Path integration provides a scaffold for landmark learning in desert ants.

On leaving the nest [1-9] or a newly discovered food site [10-12] for the first time, bees and wasps perform elaborate flight maneuvers to learn the location of their goal and the lay of the land surrounding it. In all these orientation flights the insects turn back and look [13] at the goal, which they can visually locate by landmark cues directly defining the goal. Here we show that Namibian desert ants, Ocymyrmex, when learning new landmarks in the neighborhood of the goal, acquire this landmark information when they cannot see the goal. They do so by performing well-choreographed rotation movements integrated in spiral-like "learning walks." Within these rotations, short (about 150 ms) stopping phases occur, during which the ants orient themselves in the direction of the nest entrance. On the barren sand surface the nest entrance is invisible, so the ants can aim at it only by reading out the current state of their path integrator [14-17]. Hence, they could associate "snapshot" views [18-20] taken of the nest surroundings during the stopping phases with path integration coordinates. In bees and ants such associations have often been discussed, but evidence has not been obtained yet [15, 20-22]. Copyright (c) 2010 Elsevier Ltd. All rights reserved.