原文
Tina Hesman Saey
Science News
Fri, 09 Oct 2009 11:00 CDT
Print
Silent activity
© Laurent
In sleep, the body may be still, but the brain is active.
Brain studies may reveal the purpose of a behavior both basic and mystifying

In a lab at MIT, a small black mouse named Buddy sleeps alone inside a box. A cone resembling a satellite dish sits atop his head. But the dish doesn't receive signals from outer space. Instead it sends transmissions from deep inside Buddy's brain to a bank of computers across the room.

Scientists like Jennie Young eavesdrop on the transmissions, essentially reading Buddy's mind, or at least that part of his mind occupied with a recent trip along a Plexiglas track littered with chocolate sprinkles. Young and her colleagues in Susumu Tonegawa's laboratory are monitoring nerve cells inside the hippocampus, one of the brain's most important learning and memory centers. Some of the cells in the sea horse - shaped hippocampus fired bursts of electrical energy as Buddy moved along the track. As he sleeps in his black box, those same cells spark to life again, replaying progress along the track in fast-forward or rapid reverse.

By recording the slumbering Buddy's brain cell activity, the scientists hope to glean clues to one of biology's greatest mysteries: the reason for sleep. Although sleep is among the most basic of behaviors, its function has proved elusive. Scientists say sleep's job is to save energy, or to build up substances needed during waking or to tear down unneeded connections between brain cells. Some emphasize sleep's special role in learning and memory. Others suggest that sleep regulates emotions. Or strengthens the immune system. And some scientists believe sleep is simply something that emerges naturally from having networks of neurons wired together.

"There are as many theories of sleep's functions as there are sleep researchers," says Mehdi Tafti, a geneticist at the University of Lausanne in Switzerland.

None of the many models for why people (and other animals) sleep can explain all of its complexity, says Robert Stickgold of Harvard Medical School in Boston. He equates proponents of the different sleep theories to blind men describing an elephant. It's a snake, or a tree or a wall, depending on which part of the elephant the men touch. Similarly, the answer to sleep's function seems to depend on what approach a given researcher takes. And each proposed idea contains inconsistencies that keep other sleep researchers from embracing it.

"There's no one theory that has enough unified evidence for it to be widely accepted," says Paul Shaw of Washington University in St. Louis.

Many sleep theories have been widely tested, though. Using brain wave recordings, genetic analyses, word tests, video games and various other methods, researchers have uncovered many of the pieces to the puzzle of sleep, even if they don't yet all fit together.

Asleep and Fired Up

REM sleep
© Arthur Tucker/Science Photo Library
In REM sleep, "the brain is on fire."
Not knowing why humans spend a third of their lives unconscious hasn't prevented scientists from describing five different stages of sleep from recordings of brain waves. Stage one, marking the transition between awake and asleep, is shallow. Stage two, which lasts the longest, features two forms of brain waves known as spindles and K-complexes (SN Online: 5/21/09). Stages three and four are the deepest, often referred to collectively as slow-wave sleep. Fifth is REM, the stage accompanied by rapid, jerky eye movements.

REM is the stage most often associated with dreaming, but plenty of dreaming

occurs in the other sleep stages, too. These stages are repeated in roughly 90-minute cycles throughout the night, with more slow-wave sleep early on and more REM toward morning.

In the first four stages of sleep, heart rate, body temperature and brain activity drop, supporting the view that sleep serves to save energy. But then REM comes along. During REM sleep, the body becomes paralyzed, which keeps people from acting out their dreams. Although the body is still as stone, the flight-or-fight response system is in overdrive, says Michael Perlis of the University of Pennsylvania in Philadelphia. Brain activity is as high or sometimes even higher than during wakefulness. "The brain is on fire," during REM, Perlis says. "The brain is cooking, but the body is a cold fish."

Since the brain burns up to 20 percent of the calories used by the body, REM may consume many of the calories saved during other stages of sleep.

Still, because morning brings renewed vigor, many people believe that sleep must save energy. And sleep certainly feels restorative. Recent genetic work suggests a molecular basis for that refreshment.

Sleep Wave
© Unknown
People oscillate between two types of sleep each night, REM (short for rapid eye movement) and non-REM. Non-REM sleep occupies about 80 percent of sleep time and consists of ever-deepening stages (shown in EEGs below left).
Allan Pack, a geneticist at the University of Pennsylvania, and his colleagues have been keeping mice up for hours after their normal bedtime. Activity of 2,000 to 3,000 genes differs in the sleep-deprived mice compared with mice that slumber undisturbed. None of the changes are dramatic, Pack says; they just nudge gene activity up or down a bit. Activity of genes involved in making large molecules consistently goes up during sleep. Examples include genes needed to build cholesterol and the oxygen-carrying substance, called heme, in red blood cells. Genes for molecules that help remodel neural connections are also revved up in sleep.

Studies in mice, rats, fruit flies and white-crowned sparrows have found similar patterns of gene activity, Pack and colleagues noted in a review in the February Trends in Molecular Medicine.

Sleep and wake are part of the metabolic cycle in the brain, Pack says. In this view, sleep is a time for replenishment and construction of cellular parts. "So when wakefulness comes along, you have the building blocks to make synapses," the junctures between neurons through which signals flow.

From Calories to Connections

To test the hypothesis that sleep alters metabolism, Amita Sehgal and Susan Harbison of the University of Pennsylvania left the lights on for some fruit flies. Each night for a week, the light deprived the flies of about two hours of sleep. Males made up for the loss by sleeping far more than usual the next day. Most females, though, just lost sleep and didn't make up the difference. The researchers also perturbed the flies' sleep by mechanical stimulation, which involved randomly jerking the flies' test tubes. Other flies were bumped while awake during the day, but their sleep was not disturbed.

Whether applied during day or night, mechanical stimulation resulted in decreased stores of glycogen, a starch, and increased triglycerides, a type of fat, the researchers reported in July in PLoS One. Light didn't affect stores of either substance much.

Fly
© Chiara Cirelli
Proteins (orange) that help forge connections between neurons build up in the brain while a fly is awake (left) but are depleted after sleep (right), suggesting sleep prunes neural connections, perhaps ensuring only the day’s strongest memories remain.
The stress of being jostled, rather than losing sleep, is probably what alters metabolism, at least in these fruit flies, Sehgal says. The study represents a growing trend in sleep science - the idea that sleep offers some advantage besides altering metabolism and saving energy.

"We're moving away from historical ideas of sleep saving calories," says James Krueger, a sleep researcher at Washington State University in Pullman. "It does do that, no question. But that's probably not why sleep evolved."

Sleep saves about 110 calories - about a cookie's worth - each night, Krueger says. That's not enough to make up for missing out on eating, mating or any of the other waking activities an animal does to survive. "It's a few more nuts. It's not worth it. You'd rather be awake avoiding predators," he says.

But sleep must provide some benefit that outweighs waking activities,

Krueger says - such as, perhaps, forging connections between neurons.

Krueger, in fact, suggests that sleep itself is an unavoidable result of having neurons wired together in networks. Nerve cells that work hard, electrochemically signaling neighbors, eventually need to rest and recharge. Neural quiet can spread through the brain as neurons pull their wired partners along with them over the brink into sleep, Krueger argues in a December 2008 paper in Nature Reviews Neuroscience. The quiet time may allow neurons to strengthen or weaken connections with partners.

Of course, neural remodeling is also important for learning and memory - processes often suggested as sleep's raison d'être (SN: 4/28/07, p. 260).

Breaking Bonds

Brain
© Charles Floyd
The brain orchestrates the daily sleep-wake cycle by responding to external cues, such as sunlight, and the body’s own rhythms.
But even sleep's role in learning and remembering evokes much dispute. One controversial theory, for instance, suggests that sleep, especially the powerful slow-wave variety, weakens synapses. That keeps the brain from filling up with useless connections, say sleep researchers Giulio Tononi and Chiara Cirelli of the University of Wisconsin - Madison. Their theory, known as synaptic homeostasis, is a sort of neuronal version of survival of the fittest. As an animal or person learns things throughout the day, connections between neurons get strengthened. All synapses are weakened during sleep, so tenuous connections are severed altogether and only the strongest bonds between neurons remain. This erasing of the blackboard makes room and preserves resources for the next day's learning, Cirelli and Tononi contend.

Some experiments seem to support the theory. While awake, rats build up levels of the protein GluR1, which helps increase the strength of synapses, the team reported in the February 2008 Nature Neuroscience. Levels of that protein drop when the animals sleep.

Studies in fruit flies show that snoozing leads to losing synapses. Cirelli and Tononi's group reported in the April 3 Science that proteins that help determine the strength of synapses build up while flies are awake and during sleep deprivation. Protein levels drop as flies slumber.

And while fruit flies sleep, they also lose synapses formed during social interactions, another study in the same issue of Science reported. When flies socialized, synaptic connections formed between neurons. Flies allowed to sleep after the exhausting social encounters pruned away some of the connections, but flies forced to stay awake retained the connections, researchers in Shaw's lab at Washington University found. Downsizing the number of neuronal connections could keep brain circuits from being overwhelmed by all the exciting information gathered from social interactions, Shaw says.

On the other hand, experiments with kittens suggest the opposite. In kittens with one eye sewed shut, connections between the closed eye and the brain's visual centers weakened while the kitten was awake, Marcos Frank of the University of Pennsylvania and colleagues reported in the Feb. 12 Neuron. The open eye showed stronger connections to the visual center, but only after the kitten slept. Frank says his data show that sleep strengthens connections between brain cells rather than weakening them.

Mult-Tasking
© Unknown
Studies show sleep probably serves many different functions, including enhancing learning and memory, regulating emotions, stimulating creativity and boosting the immune system.
Studies of Buddy and other mice, using electrodes implanted in their brains, tend to support the results from kittens. Recordings of the activity of brain cells sensitive to the mice's location, called "place" neurons, show that sleep allows the brain to replay events, strengthening connections between neurons and preserving long-term memories.

When Young records the firing of Buddy's neurons, a speaker crackles with what sounds to the untrained listener like radio static. To Young's ear the static is the sound of memories being made. Each time an electrode detects electrical activity in one of the neurons, it translates the activity to those audible crackles and to tracings on a computer screen.

While Buddy is awake and moving around in his box, running a maze or exploring new objects, his brain cells fire in a rhythmic pattern. As he sleeps, his brain waves slow down. But small, rapid spurts of brain cell activity, called ripples, interrupt the slow-rolling waves of sleep and burst above the background static. During those ripples, which last a fraction of a second, the place-denoting neurons fire in the same order as when the mouse was awake and exploring.

MIT's Matthew Wilson was among the first to discover these ripples. Ripples during slow-wave sleep replay the day's events, but the timing is compressed. During REM sleep, he says, rats and mice also replay events, but in real time, and not always in the same order or way they actually happened.

Cells in the hippocampus fire off a burst of ripple waves first. Then, 100 milli­seconds later, cells in the prefrontal cortex, commonly considered to be the seat of the brain's "executive centers," take up the refrain, Caltech researchers reported in the Feb. 26 Neuron. Such bursts of activity could represent transfer of information from temporary memory storage in the hippocampus to long-term storage in the cortex, Wilson says. In REM sleep, the timing of the firing between the hippocampus and the cortex is not as tightly coordinated as it is in slow-wave sleep.

Rats relive memories while awake, too, and that replay can help the animals plan their next move, Wilson and colleagues show in the Aug. 27 Neuron.

Scientists have speculated that such replay is also important for forming long-term memories. Researchers in Tonegawa's lab tested this theory directly: They blocked ripples by essentially paralyzing part of the hippocampus with tetanus toxin. Apart from diminished ripples, the mice slept normally and could remember tracks they had run for a short time. But the mice were unable to form long-term memories, the team reported in the June 25 Neuron. When researchers reversed the effect of the toxin, the ripples returned, along with the ability to form long-term memories, indicating that replaying and rehearsing memories during slow-wave sleep is a key step in solidifying them.

Across the Charles River in Boston, Harvard Medical School researchers have some evidence that replay may also be important for humans. Stickgold and Erin Wamsley have recruited volunteers to play a maze video game. After playing the game, some volunteers take a nap and some stay awake watching videos. The participants are awakened at the first sign that they are about to enter REM sleep, but some still report vivid dreams - some tangentially related to the game, such as hearing the music or exploring bat caves reminiscent of the maze. Preliminary results indicate that people who report game-related dreams improve their performance more when tested again. The dreamers improve more than either people who remained awake and thought about the game or people who slept, but didn't remember dreaming about the game.

"To us it's an indication that some of the networks related to that learning are active" during sleep, Wamsley says.

Its importance for memory is the only proposed explanation for sleep that contains a clear reason why consciousness must be shut down, says Stickgold. Human brains don't have TiVo, with the ability to record one thing while watching another. People use the same brain areas to perceive the world and then process what is happening. To fully digest information gathered throughout the day, at some point the brain has to block more input, he speculates.

In slow-wave sleep, the hippocampus shows home movies of the day's events to the cortex. During REM sleep, the hippocampus is issued a gag order, leaving the cortex to freely associate different pieces of information without the detail-oriented hippocampus stepping in to say, "no, this is what really happened." That free association may allow the brain to tie disparate experiences and facts together, making them easier to remember, or prompting new solutions to problems encountered during the day.

Learning and memory studies also suggest that sleep helps extract the gist of memories, enabling them to be filed under the correct headings, Stickgold says. How the brain does this is illustrated by studies in which participants "remember" that they learned a word such as hospital when actually the list of words they memorized contained doctor, nurse, stethoscope, bed and patient, but not hospital. Such associations give memories context and meaning.

"What your brain is leaving you with in the morning is a memory that is less accurate, but more useful," Stickgold says.

Sleep researchers still don't know how the brain decides which memories to review, edit and save, and which are junk, says Matthew Walker of the University of California, Berkeley. Emotion-associated chemicals may mark memories as important and worth saving, or send up a red flag to the brain that the memory is problematic. Over time, as sleep extracts the informational core of memories, it may also strip away the emotional blanket surrounding them, so that a person learns the lesson of the memory without all the drama of emotion. REM sleep in particular "is like group therapy for memories," he says.

Walker theorizes that this process may go awry in post-traumatic stress disorder. He lays out his case for sleep's role in processing emotional memories in the Annals of the New York Academy of Science's Year in Cognitive Neuroscience 2009. Removing the emotional blanket from memories is probably possible only during sleep, when outside stimulus is shut off, he says.

Wilson agrees that sleep can be an unfettered time to come up with new solutions. "The 'problem' with the awake state is that it is being influenced by the outside world," he says. "It is constrained by what you're currently experiencing. During sleep you can explore. The breadth of experience one has access to is much greater. I think it's very likely that during sleep you have the flexibility to evaluate and solve problems in novel ways."

REM sleep may be just what is needed to get creative juices flowing, suggests a study in the June 23 Proceedings of the National Academy of Sciences. People who had a nap with REM sleep performed almost 40 percent better on a word test requiring a creative solution than people who didn't nap or had only non-REM naps, researchers led by psychologist Sara Mednick of the University of California, San Diego show. The improvement happened only when participants drew information from a seemingly unrelated word test administered earlier in the day to solve the new problems. REM sleep seemed to help make that otherwise unrecognized connection.

"People in the REM group were able to use information they didn't know they had in their brains," Mednick says. Still, she doesn't believe all dreams mean something or that "sleeping on it" will solve every problem.

"Some dreams are going to be very, very meaningful, and some dreams are just your brain rooting through things that don't mean anything," she says.

Despite the evidence of sleep's role in brain performance, not all researchers believe that aspect to be the end of the sleep story.

"The notion that sleep is by the brain, for the brain - which is a motto in the field - is outdated," says Eve Van Cauter of the University of Chicago. "Sleep affects everything in the body and everything in the body affects sleep."

Short-term studies show that cognitive problems follow sleep deprivation, but scientists have no idea whether those problems relate to longer-term decline in memory or degenerative brain disorders, Van Cauter says.

Nearly 100 studies link sleep loss to cardiovascular disease, she says. "But we don't even have 10 studies on whether short sleep contributes to cognitive decline or dementia."

Others agree that sleep plays an important role in regulating the immune system. In fact, sleep may have evolved to improve the immune system's ability to fight off parasites, argue Patrick McNamara of Boston University and his colleagues in the Jan. 9 BMC Evolutionary Biology.

Species of animals that spend more time sleeping each day tend to have higher counts of infection-fighting white blood cells, a database analysis revealed. The more sleep on average a species gets, the fewer parasites plague its members, and the parasites that do infect longer-sleeping species are not as prevalent in their populations as parasites that sicken shorter-sleeping species.

Still, whether sleep's purpose is fighting parasites, making memories or modifying metabolism remains as much a matter of dispute as the blind men's competing images of the elephant. But perhaps that parable suggests a strategy for progress.

"The only mistake the blind men made is that they argued with each other," says Stickgold. If sleep researchers are willing to take a step back, confer and concede that others may have a point, perhaps one day the mystery of sleep will be solved.
Print this article

经研究，Ph.D不是Turing Award和Nobel Prize的必要条件。 尤其是Turing Award.

经研究，Ph.D不是图灵奖和诺贝尔奖的必要条件。

尤其是图灵奖

这个节目提供了很多关于人体的数据，可以作为衡量人工智能性能的依据。

Discovery的科普真是有启发性。美国人民真是幸福。

在这个世界上，总会有几个人同时意识到真相，但是荣誉可能只能属于一个人。

在人类认识自身的科学上，我们是在亚里士多德的岁月，还是在霍金的时代？

在亚里士多德的岁月，我们离这个世界的真相很远，但是更有可能成为一门崭新科学的奠基人。

“科学家并不是真正热衷于探索世界的真相，如果别的科学家抢在你的前面得到了你苦苦研究但是没有得到的结果，你应该感到高兴。但是事实上并不是这样的。科学家只是为了在同行之间争个虚荣而已。”

“科学家很有意思，对一个研究领域的执着却让他们与新发现擦身而过，而横跨多个领域却能成为Nobel的宠儿。”

仿佛看到自己的将来，一个宿命，一个刻在墓碑上的前缀。

“如果佛洛伊德能生在一个可以靠研究而谋生的时代，那么历史给他的称谓可能不仅仅是“心理学之父”。”

我今天在看一个很老掉牙的图片。

人类的视觉识别大致可以分为两种

1. 静态事物的识别，如形状，大小，颜色。大脑会将这些特征通过匹配后天植入的概念
2. 动态事物识别，如方向，远近等空间感要素

上面的图片属于动态事物识别的范畴

1. 识别的过程发生在大脑，而不是在眼睛。
2. 眼睛只是具有简单的感光功能：如调节焦距，颜色/亮度转换为电信号供大脑读取，以及其他一切机械能力(如转向，瞳孔调节等），这些能力供大脑调用

人类视觉匹配的过程:大脑会缓存前一个状态的物体位置（1/24秒内），然后获得当前物体的位置，通过这两个参数的比较，来得出位移量，速度，方向等变量。

上图，

1. 因为是个影子，所以你会忽略掉黑色的部分，而关注女生的特征点（如手，脚，头），关注轮廓
2. 因为是个平面图，所以你被剥夺了纵深，也就是Z轴上的空间感。

所以你只能分辨出特征点左右的方向，但是两个变量并不能决定旋转的方向，最关键的Z轴坐标就要靠运气了。

The cutting edge sciences do not come from the lab, it is right beside you.

The true faiths do not live in the temple, it is in the deep your heart.

大脑

Despite rapid scientific progress, much about how brains work remains a mystery. The operations of individual neurons and synapses are now understood in considerable detail, but the way they cooperate in ensembles of thousands or millions has been very difficult to decipher. Methods of observation such as EEG recording and functional brain imaging tell us that brain operations are highly organized, while single unit recording can resolve the activity of single neurons, but how individual cells give rise to complex operations is unknown.

虽然科技日新月异，但是关于大脑的大部分工作机制仍然是一个谜。我们已经非常了解单个神经元和突触的工作机制，但是成千上万亿的个体如何配合工作仍然难于解释。EEG技术，脑功能成像告诉我们大脑是高度有条理地协作的。这些单位单元记录可以告诉我们单个神经的活动，但是简单的单个细胞能够出现这么复杂的功能的原因仍然未知。

Actually, that is the mystery that biologists will never know.

John von Neumann告诉我们一个优秀的架构师的思维方式，他们总是善于找出程序和现实之间的鸿沟并且搭起桥梁。另：Neumann的东西不是无中生有的，是高科技山寨版。

Alan Turing图灵给的启发是最大的，他告诉我们上帝是个不太聪明的一般人。

Nils Aall Barricelli这个人50年前就在做那个事情了。。。。

睡前，还看了下历史

看完，还对普林斯顿大学再次新生敬意，计算机，博弈论，都是在那里出现的。。。。。而我到他门口转了一圈没进去，直接奔超市去了。 我捶胸啊

以下是我对他们的观点的总结

唐纳德·A·诺曼（Donald A.Norman）是个心理学和设计专家：人脑在不同情境之下会调整自己的工作方式，当人们轻松愉快时，大脑会执行宽度优先模式，也就是有很多乱七八糟的想法，会希望寻求各种不同的解决方法，当人们在危急时刻，大脑会执行深度优先模式，这个时候大脑会高度集中注意力，来帮助你完成任务。

这个也可以解释为什么很多天才都成不了事情，想法太多，实践太少，动脑子不动手。脑子想法多，手脚一定要有阿甘的精神，笨但是要持之以恒，想到了一定要做出来！！！！！

TED LINK

詹姆斯卡麦隆，大导演及民间科学家说

最后，总结一下。我学到了什么？第一：好奇心，这是你拥有的最重要的东西；第二：想象力，这是你创造现实最重要的力量；第三：对团队的尊重，这是比世界上其他定律更重要的定律。

有不少年轻电影导演向我讨教成功经验，我对他们说：“不要给自己划定界限。别人会为你去划边界，但你自己千万别去。你要去冒险。失败是你其中一个选项，但畏惧不是。从来没有一次探险是在有完全安全保障的情况下完成的。你必须愿意承担这些风险。”

另外，他讲了一个故事，当他遥控潜水艇的时候，他仿佛感觉自己的意识投射到了那个机器上面。。呵呵呵呵，这个可以算是意识投射的雏形吧。 TED LINK

互联网发明人蒂姆·伯纳斯-李（居然是英国人发明互联网，不过仍然是在MIT）数据关联，这个是我做报表得到的最大的收获：大量数据在一起呈现，会更加有趣。知道大量数据，但是不知道“呈现”，于是遗失了最有魅力的一环。

TED LINK

TED， IDEAS TO SPREAD

WPP，最好的广告公司之一，全称叫电线与塑料产品（Wire&Plastic Products），卖手推车。

IBM，世界上最好，历史最悠久的信息解决方案提供商，以前做的是打孔机，以及绞肉机。

Sony，世界上最好最有创意的消费电子产品企业，以前做过电热毯。

刚刚参加完公司首席架构师的会议，算是有几个收获 。黑体是我赞成的

1.  统一用户授权登录
2. 所有可以提供的服务都在云里注册并开放
3. 虚拟的用户界面，也就是和Google的企图一样
4. 用HTTPS 来发布数据，面向request -response的结构，这样有利于和其他系统整合
5. 后台进程处理并发处理数据，对于耗时的处理放到queue
6. 所有的商业逻辑放到Rule Engine里面处理
7. 日志统计和储存
8. 把实时数据存储和warehouse数据存储分开
9. 用存储过程来获取数据数据，而不是开发使用的SQL,这样可控性更高，不会出现业余SQL写出不合格的SQL,以致于拖了整个系统的后腿
10. 所有的配置文件都需要放到关系数据库中，这样有利于分布式管理

经常到 MIT之类的地方去，可以知道原来世界上想同样问题的不仅仅只有自己。他们叫 Multi-agent System，简称MAS

这是Artificial Intelligence: A Modern Approach 在2002年发表的。 和我在去年的想法相比，out了点。

所以说， 等我发文的时候，我就顺便把你的文字收录引用下吧，虽然这个想法的起源压根和你没关系。

1. 有个帮你联系杂志的中介，比如说业界出名的教授或者经常发水货论文的教授。
2. 有很多无所事事的时间让你看人家的论文，然后把这些论文放到自己的参考文献里。当然，我觉得好的论文都是自己写完了，然后去找些和自己观点类似的文章贴上去。
3. 熟知行话！如果我发论文，我跟他们说Self-Learning System他们肯定不懂或者说不屑地鄙视我， 如果我说Multi-agent system， 他们就会理解了。爱因斯坦发论文的时候就是有这个问题，所以好久才被学界承认。

以上文字竟然包含敏感词汇。

• MIT Computer Science下面的实验室和研究项目列表
• Princeton University 研究列表
• Harvard Computer Science 实验室及方向列表

MIT的计算机科学实验室和研究项目远远超过于该校的其他学科，是名副其实的超级热门学科，另一个特点是很多计算机的学科都是与其他学科如天体物理，人体感知，建筑，环境，材料等交叉的实验室，可以看出他们非常注重实践和应用

PU的列表中材料科学研究（物理）和计算机科学平分秋色，这个和该校传统的物理强者地位有关

Harvard是偏文科，所以计算机方面是中规中矩。博弈论的作者纳什是在Harvard发家的，所以Harvard好多项目都是和game theory有关....

他们人工智能的使命的描述是：

• Developing techniques for improving human–computer communication and collaboration in problem solving优化人机接口接口

• Modeling the behavior of intelligent communication systems 山寨人类自己

• Developing algorithms and representation languages for probabilistic and game-theoretic reasoning 为博弈论开发算法和表达语言。

• Mechanism design and automated negotiation for bounded-rational agents 本人不懂 (补充：博弈论一应用)
• Analyzing social and organizational systems 社会/群体/组织研究
• Game theory 博弈论（注：用自己的话讲就是决策路径）
• Peer-to-peer networks 对等网络

里面的几个人都很牛逼，至少履历是这样，比如这个 陈 伊琳（音译），人大本科，清华研究生，然后去Penn State University读了博士生，等等等等。

再说句酸葡萄的话：理工科博士，如一台装有超强CPU的PC，计算能力指数光鲜亮丽，但是！！！上面装的可能是DOS，甚至是操作系统都没有的。

 写论文最重要的部分，第一，导师名字不要写错， 第二，引经据典，下面给我最近的观点加点证据。

• 知识不值钱，智慧才值钱

《开放教育技术2008大会（Open EdTech Summit 2008）报告》的作者Marie Glenn指出：知识不再是力量，能够将知识转换成洞察的能力才是力量。

• 在大脑中保留知识点的索引（Index）而不是知识，才是最高的效率的体现（同样的道理可以应用到任何一个电脑系统的设计中，绝对不会错！！！！）

利用大量的信息工具及主要信息源使问题得到解答的技术和技能，被称做信息素质（Information Literacy）。具有信息素质的人，必须具有能够充分认识到何时需要信息，并有能力有效的发现、检索、评价和利用所需要的信息，解决当前存在的问题的能力。今天，信息素质已经成为成功者的必备条件之一。培养和锻炼信息素质市场前景广阔。

•   观点一“社会”是分布式集群计算机，观点二“规则”(Rules)是简单的法则，你可以称之为“佛法”

从生物角度，群居性生物通过协作表现出的宏观智能行为特征被称为群体智能。例如，一只蚂蚁智能低下，而一群蚂蚁通过共同遵守简单的行为规则，却能够表现出令人惊叹的智力水平。而从科学的角度，群体智能是通过模拟自然界生物群体行为来实现人工智能的一种方法，利用的就是群体行为往往优于个体行为这一特性。例如，我们在当当上选择图书时，往往会看到左侧有一栏标题为“购买这本书的读者也购买了如下书籍”，在这一栏中，我们往往会发现一些自己很想买、但以前不知道的书籍，这，就是群体智能的一个简单应用。

另外，作者提出了一个让我思考的命题：双语训练有助于思维发展 - 由于学习语言其实是对语言背后的模式进行识别的过程，从小学习双语可以避免成人容易出现的对模式识别的僵化现象。这个和写程序一个道理，几种语言写过了，你才能了解语言背后的思想。我想问题的时候，的确倾向于两种语言交叉思考， 写英文符号，我也不知道为什么， 只是觉得英文的单词让我的思维负担很轻。

对照我以前的观点：中文英文，世界各种语言，就像是XML一样， 都是结构化的描述性符号，本身并没有什么价值，文学的价值，在于思想，是杰作，翻译成任何语言都是杰作。这个和写程序一样，高手不会care用的是actionscript, javascript还是java, c/c++。

查看作者原文

用途一：科学计算以及图像化——超级计算机模拟图：黑洞合并产生巨大能量(图),

造物主，在我们的文明中，有很多很多别名，我们叫他神，我们叫他为上帝，我们给他的名字还有很多很多。我倾向于把他称之为造物主，这样更能反映他在这个科学世界应该有的地位和作用，而没有任何教派所赋予他的，和他捆绑在一起的其他教义（如爱因斯坦所为）。

造物主，可能是一个人，但是我觉得更大可能是一群人，一个种族，他们用他们的技术和时间，给我们创造了一个平台，支持我们的存在和发展。

造物主给我们很多线索，来暗示他的存在，我们把他叫做神谕。我们有很多成员热衷于追随神留下来的线索，希望获得神一样的力量，这些人被称为科学家。

造物主为什么创造这个世界？造物主有很多方式去利用我们这个世界的资源，其中一个就是利用我们的计算能力，在他的眼里，我们种族应该是个有巨大计算能力的超级电脑，不过这里有个矛盾点，既然他们能造出这个世界，说明他的计算能力已经超越我们本身的能力。假设一个人是一个线程，那么这个宇宙的计算资源会分配给不同的线程，额外的部分，需要来维护和更新各个资源间的同步之类的。按理说不需要利用人的计算能力来做事情了呢。。。。这个目前无解。需要有更高Level的理论来解释这个事情。

人类在造物主的规则下是不可能超越神的力量的，就像你的程序运行在虚拟机里面，你不要指望这个虚拟机能够超越这个系统的最大的计算能力，但是，计算能力不是这个决策的全部，也就说，我们有能力超越这个宇宙的范畴，利用神的力量来完成我们的意愿（大逆不道啊大逆不道啊！！）

未完

接下来

神的传说-可以当历史看。

神的世纪-程序员直接跳到Production环境里面修Bug。

Lorentzian Function

Euclidean geometry

虫洞

在数学上解释的通的，就是宇宙中事实存在的，你所要做的，就是把环境参数推到极限大或者极限小。

科学改变的只是速度和效率，但是并不会改变这个世界运行的规则。因为科学知识只是经验集合的一种体现，他是建立在规则之上的归纳总结和抽象，但是本身不能通过这个归纳总结抽象来改变基础的东西。

比如，你本来摆地摊能挣钱，那你上淘宝，你挣得更快更多。但是你仍然遵循着低买高卖的不变的原则，科技只是帮你缩短这个买卖循环的时间，而时间，恰好就是你最大的成本。

如果你本来喜欢借钱过日子，那网银提高你借钱的效率和降低你的门槛，你会越借越多，直到超出你的支付水平，把你拉入债务的泥潭。

如何成为科学家呢？是不是人人都可以成为科学家呢。

首先得了解什么是科学。

科学是只是人类正确经验的集合。

没错，说到底科学也是靠经验，本质上和中药没有区别。在过去的无数次尝试中，人们没有发现任何反例，那就可以归纳成为一个定理，那就是科学。

所有“走在前沿的科学家”进行的所谓的“高精尖的科研”，其实只是在一点点地拓展人类的知识而已。

经验总结么，谁不会？所以，谁都有成为科学家的可能性。

但是科学和中药又有不同，唯一的不同在于方法，对于科学

1. 要有合理的假设assume，你的假设必须合情合理，也许是一个公认的定理，也可能是基础的规则
2. 在合理的假设上，要有合理的限制：constraint
3. 在合理的假设和限制前提下，假定他们都是不变的，然后设定你需要研究的变量：Variant
4. 以你的变量为输入条件，不断地做不同的输入条件：这个叫实验
5. 观察你的输出
6. 得出变量/输出变化和输出/输出的关系

最近喜欢Finge，翻译过来就是边缘的意思，具体在科学上就是指边缘学科的研究。

我一直很崇拜做边缘研究的人，他们有一种内在的信仰让他们包容和开放地去认知这个世界，他们有最大的勇气让他们到未知的地方探索，他们有最有生命力的好奇心去不断拓展人类的认知领域，他们有最强的意志让他们抵挡诱惑和孤独，他们有最伟大的良知来维护科学技术最根本的教义-让人们有更好的生活。

18世纪初，化学学科和机械研究的交叉造就了人类文明史上第一次大型的生产力飞跃，19世纪初，电和磁的转化，20世纪中期，电子和数学的交叉发明了计算机，从那个时候，人类开始慢慢接近于上帝的能力了，这个世纪，将是计算机技术和生物技术最终融合的阶段，人类要完成对自身的逆向工程的最后一步——对大脑以及大脑运行模式的逆向工程，我们能够了解大脑的数据结构，大脑的处理数据的算法，我们能够精确度量每个大脑运行的速度，就像我们以GHZ来度量CPU一样，我们能够了解所有机体的接口，从此能轻而易举地更换身体的每个部件，用生物技术，或者是用纳米技术。

然后，我们开始孕育接替人类种群的生命，我们送他们离开这个星球，到宇宙的其他地方去，就像父母送儿子看着自己的儿女长大离开家，去寻找他自己的归宿一样。到那个时候，人类的文明完成自己的使命，开始慢慢地衰退，就像我们生活在大海里面的祖先的文明一样。一个轮回过去了，我们从海里走到了陆地，又从陆地飞向了宇宙。

有一天，人类文明的继承者回首这片土地的时候，就像我们看到大海一样，有种与生俱来的亲切感，让自己平静，让自己感到安全感，但是却再也回不去了。

CPU的主频指其时钟频率，理论上，最高的频率应该可以达到光的频率.不知道怎么算光的频率。。。。。。