The role of numerical and non-numerical stimulus properties in a visual numerosity estimation task with two intermixed sets

The approximate number system allows humans and other species to quickly estimate the numerosity of a set of objects. Much research has focused on its possible role as a basis for mathematics achievement and learning difficulties, relying importantly on the numerosity comparison task. In contrast, less research has focused on its underlying mechanisms and other tasks. In this study we used a numerosity estimation task, where participants had to focus on a target set of dots while ignoring a non-target set of dots of a different color. The target and non-target sets’ numerical and non-numerical properties (total area, convex hull area, color light output) were manipulated. Using a small-N design, we asked participants (N = 9) to attend several testing sessions (Nsessions = 8 per participant), to measure the magnitude of intra-participant variability across sessions and reduce error variance. Results showed that numerosity estimates were significantly affected by the target numerosity, with a magnitude about 6 times larger than both target set’s non-numerical properties. Color cues and the properties of the non-target numerosity showed minor contributions that were not consistent across participants. Participants’ performance variability across sessions was 3-4 times smaller than variability across participants. Our results support the role of numerosity as the main driver of participants’ numerosity estimates, against proposals that focus on the role of non-numerical properties such as area measures or texture density.