Abstract
Localizing objects is a fundamental function of vision, a prerequisite for most other high-level perceptual and motor goals. Localization could be object invariant: the perceived position of an object may not depend on its particular identity. This would support the commonly held assumption that localization emerges from a unified process or a single module in the brain (e.g., constant retinotopy maintained throughout the visual hierarchy, or a privileged area like FEF that signals perceived position). An alternative is that object representations do influence position assignment. To test this, we measured individual differences in the localization of upright and inverted faces. We employed a dual-task localization and recognition paradigm. On each trial, a grayscale face was briefly presented (either upright or inverted) enveloped within a Gaussian contrast aperture. The face was shown at one pseudo-randomly chosen angular location with fixed eccentricity. Upon its disappearance, observers first moved the cursor to indicate the center of the face, and then they reported its gender (female/male). Collapsed across subjects, at the group level, there was no obvious difference in localization performance for upright and inverted faces. However, the results revealed significant individual differences: each observer mislocalized the stimuli at different locations of the visual field, and these localization biases were stable, consistent with previous findings (Kosovicheva & Whitney, Curr Bio, 2017; Wang et al., Proc Roy Soc, 2020). More importantly, we found that observers showed different idiosyncratic localization biases for upright and inverted faces. The precision of the localization was similar for upright and inverted faces, suggesting that attention differences and lapsing were not likely responsible. Instead, stimulus-specific (for example, face-specific) position coding mechanisms likely exist. The results undermine the idea of a modular or single retinotopic pathway to localization, and instead suggest that object identity interacts with the assignment of perceived position.