For the study, which recruited 40 pet dogs of varying breeds, psychologists from Yale’s Canine Cognition Center placed a treat inside a puzzle, then demonstrated to their subjects how to get it out. In reality, the puzzle was just one step — all the dogs had to do was lift the lid of a box — but the researchers added an extra, unnecessary action to their demo, pushing a lever attached to the box that didn’t actually do anything. To make sure the dogs were really trying to solve the task in front of them, rather than following a perceived command, the study authors then left the room and left the animals to their own devices.

The dogs, who each went a couple rounds with the puzzle, proved adept at figuring out not only what they needed to do, but also what they didn’t: As the experiment progressed, they began disregarding the lever, going straight for the step that would get them their treat.

The study offers an interesting insight into dog cognition in its own right, but it also has another layer: The authors based their study on a similar one from 2005 that focused on children instead of dogs — and compared to the dogs, the kids weren’t nearly so savvy. Their puzzle was more complicated, but they tended to repeat the experimenters’ actions step for step each time, without ever pausing to think through or weed out the irrelevant ones. It’s a tendency the authors of this latest study refer to as “overimitation,” writing: “This pattern of results suggests that overimitation may be a unique feature of human social learning,” possibly because by uncritically copying what they see, “children generally limit the amount of time they need to spend learning through repeated trial and error.”