Social campaigns are an important tool for promoting positive change in social attitudes (in ecology, health prevention, tolerance promotion, etc.). Improving their effectiveness may therefore have a very tangible effect on many aspects of life – both for individuals and for whole societies. Among the most widespread activities that are undertaken within the framework of social campaigns, one can mention advertising through different types of media – television, radio, internet and print. Assessment of this element is done mostly on the basis of questionnaires and focus groups. Research of this type relies on measures that are proximal, such as perceived effectiveness. Although it is assumed that perceived effectiveness is causally antecedent to actual effectiveness, it would be advisable to find the way of assessing actual effectiveness more directly. One of the approaches, that aim to solve this problem is application of the tools of cognitive neuroscience. The use of cognitive neuroscience techniques for pretesting media messages in social campaigns requires properly designed research. In order to do that, the pivotal stage of each experiment design is the choice of media stimuli that will be presented to the participants. Different media and contents will lead to different patterns of responses in viewers, determining a great deal about how a message is processed, including which parts of the message are attended to, and how the message is evaluated and liked. Stimuli presented during the experiment can be static, like pictures or dynamic, like video. However, such stimuli may not be representative of real-life situations. To avoid problems with picture and video stimuli, one can use virtual environments. The aim of the article is to present a concept of virtual reality system that could enable the research of media messages effectiveness in social campaigns.