UC Merced

Foraging is an essential process for all mobile organisms. It allows organisms to locate resources such as food and mates. There is a long history of research on animal foraging in the ecology literature and recent work in cognitive science has revealed similarities between cognitive search behaviors and animal foraging behaviors. This gives rise to the possibility of bringing the rich animal foraging literature to bear on cognitive search processes. Historically there have been several major approaches to the study and modeling of foraging animals. One approach is known as optimal foraging theory which is focused on optimizing the amount of time an organism spends foraging a single location before moving on to another location. Another is an expansion on that approach which operationalizes those ideas into a spatial model known as area-restricted search. A third approach is known as the Lévy flight hypothesis. It focuses on the longer term distributional properties of foraging animals, and optimizes coverage of a given search space.

These approaches all make assumptions about environmental conditions faced by foraging organisms. The degree of resource sparsity and clustering in a foraging environment are believed to be important, but it is unclear how they affect foraging behaviors. Additionally spatial memory is a key concept important to animal search strategies, but is frequently ignored in the existing literature. For this dissertation, a series of experiments utilized a web-based foraging game to test how these variables affect foraging behaviors. The first experiment demonstrated that the degree of clustering in the environment had a significant impact on search strategies, and provided qualitative evidence that memory played a role in people’s search behaviors. The experiments revealed distributional patterns very similar to those predicted by the Lévy flight approach. The second experiment refined the method and directly tested memory cues and a broader range of resource densities. This experiment revealed results similar to the first with the addition of significant effects of both memory cues and resource density. This dissertation then discusses a model that combined key concepts from both optimal foraging and Lévy foraging to produce results very similar to those produced by human participants, but with significantly higher performance. Experiment 3 examined how human performance changes when specific advantages are provided that can be found in our model, including perfect memory and accuracy.

Finally a continuation of the model is discussed that explores the dynamics of multiple foragers searching the same space. Overall, I demonstrate that people will generate Lévy-like search distributions in a wide variety of environmental conditions, but that search strategies will alter based on the current environment. I also demonstrate that spatial memory is a key factor in foraging, and provide a simple memory-based model that produces foraging behavior very similar to those utilized by people.