Iain D. Couzin, Ph.D. - US grants

In this project the PI will study swarms experimentally and theoretically from the perspective of non-equilibrium statistical physics. The project will implement a fully controlled laboratory setup for interactive swarming experiments with fish and also use previously collected data on the collective motion of locusts. The project will contrast both systems using detailed, and idealized, swarming models to study the specificities and universality of their dynamics. Statistical physics concepts will be applied to search for generic characteristics of swarms that stem from common survival constraints. Detailed digital tracking of all individuals in a swarm will be used to investigate biologically meaningful swarming states. Comparisons will be made with equilibrium and non-equilibrium physical systems such as liquid crystals, spin-systems and granular materials. The project will achieve this by: 1. Building controlled laboratory experiments where hundreds, or thousands, of fish will be tracked concurrently and their response to stimuli recorded. 2. Developing detailed, and idealized, models to capture specific and universal aspects of swarms, respectively, and to develop new (physics-inspired) analytical tools for swarms such as correlation lengths, clustering, information flow, defect dynamics etc. 3. Investigate collective effects that are typical of non equilibrium dynamical systems such as scaling-laws, pattern formation, or transitions. 4. Testing if some swarming states follow a generic behavior due to the system adaptations and, if so, exploiting the consequences of this for theoretical analysis. The research will improve our understanding of swarming systems in particular and self-organization in general by combining experimental and theoretical approaches from biological and physical perspectives. It will set up new experiments and statistical measurements that can be used beyond the proposed context. It will explore the interactions between mechanical effects and behavioral adaptation across scales, furthering our understanding of non-equilibrium systems. In addition we will pursue a universal description of certain swarming states by hypothesizing that evolved critical-like behavior may exist. The interdisciplinary and collaborative nature of this project will be emphasized through the development of a strong outreach program for the public understanding of science. Swarming demonstrations capture the imagination and allow meaningful and creative interactions allowing the fundamental principles of collective behavior to be explored from a range of complimentary perspectives.

The oceans are one of the most dynamic environments on Earth, presenting a profound challenge for understanding the complex social, ecological, and physical interactions that occur within them. Fishers are naturally tuned to this complexity and meet their individual goals by targeting their efforts towards particular locations and types of catch, and by adapting their social interactions. Yet we - the scientific community - lack an understanding of how social behavior and ecological dynamics are coupled. Further, these feedbacks are largely ignored in present management approaches. We aim to fill this knowledge gap by answering three questions: (1) How does fisher social behavior (defined as the level to which individual fishers share information) change in response to ecological, technological and management factors? (2) What effect does social behavior have on fisher spatial dynamics and on the social structure of the fishing community? (3) How can management strategies be designed to account for the social behavior of fishers? To answer these questions we will conduct a comprehensive research project involving data gathering and analysis, theoretical modeling and the development of novel mathematical theory. This project will have direct implications for agencies responsible for managing marine resources along the west coast of the U.S. and in Hawaii, some of whom are collaborating in the research. By increasing our understanding of how humans using a natural resource interact with one another and how this in turn affects that resource, the results of this study will be relevant to fields as diverse as finance and global food security.

We propose to gather data on the spatial and behavioral dynamics of fishers along the U.S. west coast, in Brazil and in Fiji - three marine systems with contrasting social, ecological, technological and management characteristics. U.S. data will come from collaborations with the NOAA National Marine Fisheries Service, and data from Brazil and Fiji will be obtained using economic field experiments. All three data sources will be used to develop agent-based models that simulate both the dynamics of fish and individual fishers. We will distinguish ourselves from traditional modeling approaches by adopting a Complex Adaptive Systems (CAS) perspective. With a CAS perspective the spatial dynamics of fish and fishers and the social structure of fisher communities, along with other macroscopic properties, emerge from processes operating at low levels of organization, namely the actions of individual fishers and their target species. Our agent-based models will have the CAS perspective at heart, with fisher agents able to adapt and learn different behavioral strategies (e.g. sharing or not sharing information). A combination of variation, fitness and reproduction will create a selective mechanism whereby fisher behaviors converge to evolutionary stable types. We will complement our agent-based modeling with evolutionary game theory and investigate why certain social behaviors are evolutionarily stable in some marine systems and not in others. Last, we will use mechanism design theory to develop management strategies that account for changes in fisher social behavior.

This project examines how animals in groups collectively learn about their environment and how this learning process may result in "collective intelligence", whereby the group can make more accurate decisions than any single individual can. Most social animals, such as fish, have very limited means of communication and coordination, and it is currently unknown whether these animals can learn to exploit the potential collective intelligence under these constraints. This project will develop a theoretical model of collective learning, and preliminary results suggest that the same learning rules known to be ubiquitous in solitary animals can also be highly efficient in a group context and can lead to near-optimal levels of collective intelligence, even with limited coordination between individuals. The specific predictions of the collective learning model will then be experimentally tested using two social animal species: fish and humans. By using two very different species, the generality of the model can be tested, and general mechanisms of collective learning will be uncovered. These experiments will provide important insights into the learning process in animal groups, which has previously been unexplored. They may also give specific insight into how humans learn in a group context and may suggest techniques to increase the efficiency of collective learning in human groups.

This is an Institutional Infrastructure proposal to build a GPU cluster to support research in data-parallel code development and optimization, as well as research applications, in three scientific domains, namely, seismology, biology and astrophysics. These goals build on a close collaboration with an expert team in GPU computing from computer science. The proposed cluster will serve not only as an invaluable resource for computation, but will also aid cross-fostering of techniques and concepts between disciplines and will be used to stimulate collaboration and synergistic research activity in a wide range of areas.

Even though domain scientists are increasingly dependent on computation to achieve their research goals, most are not experts in parallel programming or GPU architectures. The difficulty of parallel programming for GPU clusters is an impediment to scientific progress. In order to relieve scientists of the burdens of parallel programming, computer scientists at Princeton have developed systems for automatically parallelizing programs for GPU. Building on this success, the PIs plan to extend these techniques to GPU clusters and work closely with the seismologists, biologists and astrophysicists to accelerate the pace of science.

A fundamental problem in biology is understanding how complex systems work from interactions among its components. In many animal groups, such as schooling fish or flocking birds, coordinated movements are thought to result from self-organizing social interactions among individuals. However, little is known about the social cues that the organisms within groups pay attention to and how they integrate them during decision-making, or how these interactions produce collective behavior. The proposed research combines novel experimental and computational approaches to investigate the relationship between sensory input and collective behavior using zebrafish as an experimental model. The research focuses on vision, which is essential in the coordination of many animal groups. The researchers will: (a) characterize key dimensions of the visual system, (b) use this sensory information to model the visual saliency of social cues, (c) incorporate the zebrafish visual information in current models that track the motion of individuals in real shoals, (d) develop and implement new technologies to investigate behavioral responses, and (e) measure these behavioral responses under experimental conditions that manipulate the visual saliency of group mates. Ultimately, this project will increase our understanding of how and why organisms coordinate their behavior in groups. The PI and collaborator will disseminate the results and technologies to the broader scientific community through publications, and give public lectures to broad audiences and schools.

Effective information transfer is essential for the coordination of behavior within intracellular, neuronal, social and economic networks. In many animal groups, such as schooling fish and flocking birds, the degree and speed of inter-individual communication allows individual group members to make fast and accurate collective decisions across a range of contexts, and often under conditions of considerable risk. Such emergent properties are highly desirable for many technological applications, including coordinated search, control and response by groups of robotic agents. This project will employ an experimental approach to map the relationship between sensory input and behavioral output in schooling fish under a range of ecologically-relevant scenarios in order to identify the dynamic networks of sensory communication in groups and relate this to the elementary movement behaviors exhibited by individuals, and the highly effective collective behavior exhibited by groups. These results will directly inform collective robotics. Undergraduate and graduate students will be involved in all aspects of this project and it will be integrated into classes taught by Dr. Couzin. Dr. Couzin will also develop a summer learning module on collective behavior for high-achieving, low-income high school students from local school districts in NJ through the Princeton University Preparatory Program (PUPP) and continue to work with National Geographic digital media and National Geographic Learning to engage the public


This project will reveal the complex structure of the networks underlying information flow in groups. The predominant paradigm has been to consider individuals in such groups as "self-propelled particles", which interact with neighbors through "social forces". A major limitation of this approach is that it neither considers the sensory information available to animals when making movement decisions within groups nor considers that organisms make decisions in a state - dependent and probabilistic fashion. To map the relationship between visual and lateral line sensory input and resulting behavior in schooling fish under ecologically-relevant conditions that vary in timescale and type of response, the PI will use custom software to determine the location, body posture and eye positions of members of the group to reconstruct the visual fields of all individuals in groups of up to several hundred fish. Bayesian, unsupervised learning and inverse methodologies will be used to identify the visual information used by individuals and to map the structure of social response facilitated by the lateral line. Multi-scale network analysis will be used to identify important properties and meaningful motifs/substructures within groups, and to relate these to collective capabilities. Information transfer across sensory networks will be quantified using information theoretic techniques under the different ecological contexts. These data will inform subsequent manipulations of individual behavior to test predictions about how groups filter noise and respond to extraneous cues. From this work the researchers will create new models of collective animal behavior.