Open AccessAN AGENT‐BASED EVENT DRIVEN FORAGING MODEL
Author(s) -
ANDERSON JAMES J.
Publication year - 2002
Publication title -
natural resource modeling
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.28
H-Index - 32
eISSN - 1939-7445
pISSN - 0890-8575
DOI - 10.1111/j.1939-7445.2002.tb00080.x
Subject(s) - foraging , event (particle physics) , moment (physics) , computer science , mode (computer interface) , agent based model , term (time) , animal behavior , property (philosophy) , point (geometry) , ecology , artificial intelligence , mathematics , biology , human–computer interaction , zoology , philosophy , physics , geometry , epistemology , classical mechanics , quantum mechanics
. A game‐theoretic foraging model is developed in an agent‐based framework. The animal's environment is described in terms of agents representing prey, predators and the habitat. The animal has two modes of behavior for each agent: tactical mode behaviors alter the outcomes of the events and strategic mode behaviors alter the probability of future events. Possible event outcomes are represented as utilities which have both benefit and cost components. The probabilities of successful behaviors are tracked via short‐term (tactical), intermediate‐term (strategic) and long‐term (reference) memories. Probabilities change continually through the moment‐to‐moment events, and switches in behavior occur when the expected utility of one behavior exceeds that of another. At the switch point, the expected utilities of the behaviors are equivalent; through this property the model parameters can be calibrated from behavioral experiments. A bluegill foraging study on giving‐up time is used to calibrate the model and then habitat foraging is simulated. Problems with interpreting behavioral studies without fully characterizing the multiple scales of interactions with agents are discussed.