Fish pool their experience to solve problems collectively, according to new research.
They might only have a little bit of information about their environment, but in a group, different animals might have separate but complementary information about a particular problem.
Some may know where to find food but not how to access it.
Others might know how to get at it but not where it is hidden, scientists at St Andrews University found.
In a set of experiments, scientists at the university's school of biology set out to determine whether leadership - the pulling of the group by informed members - could allow groups of animals to pool their experience in order to solve problems collectively.
Their findings, which could have implications for businesses and even bio-inspired swarm robotics, are published on the Nature Ecology & Evolution website.
Dr Mike Webster, of St Andrews University, said: "To tackle this question we presented shoals of stickleback fish with a two-part problem, in which they had to first find, and then access, some hidden food.
"Individual fish were either inexperienced or had experience of just one of the stages.
"We found that in shoals that comprised individuals trained in each of the stages more fish did indeed access the food, and did so more rapidly, compared with other shoal composition which only contained fish trained to one or to neither of two parts of the problem.
"Supporting our idea that leadership played a role in this, we found strong effects of having experienced members in the group, with the presence of these greatly increasing the likelihood of untrained fish completing each part of the problem."
'Lessons to be learned'
Researchers have known that larger groups tend to outperform smaller groups and lone individuals when completing certain tasks.
The new study shows that experience pooling, where subsets of the group assume leadership roles when completing the specific part of a task in which they have knowledge or competence, is a plausible mechanism by which this might happen.
Professor Kevin Laland, of St Andrews University, said: "There may be lessons to be learned for human behaviour too.
"Businesses and institutions already make good use of teams with diverse skills sets, and the natural world might provide further inspiration for how these groups might be put together and organised.
"Finally, artificial intelligence researchers are focusing heavily on bio-inspired swarm robotics, and the kinds of collective information processing mechanisms uncovered by this study might potentially be deployed by other researchers designing software and behaviour rules for fleets of drones."