Mikayla Shirley¹, Jason Haberman; ¹Rhodes College

A prevailing question in the study of ensemble perception is whether it is supported by domain-general or domain-specific mechanisms. Evidence for domain-specificity comes from Haberman et al., 2015, which showed that error in low-level ensemble representation does not predict error in high-level ensemble representation. This raises another question: what happens to that relationship when high- and low-level ensemble features are embedded within a single object? We created stimuli in which two features, orientation and emotion, appeared within a single face. Dimensions could be manipulated independently, depending on the task. In separate blocks, observers (N = 30) adjusted the test stimulus to match either the average orientation or average expression of a set of faces. In expression blocks, the orientation of the faces varied, but the test stimulus observers adjusted was always upright. In orientation blocks, the emotion of the test stimulus was orthogonal to the average emotion of the set. Orientation parameters prevented individual faces from tilting below the horizontal meridian, ensuring judgements of emotion would not be subjected to face inversion effects. In line with previous work, preliminary results revealed a non-significant relationship (r = .12, p>.05) between average orientation precision and average emotion precision, suggesting that ensemble representations do not become entangled when combined within a single stimulus. Ensembles remain hierarchically organized and independently represented.