|
Multi-Scenario World Description |
 |
 |
 |
|
| Model name: | Multi-Scenario World | Model title:
| Norm Recognizers acting in a Multi Scenario World | Replicated model?
| Yes | | Keywords: | Agent Architecture, Normative Agent, Norm Recognition Module, Simulation | Model authors:
| Marco Campennì, Giulia Andrighetto, Federico Cecconi, Rosaria Conte
| | Programming language? | Matlab |
In our simulation model, the environment consists of four scenarios, in which the agents can produce three different kinds of actions. We define two context specific actions for every scenario, and one action common to all scenarios. Therefore, we have nine actions. Each of our agents is provided with a personal agenda (i.e. a sequence of contexts randomly chosen), an individual and constant time of permanence in each scenario (when the time of permanence is expired, the agent moves to the next context) and a window of observation (i.e. a capacity for observing and interacting with a fixed number of agents) of the actions produced by other agents. Agents are also provided with the three-layer architecture described above, necessary to analyze the received information, and a normative board in which the normative beliefs, once arisen, are stored. The agents can move across scenarios: once expired the time of permanence in one scenario, each agent moves to the subsequent scenario following her agenda.
DOCUMENTATION
1. Purpose The Model aims to show the effects of the norm recognition module use in a multi scenario world. 2. State variables and scales World: four different scenarios defined by 9 types of actions (2 scenario-specific + 1 common action); Agents: norm recognition module implemented by a three layers architecture + a normative board.
3. Process overview and scheduling
Agents are randomly paired at each time tick and they exchange messages; they can process received information to generate new normative beliefs (that will affect their future actions, i.e. the subject of future messages)
4. Design concepts
4.1 Emergence The convergence to a single normative belief can emerge 4.2 Adaptation
4.3 Fitness
4.4 Prediction
4.5 Sensing
Agents exchange messages defined by a modal, a subject, a sender and an addressee. 4.6 Interaction Agents are randomly paired at each time tick 4.7 Stochasticity A stochastic approach to initialize the modals sent at the beginning of the simulation (when agents have no normative beliefs in their minds) is used
4.8 Collectives
4.9 Observation
5. Initialization
Agents properties are randomly initialized (agenda, time of permanence)
6. Input
7. Submodels
|
|
|
Multi-Scenario World Results |
|
|
|
|
Figures File
| Title | Description
| Figure 1
| Social input | | Figure 2
| The norm recognition module (in action) | On the right side of the figure, from the bottom the Input and the two layers of the module (layer 0 and layer 1) plus the normative belief (generated or recognized); on the left side, the normative board. Vertical arrows in the block on the right side indicate the process regulating the generation of a new normative belief. The input action ( alpha) can match with a norm present in the normative board (see the arrows path on the left side of the figure); or a new normative belief can be formed if the agent receives an input action ( alpha) (at least one time as deontic or normative valuation) for a given number of times (as fixed by the normative threshold; see the arrows path on the right side of the figure). If the agent receives no other occurence of the same input action ( alpha), after a fixed time t action exits from the higher level and the process is finalized (see Exit). | | Figure 3/a | Number of agents | Number of agents in each context runtime - with external barrier. | Figure 3/b
| Number of agents | Number of agents in each context runtime - without external barrier. | | Figure 4/a | Overall number of new normative beliefs | Overall number of new normative beliefs generated for each type of possible action - with external barrier. | | Figure 4/b | Overall number of new normative beliefs | Overall number of new normative beliefs generated for each type of possible action - without external barrier. | Figure 5/a
| New normative beliefs generated runtime | New normative beliefs generated runtime - with external constraint. | Figure 5/b
| New normative beliefs generated runtime | New normative beliefs generated runtime - without external barrier. | Figure 6/a
| Number of performed actions | Number of performed actions - case with barriers. | Figure 6/b
| Number of performed actions | Number of performed actions - case without barriers. | Figure 7/a
| Convergence rate | Convergence rate - case with barriers.
| Figure 7/b
| Convergence rate | Convergence rate - case without barriers. | Figure 8/a
| Number of performed actions | Number of performed actions - case with barriers.
| | Figure 8/b | Number of performed actions | Number of performed actions - case without barriers. | Figure 9/a
| Convergence rate | Convergence rate - case with barriers.
| Figure 9/b
| Convergence rate | Convergence rate - case without barriers. | Videos Textual Results File
| Description
| results1 (.mat MATLAB 5.0 file)
| Results using parameter array: 100, 100, 4, 1, 3, 10, NR, ND, EB
| | results2 (.mat MATLAB 5.0 file) | Results using parameter array: 100, 100, 4, 1, 3, 10, NR, YD, EB | | results3 (.mat MATLAB 5.0 file) | Results using parameter array: 100, 200, 4, 1, 3, 10, ND, EB | | results4 (.mat MATLAB 5.0 file) | Results using parameter array: 100, 200, 4, 1, 3, 10, YD, EB | Publications - Campenni, M., Andrighetto, G., Cecconi, F., Conte, R. (2008) Normal = Normative? The Role of Intelligent Agents in Norm Innovation. ESSA 2008.
|
|
|
Multi-Scenario World II. Description |
|
|
|
| Model name: | Multi-Scenario World - EMIL-S | Model title:
| Norm Recognizers acting in a Multi Scenario World | Replicated model?
| Yes | | Keywords: | Norm emergence, normative agent | Model authors:
| Marco Campenni, Steffi Henn, Peyman Jazayeri, Ulf Lotzmann, Michael Moehring, Magnus Oberhausen, Mehmet-Hadi Tohum and Jannik Weyrich | | Programming language? | EMIL-S, TRASS
|
This model is the EMIL-S replication of the Multi-Scenario World model. In this model, the agents repeatedly traverse through four different stations of a virtual airport. At each station the agents have to choose between two possible behaviours. Two types of agents are defined: social conformers and norm recognizers.
DOCUMENTATION
1. Purpose Simplifying the highly complex process of behaviour by different actors in the social context, this scenario is a simulation model of the interaction of agents of two different kinds. On the one hand, there are social conformers who strictly react on their environment and copy the commonly observed action of others without recognizing the norm from which the decision is arising. On the other hand, norm recognizers have knowledge about the existing norm in social matters, and their actions are based on this knowledge. Joining the different agents in a simulated environment reveals the consequences of acting based on different principles. 2. State variables and scales In this model the social behaviour of two types of agents, the “norm recognizers” and the “social conformers”, is analysed and compared. The agents are interacting within a shared environment, representing several stations of an artificial “airport”, and for each of the stations two context-depended actions are defined:
- baggage claim:
- take luggage
- do not take luggage
- customs:
- declare goods
- do not declare goods
- taxi stand:
- queue in the line
- pass the queue
The agent population consists of varying proportions of the two agent types.
3. Process overview and scheduling The agents repeatedly pass through the three stations in a fixed order. Each cycle is equal with one single visit of the airport.
The norm recognizer agents make the decisions for actions to perform on base of their experiences (induced by norm invocations) from former visits. The social conformer agents draw on the current behaviour of the majority of the agent population.
4. Design concepts 4.1 Emergence Emergence of norms in the artificial society of norm recognizer regarding the behaviour within each of the three stations. 4.2 Adaptation Agents adapt their behaviour by changing their action preferences, based on received feedback from the environment or received messages from other agents (punishment). 4.3 Fitness The fitness of an agent relates to it’s norm compliant behaviour. 4.4 Prediction No prediction in this model. 4.5 Sensing Agents are able to perceive their environment consisting of topographical elements and other agents. 4.6 Interaction Individuals interact “physically” within the topographical environment and by sending and receiving norm invocation messages. 4.7 Stochasticity The action selection involves a stochastic process. Furthermore, the agent mobility is influenced by random variables.
4.8 Collectives Group of norm recognizers, group of social conformers. 4.9 Observation Results are shown as animation sequences (during the simulation run using TRASS) and as output graphs (using the experimentation environment MEME). 5. Initialization All agents are initialized with uniform distributed selection probabilities.
6. Input The input parameters of a simulation run are:
Number of agents. Default: 15 agents.
Social conformer proportion. Applied values: 0.0, 0.33, 0.66. 7. Submodels None. |
|
|
|
|
<< Start < Prev 1 2 3 4 5 6 Next > End >>
|
|
Page 5 of 6 |
