John Hopfield wants computers to act more like humans.
Tuesday, a full house gathered in Heilmeier Hall to hear the Princeton University Molecular Biology professor speak about his award winning research in neurobiology.
Hopfield explained that while the computer can perform a number of computational tasks, it cannot perform the simple function of recognition, a task that a pigeon is capable of --with a brain the size of a walnut.
Hopfield said that by applying knowledge about the workings of the brain to the internal structure of the computer, this type of logical reasoning and recognition may soon be possible. His work bridges the fields of neurobiology, electrical engineering and the physical sciences.
"He very nicely connected the computational approach to understanding neuroscience and biological issues," Neuroscience Department Chairman Marc Dichter said.
This type of research will push the field of artificial intelligence to new levels. Future possibilities include improving the vision, language and understanding capabilities of machines.
Hopfield pointed out that the human brain sees objects by recognizing things that move together. By applying this same rationale to computers, machine vision is certainly within reach.
He extended this manner of reasoning to linguistics, as well. Hopfield said he believes that the associations that a brain makes with sounds that it hears can be imitated within the inner workings of the computer itself.
Often, Hopfield said, computers accomplish things in a circular manner. This requires more hardware and time for a given task. The key, Hopfield said, is in finding a more clever, simpler way of solving these problems.
"A very small, low power processor can solve these problems through the use of algorithms," Hopfield said.
Hopfield looked to the field of neurobiology and psychology for his answers. Unlike the computer, the brain approaches reasoning with some degree of flexibility rather than adhering to a list of strict rules.
"The logical system is much easier to implement than ordered rule based A.I.," Hopfield said.
Hopfield pointed out that the problematic aspect of psychological replication remains that the human brain is based on experience and emotion. Limitations in artificial intelligence lie in the incapability of a computer to feel human emotions.
"A machine can never play the piano with the same emotion as a human being," Hopfield said.
Students who attended the lecture said that although the subject matter is complex, Hopfield was successful at making the information accessible for an audience of amateur scientists.
"He presented it really well," Engineering senior Elizabeth Bucholz said. "It's a very physics-based science, and he made it simple and understandable."
Hopfield's research also earned him a major honor on Tuesday from the School of Engineering and Applied Science.
He was presented with the 2002 Harold Pender Award, given by the faculty of the Moore School "for pioneering contributions to computational neuroscience and neuroengineering."
By taking home the award, Hopfield is in good company. Last year's award was given to Jack St. Clair Kilby, who received the 2000 Nobel Prize in physics for his contribution to the invention of the integrated circuit, better known as a microchip.
