Optimal foraging theory has stimulated numerous theoretical and empirical studies of foraging behavior for >20 years. These models provide a valuable tool for studying the foraging behavior of an organism. As with any other tool, the models are most effective when properly used. For example, to obtain a robust test of a foraging model, Stephens and Krebs (1986) recommend experimental designs in which four questions are answered in the affirmative. First, do the foragers play the same "game" as the model? Sec- ond, are the assumptions of the model met? Third, does the test rule out alternative possibilities? Finally, are the appropriate variables measured? Negative an- swers to any of these questions could invalidate the model and lead to confusion over the usefulness of foraging theory in conducting ecological studies.

Gaines (1989) attempted to determine whether American Tree Sparrows (Spizella arborea) foraged by a time (Krebs 1973) or number expectation rule (Gibb 1962), or in a manner consistent with the predictions of Charnov's (1976) marginal value theorem (MVT). Gaines (1989: 118) noted appropriately that field tests of foraging models frequently involve uncontrollable circumstances; thus, it is often difficult to meet the assumptions of the models. Gaines also states (1989: 118) that "violations of the assumptions are also in- formative but do not constitute robust tests of predicted hypotheses," and that "the problem can be avoided by experimental analyses which concurrently test mutually exclusive hypotheses so that alter- native predictions will be eliminated if falsified." There is a problem with this approach because, when major assumptions of models are not satisfied, it is not justifiable to compare a predator's foraging behavior with the model's predictions. I submit that failing to follow the advice offered by Stephens and Krebs (1986) can invalidate tests of foraging models.