Elastomer compositions containing various particulate fillers can be formulated according to the specific functions required of them. Stress softening-which is also known as the Mullins effect-occurs during high loading and unloading paths in certain supramolecular elastomer materials. Previous experiments have revealed that the load-displacement response differs according to the filler used, demonstrating an unusual model of correspondence between the constitutive materials. Using a spherical indentation method and numerical simulation, we investigated the Mullins effect on polyurethane (PU) compositions subjected to cyclic uniaxial compressive load. The PU compositions comprised rigid particulate fillers (i.e., nano-silica and carbon black). The neo-Hooke model and the Ogden-Roxburgh Mullins model were used to describe the nonlinear deformation behavior of the soft materials. Based on finite element methods and parameter optimization, the load-displacement curves of various filled PUs were analyzed and fitted, enabling constitutive parameter prediction and inverse modeling. Hence, correspondence relationships between material components and constitutive parameters were established. Such relationships are instructive for the preparation of materials with specific properties. The method described herein is a more quantitative approach to the formulation of elastomer compositions comprising particulate fillers.