Collective Matter
Mechanics of Collective Matter
What happens when thousands of living agents such as cells, insects or microbes, pack together so tightly that they begin to behave not as individuals, but as a material? The answer, it turns out, is one of the most fascinating open problems at the intersection of physics, biology, and engineering. We call this regime collective matter: networks of active agents that move, reorganize, and respond to their environment as a unified whole. The examples span an extraordinary range of scales. A confluent cell layer, where cells are packed edge to edge with no gaps, behaves like a soft solid that can flow, fracture, and heal. A swarm of honeybees clusters into a living droplet that sways and stabilizes against wind, with no individual bee in charge. A raft of fire ants, swept into floodwater, assembles into a buoyant, self-healing material in minutes, a collective structure whose dynamics and properties emerge entirely from the behavior of its members.Â
Different types of collective matter (a) flock of sheeps, (2) fire-ant raft, (3) bacterial aggregation and (4) cell monolayer
Collective matter differs from conventional materials in two fundamental ways. First, it is fueled. Unlike a polymer network driven by thermal noise, a collective of living agents is powered by chemical energy (ATP in cells and metabolic activity in insects). This energy input allows the system to do something thermodynamics would otherwise forbid: consume energy to spontaneously create order, generate motion, and maintain organized structures far from equilibrium. Second, it is informed. Each agent is not a passive particle but a small computational unit which receives signals from its neighbors, processes them, and acts accordingly. A cell senses the stiffness and motion of adjacent cells and adjusts its own contractility and migration. A fire ant detects the local density and mechanical state of the raft and decides whether to hold, move, or release. This local information processing, replicated across thousands of agents simultaneously, produces a completely decentralized form of intelligence, one in which global organization emerges from local rules, with no central coordinator and no blueprint. The result is matter that can self-organize, adapt, and morph in ways that no engineered material can yet match.
Understanding collective matter requires a framework that couples physical forces and motion with the decision-making of individual agents. This is not classical mechanics ( as agents are not passive) Â nor is it pure biology (as interactions are fundamentally mechanical). It sits at the boundary between the two, and it demands new theoretical tools. In our approach, each agent is described by its mechanical state (including position, velocity, deformation) and its internal state, which governs how it processes information and makes decisions. The coupling between these two levels produces the rich phenomenology of collective matter: jamming transitions, where a flowing aggregate suddenly arrests into a solid-like state; collective migration, where thousands of cells align and move in the same direction without any leader; and morphogenetic flows, where a tissue spontaneously folds, extends, or branches into a new shape. These behaviors are not programmed into any individual agent, rather they emerge from the network of interactions between them. Our work concentrate on two systems: Confluent monolayer of endothelial cells and fire-ant rafts, each with their own medical/engineering applications.Â
Confluent Cell Layers and Tissue Morphogenesis
In a developing tissue or a healing wound, cells are often densely packed into confluent monolayers, where every cell is in permanent mechanical contact with its neighbors. This seemingly simple geometry conceals extraordinary complexity. The layer can behave like a solid, resisting deformation and maintaining shape, or like a fluid, flowing and reorganizing over time. The transition between these states (a.k.a the epithelial jamming transition) is controlled by a delicate balance between cell-cell adhesion, cortical tension, and the active forces that cells generate internally. We concentrate on endothelial cells, the cells that line the inner surface of blood vessels. Their collective organization is not merely structural as it directly controls vascular permeability, inflammatory response, and the remodeling of the vessel wall. Understanding how endothelial monolayers self-organize, how they respond to flow and mechanical strain, and how disruptions in their collective behavior contribute to vascular disease is a central question of our research.
Fire Ant Rafts and Decentralized Swarm Robotics
Fire ants provide one of nature's most striking demonstrations of collective matter. When floods strike, fire ant colonies self-assemble into buoyant rafts within minutes and can persist for days on the water surface. Despite the absence of any centralized control, the raft behaves like a superorganism: it contracts and expands, extends and retracts long protrusions to explore its environment. After detecting a solid surface within reach, the raft can further morph into a bridge, creating a living passageway for the colony. All of this unfolds while the raft maintains its solid-like mechanical integrity, continuously adapting its shape without ever losing cohesion. Even more striking, the bonds between individual ants are remarkably weak, yet the raft resists fracture and heals any defect almost instantaneously.
No central conductor is responsible for this complex behavior. Instead, the intelligence of the raft emerges entirely from local interactions between its members. Finding those rules is the central question of our research. In other words, can we find the minimal set of local behaviors that, replicated across the colony, gives rise to a material that can think, move, and heal?
This decentralized intelligence has direct implications for robotics. A swarm of simple robots, each following local rules, could collectively assemble into structures, navigate complex environments, and respond to damage, without centralized control, without communication overhead, and without a single point of failure. With teams of robotics collaborators, we aim to use fire ant rafts as a model system to extract the physical and behavioral rules that govern collective assembly, with the goal of translating them into design principles for decentralized robotic swarms.