A computer chip has been connected to a network of living brain cells, scientists reported Sunday, a development that could lead to the integration of computers with the human nervous system.

The chip should also help scientists learn how nerve cell connections are made, modified or maintained in the brain, processes thought to be the basis for learning and memory.

Even though the so-called “neurochip” can hold only 16 cells and a human brain contains billions, scientists say starting small is necessary to understand basic properties of brain cell connections.

“It’s perhaps not different from studying a big computer with a half-million transistors,” said Jerry Pine of the California Institute of Technology in Pasadena. “If you didn’t know how the transistors worked, it would be hard to understand how the computer worked.”

Pine, along with Michael Maher, of the University of California, San Diego, presented the research in New Orleans at the annual meeting of the Society for Neuroscience. Caltech scientists Yu-Chong Tai and John Wright also helped develop it.

The neurochip contains 16 wells arranged in a four-by-four grid. The wells are about four one-thousandths of an inch apart, and the entire chip is immersed in a liquid that allows the brain cells to grow.

After the researchers place a cell from a rat embryo’s brain in each well, the cells connect to each other via long, thin structures. Because the chip is connected to a computer, scientists can send and detect electrical impulses to and from each cell, mimicking the barrage of electrical activity that occurs naturally.

The chip is “probably the best available now,” said Mu-Ming Poo, a neurobiologist at the University of California, San Diego. Other than the chip, the only way to study individual nerve cells in a network is to stimulate and record from them with glass needles, a destructive process that kills the nerve cells within a few hours, Poo said.