Abstract
Purpose: When people make an eye movement toward an object, they can use both the current retinal image of the object and spatial memory to plan the saccade. Recent work has shown that people use memory in their estimate of object location when planning finger movements, and they give a higher weight to memory when there is greater uncertainty of the current visual estimate due to low contrast (Brouwer & Knill, JOV, 7(5):6, p1–12) or greater retinal eccentricity (Issen & Knill, VSS Abstract 2008). We asked whether people would show a similar pattern when issuing eye movements. Methods: In a virtual environment, two objects appeared on the right side of the screen: a “weapon” and a circular “target” region. Subjects had to move the weapon to the “reload station” located 30 degrees to the left, then touch the circular target. While subjects moved the weapon to the reload station, they typically fixated that area, and thus the circular target was in the periphery. In a third of the conditions, the position of the circular target shifted by about one degree during this transition. Eye movements were recorded with an Eyelink using corneal reflection. Results: We regressed the saccade endpoint on the remembered and updated visual locations of the target. Although the target was visible in the periphery when subjects planned their eye movements, the saccade endpoints fell between the visual and remembered locations, as evidenced by a significant regression term for memory. Conclusion: This is evidence that eye movements, like other motor commands, are not issued based entirely on the current retinal map, but rather a more complex integration of cues including the updated visual information and the remembered target location.