Fig 1.
Example of AOI dwell time and WSD calculation.
a) An example of a participant’s fixations (Bluish-green dots where bluish-green numbers denote the fixation duration for each fixation), AOI (orange ellipse) definition, and POI (sky blue X) definition are shown on a subset of an exemplar image. Although we present a partially-blurred image here to protect privacy, the participants saw real, unblurred images in the experiment. b) The uniform kernel weights used for AOI dwell time calculations. c) An example of AOI dwell time calculation. d) The Gaussian kernel weights used for the WSD calculation. e) Example of WSD calculation. Note that the original fixation durations have been reweighted based on their proximity to the POI.
Fig 2.
Example of AOI/POI definitions and selected mixture components.
a) AOI definitions from the original study (ellipses) are illustrated in an exemplar image. b) The average BIC for GMMs fitted with different numbers of components (k). The error bar denotes the standard deviation of BIC across 50 different fitted GMMs. The reddish-purple dotted line denotes the number of component with the least average BIC. c) An example of visualization of Gaussian components of the fitted GMM with the selected number of components that had the lowest BIC. The yellow dots represent the offset- and drift-corrected locations of all fixations across all participants to the exemplar image. The components are visualized as semi-transparent colored circles with means at the means of the components and radii of 2σ. The components selected to match the AOI have black boundary lines. d) An example of selected mixture components that correspond to the AOIs defined in the original study. The means of selected mixture components (Xs) were used as the POI.
Fig 3.
Correlation between AOI dwell time and WSD.
a) Correlation between dwell time on face AOIs and face WSDs. b) Correlation between dwell time on point of contact AOIs and point of contact WSDs. The sky blue line denotes the fitted regression line. The shaded region denotes the 95% confidence interval (due to the large number of data points, it is not very visible). Each point denotes a trial. The orange rectangle highlights the trials with zero AOI dwell time. The number of trials for b) is smaller than a) because there are fewer stimuli with point of contact AOIs/POIs while all stimuli had faces AOIs/POIs.
Fig 4.
Fixation durations on faces by interaction type.
a) AOI dwell time on Face AOIs by predefined interaction types. b) WSD on Face POIs by predefined interaction types. Error bars represent ±1 SD. Although the values look similar, WSD and AOI dwell time uses different metrics and cannot be compared directly.
Fig 5.
Fixation durations on faces by violence rating.
a) AOI dwell time on Face AOIs by violence rating determined by individual participants. b) WSD on Face POIs by violence rating determined by individual participants. Error bars represent ±1 SD. Note that for the LMM analysis violence ratings were z-scored within participants, but the original ratings were shown here for illustrative purposes. Although the values look similar, WSD and AOI dwell time uses different metrics and cannot be compared directly.
Fig 6.
AOI dwell time and WSD on point of contact by interaction type.
a) AOI dwell time on points of contact by predefined interaction type. b) WSD on points of contact by predefined interaction type. Error bars represent ±1 SD. Although the values look similar, WSD and AOI dwell time uses different metrics and cannot be compared directly.