In bedside testing of patients with hemispatial neglect, we have found that extinction for contralesional stimuli is less when the contralesional and ipsilesional items are different on the dimension to be reported relative to when they are the same. Importantly, a study that investigated this observation found that similarity on visual features that are not necessary for response does not impact the amount of extinction. These findings suggest that response requirements may determine what stimuli will and what stimuli will not gain access to awareness. In a related study, we found that extinction of contralesional stimuli was not determined by perceptual similarity of the ipsilesional and contralesional items but by whether they shared the same semantics (e.g., ONE + 1 ) or the same response (e.g., ONE = WON). Here, we report a single case study in which extinction was determined by whether the competing items shared the same response, regardless of whether they shared or differed in their visual features or semantics. When asked to read the item in each field, there was equivalent extinction in the conditions (ONE + ONE) and (ONE + 1) but less extinction in the condition (ONE + TWO). By contrast, when asked to count the number of characters in each field, there was more extinction in the condition (ONE + TWO) than (ONE + 1). When asked to categorize each item as either a word or digit, the degree of extinction was determined both by whether the items shared the same semantics and by whether they required the same response. The results are consistent with a biased competition model in which competition for selection is resolved flexibly depending on response requirements. Furthermore, the data provide evidence that unattended stimuli are processed to the level of response.