Vervet monkey (Chlorocebus pygerythrus) behavior in a multi-destination route: Evidence for planning ahead when heuristics fail

Animal paths are analogous to intractable mathematical problems like the Traveling Salesman Problem (TSP) and the shortest path problem (SPP). Both the TSP and SPP require an individual to find the shortest path through multiple targets but the TSP demands a return to the start, while the SPP does not. Vervet monkeys are very efficient in solving TSPs but this species is a multiple central place forager that does not always return to the same sleeping site and thus theoretically should be selected to find solutions to SPPs rather than TSPs. We examined path choice by wild vervets in an SPP experimental array where the shortest paths usually differed from those consistent with common heuristic strategies, the nearest-neighbor rule (NNR-go to the closest resource that has not been visited), and the convex hull (put a mental loop around sites, adding inner targets in order of distance from the edge)-an efficient strategy for TSPs but not SPPs. In addition, humans solving SPPs use an initial segment strategy (ISS-choose the straightest path at the beginning, only turning when necessary) and we looked at vervet paths consistent with this strategy. In 615 trials by single foragers, paths usually conformed to the NNR and rarely the slightly more efficient convex hull, supporting that vervets may be selected to solve SPPs. Further, like humans solving SPPs, vervets showed a tendency to use the ISS. Paths consistent with heuristics dropped off sharply, and use of the shortest path increased, when heuristics led to longer paths showing trade-offs in efficiency versus cognitive load. Two individuals out of 17, found the shortest path most often, showing inter-individual variation in path planning. Given support for the NNR and the ISS, we propose a new rule-of-thumb termed the "region heuristic" that vervets may apply in multi-destination routes.