BERKELEY -- Ghost images that continually flit across the retina while we're still in the womb are critical to setting up the visual system to make sense of the world when we finally open our eyes, according to a definitive new study by neuroscientists at UC Berkeley.

In a report published in this week's issue of Science, postdoctoral researcher Anna Penn and Carla Shatz, professor of molecular and cell biology and investigator in the Howard Hughes Medical Institute at the University of California, Berkeley, describe the effects of blocking these fleeting images in one or both retinas during early development, before the eyes have opened.

They found that blocking input from one eye disrupts the normal formation of layered structures in a part of the brain that relays signals from the eye to the visual cortex. This layered pattern shows up in all mammals that have been studied, and thus is apparently crucial to the ability to process visual information.

Competition between signals from both eyes is the key to helping this area of the brain segregate into regions devoted to each eye, the researchers say.

"The balance of activity coming from both eyes is critical to making this pattern needed for binocular vision," Penn says. "If you block spontaneous retinal activity during this period of development you can adversely affect the pattern."

The finding is important to understanding how the brain wires itself during development, and how outside influences may interfere with normal development before and after birth.

The ghostly images - waves of activity generated spontaneously by eye cells, without stimulation from light - wash across the retina in random patterns during early development in many if not all mammals. Shatz and her colleagues discovered this several years ago, and others have found that similar kinds of spontaneous neural activity occur in the developing auditory system and spinal cord, and may be important in preparing the brain to hear a sound or make coordinated movements.

Even before there is vision, this spontaneous retinal activity has organized the lateral geniculate nucleus - a part of the thalamus at the base of the brain which relays signals from the eye to the visual cortex of the brain - into layers. Each layer is dedicated to carrying signals from only one eye.

Shatz, Penn and their colleagues found that preventing spontaneous activity in one eye allowed the other eye to take over all the area that would normally be divided between the two eyes. The effect may well be irreversible, they say.

"If you block the spontaneous generation of signals in one eye real early, the layers don't form. One eye takes over all the area to form one big layer, and the other eye gets gypped," Shatz says.

"It is the competition between the two eyes that creates the pattern," she adds. "Just as muscles get stronger with use, the eye that sends the strongest signals gets the larger area of the brain. You either use it or lose it."

Penn and Shatz blocked the spontaneous activity with a drug related to nicotine called epibatidine, which overloads the receptors for the neurotransmitter acetylcholine and shuts them down. The effect of epibatidine on the development of critical early connections in the eye suggests that use of related drugs could have a similar effect in utero.

"There is a growing realization that in the fetus many of our senses are fine tuned by spontaneous activity in the nervous system, and that anything interfering with that activity, like nicotine, may also cause harm," says Penn, who recently received her M.D. and Ph.D. from Stanford University and plans to embark on a pediatric internship at UC San Francisco this summer.

"We don't know for sure, but this could be one explanation for smoking's effect on prenatal development - it may affect the spontaneous generation of signals that are used to form early connections in the brain," Shatz adds.

One of the puzzles of development is how billions of nerve cells in the body make the proper connections. The number of genes in the body are insufficient to specify exactly how each nerve cell is hooked up in the body, Shatz says. There are perhaps one trillion nerve cells in the brain and 1,000 trillion interconnections among them.

The most popular theory is that crude wiring is specified by our genes, but that input from our senses determines the final precise interconnections. If the nervous system were a network of telephone lines, the genes would lay down the trunk lines between cities and experience would refine the connections to reliably reach individual homes.

In fact, during early development many more connections are made between and among nerve cells than are necessary. The excess are then pruned by experience to yield the final network connections. Shatz likens the process to random telephone dialing from the retina to the brain. When the correct brain cell answers, the connection is strengthened. Connections that yield wrong numbers fade away.

David Hubel and Torsten Wiesel won the Nobel Prize in 1981 for showing that input from the eye is necessary in infancy and early childhood to prune connections so that the network can interpret the images falling on our retina.

Since Shatz and her lab at UC Berkeley discovered spontaneous activity in retinal cells in 1990, they have been trying to show that this activity before we first open our eyes is also crucial to pruning nerve connections.

Their studies have been done primarily on ferrets, which are born with an immature visual system equivalent to that of the human fetus during the second trimester.

Based on these experiments, Shatz has proposed that competition between the signals from the two eyes is what sets up the proper patterns in the brain.

The study reported this week is definitive evidence that this is true.

"Others have proposed that you need a lock-and-key mechanism to establish the right pattern of connections in the brain -- that a growing nerve cell must carry a set of molecules or a key that fits into a lock or receptor on the cell it is deciding to connect with," Shatz says. "Our work shows that you can get a distinct pattern just by an activity-dependent competitive mechanism."

"This is a clear demonstration that spontaneous activity can form connections using competition," Penn says.

The work was supported by funds from the Howard Hughes Medical Institute and the National Institutes of Health. Other co-authors include Marla B. Feller, a former postdoctoral fellow now at the National Institutes of Health, and undergraduate student Patricio A. Riquelme.