The complexity of situations makes individuals use emotions to make sense of their environment and interdependent others. In this paper, we build on the idea that physiological reactions give emotional information about the subject and we focus on Electrodermal Activity (EDA), an index of arousal, to inspect deep processes of a dyadic interaction in a mixed-motive game. Our interest lies on how conflict episodes unfold, to design intelligent agents that are more socially aware and thus able to express and recognise dyadic forms of conflict. A qualitative analysis of the data allowed us to identify moments where players made choices to cope with ongoing conflict or prospects of it in the future.