Studies investigating the same paradigm but employing different methods are often directly compared in the literature. One such paradigm used to assess behavioural flexibility in animals is reversal learning. Commonly, these studies require individuals to learn the reward contingency of either solid objects presented on the ground or images presented on a touchscreen. Once learned, these contingencies are swapped. Researchers often refer to trials required to reach learning criteria from different studies, to compare the flexibility of different species, but rarely take methodological differences into account. A direct evaluation of the validity of such comparisons is lacking. To address this latent question, we confronted kea, an alpine parrot species of New Zealand and known for its behavioural flexibility, with a standard reversal learning paradigm on the touchscreen and a standard reversal learning paradigm with solid objects. The kea required significantly more trials to reach criterion in the acquisition and the reversal on the touchscreen. Also, the absolute increase in the number of trials required for the reversal was significantly greater on the touchscreen. This indicates that it is not valid to compare learning speed across studies that do not correspond in the addressed methodology. Taking into account the kea's ecology and explorative nature we discuss stimulus abstraction (limited depth cues and tactile stimulus feedback) and the spatial relation between reward and stimulus on the touchscreen as possible causes for decreased inhibition in this condition. Contrary to the absolute increase in number of trials required for the reversal, the increase in relation to the acquisition was greater with solid objects. This highlights the need for further research on the mechanisms involved causing methodology-dependent differences, some of which we discuss, in order to increase the validity of interpretations across studies and in respect to the subject's ecology.
Keywords: Nestor notabilis; behavioural flexibility; discrimination learning; kea; reversal learning; touchscreen.