By simulating neglect-like effects in neurologically intact observers, we evaluated whether normal attentional allocation can be object centered. In a series of three experiments, observers detected a small gap on the left or right side of a configuration presented in either the left of right visual field. The figures were positioned so that on different trials, the left and right sides would fall in the same retinotopic, hemispatial, and environmental location. Thus, only the location with respect to an object-centered frame varied. We found opposite patterns of bias within each visual field: For figures in the left visual field, left gaps were detected better than right gaps, whereas in the right visual field the opposite pattern was evident. Control conditions indicate that these biases are not due to masking from eccentric contours and depend on the left and right segments being united into a single form. These results indicate that opposing orientational biases of the left and right hemispheres can operate within an object-centered frame in the normal brain. This evidence converges with patient studies and single-unit electrophysiology to reveal the importance of a relatively late, abstract locus for visual selection.