Konkle and Oliva (in press, Journal of Experimental Psychology: Human Perception and Performance) found that the preferred ('canonical') visual size of a picture of an object within a frame is proportional to the logarithm of its known physical size. They used within-participants designs on several tasks, including having participants adjust the object's size to 'look best'. We examined visual size preference in 2AFC tasks with explicit aesthetic instructions to choose: "which of each pair you like best". We also used both within- and between-participants conditions to investigate the possible role of demand characteristics. In experiments 1 and 2, participants saw all possible image pairs depicting the same object at six different sizes for twelve real-world objects that varied in physical size. Significant effects of known physical size were present, regardless of whether participants made judgments about a single object (the between-participants design) or about all objects intermixed (the within-participants design). Experiment 3 showed a reduced effect when the amount of image detail present at different visual sizes was kept constant by posterizing the images. The results are discussed in terms of ecological biases on aesthetic preferences.