Training material properties of biological tissues via mechanical memory

Cells in multicellular organisms maintain homeostasis in dynamic environments by sensing mechanical cues and adapting their internal processes. Such adaptive responses, modifying behavior to optimize function, parallel emerging concepts of physical learning. We present a minimal model for mechanical learning in epithelial tissues using a vertex model extended to include dynamic tension remodeling, mimicking active cellular contractility. In our model, tensions evolve in response to local strains exceeding a threshold, capturing key learning features such as habituation under repeated pulsatile activation, non-local memory and cooperative learning. After the system remodels under mechanical signals, we quantify changes in shear modulus, bulk modulus, and Poisson’s ratio. We find that tissues can soften, stiffen or behave as auxetic materials based on the model parameters. Our results reveal how simple local rules can encode adaptive mechanical responses in tissues, offering a mechanistic link between cellular contractility dynamics and emergent rheological properties.