Abstract
The visual system efficiently and accurately extracts summary statistics (e.g., the mean) from sets of similar objects through ensemble perception. Evidence for this phenomenon has been demonstrated in various features such as orientation, size, and faces. However, prior work suggests the visual system is biased to overestimate variability in ensembles, particularly when objects in an ensemble are more similar and have less variability. This has been referred to as the variability overestimation effect and has been found when perceiving low-level features (e.g., line orientation, color) and mid-level features (e.g., size). To further explore this bias, we examined the role of attention when perceiving variability within a set of objects. Using the method of adjustment, participants judged either the variability of size or color value in a set of nine sequentially presented circles that varied in both size and color value. For half of the trials, participants saw a pre-cue indicating to focus on either the set's size or color. For the remaining trials, participants focused on both size and color until the response phase, which presented a post-cue indicating which feature variability they were to estimate. Participants vastly overestimated variability in size (172%) and color value (139%) when objects were most similar. For color value, the pre-cue reduced this bias (115%) compared to attending to both features (168%). Similarly, participants overestimated size variability to a greater extent when focusing attention on both features (188%) than on one feature (158%). Although the pre-cue reduced the variability overestimation effect, participants still greatly overestimated the variability of both features when objects were most similar. Focusing attention on a specific feature did not eliminate the bias, further suggesting the visual system is fundamentally biased to overestimate variability. These findings have implications for real-world tasks requiring judgments about sets of similar objects (e.g., information visualization).