Learning to learn: advanced behavioural flexibility in a poison frog

Behavioural flexibility is essential for survival in a world with changing contingencies and its evolution is linked to complex physical and social environments. Serial reversal learning, in which reward contingencies change frequently, is a key indicator of behavioural flexibility. While many vertebrates are capable of serial reversal learning, only birds and mammals have previously been shown to use rule-based decision strategies (e.g. win-stay/lose-shift) to become better at learning changes in reward contingencies across reversals. While the lifestyles of many amphibians have a degree of complexity, the evidence to date suggests limited levels of behavioural flexibility. Here, we show that the poison frog Dendrobates auratus, which has evolved complex parental behaviours that likely depend on remembering locations in a flexible manner, can use a win-stay/lose-shift strategy to increase their behavioural flexibility across sequential changes in the reward contingencies in a visual discrimination task. Furthermore, probe trials demonstrate that the frogs used the provided visual cues to spatially orient in the maze in a manner reminiscent of complex spatial cognition. Our study provides the first evidence of serial reversal learning in frogs and is the first to demonstrate the use of a rule-based learning strategy in a nonavian, nonmammalian species.