Leaping to a blackboard that covers a wall in his corner office, William Bialek scrawls out a mathematical formula. Bialek, one of the top scientists at NEC Research Institute Inc. in Princeton, N.J., is wrestling with a daunting challenge. He's trying to understand flies. How do they fly so straight, even in the wind? How do signals move from their elaborate eye cells, through tiny nervous systems, to their fast-reacting flight muscles? "I'd like to know whether there are some general principles working here," says the burly bio-physicist.
That kind of unfettered curiosity is just what the Japanese electronics giant wants from the American scientists it has assembled at its two-year-old, $32 million research complex. At a time when many U.S. companies are cutting back their research and development budgets and pushing to involve scientists more directly in product development, the Japanese are taking a much different tack: They want to overcome their long-standing weaknesses in basic research.
Tapping the talents of scientists in the U.S. is one key to that strategy. Japanese companies from Hitachi, Toshiba, and Sony to Shiseido, a health care concern, have set up American labs. Every major U.S. research university now has at least one Japanese corporate research facility nearby, says Stephanie Epstein of the Center for Public Integrity, a Washington-based group critical of the growing scientific links between the U.S. and Japan.
At Princeton, NEC Corp. has tried to create an environment where scientists are free to explore fundamental questions in physics, computer science, and materials. Although the research clearly follows the lines of NEC's main businesses in computers, electronics, and communications, the lab urges its scientists not to see themselves as product developers. "The emphasis is on breaking away from incremental progress and moving in new directions that are expected to be risky," says physicist Richard A. Linke, a senior research scientist who left AT&T Bell Laboratories for NEC. The $28 billion company has a global network of applied scientists and engineers, including some in Princeton, to follow up on any breakthroughs.
To build its research hothouse, NEC recruited some 45 physicists, computer scientists, and materials researchers from the likes of Bell Labs, the University of California at Berkeley, Massachusetts Institute of Technology, and other top R&D centers. Set in a wooded corporate park near Princeton University and SRI International's Sarnoff laboratories, NEC's new facility operates more like a university than anything else. In place of a rigid bureaucracy, a five-member board of scientists allocates funds and oversees projects. This arrangement was the brainchild of the lab's founding president, Dawon Kahng, a transistor and semiconductor-memory pioneer who worked for many years at Bell Labs. Kahng, who died in May at age 61, wanted to be sure his scientists weren't stifled by the detailed five-year plans Japanese companies typically use to manage research.
SERIOUS GAMES. In the lab's freewheeling atmosphere, physicist Eric B. Baum challenges computers to games of chess and go. "There's been very little conceptual progress in computer chess since 1956," says Baum, who got his PhD from Princeton University. Such games offer insights into how humans and machines "reason" differently. Even the best computer programs must rely on brute force--processing a myriad of options--instead of the more elegant, seemingly intuitive method that humans use. Baum's goal: discoveries that lead to machines capable of subtle forms of reasoning.
Advances in computers and communications are the ultimate aim of other projects, too. Physicist Karvel K. Thornber, a 20-year veteran of Bell Labs, works on fuzzy logic, a form of mathematics that helps machines deal with complex, uncertain tasks, such as stock market forecasting. Linke is exploring properties of microscopic lasers that may some day be a key to high-speed opticalcomputers that rely more on light waves than on electronic circuitry.
NEC tries to spur collaboration between scientists with noon volleyball games and bridge. It also works hard to burnish its American image. School groups tour the lab often, and in July, 80 fifth-graders from inner-city Trenton schools will take two weeks of science classes. "It's part of being a good corporate citizen," says physicist Tineke Thio, who is doing superconductivity research.
COZY CONNECTION? For all its efforts to appear as American as possible, the lab has many contacts with Japan. Linke is using lasers developed by NEC there. Thornber works on fuzzy logic with NEC development labs in the U.S. and Japan. Physicist Peter A. Wolff, a former head of the Francis Bitter National Magnet Laboratory at MIT, is collaborating with Japanese applied scientists in optical computing. Still, company officials are wary of a one-way relationship with Tokyo. "We've tried to avoid appearing to be just a channel for American technology back to Japan," says Arthur Torsiglieri, general counsel for the lab.
Most of the Princeton scientists could care less about that. "I think what I'm doing here is more or less for the good of humanity. Everything I'm doing is published in the open literature," says Baum. "I wish American companies were contributing to basic research like this." Indeed, some at NEC feel they've got a big edge over their colleagues in Corporate America's labs, where many scientists have been shifted in recent years to applied work. "The rest of the world is very, very urgently trying to make research profitable. It's a national panic," says Linke. He likens the atmosphere at Princeton to the glory days at Bell Labs several decades ago. The NEC lab, laments Berkeley computer scientist Michael A. Harrison, is now the "only corporate lab devoted to unrestricted, undirected research in the U.S."
Whether NEC can keep this spirit alive remains to be seen. Officials point to the company's consistent support for R&D, even in tough times. But NEC's net income fell 43% in the last fiscal year, and it has delayed plans to expand the lab's staff. "We've just made a little slowdown," says Daizaburo Shinoda, acting president of the lab, which has an annual budget of $26 million.
For now, though, curiosity is flourishing in Princeton. Posters in the corridors highlight topics that NEC investigators are probing. "What is the interplay between knowledge representation and machine or human learning?" asks one. Plumbing such questions should keep NEC's U.S. scientists busy well into the next century.
RESEARCH TARGETS AT NEC'S PRINCETON LAB NEURAL COMPUTATION By studying insects, researchers hope to discover how nervous systems process information FUZZY LOGIC Scientists are experimenting with software that may help computers better handle ambiguity OPTOELECTRONICS Researchers want to build powerful computers that use light beams and microscopic lasers instead of electronic circuitry SUPERCONDUCTIVITY To try to unravel its mysteries, scientists are studying how gallium arsenide and other materials behave when lasers are fired through them at temperatures approaching absolute zero DATA: BW