Learning Foraging Thresholds for Lizards:
An Analysis of a Simple Learning Algorithm

Leslie Ann Goldberg, William E. Hart, and David Bruce Wilson

This paper gives a proof of convergence for a learning algorithm that describes how anoles (lizards found in the Caribbean) learn a foraging threshold distance. An anole will pursue a prey if and only if it is within this threshold of the anole's perch. The learning algorithm was proposed by Roughgarden and his colleagues. They experimentally determined that this algorithm quickly converges to the foraging threshold that is predicted by optimal foraging theory. We provide analytic confirmation that the optimal foraging behavior as predicted by Roughgarden's model can be attained by a lizard that follows this simple and zoologically plausible rule of thumb.

Journal of Theoretical Biology, 197:361--369.
DVI version
PostScript version

A preliminary version appeared in Proceedings of the 9th Conference on Computational Learning Theory, pp. 2--9, 1996, and is Copyright © 1996 by ACM, Inc.