Drugs designed to relieve depression may be working because they stimulate the growth of new nerve cells in a key area of the brain, say researchers.
A key part of the brain is affected
It had been thought that the drugs operate by influencing the way brain chemicals work, but evidence is emerging that the changes they produce are more far-reaching.
Research was carried out on mice placed into a new living environment - normally their anxiety at the change means they do not eat for a while.
However, antidepressant treated mice tend to take a shorter period to overcome their anxiety.
The results were published in the journal Science.
The researchers found that when these animals given a modern "SSRI" antidepressant called fluoxetine the number of neurons in an area of the brain called the hippocampus doubled.
This area is known to be important in mood and memory, and accordingly, as expected, the mice showed fewer signs of anxiety.
However, when these mice were given selective x-ray treatment which prevented the development of the new cells, the result was different.
The mice enjoyed no improvement in their mental state despite being given the antidepressant.
This suggests, say the researchers, that the arrival of the new neurons may be as important for antidepressant users as other chemical effects on the existing brain cells.
It is possible that future drugs could be targeted more precisely to stimulate neuron growth in the hippocampus, they say.
Dr Thomas Insel, a director of the National Institute of Mental Health in the US, said: "This is an important new insight into how antidepressants work.
"We have known that antidepressants influence the birth of neurons in the hippocampus - now it appears that this effect may be important for the clinical response."
However, scientists are still a long way from figuring exactly how the antidepressant triggers this cell growth.
Experiments with genetically engineered mice have suggested that a particular "receptor" on the neuron plays a key role in this process, but much work remains to confirm this.
There is no certainty that a human brain will respond in a similar way to a mouse brain.
However, Dr Tony Cleare, from the Institute of Psychiatry told BBC News Online that the theory made sense.
He said: "We know that the hippocampus can be smaller in people with chronic depression.
"What we don't know is if it can recover."