Perceptual systems can improve their performance by integrating relevant perceptual information and segregating away irrelevant information. Three studies exploring perceptual integration and segregation in audition and vision are reported in this thesis. In Chapter 1, we explore the role of similarity in informational masking. In informational masking tasks, listeners detect the presence of a signal tone presented simultaneously with a random-frequency multitone masker. Detection thresholds are high in the presence of an informational masker, even though listeners should be able to ignore the masker frequencies. The informational masker's effect may be due to the similarity between signal and masker components. We used a behavioral measure to demonstrate that the amount of frequency change over time could be the stimulus dimension underlying the similarity effect. In Chapter 2, we report a set of experiments on the visual system's ability to discriminate distributions of luminances. The distribution of luminances can serve as a cue to the presence of multiple illuminants in a scene. We presented observers with simple achromatic scenes with patches drawn from one or two luminance distributions. Performance depended on the number of patches from the second luminance distribution, as well as knowledge of the location of these patches. Irrelevant geometric cues, which we expected to negatively affect performance, did not have an effect. An ideal observer model and a classification analysis showed that observers successfully integrated information provided by the image photometric cues. In Chapter 3, we investigated the role of photometric and geometric cues in lightness perception. We rendered achromatic scenes that were consistent with two oriented background context surfaces illuminated by a light source with a directional component. Observers made lightness matches to tabs rendered at different orientations in the scene. We manipulated the photometric cues by changing the intensity of the illumination, and the geometric cues by changing the orientation of the context surfaces. Observers' matches varied with both manipulations, demonstrating that observers used both types of cues to account for the illumination in the scene. The two types of cues were found to have independent effects on the lightness matches.
Adviser: David H. Brainard. Thesis (Ph.D. in Psychology) -- University of Pennsylvania, 2012. Includes bibliographical references.