Overlapping gaze shifts reveal timing of an eye–head gate

The ability to dissociate eye movements from head movements is essential to animals with foveas and fovea-like retinal specializations, as these species shift the eyes constantly, and moving the head with each gaze shift would be impractical and energetically wasteful. The processes by which the dissociation is effected remain unclear. We hypothesized that the dissociation is accomplished by means of a neural gate, which prevents a common gaze-shift command from reaching the neck circuitry when eye-only saccades are desired. We further hypothesized that such a gate would require a finite period to reset following opening to allow a combined eye–head saccade, and thus the probability of generating a head movement during a saccade would be augmented when a new visual target (the ‘test’ target) appeared during, or soon after, a combined eye–head saccade made to an earlier, ‘conditioning’ target. We tested human subjects using three different combinations of targets—a horizontal conditioning target followed by a horizontal test target (H/H condition), horizontal conditioning followed by vertical test (H/V), and vertical conditioning followed by horizontal test (V/H). We varied the delay between the onset of the conditioning head movement and the presentation of the test target, and determined the probability of generating a head movement to the test target as a function of target delay. As predicted, head movement probability was elevated significantly at the shortest target delays and declined thereafter. The half-life of the increase in probability averaged 740, 490, and 320 ms for the H/H, H/V, and V/H conditions, respectively. For the H/H condition, the augmentation appeared to outlast the duration of the conditioning head movement. Because the augmentation could outlast the conditioning head movement and did not depend on the head movements to the conditioning and test targets lying in the same directions, we could largely exclude the possibility that the augmentation arises from mechanical effects. These results support the existence of the hypothetical eye–head gate, and suggest ways that its constituent neurons might be identified using neurophysiological methods.