We demonstrate that group size reshapes the effective interaction network in confined biological collectives. Using minimal groups of the shrimp Neocaridina davidi $(N=2,3,4)$ as a representative nonequilibrium system, we reconstruct the asymmetric coupling matrix via an inverse asymmetric Ising model. This mapping is physically enabled by thigmotactic behavior, which restricts trajectories to a discrete decision space, allowing for a maximum-entropy description appropriate for steady states characterized by broken detailed balance. Our analysis reveals a structured progression with increasing group size: Mean coupling strength undergoes social screening, while heterogeneity, structural frustration, and nonreciprocal asymmetry emerge as defining architectural features. Group-size identity is encoded in a subspace orthogonal to the global coupling scale and statistical volatility. Furthermore, the eigenvalue spectrum signals a stationary log-probability distribution in which collective rotational order is amplified while statistical barriers between configurations are lowered, thereby expanding the repertoire of accessible states. Confinement thus drives a well-defined architectural transition from simple reciprocal coordination toward a complex, heterogeneous, and nonreciprocal nonequilibrium steady state.

Close

PrevNext

View figure in article

• Collective behavior in networks
• Inverse problems
• Living matter & active matter
• Random forests
• Data analysis
• Ising model
• Machine learning
• Regression analysis
• Statistical methods