The handwriting signature is one of the most popular behavioral biometric traits for person recognition. Such recognition systems capture the personal signing behaviour and its variability based on a limited number of enrolled signatures. In this paper a cognitive inspired model based on motor equivalence theory is developed to duplicate off-line signatures from one real on-line seed. This model achieves duplicated signatures with a natural variability. It is validated with an off-line signature verifier based on texture features and a SVM classifier. The results manifest the complementarity of the duplicated signatures and the utility of the model.