Publication
The real-world size of one object may affect the observer's evaluation of the size: participants overestimate the size of the familiar object if that object has a big size in the real world. However, would the same biases be observed for ensemble representation, where access to the features of each object is highly limited? We conducted a series of experiments where we manipulated the real-world size of the objects in the set. In our previous experiments (Tiurina, Markov, Paramonova, 2020), we used objects of different shapes and found inconsistent results; thus, here, we used objects only with round shapes to estimate the mean size less noisy. We compared averaging for sets of 8 images of objects with big real-world size (e.g., road signs, sewer hatches) and small real-world size (e.g., coins, donuts). We used 5 categories with small real-world size and 5 with big (24 images for each category). In Experiment 1, the set consisted of items from one category (e.g., 8 different coins). In Experiment 2, the set consisted of items belonging to various categories according to three conditions: small real-world size, big real-world size, and mixed, as the control condition. The results of both experiments showed bias in mean size estimation - observers tend to underestimate sets consisting of items with small real-world size compared to control conditions. Complex properties of real-world objects affect not only object perception but also ensemble summary statistics representation.