Route following by one-eyed ants suggests a revised model of normal route following

Abstract

The prevailing account of visually controlled routes is that an ant learns views as it follows a route, while guided by other path-setting mechanisms. Once a set of route views is memorised, the insect follows the route by turning and moving forwards when the view on the retina matches a stored view. We have engineered a situation in which this account cannot suffice in order to discover whether there may be additional components to the performance of routes. One-eyed wood ants were trained to navigate a short route in the laboratory guided by a single black, vertical bar placed in the blinded visual field. Ants thus had to turn away from the route to see the bar. They often turned to look at or beyond the bar and then turned to face in the direction of the goal. Tests in which the bar was shifted to be more peripheral or more frontal than in training produced a corresponding change in the ants paths, demonstrating that they were guided by the bar, presumably obtaining information during scanning turns towards the bar. Examination of the endpoints of turns away from the bar suggest that ants use the bar for guidance by learning how large a turn-back is needed to face the goal. We suggest that the ants zigzag paths are an integral part of visually guided route following. In addition, on some runs in which ants did not take a direct path to the goal, they still turned to face and sometimes approach the goal for a short stretch. This off-route goal facing indicates that they store a vector from start to goal and use path integration to track their position relative to the endpoint of the vector.