Abstract: Huber, Taborsky, and Rechberger (2001) reported an experiment in which the efficiency with which captive keas opened a complex food container was increased by observation of a skilled conspecific. However, only testing social learning in free-ranging animals can demonstrate social learning in natural conditions. For that purpose, a tube-lifting paradigm was developed and tested on keas both in captivity and in Mount Cook National Park, New Zealand. The task was to remove a tube from an upright pole in order to gain access to a reward inside the tube. The top of the pole was higher than a standing kea, so that, to remove the tube, an individual had to simultaneously climb onto the pole and manipulate the tube up the pole with its bill. Because only 1 naive bird managed to remove a tube twice in 25 halfhour sessions and disappeared after success, another bird was trained to solve the task and to provide demonstrations for others. Even under such conditions, only 2 of at least 15 birds learned to remove the tube in 28 sessions. There was no indication that observer birds' use of bill and feet when exploring the tube changed as the number of observations of tube removal increased in a way that would, in principle, increase the likelihood of tube removal. The results suggest a dissociation of social learning potential as assessed in laboratory animals, and social transmission of foraging techniques in natural populations.