Deborah Gordon's long-running study at a site in Arizona followed marked Pogonomyrmex barbatus colonies for over three decades. The finding that emerged is that foraging activity is regulated locally. A returning forager carries a seed and a cuticular hydrocarbon signature that differs from that of ants doing other tasks. An ant waiting in the tunnel just inside the entrance samples these contacts, and a higher rate of returns raises her probability of going out.
Because a forager returns only when she finds a seed, the contact rate is a proxy for how well foraging is going. In dry conditions, when searching costs water, colonies that restrain foraging under low return rates do better, and Gordon's work found this restraint is heritable and associated with reproductive success.
The result overturns the intuitive picture. There is no signal from the queen, who lays eggs and does not direct work. What looks like a decision by a colony is a statistical outcome of thousands of brief antennal contacts, none of which carries an instruction.