Abstract:
Ensemble representations are argued to benefit efficient outlier processing. However, limited research has investigated the relationship between outlier and ensemble representations which was the goal of this thesis. Eight heterogeneously-sized circles were presented and viewers reported the mean size or the largest/smallest size across different blocks. The largest/smallest size could be the outlier or not in the particular distribution of a display. In a separate block that only had displays with outliers, participants reported either the mean or the largest/smallest size. Since the relevant task was indicated after display, in this mixed block, viewers were encouraged to concurrently focus on the ensemble along with the outlier. I specifically investigated how the presence of an outlier impacts the ensemble representations; whether outliers are better represented than non-outliers; whether people can form ensemble and outlier representations simultaneously. Results showed that an outlier was always excluded from mean estimations. Only a larger outlier decreased the mean size accuracy. Also, representations were more veridical when the largest/smallest item was an outlier than when it was not. Last, attending to the mean and outlier size concurrently did not yield a cost in the mean size and larger outlier precision suggesting that viewers were able to process outlier and mean size concurrently. However, when the outlier was smaller, the outlier estimates were less accurate. I conclude that an outlier is efficiently processed and somewhat downplayed in ensemble representations. Nevertheless, ensemble representations may not foster the processing of outliers that are perceptually harder to detect.
