In order to test (1) whether kea show signs of insight in a novel situation where the
intermediate step does not provide immediate visual feedback and (2) whether systematically
training the birds for the separate steps is sufficient to solve the complex problem, as
suggested by Epstein’s automatic chaining hypothesis, I developed a new method to test
insightful problem solving and systematically manipulated our subject’s training experience.
The task required the subjects to cover a trap hole with a plastic lid lying next to the apparatus
before releasing treats from a mechanical treat dispenser.
Out of the ten kea enrolled to the study, seven finished all the training and testing phases. The
study proved that the developed task fulfils the criteria for an insight problem task. It was challenging enough for kea to show signs of impasse, and reorganisation of their behaviour is
required, since all our subjects operated the dispenser without covering the hole first.
Compared to previous designs, it has the potential to overcome the issues of visual feedback
and confounding pre-training. As a proof of concept, two of our subjects displayed the
necessary actions in the correct order, showing that the task does not exceed their
I found no evidence of insight or insightful solution, although five out of eight animals
managed to cover the trap hole with the lid.
The experimental design could be used to investigate insight in another manipulative species
and on a larger number of kea.