The automatic visual attentional procession of threatening stimuli over non-threatening cues has long been a question. The so-called classical visual search task (VST) has quickly become the go-to paradigm to investigate this. However, the latest results showed that the confounding results could originate from the shortcomings of the VST. Thus, here we propose a novel approach to the behavioral testing of the threat superiority effect. We conducted two experiments using evolutionary relevant and modern real-life scenes (e.g., forest or street, respectively) as a background to improve ecological validity. Participants had to find different targets in different spatial positions (close to fovea or periphery) using a touch-screen monitor. In Experiment 1 participants had to find the two most often used exemplar of the evolutionary and modern threatening categories (snake and gun, respectively), or neutral objects of the same category. In Experiment 2 we used more exemplars of each category. All images used were controlled for possible confounding low-level visual features such as contrast, frequency, brightness, and image complexity. In Experiment 1, threatening targets were found faster compared to neutral cues irrespective of the evolutionary relevance. However, in Experiment 2, we did not find an advantage for threatening targets over neutral ones. In contrast, the type of background, and spatial position of the target only affected the detection of neutral targets. Our results might indicate that some stimuli indeed have an advantage in visual processing, however, they are not highlighted based on evolutionary relevance of negative valence alone, but rather through different associational mechanisms.