Logo IMG


Even Birdbrains Learn from Experience

Cockatoos open a complex series of locks to achieve a distant goal.

Jenny Jennings Foerst

2013-11SciObsJenningsF1.jpgClick to Enlarge ImageFigaro was only a birdbrain: a captive cockatoo (Cacatua goffini). But when Alice Auersperg, cognitive biologist at the University of Vienna, spied him wielding 10 tools he had spontaneously invented from twigs and splinters, she was amazed. Such innovative tool use was known to exist only in primates and New Caledonian crows.

Auersperg first saw Figaro make, shorten, and bend stick tools to rake in a small stone he had dropped through a mesh partition. She replaced the stone with a series of cashew nuts and observed Figaro repeat his performance. Auersperg’s observations culminated in a recent PLoS One study showing that cockatoos can solve elaborate multistep lock puzzles, without intermediate behavioral reinforcements, and immediately transfer their new knowledge to a novel challenge.

See a video of a cockatoo solving a complex lock puzzle.

Figaro shows that cockatoos can “plan interventions in the physical world well ahead of being reinforced,” says Auersperg’s coauthor, Alex Kacelnik, a behavioral ecologist at the University of Oxford. Figaro’s achievement raised the question, To what extent can cockatoos plan ahead, completing a series of actions toward a mentally represented distant goal—without support? Auersperg and colleagues reasoned that if a cockatoo must complete a chain of actions to receive a substantial reward at the end, and if each action leads only to the possibility of achieving the next action, then the bird would be unlikely to attain the final goal by mere chance. Randomly attempting the actions in the wrong order, for example, would lead to failure.

To test cockatoos’ planning and mechanical capacities, Auersperg designed a box housing a visible cashew nut blocked by five interlocking devices. The locks were concatenated so that the bird would have to solve the lock puzzle farthest from the reward before gaining access to the next, and so on. The team made each lock unique so that the knowledge required to open it would be novel.

Auersperg’s team began by testing the cockatoos’ learning process to discern whether and to what extent the 10 birds in their study could solve all five puzzles without intermittent rewards or observation of birds already trained. The researchers had no precedent for predicting how much time and support the cockatoos would require. In the end, only one bird, named Pipin, solved all five locks within the number of sessions, number of trials, and amount of time that Auersperg used to define success. But Pipin completed the entire puzzle, utterly unassisted, in only 100 minutes—a “rapid route to perfection,” Auersperg notes. The team supported the other nine birds with reinforcement until they opened the five locks.

The team then subjected all 10 birds to an experiment designed to reveal the nature of their newly acquired knowledge: whether it was rote, or it indicated understanding of mechanical functions. The scientists removed or resequenced parts of the original puzzle. In response, “the birds immediately approached the now most relevant piece of equipment downstream from the goal,” says Kacelnik. Instead of rigidly adhering to the original lock sequence, in other words, they strategized according to the various functions of the interlocking mechanisms.

The cockatoos explored the locks playfully, using their bills, tongues, and feet; individuals differed in how they opened the locks. However, all 10 birds progressed logically: Once they had solved a lock, they spent no time on that lock in future sessions, but focused on the next challenge in the chain. This stepwise approach to the complex sequence shows that cockatoos can, without reinforcement, work backward from a distant goal.

Why did the cockatoos differ in their methods and rates of success? “All sorts of questions are raised by the wide variation of performance among individuals,” Kacelnik says. “It’s important to show that one bird can meet the challenge, because that success sets the capability limit for the species,” and “the more variation we see within a species, the more important experience—as opposed to inborn traits—is to the knowledge in question.” Figaro and Pipin have set a high bar for their species, while teaching our own something new and remarkable.—Jenny Jennings Foerst

comments powered by Disqus


Subscribe to American Scientist