Comparison between a global and a limited sampling strategy in size-averaging a set of items M Tokita, A Ishiguchi |
|---|
Ochanomizu University, Japan
|
| Many studies have shown that people can accurately perceive and estimate the statistical properties of a set of items (Ariely, 2001, Psychol Sci, 12(2), 157-162; Chong & Treisman, 2005, Vision Res, 45(7), 891-900). As the accuracy with which people can judge the mean size of a set is consistent across set size, some have proposed that average size can be computed in parallel across all items. At the same time, it has been shown that the accuracy can be predicted by a strategy of sampling the sizes of limited number of items in a set using focused attention (Myczek & Simons, 2008, Percept Psychophys, 70(5), 772-788). In this study, we tested two ideal observer models (i.e., a global and a limited sampling models) to compare them with human observers. The global averaging model posits that people average over all items in a set: the limited sampling model posits that people sample two to four items randomly chosen form a set. In our behavioral experiments, participants were asked to estimate the average size of the items in a set and compare it to a reference item. The results implied that the limited sampling model could predict behavioral data. |
| Up Home |
|---|