What information could observers be extracting from frontal-plane still images? In principle, a snapshot from a continuous motion may happen to be taken at the instant that a maximum excursion of a joint is reached, and this maximum excursion may, on average, be different for males than females. For example, the still image could give a clue as to the amplitude of lateral body sway in shoulders and hips in males and females. However, we see no sign of that in
Figure 7, which shows the average male and female still images used in our study. What we do see in
Figure 7 is that, even though at the instant of right heel strike a human walker is not in the standard anatomical position (i.e., standing erect and at rest with the feet together), our images contain a lot of anthropometric information. Most obviously, the images of the males were simply bigger on average than the images of the females. In the course of observing a series of images one by one, observers could work out a running average size and classify all images larger than that as male and all images smaller than that as female. In the images that we presented to the observers, this strategy to distinguish men from women would be more successful using shoulder width as cue than using hip width or body height as cues (
Figure 4). It is of note, therefore, that observers who reported using shoulder width had higher (
p = 0.003) mean success rates in identifying the sex of SS (77%) than observers who reported using cues other than shoulder width (69%). We chose to preserve size information in the images because we felt that this was ecologically valid. After all, body size is part of the structural information that can be used for sex identification in real life. However, one can also choose to scale the images to a fixed height so that only proportional differences between men and women remain, as was done by
Davis and Gao (2004). This difference in choice could explain why our observers achieved on average a 6% higher success rate in identifying the sex of SC images than observers in the study by
Davis and Gao (2004). When it comes to proportional differences, it has been suggested in the literature that the ratio of shoulder width to hip width is an important sex clue. And, in fact, 33% of the observers reported using (the width of) both shoulders and hips as cues when identifying the sex of SS images, and half of these observers mentioned specifically the width of the shoulders relative to the width of the hips as a cue. This is in line with
Saunders, Williamson, and Troje (2010), who studied eye movements of observers and concluded that, for the perception of sex, the shoulders and hips are particularly important. In the images that we used in our study, however, there was considerable overlap between the men and women in the ratio of shoulder width to hip width (
Figure 4), rendering this cue less reliable. In general, this indicates that in studies on sex identification of point-light walkers in which kinematic data of only a few men and women are used, the reliability of cues and the absolute success scores of observers will also depend on the particular walking subjects selected and therewith the overlap in absolute dimensions and proportions between the men and the women.