In multi-agent systems like in online user-centred societies, bonds between individuals often suffer from unimportant or non-significant offences. A solution to this problem has been proposed by Asimina Vasalou in the form of a theoretical model for forgiveness, that could be an efficient means to restore affected communication and even improve performance. Our work has been to develop a formal computational model for forgiveness, to implement it in the form of a generic component, and to test it in a social simulation environment. This allowed us to realize that the social and psychological context had a significant impact on the forgiveness behaviour and the “social performance”.