Participation in experimental studies can be conceptualized as Goffmanian frames, i.e. a set of rules which include the fact the experimenter will be observing participant behavior through (the recording of) the experiment. This study is focused on frame breaches in 16 video- and audio-recorded dyadic conversations taking place in an experimental setting. Our main conclusion is that the experimental frame is conceptualized by participants as including constraints that go beyond non-experimental interactions, and in particular the need to mitigate frame breaches, (...) which are seen as face-threatening. Analyses revealed that participants only broke the research frame after they completed the task they were assigned by the researcher, and that breaches did not necessarily correspond to changes in key. Insights gained in relation to face and mitigation are discussed, as well as the participants’ need to determine their next steps once the research purpose has been perceived complete. (shrink)