Coordination, Organizations, Institutions, Norms, and Ethics for Governance of Multi-Agent Systems XV: International Workshop, COINE 2022, Virtual ... (Lecture Notes in Computer Science, 13549)

This book constitutes the thoroughly refereed post-conference proceedings of the International Workshop on Coordination, Organizations, Institutions, and Norms for Governance of Multi-Agent Systems, COINE 2022, which was held in Auckland, New Zealand, on May 9, 2022. The 14 papers included in these proceedings were carefully reviewed and selected from 15 submissions. They deal with autonomous agents and multi-agent systems, focusing on the scientific and technological aspects of social coordination, organizational theory, artificial (electronic) institutions, and normative and ethical MAS. Preface Organization Consent as a Foundation for Responsible Autonomy (Abstracts) Contents Designing International Humanitarian Law into Military Autonomous Devices*-12pt 1 Introduction 2 The Law and the Operational Framework 2.1 The Targeting Cycle 2.2 Integrating the Law into the Cycle 3 Timing and Application of Legal Tests 4 The General Framework of the Autonomous Targeting System 5 The Structure of the System 5.1 The Basics of the Model 5.2 The Structure 6 Discussion and Conclusions References Epistemic Diversity and Explanatory Adequacy in Distributed Information Processing*-12pt 1 Introduction 2 Background: DIP, RTSI and Plato's Cave 2.1 Distributed Information Processing Units (DIPs) 2.2 Regulatory Theory of Social Influence (RTSI) 2.3 Problem Specification 3 Experimental Design 3.1 Initial Conditions for the DIP (Independent Variables) 3.2 Epistemic Condition and Explanatory Adequacy 4 Formal Specification 4.1 The Environment 4.2 Agent Specification 4.3 Algorithm 5 Experimental Results 5.1 Experiment 1: Static and Dynamic Populations 5.2 Dynamic Population, Progressive Addition, non-Persistence 5.3 Dynamic Population, Progressive Addition, Persistence 5.4 Summary of Experiments 6 Related and Further Research 7 Summary and Conclusion References The Complexity of Norm Synthesis and Revision 1 Introduction 2 Preliminaries 3 Complexity of Norm Synthesis 3.1 Prohibition Synthesis 3.2 Obligation Synthesis 4 Complexity of Synthesising a Set of Norms 5 Complexity of Minimal Revision 5.1 Complexity of Minimal Prohibition Revision 5.2 Complexity of Minimal Obligation Revision 6 Related Work 7 Conclusions References Embracing AWKWARD! Real-Time Adjustment of Reactive Plans Using Social Norms 1 Introduction 2 Background 2.1 Behaviour Oriented Design 2.2 OperA 3 The AWKWARD Architecture 3.1 The AWKWARD Planner 3.2 The OperA Module 3.3 Behaviour Library 4 Implementation in DOTA2 4.1 DOTA2 4.2 The Reactive Planner Module and Behaviour Library 4.3 The OperA Module Implementation 5 Results 6 Related Work 6.1 Behaviour-Based AI 6.2 Normative Agents and Self-organisation 6.3 Hybrid Approaches with Reactive Planning 7 Conclusions and Future Work References Self-learning Governance of Black-Box Multi-Agent Systems 1 Introduction 1.1 Motivation 1.2 Illustration of the Governance Approach 1.3 Contribution 2 Related Work 3 Model 3.1 Notation 3.2 Multi-Agent System 3.3 Governance 3.4 Sequence of Actions in a GMAS 3.5 Degree of Restriction 4 Model Analysis 4.1 Fairness 4.2 Learning 4.3 Stationarity 5 Experimental Evidence 5.1 The Dining Diplomats' Problem 5.2 Reward Functions 5.3 Configurations 5.4 Frameworks and Algorithms 5.5 Reproducibility 5.6 Results 5.7 Discussion 6 Conclusion and Future Work References Computational Theory of Mind for Human-Agent Coordination 1 Introduction 2 Abstraction Heuristics and Affinity 2.1 Computational Affinity 2.2 Computational Affinity and ToM 3 Negotiation with Computational Affinity and ToM 3.1 Negotiation Framework 3.2 Negotiating with Affinity and ToM 4 Experiments and Results 4.1 The Effect of Affinity on Agreements 4.2 Affinity Types and Agreement Rates 4.3 Roles of ToM and Affinity in Agreements 5 Discussion and Future Work References Computational Discovery of Transaction-Based Financial Crime via Grammatical Evolution: The Case of Ponzi Schemes 1 Introduction 2 General Framework 3 Generation of Ponzi Schemes 4 Experiments and Results 5 Discussion 6 Conclusions References Centralized Norm Enforcement in Mixed-Motive Multiagent Reinforcement Learning*-12pt 1 Introduction 2 Related Work 3 Normative Systems and the ADICO Grammar of Institutions 4 Reinforcement Learning (RL) 4.1 Single-Agent Reinforcement Learning 4.2 Multi-Agent Reinforcement Learning (MARL) 5 Centralized Norm Enforcement in MARL 6 Tragedy of the Commons Experiment 6.1 A neMG of a Tragedy of the Commons Environment 6.2 Results and Discussion 7 Conclusion References Supporting the Reasoning About Environmental Consequences of Institutional Actions 1 Introduction 2 Artificial Institutions 3 The Purposes of Status-Functions 3.1 Definitions 3.2 Functions and Algorithms 4 Implementing the Purpose Model 5 Conclusions and Future Work References Social Motives and Social Contracts in Cooperative Survival Games 1 Introduction 2 Scenario and Related Work 3 Simulator Design 3.1 Health Modelling 3.2 Global Utility 3.3 Treaties 4 Agent Design 4.1 Social Motives 4.2 Food Consumption 4.3 Handling Treaties 5 Experimental Results and Discussion 5.1 Simulation A 5.2 Simulation B 5.3 Simulation C 5.4 Simulation D 6 Summary and Conclusions 6.1 Summary 6.2 Future Work 6.3 Conclusion References Evaluating Human and Agent Task Allocators in Ad Hoc Human-Agent Teams 1 Introduction 2 Related Work 3 Hypotheses Development 4 Collaborative Human-Agent Taskboard (CHATboard) 5 Methodology 5.1 Interaction Protocols 5.2 Agent Characteristics 5.3 Evaluation Metrics 5.4 Experimental Configurations 6 Experimental Results 7 Discussion and Future Work References Fleur: Social Values Orientation for Robust Norm Emergence 1 Introduction 2 Related Works 3 Fleur 3.1 Cognitive Model 3.2 Emotion Model 3.3 World Model 3.4 Social Model 3.5 Decision Module 4 Experiments 4.1 Experimental Scenario: Pandemic Mask Regulation 4.2 Experimental Setup 4.3 Hypotheses and Metrics 4.4 Experiments 4.5 Threats to Validity 5 Conclusions and Directions References Reasoning About Collective Action in Markov Logic: A Case Study from Classical Athens*-12pt 1 Introduction 2 Common Knowledge 3 Points of Lycurgus During The Trial 4 Markov Logic Networks (MLNs) 5 An MLN Model of Lycurgus's Argument 6 Results and Discussion 7 Conclusion and Future Work References Design Heuristics for Ethical Online Institutions 1 Introduction 2 The Conscientious Design Story so Far 2.1 The WIT Design Pattern 2.2 Conscientious Design Value Categories (CD-VCs) 3 The Easyrider Online Institution 3.1 Goals and Values 4 Making Values Operational 4.1 Three Heuristics for Structuring Value Operationalisation 4.2 Heuristics for the Choice of Values 4.3 Heuristics for Value Imbuing 4.4 Heuristics for Value Assessment 5 Closing Remarks References Author Index