This work is a twofold contribution to the analysis of conflict detection process in air traffic controllers (ATCos). The first one addresses methodological aspects and proposes a way to get responses as close as possible to controllers' actual expertise without using artifacts such as rating scales or inferring judgments from verbal material. The second objective is to compare the influence of three geometrical features of aircraft encounters and their capacity to alter an accurate perception of conflicts. The proposed methodology appeared to be useful for collecting expertise as controllers quickly appropriated it, and led to get coherent data. Its use can be envisaged when a reliable representation of mental picture of ATCos is essential. Concerning the geometrical features of aircraft trajectories, aircraft attitudes i.e., the fact they are stable, climbing of descending, entailed significant differences on detection accuracy. To a lesser extent, catch-ups and segmented trajectories showed a capacity to make an accurate perception of conflicts more difficult. These results must be interpreted as tendencies more than precise or quantified results. As the objective of this experiment was to be a pre-experiment in preparation for future collecting in the framework of the European project SESAR, a few different choices concerning the trajectories to be used in the traffic scenarios will help to precise these results.