How do people perceive the variability of multifeature objects?

Humans can judge the summary statistics of various feature dimensions from multiple objects, but it remains unclear whether and how ensemble perception occurs for multifeature objects. The present study investigates how people perceive the overall variability of multifeature objects. Participants estimated the overall variability of a set of stimuli having various orientations and colors, with each feature's variability randomly determined in each trial. Across three experiments, we found that most people considered both dimensions when estimating variability. To explore how people consider both features, we manipulated the interfeature correlation to examine whether perceived variability relies on the combination of marginal distributions or a joint distribution. The interfeature correlation does not influence the marginal variability of each feature but does reduce the overall variability of a multidimensional joint distribution. Our results showed that the interfeature correlation did not influence the perceived variability, consistent with the prediction based on marginal distributions. When similar features were spatially adjacent, however, interfeature correlation reduced perceived variability, and the contribution of orientation diminished, suggesting that spatial regularity modulates how different features are combined for variability judgments. These results indicate that multiple feature information contributes to variability perception, supporting the idea of a domain-general variance processor. (PsycInfo Database Record (c) 2025 APA, all rights reserved).