Frequent errors are the worst: robustness to individual failures in collective foraging

Abstract

The collective behavior of complex systems emerges from the actions and interactions of their individual components. But what if these individuals make mistakes, or deviate from behavior tuned to lead to collective success? Here, we explore the effects of individual errors on collective outcomes. We investigate in particular whether information flow among individuals exacerbates or mitigates such individual failures. We use an agent-based, spatially explicit model inspired by collective foraging in social insects. Social insect colonies forage for food with autonomous workers who search for and exploit resources around the central nest, as well as share information about discovered resources. We find that the errors that have the most chances of occurring had the strongest impacts: for example, false positive detections can occur at any time during search, and each such error derailed exploration activity. Similarly, forgetting errors are potentially frequent and detrimental to resource exploitation. Despite the fact that communication inherently may narrow the breadth of information used by a colony, we found, in contrast, that it enhanced spatial exploration in our model. Communication in our model also reduced the effects of individual errors, instead of permitting erroneous information to spread. Our model thus illustrates that communication plays a central role in error management in complex systems, and that the evolution of communication systems in social insects may be shaped by selection on exploration and error mitigation as well as on efficient food retrieval.