We investigated effects of element features on statistical description of relative frequency. In Experiment 1, we manipulated the proportion of two elements types forming a set and the set size and measured response time and the accuracy of relative numerosity discrimination. We did this with element pairs that had been shown to produce search symmetries or asymmetries in visual search task. We found that pop-out elements in the search asymmetry pair were numerically overestimated. In Experiment 2, we used sets of circles and circles with gaps to eliminate the possibility that the overestimation found in Experiment 1 was due to larger contour length of pop-out elements. In Experiment 3, we manipulated proportion of two elements types and the set size to measure point of subjective equality (PSE) and the slopes of z-score functions in relative numerosity discrimination to support the results of Experiments 1 and 2. The results generally showed that the proportion of pop-out elements is likely to be overestimated and that set size had no effect, suggesting that the types of features characterized by visual search could influence the accuracy and precision in discrimination of relative numerosity.
