Fig 1.
Examples of stimuli used in both experiments.
Fig 2.
Shows example scan paths for six abstract art works for six different participants.
The three on the left are from Experiment 1 (labelled for each participant as: “Landscape”, “Portrait” and “Abstract” from top to bottom) and those on the right are from Experiment 2 (labelled “Still Life”, “Action” and “Abstract” from top to bottom). Fixations are plotted as pale blue circles, the center of each is the fixation location while their diameter represents their duration which is also given by the number (in milliseconds) next to each circle. Saccades are shown as straight yellow lines between their start and end points with the yellow number indicating the point at which each saccade as executed in the path sequence (i.e., 1 is the first saccade, 2 is the second and so on).
Table 1.
Overall average saccade and fixation parameters are shown for each art label type for Experiment 1 and 2.
Fig 3.
Shows eye movement parameters across the viewing period.
The upper row from left to right shows A average fixation counts and B average fixation durations. The lower row, left, shows C average saccade amplitude in degrees of visual angle. These show little effect of art content label but clear effects of viewing time. The Lower row, right, shows the D average saccade counts for those classed as “horizontal” and “vertical” as a function of the label. Again, little effect of art content label is seen, but there is a clear bias for horizontally directed saccades. Error cars show repeated measures estimates of error [53].
Fig 4.
Shows eye movement parameters across the viewing period for each categorical label (Abstract, Still life, and Action).
The upper row from left to right shows A average fixation counts and B average fixation durations. The middle row shows C average saccade amplitude in degrees of visual angle and D shows the average saccade duration in milliseconds (ms). The lower row E shows the average saccade peak velocity in degrees per second. Each of these eye movement parameters shows little effect of CategoryLabel but clear effects of viewing time. Error cars show repeated measures estimates of error [53].