Skip to content
MAGICS Lab

Northeastern University

Design & Governance of Intelligent Complex Systems

MAGICS = Multi-AgentIntelligentComplexSystems

MAGICS Lab takes a sociotechnical perspective on the design and governance of complex engineering systems. We model how agents co-evolve and adapt — to each other and to their environment — and how individual behavior and collective system-level dynamics shape one another, bringing together technical design, behavioral modeling, and policy.

Explore the research Key papers

Mission

A sociotechnical lens on complex systems.

MAGICS Lab takes a sociotechnical perspective toward the design and governance of complex engineering systems, bringing together technical design, models of human behavior, and policy and regulation.

A throughline of our work is modeling the behavior of complex systems through co-evolutionary and adaptation models — how agents adapt to one another, and how their behavior and interactions shape collective system-level dynamics, which in turn reshape the agents themselves.

Founded in 2019, the lab evolved from CENS Lab (Complex Evolving Networked Systems), active 2011–2018. Our work spans multi-agent reinforcement learning, network science, behavioral modeling, and the strategic behavior of human and AI agents in sociotechnical settings.

MAGICS Lab overview — three research areas (Human and AI Agents; Network and Interaction Architecture; System Behavior and Dynamics) arranged around a central MAGICS Lab network motif, with a four-stage approach strip below (Understand, Model, Design, Govern).

The lab in one figure — three research areas around a multi-agent network motif, anchored to a four-stage approach.

Three research areas

Where we work.

Three intersecting research areas, each with its own methods and questions, and each feeding into the others.

Human & AI Agents

Modeling, learning, and decision-making of social and AI agents — from boundedly-rational humans to large language models acting under strategic incentives.

Network & Interaction Architecture

Structure, modularity, information flow, and interaction architecture in networked sociotechnical systems — and how those choices shape what the system can do.

System Behavior & Dynamics

Emergent structures and behavior, social norms, resilience, stability, coordination, and cooperation — the dynamics that ride on top of structure and agent design.

Key papers

Highlights.

A reinforcement-learning manager observes aggregated outcomes from a network of LLM agents and, each step, selects an information-disclosure policy that modulates what each agent sees in its prompt. The agents' interaction graph is fixed; only the information layer is reshaped.

ASME Journal of Mechanical Design, 148(4): 041708 · 2026

Adaptive Information Modulation: Designing Governance Mechanisms for Multi-Agent Artificial Intelligence Systems

Govern a multi-agent LLM system by modulating what each agent sees — not who it talks to.

Qiliang Chen, Sepehr Ilami, Nunzio Lorè, Babak Heydari

Three-layer model diagram: inventor types layer (heterogeneity in search distance and adoption approach), organization-design layer (core-periphery networks for information sharing), and performance-landscape layer (a rugged NK-style landscape), with arrows describing how types are allocated to network positions and how agents search and feed back performance scores.

Strategic Management Journal · 2026

Core or Periphery: Examining Where to Allocate Heterogeneous Inventors and the Impact on Firms' Innovation

Where heterogeneous inventors sit in a firm's network changes what the firm can discover.

Babak Heydari, Shinjinee Chattopadhyay, Soumyakant Padhee, Samina Karim

Three-panel summary: (1) study goal — what shapes LLM strategic behavior more, game structure or contextual framing; (2) experimental setup — four games × five framings × GPT-3.5, GPT-4, LLaMa-2 producing cooperate-or-defect choices; (3) key findings — GPT-3.5 is context-driven, GPT-4 is structure-centric and bimodal, LLaMa-2 balances both.

Nature Scientific Reports · 2024

Strategic Behavior of Large Language Models and the Role of Game Structure versus Contextual Framing

Do LLMs reason strategically, or just react to framing? Both — and the mix depends on the model.

Nunzio Lorè, Babak Heydari

All highlights

Updates

Latest news.

  • Prof. Heydari joined the Network Science Institute as Core Faculty.

    Babak Heydari has joined Northeastern's Network Science Institute as Core Faculty, deepening the lab's ties to the broader NSI community.

  • Prof. Heydari promoted to Full Professor.

    Babak Heydari has been promoted to Professor in the Mechanical & Industrial Engineering Department at Northeastern University's College of Engineering.

  • New publication in the Journal of Mechanical Design: "Architecting adaptive networks."

    Q. Chen and B. Heydari publish in the Journal of Mechanical Design (148(1): 011701) — reinforcement learning with generative policies for multi-agent governance.

  • New paper accepted at the Journal of Mechanical Design: "Adaptive Information Modulation."

    Q. Chen, S. Ilami, N. Lorè, and B. Heydari design governance mechanisms for multi-agent AI systems — accepted to the Journal of Mechanical Design.

All news