Win shifting in nectarivorous birds: selective inhibition of the learned win-stay response

A variety of nectarivorous species have demonstrated a bias to ‘win-shift’ (shift away from/avoid locations that have recently yielded food, as opposed to ‘win-stay’ behaviour where the animal returns to such locations). Since recently exploited flowers contain no nectar, the win-shift bias is a candidate for an adaptive specialization of cognition. This bias tends to manifest as faster learning and/or better performance on a win-shift than a win-stay task. In standard win-shift/win-stay tasks (multiple two-phase trials where animals first find rewards in particular locations, and must subsequently avoid, or return to, such locations) noisy miners, Manorina melanocephala, and rainbow lorikeets, Trichoglossus haematodus, developed patterns of revisits/errors in the exploration phases that corresponded to the spatial contingency they experienced in the test phases: birds reinforced to shift revisited locations in the exploration phase that were unrewarded; birds reinforced to stay revisited rewarded ones. This was true even for birds that failed to shift or stay appropriately in the test phase itself. This suggests that the poorer performance of nectarivorous birds in win-stay than win-shift tasks may not be a consequence of an inability to learn the win-stay contingency. Our results suggest that these birds are equally sensitive to the win-shift and win-stay contingencies. This implies that, if the win-shift bias previously reported in nectarivorous birds is an example of a cognitive adaptation to the depleting nature of nectar, then the specially adapted mechanisms may have more to do with inhibition of the win-stay response than insensitivity to the win-stay contingency.

► Biases to win-shift are common in species exploiting readily depleting resources.
► We examine exploration behaviour of two bird species on shift and stay tasks.
► Birds unable to win-stay after a delay showed ‘stay’ behaviour during exploration.
► Better win-shift performance is not due to insensitivity to win-stay contingencies.