The dynamics of landmark navigation in wood ants

Lead Research Organisation: University of Sussex
Department Name: Sch of Life Sciences


Some ants follow long, visually guided foraging routes. Study of their route following shows that they acquire multiple visual memories, retrieve them appropriately during the route and guide their paths with this stored visual information. We will improve our understanding of these processes by analysing landmark guidance on a moment-by-moment basis. For this purpose, we have developed a route along which ants guide themselves by means of a single visual feature that we can perturb at defined points during their path. A video-tracking camera gives a precise record of the ant's path and its body orientation, and lets us infer the landmark's position on the ant's retina. This methodology also opens to experiment a neglected but significant aspect of landmark guidance. Ants are so close to the ground that small bumps will frequently obscure their view of guiding landmarks so that they must be able to maintain a path with only intermittent access to visual input. How do ants navigate over uneven terrain? The information that we will obtain by recording the ant's behavioural responses to landmark perturbations will bring us closer to the underlying mechanisms of visual guidance, and so will be of use to neuroscientists and to computer scientists working on the navigation of autonomous robots. We have found that ants will learn a straight route to food placed at the base of a black-white vertical edge, where the black side fades slowly into the white of the walls of the surrounding experimental arena. In this case, the ant's visual task is simply to keep the edge at the front of the eye throughout its approach. Ants will also learn a straight route to food placed at a short distance to the side of the edge. This task is more taxing as the desired retinal position of the edge shifts from the front towards the periphery of the eye during the ant's approach. Our data suggest that the ant stores a sequence of visual memories (the desired positions of the edge on its retina), and that during its route it retrieves the appropriate memory and guides itself by moving to place the edge at the currently desired retinal position. The retrieval of particular desired edge positions seems to be cued by the gradient's apparent width, a visual parameter that increases reliably during the ant's approach and so can provide a robust retrieval signal. Perturbation studies with gradient edges displayed on an LCD screen will give us data to test and improve these hypotheses. The ant's reaction to abrupt changes in the position of the edge should reveal the current desired retinal position as the null point, where there is a switch in the direction of the ant's corrective response to perturbations of different magnitudes and directions. We will map how the desired edge position changes along the route. Does it change continuously or in a step-wise manner, as predicted by a sequence of discrete memories? By altering the width of the gradient, but keeping edge position constant, we can see whether the ant's desired edge position shifts as predicted by the notion that the width of the gradient determines the ant's currently active memory. Similar experiments can tell us how ants cope with intermittent visual input and the role of motor learning in this process. Thus, we expect ants to show 'inertia' and continue towards the goal for a while after the edge is made to disappear. The learning of a motor trajectory should show itself as inertial effects that increase with experience and an increased sluggishness when the ant tracks oscillating edges. Stronger evidence for trajectory learning will be sought by training ants to perform curved trajectories to an edge that always moves during their approach and examining the curvature of the path when the edge vanishes. To explore how ants behave in more natural conditions, we will analyse landmark guidance when ants walk over bumpy ground.

Technical Summary

We will investigate memory retrieval, landmark guidance and its interaction with learnt motor trajectories in wood ants using a landmark perturbation technique. Ants are so low to the ground that small bumps often obscure their view of landmarks. We will explore how landmark guidance operates in a system that has to cope with intermittent visual input. Ants will be trained to follow a route guided by a single, vertical, white-black edge which can be moved or transformed during a route. Previous experiments with fixed edges show that trained ants follow a straight path between start and food (placed 10 cm to the side of the edge), during which the ant guides itself by moving so that the edge travels in a controlled way from a frontal position on the retina towards the periphery. These and other data suggest that ants store a sequence of desired retinal edge positions, which they retrieve correctly along the route, and that retrieval is cued by some measure related to the perceived angular extent of the intensity gradient going from the black edge to the white of the background. These hypotheses will be tested by determining for what directions and magnitudes of edge displacements ants make no correction, and how the position of this null point changes during the trajectory. To test whether gradient extent controls memory retrieval, we will monitor responses when the edge position is kept constant and the extent of the intensity gradient is changed. Using rectangular landmarks with two vertical edges, we will use similar methods to test whether ants bind the two edges together and measure its width independently of retinal position. To discover how ants progress with intermittent visual input, we will investigate (1) if ants trained to a single edge continue their course when the edge disappears or oscillates; (2) whether ants can be trained to curved routes; (3) how ants navigate and use landmarks when travelling over bumpy ground.


10 25 50
publication icon
Collett M (2009) The learning and maintenance of local vectors in desert ant navigation. in The Journal of experimental biology

publication icon
Collett M (2013) Spatial memory in insect navigation. in Current biology : CB

publication icon
Collett M (2012) How navigational guidance systems are combined in a desert ant. in Current biology : CB

publication icon
Collett M (2009) Local and global navigational coordinate systems in desert ants. in The Journal of experimental biology

publication icon
Collett TS (2014) Scene perception and the visual control of travel direction in navigating wood ants. in Philosophical transactions of the Royal Society of London. Series B, Biological sciences

publication icon
Lent DD (2010) Image-matching during ant navigation occurs through saccade-like body turns controlled by learned visual features. in Proceedings of the National Academy of Sciences of the United States of America

publication icon
Lent DD (2013) Visual scene perception in navigating wood ants. in Current biology : CB

Description Our major finding emerged from the following experiment. Ants followed a short route to an inconspicuous feeder positioned at a fixed position relative to a vertical, luminance edge. Ants responded to an unexpected jump of the edge by turning to face the new feeder position specified by the edge. Importantly, the initial speed of the turn increased linearly with the turn's amplitude. This correlation implies that the ants' turns are driven initially by their prior calculation of the angular difference between the current retinal position of the edge and its desired position in their memorised view. Similar turns keep ants to their path during unperturbed routes. The neural circuitry mediating image matching is thus concerned not only with the storage of views, but also with making exact comparisons between the locations of a visual feature in a memorised view and of the same feature in the insect's current retinal image.
Exploitation Route This finding is both of theoretical interest and has provided us with a good way to investigate visual perception in ants.

The paper describing the work has been of use to others and has been cited 26 times.
Lent, D. D., Graham, P., & Collett, T. S. (2010). Image-matching during ant navigation occurs through saccade-like body turns controlled by learned visual features. Proceedings of the National Academy of Sciences, 107(37), 16348-16353.
Sectors Other

Description Results of research used by other researchers in other instutions
Sector Other
Impact Types Cultural