TOP-DOWN INFORMATION PROCESSING IN COGNITION EYE MOVEMENTS
视觉图像辨认眼动中的Top－down信息处理

During visual image cognition , the eye-movement pattern is not scanning uniformly throughout the entire image ,but only consists of saccades and local fixatin pauses.The rapid saccadic movements change the positions of the eye and the fixation pauses between saccades acquire key infonnation of the cage. hi present inveStigation, the eye-movement twbories un colded and analed while a variety of cage sthauh presented ti subjects. The experunebtak data have shown that: (1) for sable geometric patterns, the eye-movement fixation pauses concentrated on the special geometric features of the pattern,i.e.those singularities,of which the characteristics are different from their neighbor regions,(2)for complex pictures,the locations of fixations depended on subjects' conceptUal model and motivation,(3)for Chinere characters, the eye-movement patternS also depended on subjan'knOWledge and familiarity with the charaCters (i.e.the coneeptual model). These experimental adults indiCate that the visual cage coition is mainly a top-down process. For aeqchition of suffidri infonnationfrom the image efficiently,the cential nervous system(CNS)controls eye movements to fix on the points which are singularities of the geometric pattern or contain esential infonnation of the image.With top-down processing,it is efficient for image coghtion because the visual system does not have to precess each pixels of the target, but deal with the key regionsselected by CNS.