Possible uses of path integration in animal navigation

Path integration, in its simplest form, keeps track of movement from a starting point and so makes it possible to return to this point. Path integration can also be used to build a metric spatial representation of the environment, if given a suitable readout mechanism that can store and recall the coordinates of any one of multiple locations. A simple averaging process can make this representation as accurate as desired, given enough visits to the locations stored in the representation. There are more than these two ways of using path integration in navigation. They can be classified systematically according to the following three criteria: Is there one point at which coordinates can be reset to correct errors, or several? Is there one possible goal, or several? Is there one path integrator, or several? I describe the resulting eight methods of using path integration and compare their characteristics with the available experimental evidence. The classification offers a theoretical framework for further research.