Eye and Limb Tracking

Behavioral Aspect

When we fix our vision on some point in space, the direction of the line from our eyes to that fixation point is known as our gaze direction. Our sense of sight provides highly detailed information about objects along this gaze direction. Therefore, when exploring novel environments, quickly moving, unexpected, or otherwise salient objects that we might want to evaluate in greater detail will tend to attract our gaze. However, when vision is used to guide the limbs for planned action, the nature of eye movement tends to be more structured. In this latter case, eye movement appears to be optimized to obtain information that allows for the most efficient movement of our limbs. For example, if we are making a peanut butter sandwich, we might direct our eyes to a knife so we can determine its location precisely enough to pick it up. Once the location of a given target (e.g., the knife) has been gathered, the eyes initiate a saccadic movement to the next target (e.g., peanut butter jar) at about 100 to 200 milliseconds (ms) before the limb actually reaches the first target. This is likely because visual information from this time period cannot be processed quickly enough to be of use in guiding a limb to the first target, but planning for the next target can begin. In some situations, however, gaze direction remains locked to a target until the instant the limb reaches it. This is most easily seen in pointing, where the eyes fixate on the pointing target until the pointing motion is complete. In addition to target location acquisition, our eyes are used to monitor complicated motor tasks and search for errors of movement that can be corrected. Thus, we may watch what our hands are doing as we attempt to spread peanut butter on bread or cut the resulting sandwich into equal halves.

These qualitative observations have been measured and quantified experimentally. For example, when subjects are asked to perform simple tasks that involve grasping and moving of single objects, it is found that their eyes tend to fixate only on locations that will become the targets of limb motion, or on objects that could become obstacles to such motion. When such tasks involve direct reaching to a target, it is found that gaze usually begins shifting to the target at about 100 to 200 ms after the target becomes visible, and the hand begins movement about 100 ms later. Typically, a subject's gaze will reach a visual target at a few hundred milliseconds (usually less than 500 ms) before the hand arrives. This is thought to be enough time for visual feedback (discussed further later in this entry) to aid in accurate reaching.

...