swarm-swarmfest2007
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Swarmfest2007] Abstract submission


From: Steve Railsback
Subject: [Swarmfest2007] Abstract submission
Date: Fri, 01 Jun 2007 09:42:00 -0700
User-agent: Thunderbird 2.0.0.0 (Windows/20070326)

Here is my abstract, which I will withdraw if we need the space.

Steve

Name: Steve Railsback
Affiliation: Lang, Railsback & Associates and Humboldt State University
Email: address@hidden
Mailing: 250 California, Arcata CA, USA 95521

Requested presentation type: Oral

Title: State-based Predictive Theory for Modeling Adaptive Behavior of Animals (and People?)

Abstract: One of the key challenges of agent-based modeling is finding useful models of agent behavior. Ideally we would have theory: general concepts that can be applied to specific models. In ecology, adaptive behavior of individuals often requires tradeoffs between the benefits of feeding more (growth; reproducing sooner and more successfully) and the risks (greater exposure to predation). The widely-used "State-based dynamic modeling" theory assumes individuals make these tradeoffs by optimizing an explicit measure of future fitness, e.g. the expected number of offspring at a future reproduction season, equal to the probability of surviving to the season times the number of offspring produced. The fitness measure depends on both the benefits of feeding (avoiding starvation mortality, producing more offspring) and the risks (probability of dying due to predation before the reproduction season). However, the fitness measure can be optimized only if future conditions (food availability, predation risk) are completely predictable. In realistic agent-based models, future conditions are typically unpredictable and subject to feedback from individual decisions. "State-based predictive" theory solves this problem by assuming individuals use a simple prediction of future conditions to make adaptive decisions, and update their predictions and decisions as conditions change. This theory has reproduced a wide variety of observed behaviors and trophic interactions in simulated fish...could it also be useful for simulating human behavior?

--
Steve Railsback
Lang, Railsback & Associates
Arcata, California
www.LangRailsback.com



reply via email to

[Prev in Thread] Current Thread [Next in Thread]