An ant of few words: group communication with limited vocabulary
Biological individuals often interact to form cooperative societies that have functional advantages.
How the specifics of these interactions constrain collective performance is not well
understood. In this context, we study how desert ants inform each other about the presence
of food. We use automated tracking to generate a large data-base of ant trajectories and interactions
that provides us with sufficient statistics to empirically estimate the efficiency of their
communication. This is done, quantitatively, by calculating the information theoretical channel
capacity of the ants' pairwise interactions. We find that this channel is noisy to a degree
that makes it difficult for ants to tell between a recruiter reporting about food and a random
collision within the dark nest environment. To distinguish these ambiguous signals the colony
must therefore perform error-correcting on the level of the group. We demonstrate that the ants
accomplish this by exhibiting strict control of when to transmit a message and when to respond
to received information. This control leads a collective process that couples negative and positive
feedbacks and ensures reliable colony performance. Thus, the ants need no language, but
just one aptly used "word" pronounced with conviction inside a noisy environment.