If you're a New Scientist reader, you may have come across this article titled "Beauty is in the eye of your friends." The brief article (which I found via 3 Quarks Daily) describes research purporting to show that whether (heterosexual) women find a man attractive depends, in part, on whether other women find him attractive, a phenomenon called mate choice copying.
In some animal species (guppies, quail, etc.), females sometimes pick mates based in part on the choices of other females, especially more experienced ones. If an experienced female guppy likes a male, then he's probably worth mating with. Jones et al.1 wondered if something like this might go on in humans, so they designed a pretty simple experiment to look for mate choice copying behavior. The experiment had three phases. In the first phase, participants (28 females and 28 males, all heterosexual I presume, though the paper doesn't specify their sexual orientation), participants were shown eight pairs of male faces that previously tested female participants had rated as equally attractive, and they were asked to indicate which face was more attractive, and how much more attractive it was (see the figure below, from Jones et al. Figure 1, p. 2).
In the second phase, the participants were shown the same pairs of photos, but this time, in between the photos of the males, all of whom were facing the camera, were sideways-facing female faces. The female was facing one of the two male faces, and had either a smile or a "neutral" expression (as in the figure below, from Jones et al. Figure 2, p. 3).
In the third and final phase, participants again viewed the pairs of male faces, as they were presented in the first phase, and rated their relative attractiveness a third time.
Jones et al. predicted that, if human females engage in mate choice copying, then the perceived attractiveness of the male faces should be influenced by whether the female is facing and smiling at it. Specifically, for faces that are being smiled at, ratings in the third phase should be higher than those in the fist face. Male faces looked at by females with neutral expressions will rated either equally or less attractive in the third phase than the first.
Consistent with the prediction, female participants rated smiled-at faces as more attractive in the third phase than in the first, and faces looked at with a neutral expression were rated as less attractive in the third phase than the first. Males' attractiveness ratings of the male faces showed the opposite pattern. If the face had been smiled at in the second phase, males rated it as less attractive in the third phase than in the first, and if the female face was facing it with a neutral expression in the second phase, males rated it as more attractive in the third phase than in the first. Jones et al. argue that the female participants are thus exhibiting mate choice copying behavior, while the males are probably affected by whether they perceive the faces as belonging to competitors. If the pictured female appears to like one of the faces more (because she's smiling at it), the males see that face as competition, and thus rate it as less attractive.
At first sight, this seems pretty straightforward evidence of mate choice copying. If one female appears to like one of the two guys, then the female participants like him more. However, there's a potentially serious design flaw in the study. Those of you who've taken an experimental class may remember the difference between a between subject design and a within subject design. In a between subject design, different participants participate in the different conditions of an experiment. In a within subject design, each participant participates in all (or more than one) condition. Jones et al. used a within subject design, in which each participant viewed each pair without the female face and with the female face. The problem with such a design is that the purpose of the study becomes pretty obvious to many participants because the experiment is simple (in technical jargon, it has pretty strong demand characteristics). This often leads participants to attempt to answer in a way that will confirm the obvious predictions of the experimenter (by, say, rating smiled-at faces as more attractive). The problem, then, is that it's difficult to know whether the manipulations (the direction the female was facing, and whether she was smiling or had a neutral expression) caused the differences between conditions, or whether participants figured the purpose of the experiment out, and answered accordingly. So before journalists write any more articles about how women are attracted to guys because other women are attracted to them, we should probably do some more research to find out whether mate choice copying is something they're actually doing, or if Jones et al.'s participants were given just a little bit too much information about the purpose of the study.
1Jones, B.C., DeBruine, L.M., Little, A.C., Burriss, R.P., & Feinberg, D.R. (In Press). Social transmission of face preferences among humans. Proceedings of the Royal Society of London B .
Your comment is right; subjects could have guessed the aims of the study and respond guessing what was at stake. Now, the interesting question is why female participants and male participants responded in a different way: even if the experimental setting showed the goal in an obvious way, the experimental hypothesis was not explicit nor suggested in the setting, and in principle, we should expect the same responses both from male and female participants. That is, nothing in the experiment was suggesting that female participants were expected to like the men that were liked by others, and likewise, nothing was suggesting that male participants should dislike them. The very fact that even under such a clear experimental setting where the goal is obvious female and male participants decide to act according to the non-explicit experimental hypothesis suggests that the experiment might be on the right track.
I don't know whether this shows anything at all. Just because someone says one face is "more attractive" does not mean that she prefers that face for any purpose. It is just as likely that when rating faces for attractiveness, at least some women will state what they believe an objective standard is rather than their own subjective standard. And it still is not necessarily related to social preferences.
About 20 years ago I read about a study that attempted to change the notions of beauty and goodness associated with physical appearance among kindergarten aged girls. As I recall, at the outset of the school year, the girls were asked which dolls were the "prettiest" or the "nicest." Nearly all the girls chose the blonde dolls. The teacher then instructed each child that she was wrong and that the better dolls were the dark skinned and dark haired dolls. Sure enough, from that point on, when asked which dolls were "prettiest," the girls said the dark skinned dolls were. But they still preferred to play with the blonde dolls almost exclusively.
I thought this was an ugly experiment, even if it was supposed to be an exercise in racial tolerance. (I don't like screwing around with little kids' heads.) But it led me to be more critical of studies that reputedly measure internal attitudes based on a stupid quiz.