On localization and saccade programming.

Three experiments investigated if and how saccade latency and accuracy are related to the localizability of target positions in parafoveal strings of varying length. The data show that target localizability indeed determines the latency of accurate primary saccades. To overcome the rapid acuity fall-off in peripheral vision as well as the lateral masking of the target induced by the presence of more or less similar nontargets the visuomotor system must increase saccade latency by approximately 60-100 msec/deg (depending on the target), in order to attain the target accurately. It is also found that the eye's first landing position within the string for short-latency saccades (190-210 msec) is independent of target localizability. Probably these primary saccades are programmed to land near the middle of the string, not to land on the target itself. This leads to a re-interpretation of the global effect in terms of efficiency of oculomotor behavior. With regard to reading the study reveals that because of the very limited localizability of target positions in parafoveal strings and because of the timing characteristics of the localization process it seems advantageous for the eye movement control system not to wait until a given target position (e.g. the convenient viewing position) is clearly localized. Instead it should program two successive saccades: one which brings the gaze quickly somewhere into the next word and one which redirects the gaze quickly to the desired position.