interest in what he calls “bio-enterprise.” He looks at progress in the semiconductor industry in 30 years (gigantic) versus progress in treatment of Parkinson’s in 50 years (minimal) and — well, he’s pissed off.
He broke down the problems with biomedical research into three headings: speed, failure, and success.
Speed — In semiconductors, Grove said, they could test new patterns on little corners of existing, commercially sold wafers, creating a constant stream of “FedEx Trucks” sending tests out and results in to the company. Biomedical research gets packed onto trains, and they leave the station once a decade (that is, large-scale clinical trials that have to play out in humans over years). We need to find ways to speed up testing.
Failure — In short, biomedical research doesn’t spend enough time looking at its failures and trying to glean useful lessons. In semiconductors, a mistake can lead to a whole new type of memory system. In medicine, things get swept under the rug or subsumed in averages of multiple data points. Why did that one patient get better on the treatment? It might be coincidence; or it might be a breakthrough.
Success — Here, Grove talked about something a number of speakers were talking about: “We are about to experience an explosion of Alzheimer’s disease cases. Population statistics, incident rates and demographic changes indicate that the incidence of AD is doubling every five years. North America alone is going to have multiple millions of cases in a few more years, and when you look at the economic aspect of this, by 2030, the spending on Alzheimer’s disease will be as much as the total Medicare spending on everything in this country today. This is not a stochastic process. This is not a maybe. This is going to happen plus or minus a little bit.” His point was that even if we had a sure-thing treatment for AD, if it cost, say, $1 billion, it would be almost impossible to put together the funding to develop the treatment.
Summing up the difference between semiconductors and bio-enterprise, Grove offered this: “The semiconductor industry says ‘what matters is time to money’ … In the bio-enterprise, my impression is the corresponding statement is ‘good science takes time.’ Is that true? Yes. Does it help? No.”
His suggestions for improving the situation:
* More focus on biomarkers (reliable ways to measure the progress of diseases and the effectiveness of drugs)
* More open data
* A risk-multiplier for patent extensions, rewarding truly innovative drugs over “me-too” drugs
[The full speech transcript is available here.]
#4 — Press panel on emerging technologies in neuroscience
The highlight here was Mark Ellisman discussing plans for a “Whole Brain Catalog” (think “Whole Earth Catalog“) that would be something like Google Maps (crossed with Wikipedia) for the brain.
#5 — Martha Farah of the University of Pennsylvania
On Monday, Martha Farah gave a talk on neuroethics. At least, that’s how the talk was billed. To my mind, it was mostly a listing of really cool things people are doing with neuroscience, with a coda at the end that said: “And there are ethical implications to all this.” Maybe she ran out of time.
I don’t mean in any way to insult the talk, though, as it was one of my favorites.
A few tidbits I picked up:
* The fall 2007 Jack Daniels ad campaign made use of neuromarketing data. (I’ve learned, in my short time following this stuff, to be more than skeptical regarding the limitations of what neuromarketing can tell us right now … but I’m still excited to see corporate America getting interested.)
* A 2001 case in Iowa has set out the groundwork for so-called “brain-fingerprinting” (a form of lie detection) to be admitted into evidence.
* I posted the YouTube to this before, but I’ll mention it again: A company called Emotiv is getting ready to introduce a game where you move objects using EEG (i.e., your mind).
* Brain scanning could be used to measure personality traits (extroversion, etc.) and to measure things in an educational setting, such as reading ability.
Ethical issues include: distributive justice (rich folks will have access; others, not-so-much at first), privacy, safety, and freedom (is there a Fifth Amendment right not to have your brain scanned during a police interrogation?).
#6 — Newt Gingrich
I can’t get away from Newt. This wasn’t CPAC, but for some reason the former Republican Speaker of the House was there. Actually, he gave a great speech and was surprisingly well received by a bunch of presumably liberal scientists.
His first message was simple: Lobby Congress. You won’t get what you need if you don’t tell anyone. You can’t, as citizens, not engage the political world and then complain when Congress does something stupid that hinders you. My sense is the neuroscience community has been doing a good job of getting organized in this regard, so the message fit with the direction things are already going.
Other suggestions from Newt:
* A pool of federal grant money should be set aside for scientists under 40 — people too young to know that certain things can’t be done.
* A government-awarded-prize system should figure more prominently in encouraging the development of drugs.
* Help government come up with an overarching strategy as regards Alzheimer’s. (He’s starting a group to look at the issue — which, to anyone who follows Newt, could not be less surprising.)
#7 — Press panel on brain-machine interface
Last but certainly not least, on Tuesday, was the amazing presentation on brain-machine interface — which I already wrote up here. The basic point: We can record the firing of neurons in the motor cortex, figure out what movements they correspond to, and then use the translation to direct robotic limbs. From here, it’s all a problem of refinement.
OVERARCHING THEMES:
I’ll spare you. But I’ll just say the speed of developments is fast and only likely to get faster. Things that sound like science fiction now are likely to be cutting edge and then old news in most of our lifetimes — assuming you’re, say, under 50. Anything we can kind-of, sort-of, just-barely do right now, we’ll eventually perfect. I can’t wait for tomorrow.
Recent Comments