OFTEN implicit in the interpretation of visual search tasks is the assumption that the detection of targets is determined by the feature-coding properties of low-level visual processing1,2. But higher level processes have also been implicated as visual search ability is enhanced in a depth plane3 of when two-dimensional shapes are interpreted as three-dimensional forms4,5. Here we manipulate binocular disparity to degrade visual search, so that otherwise identical features become parts of surfaces through perceptual completion, rendering them less clearly distinguishable as targets and distractors. Our results indicate that visual search has little or no access to the processing level of feature extraction but must have as an input a higher level process of surface representation.