From Play, by Stuart Brown:

One biologist who studied river otters decided to train some of them to swim through a hoop by offering a food reward for completing the task. Shortly after the otters learned to do this, the animals started introducing their own twists to the task. They swam through the hoop backward and waited to see if they got a reward. They swam through and then turned around and swam back through the other way. They swam halfway through and stopped. After each variation, they waited expectantly to see if this version of the task would earn a reward or not.

… By having fun and mixing it up, the otters were learning far more about the way the world works than if they had simply performed the initial task flawlessly. … The biologist ruefully noted that he had been trying for years to get his graduate students to use such playful investigation rather than rote learning and mechanical thinking in their research.