Scientists used to believe adult brains did not grow any new neurons, but it has emerged that new neurons can sprout in the brains of adult rats, birds and even humans. Understanding the process could be important for finding ways to treat diseases such as Alzheimer's in which neurons are destroyed.

Most neurons sprouting in adulthood seem to be in the hippocampus, a structure involved in learning and memory. But they rarely survive more than a few weeks. "We thought they were possibly dying because they were deprived of some sort of input," says Elizabeth Gould, a neuroscientist at Princeton. Because of the location, Gould and her colleagues suspect that learning itself might bolster the new neurons' survival, and that only tasks involving the hippocampus would do the trick.

To test this, they injected adult male rats with a substance that labeled newborn neurons so that they could be tracked. Later, they gave some of the rats standard tasks. One involved using visual and spatial cues, such as posters on a wall, to learn to find a platform hidden under murky water. In another, the rats learnt to associate a noise with a tiny shock half a second later. Both these tasks use the hippocampus—if this structure is damaged, rats can't do them.

Meanwhile, the researchers gave other rats similar tasks that did not require the hippocampus; finding a platform that was easily visible in water, for instance. Other members of the control group simply paddled in a tub of water or listened to noises.

The team report in Nature Neuroscience that the animals given the tasks that activate the hippocampus kept twice as many of their new neurons alive as the others. "Learning opportunities increase the number of neurons," says Gould.

But Fred Gage and his colleagues at the Salk Institute for Biological Studies in La Jolla, California, dispute this. In the same issue of Nature Neuroscience, they report that similar water maze experiments on mice did not help new neurons survive.

Gould thinks the difference arose because the groups labeled new neurons at different times. Her team gave the animals tasks two weeks after the neurons were labeled, when the new cells would normally be dying. She thinks the Salk group put their mice to work too early for new neurons to benefit. "By the time the cells were degenerating, the animals were not learning anything," she says.