Simulating Heterogeneous Farmer Behaviors
The Pressure , Response , and Impact Systems
The remaining parts of the model define the environment that influences the cognitive processing of the population of agents . Each agent contributes to the Pressure system of other agents by acting as a social referent . When an agent ’ s attitude toward a behavior increases , it sends out referrals to every other agent in its social network . These referrals contain the sending agent ’ s level of belief reinforcement at the time . The receiving agent incorporates the reinforcement into its SN ( t ) calculation , not its AB ( t ). Thus , one agent ’ s AB ( t ) is impacting another agent ’ s SN ( t ). Note also that the receiving agent may not have a high motivation to comply with the sending agent , so the impact to the receiving agent ’ s BI ( t ) could be positive or negative . At the start of the model , each agent is randomly assigned a network of other agents and the configuration of this network can be controlled by the user .
The Response system has two components : the messengers and the influential figure . Each messenger is randomly connected to some subset of the population of agents and those agents in turn have a random motivation to comply with the messenger . Information sent by the messenger is incorporated into the receiving agent ’ s AB ( t ) or PBC ( t ). The idea being that persuasive messaging seeks to alter an individual ’ s attitude toward a behavior by informing the individual of associated outcomes or factors present or espousing the benefits of those outcomes or factors to increase the individual ’ s evaluation . The user can instantiate as many messengers as desired and change the start time , duration , and rate at which messages are sent . The rate is used to parameterize a Poisson process for generating periodic messages .
The influential figure is an optional addition to the Response system . The influential figure is essentially a super-connected agent that consistently sends out referrals advocating on behalf of the messengers . The user can control what percentage of the population is connected to the influential figure , along with the start and stop times , and the rate at which the influential figure sends referrals . The user can also control the maximum and minimum values for motivation to comply with the influential figure . Since the influential figure is sending referrals , it impacts the receiving agent ’ s SN ( t ) rather than AB ( t ) or PBC ( t ).
The Impact system is simply a queuing service that is only open for a user-specified interval of time . When an agent reaches its threshold for action , it submits a service request that contains a randomly selected expiration date . The Service then checks the queue of service requests periodically and fulfills any requests it removes from the queue while it is open . The queue is first-in-first-out , so fulfillment of requests depends on how other agents are reacting to the environment . Many requests will result in a full queue that the Service may not have enough time to process completely . If the Service is closed when an agent attempts to submit a request or the agent ’ s request expires before it is filled , there is a negative impact on that agent ’ s PBC ( t ). Conversely , if an agent is able to submit the request and it is subsequently completed , the result has a positive impact on the agent ’ s PBC ( t ).
Model Propagation and Run Time
A final feature of the model is that the agents can lengthen the run time by continuing to send referrals to their social network after the messaging campaign has stopped . The propagation of referrals and
210