Motion information plays only a secondary role in sex identification of walking persons in frontal view

Abstract

Observers have a success rate above chance in identifying the sex of walking persons on the basis of movies showing only point lights. It has been claimed that observers rely heavily on motion information for their judgment. Here, we studied, for the frontal plane, the added value of motion information over just form information. In the first experiment, we asked 209 observers to identify the sex of frontal-plane still images of point lights of six male and six female walkers. We used two types of point-light images: (1) cloud-like images, showing just point lights, and (2) skeleton-like images with point lights interconnected. On the basis of cloud-like still images, observers had a mean success rate of 63%; on the basis of skeleton-like still images, they had a higher mean success rate of 70% (p < 0.001). In the second experiment, we asked 273 observers to identify the sex of skeleton-like still images and skeleton-like movies of eight full walking strides. The overall success rate based on movies was 73%. Among the observers first presented with still images, the success rate based on still images was 68%, but for observers first presented with movies the success rate based on still images was 74%, not different from that based on movies (p > 0.05). Our interpretation was that motion information revealed what the point lights represented but had no additional value when this became clear. Hence, we concluded that motion information plays only a secondary role in sex identification of walking persons in the frontal plane.

Figure 1.

A point-light image can be constructed with marker dots disconnected in a cloud or with marker dots interconnected by lines to create a kind of skeleton. In this study, we used still images that were cloud like (SC, left), and still images and movies that were skeleton like (SS and MS, right). SC and SS images were extracted from MS videos at the instant that the right heel contacted the ground. We displayed the following virtual joints calculated from cluster markers: shoulder joints, elbow joints, wrist joints, C7–T1, T12–L1, L5–S1, hip joints, knee joints, and ankle joints. Also, markers at the heels and big toes were displayed. Because the locations of some virtual joints could be calculated independently from two clusters of markers, some segments may seem “dislocated.” For details, see van Leeuwen et al. (2020a). The observers had to answer the question: “Is this a man or a woman?”

Figure 2.

Design of Experiment 1. Each observer was randomly assigned to group SC–SS or group SS–SC. Observers in group SC–SS first saw 13 SC images (still images that were cloud like) and then 13 SS images (still images that were skeleton like). Observers in group SS–SC first saw 13 SS images and then 13 SC images. The first image in a block was not included in the analysis; its purpose was to get an impression of what the observer saw when he or she was still naïve (the question was posed: “What do you see?”) and give an example of the images and question that would follow in the block.

Figure 3.

Box plots of success percentages achieved by the two groups in Experiment 1. The central horizontal line in each box indicates the median, and the bottom and top edges of the box indicate the 25th and 75th percentiles, respectively. The whiskers extend to the most extreme data points not considered outliers, and the outliers are plotted individually using plus (+) signs. The black circle in each box indicates the mean value. The arrows indicate the order of presentation for easy reference. Each asterisk (*) signals that the indicated difference was statistically significant (p < 0.05).

Figure 4.

Rate of correct sex identification of the SS images of individual males and females, plotted as a function of different properties extracted from these images. Note that shoulder width, hip width, and their ratio were determined from the images presented.

Figure 5.

Design of Experiment 2. Each observer was randomly assigned to group SS–MS or group MS–SS. Observers in group SS–MS first saw 13 SS images and then 13 MS videos; observers in group MS–SS first saw 13 MS videos and then 13 SS images. The first image or movie in a block was for practice only.

Figure 6.

Box plots of success percentages achieved by the two groups in Experiment 2. The central horizontal line in each box indicates the median, and the bottom and top edges of the box indicate the 25th and 75th percentiles, respectively. The whiskers extend to the most extreme data points not considered outliers, and the outliers are plotted individually using plus (+) signs. The black circle in each box indicates the mean value. The arrows indicate the order of presentation for easy reference. Each asterisk (*) signals that the indicated difference was statistically significant (p < 0.05).

Figure 7.

Average skeleton-like point-light images of the male (n = 6) and female (n = 6) walkers at the instant that the right heel contacted the ground.