[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
- [Swarmfest2007] Abstract submission,
Steve Railsback <=