Visual neurons integrate information over a finite part of the visual field with high selectivity. This classical receptive field is modulated by peripheral inputs, which play a role in both neuronal response normalization and contextual modulations. However, the consequences of these properties for visuo-motor transformations are yet incompletely understood. To explore those, we recorded short-latency ocular following responses in humans to large center-only and center-surround stimuli. We found that eye movements are triggered by a mechanism that integrates motion over a restricted portion of the visual field, the size of which depends on stimulus-contrast and increases as a function of time after response onset. We also found evidence for a strong non-isodirectional center-surround organization, responsible for normalizing the central, driving input so that motor responses are set to their most linear contrast dynamics. Such response normalization is delayed ~20ms relative to tracking onset, gradually builds-up over time and is partly tuned for surround orientation/direction. These results outline the spatio-temporal organization of a behavioral receptive field, which might reflect a linear integration among sub-populations of cortical visual motion detectors.