Texture segmentation tasks are well-known paradigms to study human visual processing in the detection of salient stimulus items. In a recent EEG study, different event-related brain-potentials have been identified which reflect the underlying pre-attentive and attentive processes. We show that a neural model of texture boundary processing can account for the findings of the EEG study concerning the pre-attentive processes. The model identifies the functional role of these processes as well as their probable cortical origin. In addition, it is demonstrated that the proposed computational mechanisms can account for the differences between automatic segmentation and visual search.