D. M. Gordon, Dept. of Biological Sciences, Stanford University,
How does a colony allocate the appropriate number of ants to the tasks it must perform, and how is this done in the absence of a central coordinating mechanism? This problem is similar to the optimization of task distribution in human engineered parallel distributed systems (PDS), such as interacting robots or microprocessors. Much can be gained by bringing engineering knowledge to the investigation of colony behavior, and biological inspiration to the design of PDS. We show that the parameters of a theoretical model of task allocation in ant colonies can be estimated from empirical data, and that the equilibrium distribution of the model is close to the task distribution of a laboratory ant colony. We also show that the deterministic model does not behave very differently from a stochastic counterpart, even for small colonies. We suggest that colony size may have a strong influence on colony behavior, both on the speed of response to environmental change, and on the relative importance of environment and social interaction in determining task allocation.