Emergent Learning in an Embedded Spheroid System

Learning involves forming new associations that modify existing association networks to enable novel tasks. Beyond neuronal networks, non-neuronal cellular collectives, such as spheroids, can exhibit learning-like behaviors through mechanical interactions. We present a computational study of emergent learning in a spheroid embedded in a collagen network. Using a 3D vertex model to represent the spheroid coupled to a fiber network, representing the collagen, we capture both cell-cell and cell-matrix interactions. By applying mechanical strains to the fiber network, we demonstrate how correlated cellular displacements and/or deformations form mechanical association networks. Our simulations reveal that the cellular architecture and matrix properties critically influence these networks. Higher fiber densities fluidize solid spheroids, promoting nontrivial high-stress structures that facilitate long-range cell interactions and emergent learning behaviors. Additionally, the fiber network itself can achieve learning by altering its topology through dynamic bonds. These findings suggest that mechanical interactions at the cellular level can induce learning-like behaviors without neuronal networks, offering new insights into adaptive behaviors in biological systems.