Daniel Herrera

The paper develops ideal observer models to estimate 3D motion from natural binocular stimuli, showing optimal computations that match human performance patterns.

The paper investigates the role of higher-order texture statistics in natural image segmentation and finds they provide only modest improvements over spectral statistics.

The paper investigates how contextual modulation interacts with segmentation in human peripheral vision using naturalistic textures, showing that segmentation cues affect contextual modulation not captured by summary-statistics models.