Fig. 9

Crowd analyses on data from studies 1 (a) and 2 (b) presented separately for trials with one image and with many images. As each point represents the average performance of 14,000 (or 20,000 in study 2) randomly sampled groups of participants (1000 groups per target identity), standard error bars would lie within the marker point and are therefore not represented on the graph. In study 1, mean standard error = 0.004 for 1-image trials and mean standard error = 0.003 for 3-image trials. In study 2, mean standard error = 0.003 for 1-image trials and mean standard error = 0.002 for 16-image trials