Fresh research may help explain why regular exercise can improve brain power, say Cambridge scientists.
The report, which was published in the Proceedings of the National Academy of Sciences, found mice which exercised performed better on memory tests.
These mice also grew more new cells in a part of the brain linked to memory than those which did not exercise.
The authors believe the new brain cells were behind the improvement in cognitive performance.
The aim of the study, which was carried out by scientists from the Department of Experimental Psychology at the University of Cambridge and researchers at the National Institute on Aging in Baltimore, was to find out why exercise might improve brain function.
Previous research had suggested that exercise helps mental performance in both people and animals. Studies had also shown that exercise increases the number of new brain cells in rodents.
The new finding in this study is that mice which exercise are better able to distinguish between memories of similar things. The authors believe this is explained by the additional brain cells generated by exercise.
The study was conducted on two groups of mice over a period of 105 days. The mice were trained to touch a box on a computer screen to get food pellets.
One group were then allowed unlimited access to an exercise wheel. They ran over 20km (12 miles) a day on average. The control group were not able to exercise.
Both groups were then repeatedly shown two boxes on a screen, one of which provided a treat when it was touched.
The mice learned which box released the treat, and then the boxes were moved around. First the boxes were moved close together, which made it harder for the mice to remember which one to touch to get the food.
The exercising mice did better on this task than the non-exercising mice.
The task was then made easier by placing the boxes further apart so that they seemed more distinct. This time there was no difference in the performance of the exercising and non-exercising mice.
"Keeping similar memories distinct is an important part of having a good memory" says the senior author of the study, Timothy Bussey from Cambridge University.
"It is this aspect of memory that is improved by exercise, our study shows.
"The human equivalent might be remembering which car parking space you have used on two different days in the previous week. It becomes difficult to distinguish memories when events are similar."
By the end of the experiment, the animals which exercised had more than twice as many new brain cells as those that did not.
These cells were in the hippocampus, an area of the brain which is important in memory and learning.
The Cambridge team believes the results of their study may well extrapolate to humans, a view shared by another researcher who studies the impact of exercise on memory.
Stan Colcombe, from Bangor University, said: "Their data strongly suggest that new neurons created after exercise can play a role in improving cognitive function, which likely has direct implications for human research into the effects of exercise on neurocognition."
He described the research as "a very elegant experiment" which "made a valuable contribution in understanding the effects of exercise on brain health and function".