We describe a new point of view that uses contextual information of the interrelated conditions in which criminality or deviant occurs by means of computer simulation to examine its role in a progressively changing society. Based on the functionalist point of view, we model society as a multi agent system and criminality as a noise. A noise is injected as training signals to the single layer perceptrons that functions as agents (i.e. society) and all agents must comply the fitness function. Failure to comply will result in the agent being removed from the "society" and be replaced by a "newborn" that inherits some "upbringing" information from its "mother's agent". Our simulation studies point toward the constructive effects of criminality. The major contribution in our study is that new paradigm can show and measure criminality effects as in contrast to mere verbal descriptions of the social sciences. The new paradigm could be used to provide the functional explanation of atypical trends in criminality in East- and Middle European and some other countries and provide "artificial experimental data" for future studies.