Skip to main content icon/video/no-internet
icon backReturn to Entries

Agent-Based Models

Agent-based modeling is a simulation technique that relies on the capabilities of individual actors, called agents, in order to model a global behavior. In an agent-based model (ABM), complex, system-level behavior emerges from the local action of, and interaction among, a large number of heterogeneous agents. While the literature contains conflicting definitions of the word agent, researchers who work with agent-based models believe that an agent is an entity that can take action, has an internal model of the world, possesses information-processing capabilities, and has an internal logic controlling its actions. The relationships between agents, the social and spatial topology in which agents are embedded, and the logic that guides agent behavior play a crucial role in determining the overall behavior of the system. Global outcomes emerge as heterogeneous agents interact and engage in various local activities.

Intellective and Emulative Models

Though ABMs have grown in popularity in the last half-century, the technique was used in some of the earliest computers; in fact, some forecasting attempts that predate the modern computer might be considered forerunners of the modern-day, agent-based model. With the popularization of ABMs in the last 20 years has come a vast expansion in the number and types of models. A key dimension along which models differ is the intellective-emulative dimension; intellective models are designed to test a specific and narrow hypothesis or illustrate a point, while emulative models usually contain more agents and are designed to encapsulate and evaluate alternative theories and to set policy directions. Intellective models are often built using specialized languages and frameworks designed to facilitate the development of simple models (such as RePast, Net-Logo, MASON, and Swarm), while emulative models may be written in more general-purpose programming languages (such as C++ and Java).

Despite such differences, all ABMs have a number of common features. In all models, there is a set of heterogeneous agents, there is an environment, the agents are capable of processing information about the environment and making decisions, the user can alter the set of agents and the environment by altering the values on a set of parameters, and global consequences of local decisions are observed. The environment may be spatial, and so agents change behavior based on their physical location; or the environment may be social, and so agents change behavior based on their social position. Generally, there is a random element, such as the probability that two agents who find themselves in the same space will interact; hence, multiple replications of any virtual experiment are needed to generate a good understanding of the emergent behavior. Serious analysis of any model's results, or exploration of its parameter space, will require statistical estimation of the model's response surface. The higher the fidelity of the ABM, the more complex the statistical modeling required for assessment.

Advantages of the Agent-Based Model

There are a number of advantages to investigating a research problem by building or extending an agent-based model. All simulation techniques, including agent-based modeling, are key tools for theory development as they force researchers to encode their assumptions when writing models and to question previously hidden assumptions in theories. This process allows a researcher or policy maker to realize the limitations of a particular theory or solution, or conversely, to develop extensions of a theory into a new domain or to develop a solution that is more robust. When building an agent-based model, the simulation designer will have full control over which types of data will be gathered and can be modified relatively easily if followup virtual experiments are performed. The data gathered will not be subject to the kinds of cognitive or methodological biases found in empirical research.

...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles

Read next

More like this

Sage Recommends

We found other relevant content for you on other Sage platforms.

close

Have you created a personal profile? Login or create a profile so that you can save clips, playlists and searches

close
close
close Signed In to profile You are signed in as:
close
Logged in Institution Institution