An extremely pedantic exploration of an error in How the Mind Works

A few weeks ago I was reviewing some of the annotations I had made while reading Steven Pinker’s How the Mind Works, and I came across an interesting claim about hormonal changes during pregnancy. I was especially curious to learn more about changes in insulin, the hormone that promotes energy storage by causing tissues to absorb glucose from the bloodstream. Insulin is hugely important in human metabolism, and when the body cannot properly regulate insulin the result is the disease diabetes. Given my interests in endocrinology and fetal development, as well as my personal fear of developing diabetes, I was eager to investigate further. The questionable claim appeared at the end of a section about how, in spite of their apparent shared interests, a mother and her fetus compete for limited metabolic resources within the mother’s body:

“The human placenta is a tissue of the fetus that invades the mother’s body and taps into her bloodstream. Through it the fetus secretes a hormone that ties up maternal insulin, increasing  the levels of blood sugar which it can than skim off. But the resulting diabetes compromises the mother’s health, and over evolutionary time she has fought back by secreting more insulin, which prompted the fetus to secrete more insulin, which prompted the fetus to secrete more of the hormone that ties up insulin, and so on, until the hormones reached a thousand times their usual concentrations. (Pinker 443, my bold)

Early in the sentence, two hormones are mentioned: insulin and “the hormone that ties up insulin”, by which he means human placental lactogen (hPL). When Pinker writes that “the hormones reached a thousand times their usual concentrations”, I interpret this to mean that the concentration of each individual hormone undergoes a thousand-times increase relative to its own “usual concentration”, rather than the combined concentration of both hormones increasing by a thousand times. I resisted this second interpretation because it would be ridiculous to lump together two hormones with such different mechanisms-of-action into a single category.

Furthermore, it is unclear what the “usual concentration” of hPL refers to. hPL is only synthesized and released by the fetal tissue sometime early in pregnancy. I found no evidence that hPL is present in non-pregnant adults, thus making the concept of a “usual concentration” meaningless; thousands of times zero would still be zero. Presumably, the “usual concentration” of insulin refers to its concentration in nonpregnant women.

To verify these claims, I read the source Pinker cited for this section, “Genetic conflicts in human pregnancy”, a fascinating and comprehensive paper by David Haig. Regarding insulin:

“By contrast to glucose, fasting insulin remains close to nonpregnant levels during the first and second trimester, but then increases during the third trimester in parallel with growth of the fetus.(Haig 510, my bold)

“Fasting insulin” refers to the fact that blood insulin concentration is measured after a period of fasting, to control for the fact that insulin levels fluctuate depending on how recently you’ve eaten, and what kind of food you ate. Therefore, that caveat doesn’t leave room for Pinker’s claim that insulin concentration increases to a thousand times the nonpregnant level. Nor do “the increases during the third trimester” rescue the claim. An increase “parallel with the growth of the fetus” would not translate into an increase of “a thousand times the usual concentration”, regardless of how you measure fetal growth. According to this chart of average fetal growth, during the third trimester (week 27 until birth) a fetus’s weight increases by 4.24 times (1.93lb to 8.19lb) and its length increases 1.4 times (14.41 inches to 20.20 inches). With this in mind, I am comfortable asserting that Pinker’s claim that the concentration of insulin increase a thousand times its usual level is false.

Regarding hPL concentrations, Haig writes:

“As discussed in a previous section, the manipulation of maternal responses by allocrine hormones is predicted to favor an evolutionary escalation in which increased maternal resistance to manipulation favors increased placental production of the hormone. hPL (and to a lesser extent hPGH) matches these predictions. hPL is the most abundant peptide hormone produced by primates. Its concentration in maternal serum increases through- out pregnancy, reaching 5 to 15 µg/ml near term, at which stage syncytiotrophoblast secretes 1 to 3g/day. Levels of hPGH are much lower, but follow a similar temporal pattern, exceeding 15 ng/ml near term. Concentrations of hPL and hPGH in fetal serum are much lower than concentrations in maternal serum (Frankenne et al., 1988; Eriksson et al., 1989; Walker et al., 1991). For comparison, the plasma concentration of hGH, integrated over a day, is about 3 to 6 ng/ml in young nonpregnant adults (Daughaday,1989).” (Haig 511, my bold)

In this section, Haig compares the concentrations of three different hormones: hPL, hPGH (human placental growth hormone), and hGH (human growth hormone). Unlike the conflation of insulin and hPL that I criticized earlier, the comparison among these three hormones is entirely sensible given that they are in the same gene family, the result of duplications from an ancestral growth hormone gene. Moreover, they function is analogous ways. This is important because Pinker’s mysterious claim might have been due to choosing the concentrations of hPGH or hGH as the “usual concentration” of hPL. Since the peak hPL concentration mentioned in the above section (5-15 µg/ml) is approximately a thousand times the ones mentioned for hPGH (5-15 ng/ml) or hGH (3-6 ng/ml), it is quite likely that this was the root of Pinker’s mistake. Instead of suggesting that the concentration of hPL increased a thousand times its usual concentration, he should have said that the concentration of hPL reached levels a thousand times greater than any other related hormones.

Distressingly, this passage from “Genetic conflicts in human pregnancy” also has a confounding sentence in it that must be addressed before Pinker’s claim about the thousand-fold increase in hPL can be conclusively falsified. The sentence, “hPL is the most abundant peptide hormone produced by primates” seems to contradict a fact I presented earlier when I challenged the notion of a “usual concentration” of hPL. Specifically, I claimed that hPL is produced only by the fetus and is absent the nonpregnant state. Given that every other source I had consulted mentioned hPL only in the context of pregnancy (the Wikipedia entry for hPL explicitly asserts that “hPL is present only during pregnancy”), I was perplexed. I did some further research on the issue of whether hPL is present in nonpregnant adults. I tried to access the sources Haig cited in this section, but they were not available through my institution. However, I found a review of hPL function from a 1969 issue of the New England Journal of Medicine that not only clarified the confusing sentence in Haig’s paper but also further supports my diagnosis of Pinker’s error:

“[hPL] concentration in maternal serum, as measured by immunoassay in late pregnancy, reaches levels higher than any other known protein hormone at any time and may be 1000 times that of growth hormone.” (Saxena et al., 225)

In light of this, Haig’s ambiguous claim that “hPL is the most abundant peptide hormone produced by primates” is properly understood as “hPL is the most abundant peptide hormone produced by primates [but is present only during pregnancy].” My suspicion that Pinker compared the hPL concentration to that of related growth hormones is further reinforced by the corresponding mention of a “1000 times” increase.

In conclusion, Pinker seems to have misunderstood a detail he read in his source, and this incomplete understanding translated into the uncharacteristically sloppy and ambiguous sentence that motivated me to investigate the claim in the first place.

P.S. While I was doing research about insulin changes during pregnancy for this post, I came across a study that investigated the question in goats. Apparently, the concentration of insulin varies very little during a goat’s pregnancy. That being said, check out this awesome video of stampeding baby goats. (h/t Donald Huang)


Haig, D. 1993. Genetic conflicts in human pregnancy. Quarterly Review of Biology, 68, 495–532.

Pinker, Steven. How the Mind Works. (1997). Norton

Badri N. Saxena, M.D., Kendall Emerson, Jr., M.D., and Herbert A. Selenkow, M.D. N Engl J Med 1969; 281:225-231July 31, 1969DOI: 10.1056/NEJM196907312810501