Abstract:
Kea (Nestor notabilis) are a large-brained and highly social species of parrot adapted to living in the New Zealand alps. Both their ecology and evolutionary history suggest this species may have evolved flexible and complex cognition. Over the course of this thesis, I investigate several aspects of the physical, numerical, and social cognition of this species. First, I provide the first evidence of self-care tooling in a kea through repeated observations of a disabled individual. I then contextualise this behaviour phylogenetically across all parrot species, demonstrating how crowdsourcing can be used as a method to increase our detection of rare animal behaviours and trace their likely evolutionary history. Next, I find that, despite quickly learning to pull up baited vertical strings, kea have a poor understanding of connectivity in a horizontal loose-string task. The subsequent chapter demonstrates that kea simultaneously represent the identity and trajectory of two hidden objects and can predict the end points of incomplete object trajectories. I then build on these findings, and harness the signature-testing framework, to investigate kea’s ability to judge probabilistic events. I find that kea make statistical inferences based on relative frequencies and integrate physical and social information into their predictions in a domain-general manner. My next chapter examines how naïve kea perceive virtual stimuli relative to the real world and finds that kea expect physical processes in the virtual and real worlds to be equivalent and continuous, helping validate the use of virtual stimuli to test animals’ cognitive abilities. In my final two chapters, I overview how studies of avian cognition can contribute to important debates in comparative psychology and offer practical applications for how research on kea cognition can help the conservation of this endangered species.