Frozen Evolution. Or, that’s not the way it is, Mr. Darwin. A Farewell to Selfish Gene.

As I already mentioned, I began and never finished a number of studies; more often, I finished them but didn’t analyze the data, which were left lying somewhere in my desk, and eventually passed from even the back of my mind. They never got a high enough priority for me to dig them out again; there was always more important data to analyze. Sometimes this other data was more important from a scientific stand-point, meaning that I was particularly interested in how a certain study would turn out. Other times for practical reasons, when a study was part of the undergrad or graduate thesis a student had to defend, I had to give priority to this student’s study. Therefore, it is possible that the most interesting discoveries lay unrecognized, buried at the bottom of my desk drawer.

When I think back on my unanalyzed data and unfinished studies, I envision a sort of comparison. Doing science is most similar to walking through an unknown rugged landscape. One sets out on a path, sees various paths branching off, and tries to pick the most interesting of these. But the landscape is rugged, so he doesn’t know – if he’d turned on an earlier path or continued on the original – where he might find, perhaps, a spring or an enchanted valley, more beautiful than the landscape he’s currently traversing. This comparison has one flaw: the scientist himself makes these paths, as he asks himself questions. So there is some difference between roaming a real and scientific landscape; in the end, though, it’s still aimless wandering. When one chooses a path, he must accept that he won’t get to the others and will never know what he might have found at their ends. Maybe the reason I found several interesting things during my studies on toxoplasmosis, and the reason the members of my lab made a number of compelling discoveries, was that I was intuitively able to choose those paths which led to something exciting. Or it might have been just good luck. Of

course it’s also possible, that we narrowly missed things that could have been much more interesting.

One such path, whose results I never really analyzed due to time constraints, was the longitudinal study that my colleague Havlíček and I carried out on the university students (Box 41 Longitudinal and cross-sectional studies).

During this study we sent each student a letter with the same psychological questionnaire they’d completed five years ago. We expected that differences in psychological factors due to infection would have intensified in Toxo positives, over those five years between the first and second questionnaire. In the Toxo negatives,we expected on average no change. Hence we expected that Toxo positive women would have an increase in warmth (factor A) and rule consciousness (factor G); while in Toxo positive men, these factors should have decreased.

I don’t remember anymore, whether the results were even statistically significant, but in any case they were opposite of the direction we expected. In infected men, factor G increased instead of decreasing; in infected women, factor A decreased instead of increasing. The reason we obtained such unexpected results could have been trivial, a mistake in the data or even an error in the statistical program (Box 42 Errors in statistical programs).

We carried out only perfunctory analyses on data that was not checked carefully enough (Box 17 Checking and cleaning data). Therefore, it is possible some people’s data had unreasonably high or low values; these errors might have completely changed the results of the analyses. But the study could have had results opposite of those we expected for nontrivial reasons. The values for Toxo negatives could have shifted because part of them got infected during those five years (for obvious technical reasons, it was not possible to retest students for toxoplasmosis after five years). The values of both Toxo positives and negatives could have shifted due to the so-called ceiling effect. For example, Toxo positive women already had a very high factor A (warmth) in the first testing, so stochastic changes over the five years were more likely to be decreases than increases in its value. Cattell’s psychological factors take on values from 1 to 10. If a Toxo positive woman were to get a 9 for factor A, it would be much more likely that a second testing would give her a lower number (1-8), than a higher number (10). Women who got a 10 for factor A in the first testing reached the “ceiling” of this factor; in another test they can only get an equal or lesser value, for they can’t get any higher. Conversely, a Toxo negative women who got a 3 for this factor in her first test, had a greater probability of having the value decrease than increase, due to a random change, mistake, or a mood swing during the repeat test. Of course these are just theories, and I expect that modern statistical methods could eliminate the ceiling effect.

It is also possible, I think, but not very probable, that the opposite shifts in Toxo- positive subjects are real. For example, the personality changes could be only transient and begin returning to their original values after five years. However, our studies on male and female patients who have passed the stage of acute toxoplasmosis do not show any such return (see Figs. 13 and 14). It’s possible that I’ll eventually get back to the old data, or even try to add to them. It would definitely be interesting to send this questionnaire to the same people after another 5 years and see how the personal profiles of Toxo positive and negative people develop in such a long period of time.

Something else that surprised me, and is worthy of note, regards factor G (rule consciousness). So long as we used the original version of Cattell‘s Questionnaire, Toxo positive male students had a significantly lower G factor than did the Toxo negative males. But over time, new versions of the questionnaire appeared, and some of the reviewers of our manuscripts criticized us for using the outdated version. So finally, for practical reasons, to one-up the reviewers, we decided to switch to a newer questionnaire. When we used it to test a large group of soldiers, about 400-500 in total, it turned out that Toxo positive and negative men still significantly differed in factor G, but this time Toxo negative men on average reached a lower value than did the Toxo positive. In other words, the new version of the questionnaire showed Toxoplasma as having effects on men, as did the old questionnaire on women. Of course, I don’t know why this happened, but two possibilities seem most likely. The first relates to the change in Cattell‘s Questionnaire – that the new version measures men differently than does the old. The second is based on the change in people – that students and employees of the college react differently to latent toxoplasmosis than do soldiers of the mandatory military service. We can test the possibilities by either testing the students with the questionnaire’s new version, or giving the soldiers the old version. Seeing as the original study with the new version of the questionnaire was conducted on these soldiers, and in the meantime mandatory service was abandoned, we could only test the students using the new questionnaire. In a bit, I’ll explain what happened.

If it was true that the new questionnaire measured factor G (rule consciousness) differently than the old, then psychologists who habitually used the questionnaire would not be too happy, since it would mean that one the versions indicated something other than it should. The other possibility, that reaction of soldiers and students to the same factor (toxoplasmosis) was different, wouldn’t be all that serious nor surprising (see also Box 40 What all influences the results of performance tests?). It’s actually a common problem of most observational studies, which we encounter quite often in those with questionnaires. It’s an issue of the representativeness of a sample.

In our studies, we generally aren’t interested in just a particular study group, but in what happens in the entire population. For example, we don’t want to know how the 2001 students of the college of natural sciences react to toxoplasmosis, but how people in general react to it (or maybe men in general). Of course we can’t test all the people infected by toxoplasmosis and observe them along with all the uninfected people. Instead, we must use a smaller sample of the population, like 300 students or 500 soldiers; and then extrapolate the differences between the Toxo positive and negative people in the group to the whole population. The problem with conducting a study on a truly random population sample (so that it includes all age groups, usual professions and social and economic classes), is that the sample will be very heterogeneous. Thus the effects we measure in the sample – such as psychological traits – will have an enormous variability in the trait being studied. The group will have individuals with extremely high and extremely low values of this trait. As a result, if there is a difference between Toxo positive and Toxo negative people in this group, there is a good chance our tests won’t discern it as statistically significant. Statistical tests usually determine statistical significance by finding to what extent the variability in our group explains the studied factor (such as toxoplasmosis). Of course, when we’re analyzing a very heterogeneous group, then the contribution of even a fairly strong factor to its variability is relatively small. Consequently, these tests offer only negative (statistically nonsignificant) results. One way to limit this risk is to make the group as homogenous as possible – like using only the students of the college of natural sciences. In many ways, these students are quite similar to each other, for they all got into our group,already passed through several “sieves,” apparently have similar interests, are all about the same age, passed the entrance exam for the college of natural sciences, and were willing to volunteer for our tests (Box 43 Where have all the brown-eyed students gone?).

As a result, the variability (dispersion) of their characteristics is much lower, than had we chosen a representative sample of the general population. Theoretically, a sample of soldiers of the mandatory military service should also be representative of the masses. In reality, this is far from true. When we conducted our studies, it was well known that the mandatory military service in the Czech Republic would soon be cut. Therefore, the only people who went to the military were either those who didn’t mind it, those who looked forward to it, or the people unable to get a note from the doctor to excuse them until it would no longer be required.

But a homogenous sample has at least two disadvantages. Firstly, things we discover on such a homogenous study sample cannot be entirely generalized to the whole population, for it’s possible these discoveries fit only the sample group. Secondly, the sample group may be missing some characteristics present in another sample group, or in the general population. This second problem can be partially addressed by studying our particular phenomenon in several very distinct sample groups. But then one can get the situation we had with soldiers and students examined with Cattell‘s Questionnaire – in terms of the factor of rule consciousness, the relationship between Toxo positives and negatives was different in the student group than in the soldier group. In the male student sample, the infected people had a lower rule consciousness than the uninfected – but in the soldier group, the uninfected had a lower rule consciousness. One explanation would be that the phenomenon of compensation, more strictly overcompensation for psychic change, might play a role in this.

When the infected soldiers or students felt that something strange was happening with their behavior, they may have consciously or subconsciously tried to suppress or at least “cover up” the changes from themselves and those around them. They may have overdone this cover-up to the extent that, though they really had a higher rule consciousness, it showed up on the psychological questionnaire as lower. Something similar could have also happened in the case of warmth (factor A). For a long time, it was how we explained to ourselves why men and women have opposite reactions to toxoplasmosis. This happened not only with warmth, as well as rule consciousness, but with many more psychological factors. Once I sat down and counted up the factors with a shift that wasn’t statistically significant, but only suggested a trend; or where the shift was statistically significant just in one of the genders, while the other gender had an opposite trend. It turned out, that out of 16 Cattell‘s factors 12 had an opposite toxoplasmosis-associated shift in men to that in women. Eventually, it became apparent that the explanation based on overcompensation of toxoplasmosis-induced changes was false; and that there was a number of simpler explanations for the opposite shift of psychological factors in men and women. But this realization came to us only with the results of our study on testosterone levels in Toxo positive and negative people; but these results will be discussed in a later chapter.

As I already mentioned, we can no longer test soldiers of the mandatory military service (hopefully, so long as mandatory military service isn’t reinstituted). However, we did analyze data obtained from blood donors with the old version of the questionnaire. It turned out, that even in this group infected men behaved the same as infected women – they had a higher rule consciousness (factor G) and a lower vigilance (factor L) (34). Therefore, it seemed that our students, not the new questionnaire, were at fault. To confirm this possibility, (with a heavy heart) we began testing our students with the new version of the questionnaire. Try and guess the results that we got. Are you ready? An experienced scientist certainly would guess correctly that this study turned out completely differently than the previous two. In men, the differences between infected and uninfected weren’t statistically significant; though this time infected students had a lower instead of a higher factor G. In women, the differences were statistically significant; but now infected women’s factor G was statistically lower. It’s hard to decide, how we should determine these new results (which came out completely opposite to the results on students over 10 years ago). However, it’s most likely that the students of the college of natural sciences today are quite different from the students 10 or 15 years ago. Even without the questionnaire, it’s clear that our students change over time. (Of course, they get worse. Each teacher agrees that the department had its best pupils when they were students – whether the teacher was a student 5 or 50 years ago.) We’ll never put together the exact same group of students that we tested 15 years ago. So we’ll never definitely know, whether each population’s different results of Cattell’s Questionnaire were really due to distinct personal profiles in each group, or differences between the new and old questionnaire versions. It’s true, we could let our students take both the new and old version of the questionnaire, but I probably won’t dare to do that any time soon. I’m almost afraid to think what our test subjects would say to us, and I couldn’t print it in this book, if they had to fill out two nearly identical 200-question questionnaires.

By the way, the fact that we have different students than before is one of the paths I’d like to explore, had I the time. Since 1992, every year I go to the lectures of my colleagues and convince students to take part in our trials, and each time a fairly large number of students succumbed to my persuasion. And so far all these victims have, among other tests, completed the Cattell’s Questionnaire. Though we didn’t systematically follow the changes in our students’ psychological profiles, we noted some differences anyway.

Cattell’s Questionnaire has about 12 questions, which can be used to roughly estimate intelligence. Sometime in 2006, we compared how the intelligence of our students changed over the years. We found that intelligence of the male students stayed about the same; in women, however, it was slowly decreasing. I hardly think that this is due to a decrease in general intelligence in the female population, and a steady-state of intelligence in men. It’s almost certainly because, while we test about the same amount of students each year, over 17 years the number of female students has dramatically increased (about three times the original number). The increase in accepted applicants is due to the fact the college gets money from the state primarily based on number of students, so it has a tendency to increase the number of students each year. In general, I see this as good thing: the more people aged 18 to 22 we put in classrooms, the less of them will be in the streets setting cars on fire and throwing Molotov cocktails at the police. Plus, we’ll teach them to learn – whether by training their minds on the classification of sawyer beetles, theology or macroeconomics. Knowing how to learn (anything) is a skill that will almost be useful in finding a place in today’s changing world. But raising the number of students also has its dark sides. Namely, the workload of teachers is increasing. As a result of the increasing number of applicants, already many years ago we abandoned oral entrance exams. In my opinion, this is a grave mistake, and the greatest failure of us teachers.

I have no solid data to support this opinion, but I have the feeling that statistically women, unlike men, do much better in written than in oral exams. Additionally, oral exams also looked at the motivation of applicants, so our college got these enthusiastic young biologists, both men and women, who had wanted their whole life to study, for example, the behavior of Przewalski's horse in the Prague Zoo, or collect rove beetles. To a disinterested bystander, this may seem like a silly occupation, but experience shows that precisely such “enthusiastic fools” grow up to be the most interesting students or scientists, with the most original ideas. The majority sooner or later moved on from Przewalski's horses or rove beetles, and, with no less enthusiasm, turned to studying things like the characteristics of cancer cells. But because today we only give written exams, these people are much less likely to get in than before. They studied exciting Przewalski's horses, not boring high school textbooks. For this reason, they don’t make up the same proportion of the increased amount of students; instead probably about the same number gets in as before. And among “normal” high school students, i.e. in male and female students who were never interested in rove beetles nor Przewalski's horses, females score better in written exams than males do. This, I think, is the reason that average intelligence decreases specifically among female students. More and more female students are accepted, so there is a greater fraction of them who, while not that creative and motivated, were able to cram for the high school material we test in our written exams. I guess that in comparison with studying Przewalski's horses and collecting beetles, the ability to excel in written tests over high school knowledge is a relatively poor indicator of success in university studies – and probably does not correspond with formal intelligence tests.

Using written exams to test biology is a questionable method anyway. Yet even I must do so, because annually about 400 hundred students apply for evolutionary biology classes. If I had to give each person an oral exam, then I couldn’t be a scientist, but would spend half the year testing undergraduate students. So, I was forced to switch to written exams; in other words, I made myself a sort of compromise. All students take a written exam, and if they aren’t happy with their grade, then they can request an oral exam, in which their previous written score is not looked at. So the written exam is actually a test they can retake orally. The truth is that most students are content with their written exam grade and never come in to improve it. I don’t know, whether it’s because my grading is overly lenient (maybe because I don’t want too many students taking an oral exam), or because our students don’t really care what grades they get. I’m afraid that many students only care about completing their study requirements, not about how they do so or with what results.

Giving even written exams to hundreds of students, and entering grades into their transcripts and the department database, is an annoying business anyway (I am certainly much happier giving lectures and researching). To make this teaching duty more enjoyable, I usually try to carry out an experiment along with the written exam. Participation in the study, of course, is voluntary; and I always stress that it has nothing to do with the grading. But most students usually take part in it, and after the test for example, complete some questionnaire. Recently, during a written exam, I observed how the hormone levels in the saliva of students changed in relation to how they did on the test. I found that the level of testosterone rose in “victors” and decreased in those “losers,” verifying that in this regard students behave the same as bucks fighting for a doe. Furthermore, I unexpectedly discovered that a student’s subconscious may understand evolutionary biology better than his conscious (see Box 44 Does a student’s subconscious understand the material he’s learning better than his conscious mind?).

But back to Toxoplasma and toxoplasmosis. We hit upon another potentially interesting mystery already in the first years of our study on the effect of latent toxoplasmosis on human behavior. Latent toxoplasmosis is diagnosed using two independent tests:

the complement-fixation test (CFT) and the enzyme-linked immunosorbent assay (ELISA) (Box 45 How science, using the ELISA test, once again saved mankind).

Fig. 24 The relationship between testosterone levels after a written examination on evolutionary biology, and the number of errors made on said test was statistically significant. Neither the relationship between testosterone levels before the exam and the actual number of errors, nor the relationship between testosterone levels after and before the exam and the number of errors the student thought he made was statistically significant

ELISA test recognizes class IgG antibodies in the blood (see Box 22 Antibodies and myths about them). A complement-fixation test detects class IgG immunoglobulins, but only some of their subclasses, and also detects class IgM immunoglobulins. The ELISA is more sensitive in the case of a fresh infection, and can more precisely distinguish whether a person has acute or latent toxoplasmosis. A complement-fixation test is less sensitive, but its results fluctuate less and based on them one can estimate when the person was infected. In some people, the results obtains with each method differ. A person may be Toxo negative based on the ELISA and Toxo positive based on the CFT, or vice versa. In most cases the results of the two tests agreed; nevertheless, in our groups there were always several individuals (in large groups even a couple dozen people), whose results did not match. Sometimes we solved the problem by not including these people in the final analysis, counting only the people who had the same results for both tests.

What is surprising – when we split the people into Toxo positive and negative based only on the CFT results (and didn’t rule out those with opposite test results obtained by ELISA), we almost always found a stronger correlation between toxoplasmosis and psychological factors. Conversely, when we divided the persons into positive and negative based only on the ELISA results, we usually found a stronger relationship between toxoplasmosis and reaction time.

I don’t really know, how we’re supposed to explain this observation. Our working hypothesis presumes that the CFT is less specific and detects not only toxoplasmosis (infections by Toxoplasma gondii), but also infection by a related protozoan – and that both these protozoans cause the same psychological changes. Furthermore, the ELISA method might be more specific and only detect infection by Toxoplasma gondii, and not by the other hypothetical protozoan which causes no changes in performance. It is even possible that both tests detect only Toxoplasma gondii, but that the CFT detects a wider range of strains, whereas the ELISA detects only the most widespread (Box 46 There’s no Toxoplasma like Toxoplasma).

Recently, it was discovered that different strains affect the behavior of infected rats in different ways. Whether we presume that the differences in our two tests for Toxo are caused by the occurrence of another parasite species, or by the distinct characteristics of several Toxoplasma strains, it’ll probably be necessary to someday focus on the matter more closely. The two parasites (or different Toxoplasma strains) don’t have be different in only how they affect the human psyche and reaction time; they may also be distinct in other, clinically much more important characteristics.

Fig. 25 The three main groups of European and American Toxoplasma strains. Molecular taxonomic methods demonstrate that in Europe and North America these three closely related and biologically similar Toxoplasma strains are prevalent. The three groups branched from a common ancestor, and each group into several strains, relatively recently. In contrast, South America is home to several strains, which are not related to these groups; they branched from Toxo’s common ancestor much earlier. It is likely that most of the evolution of Toxoplasma gondii occurred in South America. The numbers on the graph show genetic distances. Adapted from Sibley and Ajioka, 2008.

Of course, our hypothesis, which suggests that each diagnostic test has a different specificity, isn’t the only possible explanation. Perhaps one of the tests overlooks fresh infections, whereas the other overlooks those that happened long ago. It’s possible that psychological changes develop more slowly than the changes in reaction time (or vice versa). In this case, the different risk in each test of falsely negative results, for someone freshly or long infected, could explain why the correlation between toxoplasmosis and psychological factors comes out stronger when we use the CFT to diagnose toxoplasmosis; and why the relationship between toxoplasmosis and performance comes out stronger when we use the ELISA.