This paper presents a cognitive agent model capable of showing situations where self-generated actions are attributed to other agents, as, for example, for patients suffering from schizophrenia. The mechanism underlying the model involves inverse mirroring: mapping preparation states onto sensory representations of observed actions. It is shown how this mechanism can develop based on Hebbian learning. The model provides a basis for applications to human-like virtual agents in the context of for example, training of therapists or agent-based generation of virtual stories.