Eye movements were recorded while participants viewed line-drawing pictures
of natural scenes in preparation for a memory test (Experiment 1) or to fi
nd a target object (Experiment 2). initial saccades in a scene were not con
trolled by semantic information in the visual periphery, although fixation
densities and fixation durations were affected by semantic consistency. The
results are compared with earlier eye-tracking studies, and a qualitative
model of eye movement control in scene perception is discussed in which ini
tial saccades in a scene are controlled by visual but not semantic analysis
.