ThreatSim: A novel stimuli database of threatening and nonthreatening image pairs rated for similarity

Researchers often require validated and well-rounded sets of image stimuli. For those interested in understanding the various visual attentional biases toward threatening stimuli, a dataset containing a variety of such objects is urgently needed. Here, our goal was to create an image database of animate and inanimate objects, including those that people find threatening and those that are visually similar to them but are not considered threatening. To do this, we recruited participants (N = 77) for an online survey in which they were asked to name threatening objects and try to come up with a visually similar counterpart. We then used the survey results to create a list of 32 objects, including eight from each crossing of threatening versus nonthreatening and animate versus inanimate. We obtained 20 exemplar images from each category (640 unique images in total, all copyright-free and openly shared). An independent sample of participants (N = 191) judged the similarity of these images using the spatial arrangement method. Data were then modeled using multidimensional scaling. Our results present modeling outcomes using a "map" of animate and inanimate objects (separately) that spatially conveys the perceived similarity relationships between them. We expect that this image set will be widely used in future visual attention studies and more.