XIV. Cherchez la testostérone!
When we discovered that Toxo positive and negative men differ in the length ratio of their fingers, as well as body height, we immediately considered the possibility that the relative differences in psyche and morphology could be caused by a change in testosterone levels. Testosterone levels are known to have a significant effect on a number of human characteristics, ranging from behavior to anatomical proportions. One of these characteristics is the masculinity of an individual’s appearance. Because we had many photographs of students as a by-product of our tests for Suspiciousness (Vigilance), we had material on which to test our hypothesis on that there were higher testosterone levels in Toxo positive men. We had the female students of another college, who didn’t know our students, evaluate the level of masculinity and dominance of the men on the photographs, based on an eight-point scale. From these evaluations, we calculated an average value of visually-determined masculinity and dominance for each student. True to our hypothesis, it turned out that Toxo positive men appear significantly more masculine and dominant to female evaluators than do Toxo negative men. But after conducting the same (reciprocal) test with the photographs of the female students, we found that male evaluators didn’t see any appreciable difference in femininity and dominance between Toxo positive and negative women (57).
We wanted to show the readers of our article the differences in appearance between Toxo positive and negative men, without making public the photographs of our students. To this end, we used composition photography, a technique invented by Darwin’s cousin, Francis Galton. To see what the typical criminal looked like (something that would be near-impossible today, for reasons of political correctness), he exposed a number of individual photographs of criminals on photographic plate, creating a single composite portrait. Today the techniques for composite photography are a bit more
Fig. 33 Differences in the facial features of Toxo negative and positive students. Composite photographs from 22 pictures each of (a) Toxo negative students and (b) Toxo positive. Toxo positive men on individual pictures were evaluated by female students as more dominant and masculine than uninfected men; these characteristics are also apparent on the composite photographs.
sophisticated. First each photograph is marked with dozens of reference points on the computer screen, and then a special program projects the individual portraits over each other, so that the reference points match up as best as possible. The result is a single photograph of the “average person,” in which individual features disappear, whereas facial characteristics representative of the group remain visible. So if we take 30 photographs of individual students, randomly divide them into two groups and make a composite photograph of each, we should end up with two very similar composite portraits. Considering the very racially homogenous population of the Czech Republic, 15 photographs would suffice for the individual facial features to cancel each other. To make composites of our students’ photographs, we separated them according to gender and presence of Toxoplasma infection. Hence we created four composite photographs: the “average” Toxo positive man, Toxo positive woman, Toxo negative man and Toxo negative woman. To ensure that the possible differences in the composite photographs weren’t coincidental, we used 22 instead of 15 student portraits (that was the maximum number we had available for Toxo positive male students). The resulting composite photographs of the men can be seen in Fig. 33. The composite photographs of the Toxo positive and negative women are almost identical (not shown). In contrast, the composite face of the Toxo positive male students look more dominant and masculine than that of the Toxo negative male students (57). Just curious tidbit: the evaluators also agreed that the “average” Toxo positive man looks older. In reality, the Toxo positive male students who made up the composite photograph were, on average, almost a year younger than the Toxo negatives (Box 63 How to compare photographs using a one-sample test – do we like self-similar faces?)
Just as an aside, since we already had a program set up for evaluating characteristics from photographs, along with the data from the Cattell’s questionnaire’s, we decided to see if our evaluators would be able to guess some of the characteristics of the students, based only on the photographs of these students. We had already determined the students’ characteristics using the psychological questionnaire. In her study, Anna Kotrčová-Rubešová discovered that women in particular are pretty spot-on in their guesses. Their guesses regarding the Cattell’s factors of Warmth (A), Liveliness (F) and Social Boldness (H) correlate strongly with the factors determined with Cattell’s questionnaire. Men were not as talented in this discipline, but nevertheless managed to guess some characteristics of the photographed people surprisingly well. We must also realize that the evaluators had only a single photograph of that person’s face (with the hair edited out) to work with; furthermore, the photographed people had been instructed to pose with a neutral expression. In normal, everyday life, a person gets much more information with which to evaluate others – and so it’s likely that they are also more successful in this endeavor. As Věra Pivoňková proved later on, women were even more successful in guessing psychological traits when they were given composite
Box 63 How to compare photographs using a one-sample test – do we like self-similar faces?
There are two ways to determine whether photographs of Toxo negative or photographs of Toxo positive people appear more attractive to evaluators. Either we can show the photographs individually one after another, and have the person evaluate each for a certain characteristic on a scale of 1 to 8. Or we can always project a pair of photographs, one Toxo positive and the other Toxo negative, and have the evaluator choose which he thinks is stronger in the given characteristic (such as more dominant or attractive), and maybe even decide by how much. For example, in one of our studies Jitka Lindová looked at how attractive people found faces that were similar or dissimilar to their own. First we prepared a set of 15 composite photographs, each constructed from three photographs of various English students. Because each composite photograph averages the features of only three students, each composite face was fairly distinct. The photographs, however, were not of any particular students – so that our evaluators wouldn’t recognize them. The thus created 15 composite photographs were then tailored to each evaluator we invited to take the test. Using a special program, we either added the features of the evaluator or subtracted them from the composite photograph. So when the evaluator (a student, whose photograph we had obtained from our earlier Suspiciousness test) logged on to the website under his designated username, he was successively shown 15 pairs of photographs. Each pair of photographs was very similar, but one resembled the evaluator more, whereas the other did not (Fig. 34). The differences were very subtle; none of the evaluators noticed that half of the 30 total photographs somewhat resembled their own face. For each pair of photographs, the evaluator was to decide on a scale of 1 to 8, whether he found the face on the right or left more attractive. (1: left is very clearly more attractive, 2: left is clearly more attractive, 3: left is a little more attractive, 4: left is barely more attractive, 5: right is barely more attractive, 6: right is a little more attractive, 7: right is clearly more attractive face, 8: right is very clearly more attractive). To avoid introducing a systematic error, we alternated the placement of the similar and dissimilar face as projected on the left or right side. (We simultaneously alternated the polarity of the scale, so that strong preference of the self-similar face was always indicated by 8 points, regardless of its position on the screen). Furthermore, each evaluator saw the 15 pairs of composite photographs in a different order.If the evaluators were not influenced by the similarity or dissimilarity of each face to their own, then their average evaluation should lie between a 4 and a 5. In reality, it wasn’t so. Women, for example, preferred similar or dissimilar faces based on whether or not they were in the fertile phase of their menstrual cycle. I won’t reveal any more of the results, as we haven’t yet published the study – two times since it was finished, Jitka Lindová happened to be very attracted to her husband during the fertile phase of her menstrual cycle. The statistical test that we used to analyze the study is classified as a one-sample test.It compares the calculated value of a random sample, which is subject to random error, with the theoretical value (in this case, 4.5), which is not subject to random error. A one-sample test is considerably stronger (more sensitive) than a two-sample test, which compares two values computed from two random samples (whose values are subject to random error). Usually they are the average value of the dependent variable from the experimental group and the control group.
photographs consisting of several people that achieved an extreme score for a given factor (a very high or low value).
Of course, it’s a question whether the physiognomy of the human gradually changes to match his personality, or whether the human learns to behave and think to match what his environment expects from him based on his appearance (Box 64 Are brown-eyed men more dominant than the blue-eyed, and is it because of their eyes?). When perusing the panopticon of our leading politicians, I sometimes wonder if a person who looks like a bastard at first glance, might be a bastard in real life. It’s probably not politically correct, but what if there was something to it?
Fig. 34 Use of composite photography and a morphing technique to determine preferences for self-similar persons. Each pair of pictures was created through computer modification (morphing technique) from one original composite. In each pair of composite photographs, the picture on the right was made more similar to the features of the evaluator, and the picture on the left less similar – by adding or subtracting, respectively, the characteristics features of the evaluator from face on the composite photograph. The original composite pictures were formed from three (in each case different) photographs of students from England. The evaluator was to determine which of the two faces in his pair is more attractive. Among female evaluators, the preference for self-similar faces changed over their menstrual cycle.
Box64 Are brown-eyed men more dominant than the blue-eyed, and is it because of their eyes?
The answer is yes and no. As I wrote earlier, Karl Kleisner and I studied the human eye as a possible organ for communication. The color of the iris is prominent in the human eye, especially in contrast with the very visible white sclera. In most mammals, eyes are not this prominent, so it is possible that in humans they developed the secondary function of communication – that through them, man conveys information to his surroundings. For example, the movement of the eye made visible by the white sclera can reveal precisely where a person is looking, so that a woman wearing a low-cut top on the subway can see, out of the corner of her eye, if the gaze of the man sitting across from her has finally wandered to where she, as well as the designer of the shirt, intended. But it is possible that eye-color could also reveal certain information. When we had evaluators (primarily females) decide the dominance of brown- and blue-eyed students based on their photographs, we found that on a seven-point scale the brown-eyed students were seen as more dominant (Fig. 35). For the control, we used a computer program to switch the eye colors – that is, we gave the blue-eyed photographs brown eyes, and the brown-eyed photographs blue eyes. And then we repeated the experiment with a different group of evaluators. And you wouldn’t believe what happened! Nothing whatsoever. The originally brown-eyed men were still evaluated as dominant, even though we changed their photographs to show blue eyes. This must mean that the women didn’t evaluate dominance based on eye-color, but facial features; brown eyes happened to be set in a “more dominant face.” The most likely explanation we have so far come up with, is something that we called the social feedback loop hypothesis(58). (Remember, it’s not important who first proposes the hypothesis, or who explains it the best, but rather who comes up with a name that sticks.) According to this hypothesis, all children are born blue-eyed but some of them later develop brown eyes (in the Czech Republic, about half of children). If a boy is blue-eyed, then he is treated as a child for a longer period of time. This leaves an impact not only on his behavior, but also on his usual facial expressions, which over time is inscribed into his facial features. Blue-eyed boys (at least in our central-European population) start off with a greater probability of looking submissive, as opposed to brown-eyed boys, who start off more likely to seem dominant, all because their environment expects it of them.
We decided to test our hypothesis about higher testosterone levels in Toxo positives using direct biochemical methods as well. In another series of ethological experiments, we took samples of saliva from students and used radioimmunoassay to determine the concentration of the testosterone. It turned out that Toxo positive men really do have significantly higher testosterone levels. That was expected. But what surprised us
Fig. 35 Differences in the face shape of blue-eyed vs. brown-eyed male students. The bottom two pictures are composite photographs of the average blue-eyed (left) and average brown-eyed (right) student of the College of natural sciences. The top two pictures have the typical characteristics from the bottom composites (the differences between the left and right figure) enhanced by computer. Based on these morphological features, student evaluators ranked individual photographs of originally brown-eyed students as more dominant than those of originally blue eyed students, even after we used a computer program to change the eye-color from brown to blue or vice versa. Regardless of eye-color, female evaluators perceived originally brown-eyed students more dominant. These results are surprising since eye color depends more or less on one gene, so in each generation brown eyes should “move from” one face to another, and not be connected to a certain type of face.
was that Toxo positive women had a significantly lower concentration of testosterone (59). The fact that the same factor – in this case, infection by a parasitic protozoan – can raise testosterone levels in men while lowering them in women, is pretty strange. On the other hand, this fits with the results of our previous studies, which showed that toxoplasmosis causes an opposite shift in men versus women for a number of factors. Originally we tried to find a psychological explanation for this phenomenon, and nature unexpectedly presented us with not one but two – a psychological and an endocrinological explanation. It’s clear that a number of psychological factors, including the Superego Strength (factor G), could correlate with the level of testosterone. If toxoplasmosis raises testosterone levels in men but lowers them in women, it’s logical that psychological factors related to testosterone concentration would also shift in opposite directions. The actual mechanism which Toxoplasma uses to influence testosterone levels is still unknown. Increased testosterone levels in Toxo positive men could be explained by the fact that dopamine production correlates with testosterone production. Humans and lab animals with high levels of dopamine in the brain are also known to have high testosterone levels. This phenomenon, however, does not explain why infected women, who also exhibit higher dopamine levels, have lower testosterone levels.
We repeated the same study with a different group, a fairly large group of patients from an immunological clinic who were primarily treated for allergies and immunodeficiencies. With this group, we were unable to prove a relationship between toxoplasmosis and the level of testosterone (or estradiol or cortisol). What was interesting, however, was that the level of testosterone in men increased with time after infection (this time was estimated based on the decrease of specific Toxoplasma antibodies). It’s apparent that this group experienced the sieve effect, which could have impacted our results. In the Czech Republic, toxoplasmosis is about equally prevalent in men and women. But in our group of 312 women and 228 men, toxoplasmosis occurred in 24% of the women (which is close to the normal value) but only in 10.9% of the men (ridiculously below average). As I explain in a later chapter, Toxoplasma significantly influences the immune system of both men and women. Whereas Toxo positive women show signs of a more active immune system, Toxo positive men show signs of immunosuppression. It’s possible that Toxo positive men are protected against allergy, so we find very few of them among the patients of an immunological clinic. It’s even possible that the immunosuppression in Toxo positive men is caused by the increased testosterone levels. The 14 Toxo positive men that made it into our group may have had a deficiency in testosterone production, so even Toxoplasma couldn’t save them from allergies. That would explain why we were unable to find higher testosterone levels in the infected men of this group, even though their testosterone levels increased with time after infection. Of course this is only a hypothesis, which needs to be verified in the future (Box 65 How are hypotheses tested in science?).
Box 65 How are hypotheses tested in science?
It’s difficult, but maybe that’s why we scientists enjoy it. In general, we have to comply with a rule explicitly described only in the first half of the 20th century by sir Karl Raimund Popper: In science, hypotheses cannot be proven – only refuted. We consider a hypothesis to be (conditionally) true, when it has survived a sufficiently long onslaught of intensive attempts to refute it. The word conditionally, included in parentheses, is immensely important. Scientists, and also the public, should accept that a scientific “fact,” hypothesis or theory, even one which has rarely been doubted and is commonly thought to be “safely proven,” can at any time be refuted. It doesn’t matter how many times our experiments confirm a certain hypothesis; how many times the results of our empirical studies agree with the expectations that come from our hypothesis. The same results can agree with other hypotheses that we haven’t yet thought of – and so hypotheses that we haven’t tested. But if our empirical studies give us a result which disagrees with our hypothesis, we should rightfully be excited, because we succeeded in refuting the given hypothesis.
Of course, if the refuted hypothesis is our own, the enthusiasm usually isn’t completely sincere. But a scientist who is greatly averse to refuting his own hypotheses can choose to test exclusively the hypotheses of others, and then he’s in the clear. However, the situation is a bit more complicated than depicted in basic textbooks of scientific methodology and on Wikipedia. For example, Popper’s principle about hypotheses that cannot be proven but can be refuted applies only to those hypotheses which take the form of a statement with a generalized quantifier; those hypotheses which take the form: For all x, y is true. The hypothesis that “all birds have one pair of wings” is impossible to prove (we’d have to catch all the birds in the world and check that each has one pair of wings). If we find just one bird with two pairs of wings, our hypothesis is refuted. Popper’s principle doesn’t apply to the hypotheses that take the form of a statement including an existential quantifier; hypotheses that take the form: There exists at least one x, for which y is true. The hypothesis that “there exists a leprechaun who always hides my lab keys” would be difficult for anyone to disprove based on empirical data (just this morning the terrible creature hid them twice!), but it is possible to prove – I only need to catch him in the act (he had better be ready!). Popper would probably easily handle this objection. He would probably argue that for every hypothesis there exists a superior hypothesis stating that “all empirical consequences implied by our hypothesis are true.” Since this superior hypothesis takes the form of a statement with a generalized quantifier, it cannot be proven, but with a little luck, it may be refuted. However, even this would not completely save Popper. In the second half of the 20th century, the historian and scientific theoretician Samuel Thomas Kuhn claimed that no theory can really be refuted by an empirical experiment, because the theory can always be modified to include the facts that didn’t fit with its original version. (A theory is basically the same as a hypothesis, but is usually more complicated and explains a wider range of phenomena.) In general, scientists abandon an old theory only after inventing a new one to take its place. At first the new theory is less precise, and sometimes even worse at explaining the known facts; but it’s also simpler and more elegant (if only because there hasn’t been time to make it more complicated). Switching to a new theory is generally a very gradual process. The supporters of the old theory don’t just wake up one day and say, “the new theory is better, let’s forget about the old one.” Usually proponents of the old theory have to die out (or at least go into retirement), and then proponents of the new theory take their place.
To get back to something more specific: naturally, we didn’t cast aside our hypothesis at the first sign of conflicting results. We expected that Toxoplasma raises testosterone levels in men; and we discovered a group in which infected and uninfected patients had the same testosterone levels (the Toxo positives even had them a little lower). Instead we modified our hypothesis: “in the general population, Toxoplasma-infected men have higher testosterone levels; but in a population of patients with allergies and immunodeficiencies, they have the same and maybe even lower testosterone levels as a result of the sieve effect” (see the mechanism that includes the effect of immunosuppression, described in the main text). And so we salvaged our hypothesis. When it becomes too complicated, we’ll abandon it, so you won’t have to wait around for my retirement. I’m usually not too enamored with my hypotheses and theories – I prefer inventing new ones.
We used the model of lab mice to verify the observation that Toxo positives have altered testosterone levels (60). Our experiments confirmed the results on humans, but are nevertheless difficult to interpret. We expected to see a phenomenon similar to the one we observed among our students; that Toxo positive males have higher testosterone levels and Toxo positive females have lower testosterone levels. Instead we found that both male and female lab mice had significantly lower testosterone levels after infection. We still have to confirm this result with other experiments. If we do confirm it, and discover other groups in which toxoplasmosis does not raise testosterone levels in men (we found this in two groups so far: the group of patients with allergies, and a group of 100 infected and 100 uninfected soldiers), then we’ll have to consider the possibility that something about our students is weird. For example, we might suspect that toxoplasmosis does not increase the levels of testosterone in male students, but rather that male students from the country, where the prevalence of toxoplasmosis is higher, have higher levels of testosterone than do male students from the city.
Originally we didn’t know what molecular mechanism Toxoplasma could be using to influence the behavior and psyche of its host. Currently, we sort have the opposite problem. We have several factors that could be responsible for the changes in behavior and psyche in infected people. First there is dopamine, which we think may affect the tendency towards “novelty seeking.” In this case, the psychological shift occurs in the same direction in men and women. Then there is testosterone, which could be responsible for the changes in other psychological factors. Furthermore, increased levels of testosterone in men and decreased levels in women could explain why a number of psychological and behavioral characteristics shift in opposite directions in infected men and women. Another question is whether these factors – testosterone and dopamine – are really independent of each other, or whether changes in testosterone might be caused by raised levels of dopamine.