Within a duration of 20 ms humans can compare the orientations of two test lines so as to encode and place in memory their mean orientation and orientation difference while ignoring noise lines in the space between the test lines. Furthermore, performance is not impaired by randomly varying the location of each test line from trial-to-trial. We conclude that the two test lines are not compared by shifting eye fixation or attention from one to the other, nor by attending to two spatial locations. This evidence is consistent with the proposal that the human visual system contains second-stage long-distance comparators, any one of which responds to simultaneous stimulation of two conventional first-stage spatial filters located some distance apart and is insensitive to stimuli that fall between these two first-stage filters. We suggest that our observers performed discriminations by attending to the outputs of the proposed second-stage long-distance comparators rather than by attending to two spatial locations. In addition to their mean orientation and orientation difference, humans can simultaneously encode and place in memory the separation and mean location of the two test lines while ignoring stimuli in the space between the lines. We suggest that, following each of the eye's exploratory saccades, the proposed second-stage long-distance comparators, in effect, take a snapshot of an object's retinal image that ignores the object's surface texture while encoding the shape of its boundary.