Abstract: Visual imagery is an important visuo-spatial representation, which has the same mental storage characteristics with visual perception (Romke, Kosslyn & Hamel, 1997), sharing the same brain mechanisms with visual perception (Kosslyn, Thompson & Alpert, 1997; Ganis, Thompson & Kosslyn, 2004). As is known to all, fixation is the primary source of visual perception information, and eye movement is very important to visual perception. So, does eye movement also play an important role in the visual imagery processing? There are two contrasting accounts for the phenomenon. The functional account hypothesizes that the encoding of each eye fixation during perception process participates, as an index in the location of the space in the subsequent image generation.
In contrast, the epiphenomenal account views eye movements during imagery processing as the passive “spill over” from covert shifts of attention during the stage of image inspection, subsequently to its generation (Laeng & Teodorescu, 2002). Early studies (Jacobson, 1932; Lorens & Darrow, 1962) reported increased oculomotor activity during visualization and during mental arithmetic. Recently, researchers found that subjects who fixed their gaze centrally during perception did the same spontaneously during imagery. The subjects who are free to explore during perception, but maintaining central fixation during imagery, showed decreased ability to recall the pattern. They concluded that the eye scan paths during visual imagery reenact those of perception of the same visual scene and eye movements during mental imagery are not epiphenomenal but assist the process of image generation (Brandt & Stark, 1997; Laeng & Teodorescu, 2002). Actually, whether regular eye movements promote the representation processing relies on the level of eye movement regularity of change that regular change task information accessibility of visual imagery would cause. If task information accessibility level change inevitably leads to eye movement regularity changes, it can be inferred that eye movements play a functional role in visual imagery.
Forty postgraduates, from 25 to 30 years old, attended two experiments respectively. All participants didn’t know the purpose of the experiment and had normal or corrected-to-normal vision. Stimulus presentation and data collection were controlled by two microcomputers. Eye movements were recorded by means of the Eye Tracking Device made by ASL with a sample frequency of 256Hz. Referencing Brandt & Stark (1997) and Laeng & Teodorescu (2002) study, the “perception-imagery” dual task experimental paradigm was applied to compare the change of eye-movement parameters in different task information accessibility level in Experiment 1. In Experiment 2, task information accessibility effect was dissociated from eye controlling in order to see how task information accessibility levels influence visual imagery eye movements.
The results revealed that under the low task information accessibility, the eye movements in visual imagery duplicate the pattern in perception, and eye-movement parameters, such as average fixation, saccade duration, saccade distance, will change in accordance with task information accessibility levels. The second experiment replicated successfully the findings of the first experiment. Eye-movement control and task information accessibility has different influence on visual imagery processing. The eye movements in visual imagery not only duplicate the pattern in perception, but also change depending on the level of task information accessibility. That is to say, eye movements play a functional role in visual imagery.

Key words: visual imagery, eye movement, information accessibility, visual representation