Research

My research has largely focused on collective phenomena. Given simple interactions between agents, what complex collective behavior can arise?

Rational behavior during epidemics

Faced with a dangerous epidemic, humans might spontaneously social distance to reduce their probability of infection. Early theoretical work in the field had modeled behavioral changes with heuristics, e.g. relating the strength of social distancing in a compartmental model directly to, say, the current disease prevalence.

Later work studied such behavior as a self-organized phenomenon: Behavioral change affects the course of the epidemic, and in turn again modifies people’s behavior. It is natural to investigate the problem as a game: Individuals adjust their behavior in order to optimize an objective or utility function that quantifies expected infection and social distancing costs, given the state of the epidemic. Decision making motivated by such objective functions is known as rational in the economics literature. This rational decision making self-consistently gives rise to population behavior and therefore the course of the epidemic. Such games exhibit Nash equilibria in which no individual can deviate from the population behavior and expect to increase their utility.

We are working to deepen the understanding of such epidemic games with the goal of providing insights to individuals and policymakers alike that might help with better decision making in future epidemics.

Understanding Nash epidemics

Previously, the analytic understanding of Nash equilibria in epidemics had been extremely limited, leaving us reliant on numerical solutions. Here, we identified an exact analytic expression for fully time-varying Nash equilibrium behavior and resultant disease dynamics. We leverage this analytic solution to show that the strength of rational social distancing is proportional to both the perceived infection cost and prevalence. Remarkably, this gives a posteriori justification for the sort of simple heuristics developed to understand diseases like HIV. We discuss how this and additional analytic results provide a deep and intuitive understanding of the disease dynamics, useful for both individuals and policymakers. In particular, the relationship between social distancing and infections represents a heuristic that could be communicated to the population to encourage, or “bootstrap,” rational behavior.

Epidemics with Altruism

During epidemics, social distancing allows susceptible individuals to protect themselves (and others) but confers no personal benefit for infected individuals if recovery provides immunity. However, individuals are likely to be at least weakly altruistic and may be interested in protecting others when infected. A strongly altruistic population where individuals value others as equal to themselves would be expected to self-isolate when infected. This would strongly suppress the disease, avoid Herd Immunity, and vastly improve outcomes. Still, little was known about how weaker altruism affects behavior during epidemics.

In this work, we show using game theory that even extremely weakly altruistic individuals, valuing their own lives equivalent to roughly 100,000 others, can rationally achieve almost identical outcomes. Individuals self-isolate in order to avoid setting off chains of infections that they would perceive as costly to them even at such small altruism. Our results are robust to a moderate fraction of asymptomatic cases or completely selfish individuals.

The resulting behavior, while emerging from a complex optimization problem, is simple enough that it could have evolved as a behavioral response in social animals, as well as being easy to communicate and understand for humans.

  • M.P. Lynch (co-lead), S.K. Schnyder (co-lead), J.J. Molina, R. Yamamoto, M.S. Turner, The theory of epidemics with altruism, PNAS 123 (9)
    e2518893123 (2026).

How should governments deal with spontaneous behaviour of populations and limited capacity of health services?

During epidemics, how and when should society impose behavioral restrictions in order to reduce infections? Policymakers must balance a complex spectrum of objectives, suggesting a need for quantitative tools. Whether health services might be ‘overwhelmed’ has emerged as a key consideration. We showed how costly interventions, such as taxes or subsidies on behavior, can be used to exactly align individuals’ decision making with government preferences even when these are not aligned.

In order to achieve this, we developed a nested optimization algorithm of both the government intervention strategy and the resulting equilibrium behavior of individuals within a mean-field game based on a compartmental infectious disease model. We focused on a situation in which the capacity of the healthcare system to treat patients is limited and identify conditions under which the disease dynamics respect the capacity limit.

What constitutes rational behavior in the anticipation of a vaccine?

Social distancing strategies of individuals will depend on information about the course of the epidemic but also on when they expect the epidemic to end, for instance due to vaccination. Typically it is difficult to make optimal decisions, because the available information is incomplete and uncertain. We showed how rational decision-making depends on information about vaccination timing. We assumed that the arrival time of a vaccine is uncertain and that only its probability distribution is known. We predict stronger social distancing the earlier the vaccination is expected and also the more sharply peaked its probability distribution. In particular, individuals will only exhibit stronger social distancing (as compared to when they don’t expect a vaccine to become available) if a vaccine is expected to arrive well before the epidemic would have naturally run its course. We believe these results provide meaningful guidance to people and policymakers for navigating epidemics. In particular, our work makes clear that vaccination development schedules should be communicated to the public as early and clearly as possible in order to induce self-organised social distancing.

  • S. K. Schnyder, J. J. Molina, R. Yamamoto, M. S. Turner, Rational social distancing in epidemics with uncertain vaccination timing, PLoS One 18, e0288963 (2023).

Collective dynamics in cellular tissues

Cells coexist together in colonies or as tissues. Their behavior is controlled by an interplay between intercellular forces and biochemical regulation. We develop a simple model of the cell cycle, the fundamental regulatory network controlling growth and division, and couple this to the physical forces arising within the cell collective. We analyze this model using both particle-based computer simulations and a continuum theory. 2D colonies confined in a channel develop moving growth fronts of dividing cells with quiescent cells in the interior. The profile and speed of these fronts are nontrivially related to the substrate friction and the cell-cycle parameters, providing a possible approach to measure such parameters in experiments. This model also enables a quantitative match to growing MDCK colonies. We then studied the competition of two colonies with different rates of programmed cell death (apoptosis) and characteristic cell-cycle control pressures. Synchronisation of division and apoptosis events can emerge, causing oscillations in pressure and cell-cycle activity.

Then, we developed a deep learning framework for estimating the underlying equations of motion from observed trajectories of interacting biological agents such as cells. Our framework integrates graph neural networks with neural differential equations, enabling effective prediction of two-body interactions based on the states of the interacting entities. We successfully applied this approach to synthetic data generated from a model with non-reciprocal forces that mimics the collective dynamics of the cells of slime molds.

Earlier, we had built a minimal mechanical model for migrating and proliferating cells that incorporated dynamically polarised cell shapes. Contact inhibition of locomotion (CI), a process by which cells either stop locomotion or change direction when coming in contact with other cells, was captured in the model as whether the cell is able to polarise itself by extending itself in the direction of motion by applying a force against neighbouring cells to make space. Despite the simplicity of the model, the collective migration behaviour of the cells is highly complex, while being in agreement with experiments. For instance, our work suggests that the reason why crawling cells often exhibit broad fronts is because this helps facilitate collective alignment. In a version of the model where cells exhibit ameboidal shape oscillations, we demonstrate that the interplay of directed motion, shape oscillations, and excluded volume enables cells to locally ‘‘synchronize’’ their motion and thus enhance collective migration.

  • S. K. Schnyder, J. J. Molina, Y. Tanaka, R. Yamamoto, Collective motion of cells crawling on a substrate: roles of cell shape and contact inhibition, Scientific Reports 7, 5163 (2017).
  • M. Campo, S. K. Schnyder, J. J. Molina, Th. Speck, R. Yamamoto, Spontaneous Spatiotemporal Ordering of Shape Oscillations Enhances Cell Migration, Soft Matter 15, 4939 (2019).
  • S. K. Schnyder, J. J. Molina, R. Yamamoto, Control of cell colony growth by contact inhibition, Scientific Reports 10, 6713 (2020).

Anomalous Transport in Porous Media

In heterogeneous materials, such as in biological cells, porous media or ionic conductors, anomalous diffusion is observed, i.e. the mean-squared displacement (MSD) of a tracer particle grows sub-diffusively over long periods of time. This implies that their dynamics cannot be described with the diffusion equation. A paradigm for modelling of transport in heterogeneous media is the Lorenz Model, in which a single mobile particle moves in a static matrix built of overlapping hard disk obstacles. In this system, anomalous diffusion is the signature of a localisation transition caused by the percolation of the matrix. This being a dynamic critical phenomenon, the exponent of the anomalous diffusion should have a universal value. Still, in experiments and simulations a wide range of exponents are reported, and it was unclear if those exponents represented transient behaviour, or whether they can be connected to universal behaviour in some way.

We generalised the Lorentz Model in molecular dynamics simulations by introducing interacting tracer particles and soft interactions between all particles. This made the system comparable to realistic systems. We showed that due to the soft interactions the localisation transition becomes rounded and that the universality of the dynamics breaks down. Cooperatively, particles can overcome finite potential barriers, even speeding up the dynamics when the tracer density is increased. As a result, the anomalous exponent become tuneable. Such a rounding is impossible for hard disks, thus presenting the rare situation where soft and hard interactions yield qualitatively different dynamics.

My simulations were used to interpret an experiment of a heterogeneous porous medium as formed by a binary mixture of colloids confined between plates. In this system and simulations, we directly demonstrated the rounding of the localisation transition and investigated the resulting dynamic heterogeneities.

  • T. O. E. Skinner, S. K. Schnyder, D. G. A. L. Aarts, J. Horbach, R. P. A. Dullens, Localization Dynamics of Fluids in Random Confinement, Physical Review Letters 111, 128301 (2013).
  • S. K. Schnyder, T. O. E. Skinner, A. Thorneywork, D. G. A. L. Aarts, J. Horbach, R. P. A. Dullens, Dynamic heterogeneities and non-Gaussian behaviour in 2D randomly confined colloidal fluids, Physical Review E 95, 032602 (2017).

Earlier, we discussed the Lorentz Model in the framework of the mode-coupling theory of the glass transition. We derived asymptotic results for the order parameter as well as the dynamic correlation functions and the mean-squared displacement close to the transition. The influence of an infrared cutoff is discussed. We then compared the mode-coupling theory predictions with computer simulations and found that certain aspects were surprisingly well predicted.

Classical Liquids in Fractal Dimension

We introduce fractal liquids by generalizing classical liquids of integer dimensions to a noninteger dimension. The particles composing the liquid are fractal objects and their configuration space is also fractal, with the same dimension. Realizations of our generic model system include microphase separated binary liquids in porous media, and highly branched liquid droplets confined to a fractal polymer backbone in a gel. Here, we study the thermodynamics and pair correlations of fractal liquids by computer simulation and semianalytical statistical mechanics. Our results are based on a model where fractal hard spheres move on a near-critical percolating lattice cluster. The predictions of the fractal Percus-Yevick liquid integral equation compare well with our simulation results.

Structure factors in a two-dimensional binary colloidal hard sphere system

We present a comprehensive set of measurements of the static structure factors for quasi-2D monodisperse fluids and two different binary colloidal hard sphere mixtures: a small size ratio (SSR) system with a negligibly small negative non-additivity and a large size ratio system with a significantly larger non-additivity. We compare the experimental results for the monodisperse and SSR systems to those calculated using density functional theory (DFT) for additive mixtures. Furthermore, we determine the zero-wavevector limits of the static structure factors for the monodisperse and binary hard sphere fluids directly from an analysis of number and concentration fluctuations. For the monodisperse case, this leads to the isothermal compressibility, which agrees very well with DFT, and is consistent with the scaled particle theory equation of state for hard disks. For the binary fluids, the partial static structure factors are used to calculate the Bhatia–Thornton structure factors, and we find qualitative agreement with DFT for the SSR mixture. Finally, the zero-wavevector limits of the Bhatia–Thornton structure factors are determined and directly related to the thermodynamic factor, the dilatation factor and the isothermal compressibility.

  • A. L. Thorneywork, S. K. Schnyder, D. G. A. L. Aarts, J. Horbach , R. Roth, R. P. A. Dullens, Structure factors in a two-dimensional binary colloidal hard sphere system, Molecular Physics 116, 3245–3257 (2018).

Dynamical decoupling and recoupling of the Wigner solid to a liquid helium substrate

We investigate the dynamical interaction between an electron crystal trapped above the surface of liquid helium and surface waves (‘ripplons’) excited by its motion. At rest, the electron system is ‘dressed’ by static ripplons to form ‘ripplopolaron’ states. As the electrons move, resonant ripplon scattering results in a growth of the ripplopolaron effective mass, on timescales comparable with the inverse of the ripplon frequency (∼100 ns). Under sufficient driving force, the electron system decouples from the surface waves and moves at high velocity, before decelerating sharply when the electron solid and surface excitations recouple to form a ‘new’ ripplopolaron system. The mass of the newly formed ripplopolarons is similar to that in the initial static case.

  • D. G. Rees, S.-S. Yeh, B.-C. Lee, S. K. Schnyder, F. I. B. Williams, J.-J. Lin, K. Kono, Dynamical decoupling and recoupling of the Wigner solid to a liquid helium substrate, Physical Review B 102, 75439 (2020).