Stereoscopic displays have gained much popularity, while visual discomfort during watching stereoscopic displays is becoming an increasing concern. Vergence‐accommodation conflicts are one of factors inducing visual discomfort. How to properly assess the visual discomfort by vergence‐accommodation conflicts has been the focus of researchers. Compared with previous studies, this paper aims to address two problems: (1) designing an elaborate experiment to induce visual discomfort without interference of the long performance time and (2) objectively selecting fewer electrodes and extracting electroencephalogram (EEG) features much related to visual discomfort. The results show that, better than previous experimental paradigm, the designed experimental paradigm not only causes the visual discomfort properly but also shortens the performance time to less than an hour. Moreover, for the proposed model, the visual discomfort can be evaluated well based only on six selected electrodes or even better than previous models relying on much more electrodes.