Literal grid map models for animal navigation: Assumptions and predictions.

Many animals can navigate from unfamiliar locations to a familiar target location with no outward route information or direct sensory contact with the target or any familiar landmarks. Several models have been proposed to explain this phenomenon, one possibility being a literal interpretation of a grid map. In this paper we systematically compare four such models, which we label: Correct Bicoordinate navigation, both Target and Release site based, Approximate Bicoordinate navigation, and Directional navigation. Predictions of spatial patterns of initial orientation errors and efficiencies depend on a combination of assumptions about the navigation mechanism and the geometry of the environmental coordinate fields used as model inputs. When coordinates axes are orthogonal at the target the predictions from the Correct Bicoordinate (Target based) model and Approximate Bicoordinate model are identical. However, if the coordinate axes are non-orthogonal different regional patterns of initial orientation errors and efficiencies can be expected from these two models. Field anomalies produce high magnitudes of orientation errors close to the target, while region-wide nonlinearity leads to orientation errors increasing with distance from the target. In general, initial orientation error patterns are more useful for distinguishing between different assumption combinations than efficiencies. We discuss how consideration of model predictions may be helpful in the design of experiments.