Fig 1.
Examples of AOI designs used in four eye-tracking studies with dogs.
Note that all four designs include either the infraorbital margin or upper nose area in their eyes AOI. Images are from A: [59], B: [60], C: [62], D: [64].
Table 1.
The eyes AOI design and locations of the starting position and stimulus presentation of six dog eye-tracking studies that investigated where dogs look on faces.
Table 2.
Number and % of valid trials out of total trials per stimulus type (60 in humans and 56 in dogs) included in the statistical models.
Fig 2.
The twelve stimuli presented to each dog and human subject in this study.
Fig 3.
Stimulus presentation scheme of two different example trials.
In each trial, a stimulus was displayed either to the left or right of the central starting fixation point on the screen (x = 512, y = 384). The x-coordinate of the upper vertex of the nearer edge (indicated by a green dot) of the (invisible) image frame was positioned 100 pixels away from the starting fixation point. Each stimulus was presented for seven seconds.
Table 3.
Summary of the statistical models tested.
Fig 4.
Examples of AOIs and whole stimulus area for each stimulus condition.
Each AOI is indicated as a red shape overlaying a part of each stimulus. The size of each AOI is indicated in (horizontal) visual angle below each AOI and the centroid of each of the 2 areas (the inner features center AOI and whole stimulus) is indicated as a solid yellow point. The relative size of the stimuli are not actual.
Fig 5.
The probability that the first fixation of humans and dogs was on one of the five AOIs in the three stimulus conditions.
Each panel corresponds to each stimulus condition (panel A: human face; panel B: dog face; panel C: non-face object). Solid symbols (a triangle or a circle) and error bars depict estimated means and 95% confidence intervals, respectively. Smaller symbols of the same shape in softened color in the background indicate trial data used for the analysis. Numbers on the x-axis indicate the number of hits on the corresponding AOI across dog/human participants and trials. Dashed line indicates chance level probability (0.2) and AOIs with the estimate significantly higher than the chance level is indicated with stars next to the solid symbols. Further, the AOI pairs with significant differences are annotated with lines and stars, yet only if the differences between all AOI pairs are significant.
Table 4.
Summary of the model test results (type III sums of squares ANOVA using chi-square tests).
Fig 6.
Pixel distances between the first fixations of humans and the centroid of one of the face areas.
Each violin plot, depicting kernel probability density, shows the distribution of the measured pixel distances: the inner features center AOI or the whole stimulus area. Note that in both human face and dog face conditions, the first fixations of humans were on average significantly closer to the centroid of the inner features center AOI than that of the whole stimulus area. Solid circle and error bars overlaid on the violin plots depict estimated means and 95% confidence intervals, respectively. Dots indicate trial data included in the analysis.
Fig 7.
The distribution pattern of relative total fixation duration (rate) of humans or dogs.
Each panel has a plot of each stimulus condition (panel A: human face; panel B: dog face; panel C: non-face object). Solid circle and error bars overlaid on the violin plots depict estimated means and 95% confidence intervals, respectively. Smaller symbols of the same shape in softened color in the background indicate trial data used for the analysis. Dashed line indicates chance level rate (0.2) and AOIs with the estimate significantly higher than the chance level are indicated with stars next to the solid symbols. Further, the AOI pairs with significant differences are annotated with lines and stars, yet only if the differences between all pairs in the stimulus condition are significant.