Pigeons (Columbia livia) use local compared to global spatial information during a traveling salesperson task

Past work has shown that pigeons can take efficient routes during traveling or spatial optimization problems in laboratory environments. Whereas pigeons are relatively efficient travelers during these tasks, it remains unclear whether the pigeons primarily rely upon local information in the array of locations (such as a nearest-neighbor [NN] solution that minimizes the distance their current location to their next location), or if they are able to use the global geometry an array of locations to take the most optimal routes. In the current study, we examined the paths that pigeons took that included five locations: a start box, three different feeders, and an end box. We tested the pigeons with pairs of mirrored test configurations of feeders that were similar, but for which the end box was in a different position. If pigeons were attending to the global features of the array, then they should take the most efficient global route based on the end location, compared with a less efficient NN route based on minimizing the distance of the current leg. Overall, the pigeons were efficient and took routes that were more effective than a random traveler. The results also showed that the pigeons tended to take NN rather than global routes. Different testing environments or reward structures may lead to differences in how pigeons differentially weigh the use of local versus global information.