Shortly after the publication of Angela Saini’s book Superior: The Return of Race Science, Mr. A wrote a scathing Amazon review of the book. Since I had found Saini’s analysis to be carefully researched and documented, I posted a strong criticism of Mr A’s claims. Mr. B and Mr. A then attacked several specifics of my criticisms, but as the conversation continued Mr. A seemed to consider my arguments more seriously and to attempt to formulate serious counterarguments, leading to a deeper exploration by both of us of the scientific issues. It should be kept in mind that the issues discussed in this deeper exploration concern genetic differences in IQ with regard to individuals but not with regard to racial groups, where hereditarian and racialist arguments encounter whole arrays of additional problems. (These additional problems are discussed in more detail in the next two conversations with Mr. C and Mr. D, as well as in the book reviews and other pages of this website.) As I note at the bottom of this page, my conversation Mr. A ended somewhat mysteriously, possibly because Mr. A was having second thoughts about some of his own arguments.
Mr. A’s initial review:
Mr. A’s review of Saini’s book was quite lengthy, but the core of his argument was this: He described Saini as a lay person who claimed to know more about science than the professional scientists themselves and who wrote in a popular tradition of ignoring “large swaths of biological research.” In contrast, Mr. A argued for the genetic determination of intelligence, citing several well known hereditarian authors (e.g. Robert Plomin and Richard Haier) and several large-scale studies of the human genome that had identified numerous genes associated with cognitive performance using educational attainment as a proxy for intelligence (e.g. Lee et al. (2018)–see below). Moreover, he pointed out that some of these genes had been associated with pathways in the brain identified with higher cognitive functioning (Okbay, et al. (2016)–see below). He went on to criticize Saini as seeming to have a political agenda and added his belief that Saini had a “dangerous” anti-scientific attitude in the pursuit of that agenda. Our conversation then proceeded as follows:
My response to Mr. A’s review:
Mr. A, you condemn Saini as a journalist who criticizes experts, but you fail to note that many of the problems she describes were identified by other experts in the same fields. You yourself cite researchers to support your position but consistently omit those who disagree with you (e.g., Lewontin, Flynn, Rutter, Kaufman and Cooper). So if you wish to condemn Saini as a lay person who fails to cite prominent researchers, then you must condemn yourself for doing the same thing. Your criticisms of Saini are frequently shaky. For example, you discuss studies of educational attainment and state that “educational attainment is often used as a proxy for intelligence” in GWAS [genome-wide association] studies. Yet you ignore the fact that the very research group you cite specifically warns against this interpretation and emphasizes that the method they use is particularly likely to lead to spurious genetic inferences in relation to ethnic and racial minority groups (see their FAQ page)—this is precisely the kind of problem that Saini is talking about. You assert that the genetic determination of intelligence is a “settled” and “uncontroversial” fact, like climate change. But your argument here confuses “genetics” and “heritability” (a mistake that Saini does not make). The former is a biological concept referring to causation, the latter a statistical function of both genes and environment that has no clear biological meaning. What heritability estimates actually tell us about genetics is limited and gives little support to hereditarians. Nicholas Mackintosh, perhaps the most respected intelligence researcher in the UK, reviewed the entire literature on the heritability of intelligence and noted that heritability fluctuates substantially across populations, that the environment mediates genetic effects in ways that are largely unknown, and that genes that may be linked to intelligence appear to have “only a very small effect” (IQ and Human Intelligence, p. 289). Thus, inferences from heritability to genetics always entail concealed and subjective working assumptions by those who make them, typically minimizing or disregarding environmental factors and interactions between genes and/or the environment. For these and related reasons, Turkheimer, Ceci, Richardson, and other researchers have pointed out that attempts to quantify genetic variance are highly questionable and possibly meaningless. You defend the rights of researchers to publish their findings without “suppression.” Fair enough, but this does not give them immunity from authors like Saini who expose the flaws, limitations, and misapplications so often associated with their work.
Mr. B’s response in defense of Mr. A:
Your comment calls for some remarks.
First, it is not sufficient to accuse Mr. A of “consistently” omitting those who disagree with him. You also need to demonstrate that the researchers you mention have made points undermining Mr. A’s position. Instead of doing this, you seem to be engaged in a game of “who has more famous people on his side.” You would be more convincing if you tried to make an argument of your own.
As for the GWAS, educational attainment is often used as a proxy for intelligence because they strongly correlate with each other (see Deary et al. [2007], ‘Intelligence and educational achievement’, Intelligence 35[1], 13-21), and because there are far more data for educational achievement, which means that one can get much bigger sample size by using them (a recent example of such GWAS is : Willoughby et al. [2019], ‘The role of parental genotype in predicting offspring years of education : evidence for genetic nurture’, Molecular Psychiatry, 23 August 2019).
In addition, the following study indicates that variants associated with educational attainment are also associated with cognitive health: Rietveld et al. (2014), ‘Common genetic variants associated with cognitive performance identified using the proxy-phenotype method’, PNAS 111(38), 13790-13794. This is also mentioned in the FAQ of the Social Science Genetic Association Consortium: “We also found that many of the genetic variants that affect educational attainment also affect cognitive performance and math abilities.”
Also, it is unjust to accuse Mr. A of confusing “genetics” and “heritability.” Unlike you, he understands that “heritability” implies the existence of genetic factors even if such factors are still unidentified. It is erroneous to say that heritability “has no clear biological meaning”: heritability refers to the proportion of variation in a trait that can be explained by genetic differences between individuals. It is true that “heritability” does not tell us about what happens at the molecular level, but it need not do so. One can assess the proportion of variation that is due to genetic factors (from quantitative genetics) without precisely knowing which genes are operating (from molecular genetics). The failure to differentiate between both tasks is a permanent source of confusion.
Finally, although Mackintosh was indeed a prominent intelligence researcher, the second edition of his book was published in 2011 (and the original edition in 1998). The field has evolved quite a bit since then, with polygenic scores now enabling researchers to predict part of the variance in phenotypic intelligence and other traits.
Mr. A’s comment that additionally followed:
To add to what Mr. B has said, it’s worth mentioning that I have nothing against researchers such as James Flynn and Jelte Wicherts, who argue that differences between group averages in IQ scores are environmentally caused, while fully accepting the mainstream view within psychology that general intelligence is a measurable trait with a partially genetic basis. (Flynn makes it clear in the second chapter of his 2007 book “What is Intelligence?” that he accepts the existence of large genetic influences on IQ variance between individuals, and I’m assuming you’re familiar with the reason high within-group heritability for a trait does not require between-group heritability to be equally high, if group differences are caused by an X-factor.) What I’m observing, though, is that the style of argument made by Flynn and Wicherts is increasingly going out of fashion. Nowadays, all of the most active opponents of hereditarianism are taking the approach of trying to discredit basic concepts in the fields of intelligence research and behavioral genetics. Angela Saini, Gavin Evans, and your own comments here all are examples of this new approach.
One possible outcome of this approach is that you’ll succeed at what you’re trying to do, and discredit not only the hereditarian hypothesis about group differences, but the overall fields of intelligence research and behavioral genetics. However, there’s also another possible outcome. If research about the genetics of intelligence survives, and polygenic scores continue predicting more and more of the variance in IQ and educational achievement (see last year’s literature review from Nature Reviews Genetics, “The new genetics of intelligence” [by Robert Plomin]), an increasing number of people are eventually going to realize that what they’re being told about the non-validity and non-heritability of IQ doesn’t match up with reality. And if that is the eventual outcome, I think that people like you, Saini and Evans are going to regret having redefined opposition to race research so that it’s based on rejecting these tenets of well-established fields.
The vast majority of intelligence researchers and behavioral genetics have no interest in studying group differences, and up to this point most of them have stayed out of the debates over that topic. But there is an increasing awareness among them that the more recent, broader attacks of their fields can’t be safely ignored, and that some type of defense will be necessary. . . The next few years should be very interesting, because this will be the period of time that determines whether anti-racist authors succeed at taking down these fields, or whether their effort to do so will end up backfiring.
My responses to Mr. B and Mr. A:
Dear Mr. B: You seem to have missed the point. My list of “famous people” (i.e. respected researchers) was a response to a sweeping distortion by Mr. A: the portrayal of Saini as a lay person without expert support. If you want the citations you can find them in Saini’s book and [here]. But what’s more revealing is the way you responded to the specific sources I did provide. You ignored the substantive points and attempted to deflect. So when I noted that Mr A’s own source had contradicted the very claim he was making about proxies, you responded by citing a different source reporting a different correlation (between intelligence and achievement–not EA [educational attainment]). And please note that even when treated as proxies, correlations are still associational, not causal. In fact, confusions about the nature of correlation underlie most misunderstandings about the relationship between genes and intelligence, the classic case being the noncongruence of heritability and genetics. This is the core of Macintosh’s critique and is why it still holds. The correlation-causation fallacy has not disappeared since 2011! This fallacy permeates not only the arguments of “race realists,” but also the arguments of those who overestimate genetic influence on the basis of GWAS, GCTA [genome-wide complex trait analysis], and PGS [polygenic score] studies. Your comments about both Rietveld and Willoughby fall into this category. In Rietveld’s study, genes, EA and cognitive performance are correlated, but the macro- and microenvironmental context which also shapes the phenotype, both interactively and independently, is not addressed in this study, nor, indeed, in most of this literature—your comment that there is “no need” to find the genetic pathways notwithstanding. And in Willoughby’s study, the so-called “genetic nurture” literature does identify some effects of “nurture” but the inclusion of the word “genetic” in the title is gratuitous since the actual interplay of genetic, neural, psychological, social and cultural influences on parenting behaviors is not addressed in this study, and is almost completely unknown. These are not minor details; they are the whole issue—and they have plagued behavioral geneticists for nearly 100 years. A number of these researchers have attempted to disentangle correlation and causation by using statistical controls. But Kaufman has analyzed why this strategy is both inaccurate and unavoidably incomplete (see Saini p. 190, and the more extensive treatment in “Statistics, adjusted statistics, and maladjusted statistics,” Am. J. Law & Medicine, 2017, 43, 191-208). Likewise, Turkheimer has analyzed how these inherently weak statistical controls (technically categorized as quasi-experiments–look up the word “quasi”!) are often confused with the strong controls of true experiments (RCTs). Unlike the latter, which establish causation, the former never rise above the level of predictive association (see, for example, “Commentary: variation and causation in the environment and genome,” International J of Epidemiology, 2011, 40, 598-601; also see “Applied ethics of polygenic scores,” Aug 22, 2019, Genetics and Human Agency website). Not surprisingly, the predictions they make are correspondingly weak. It should concern you that unlike work in other sciences, the predictions in this field have gotten weaker rather than stronger as the field has progressed. Failing to acknowledge these problems does nothing to advance science.
Dear Mr. A: Let me take your responses to my comment one paragraph at a time.
First paragraph: I agree with much of what you say, with two exceptions: (1) Your claim that genetic influence is “large” is problematic. In reality the “size” of this influence is indeterminate because, as Flynn notes in the same chapter you cite (p. 11), genetic influence is inextricably connected to environmental influence. There is a whole literature on how genetic variance drops substantially when environmental variance is accurately assessed, sometimes approaching zero (see Turkheimer, 2003, “Socioeconomic status modifies heritability of IQ in young children,” Psychol Sci, 14, 623-8). (2) The analyses by Saini, me, and others are hardly a “new approach.” I would simply call them “thinking critically.”
Second paragraph: I don’t have a problem with hereditarian interpretations, except when they become ideological and/or overinterpret the evidence (which happens quite a bit in this literature). Plomin is a good example. He strikes me as sincere, but his hereditarian blinders seem to skew his judgment. Jay Joseph has catalogued Plomin’s 40-year history of overstating progress in the field of genetics (see The Trouble with Twin Studies, 2015, pp 223-234). As noted above, predicted genetic variance has dropped significantly in recent years due to the “missing heritability” problem and to studies like the one by Turkheimer mentioned above. Whether future findings will reverse this trend remains an open question.
Third paragraph: Yes, many intelligence researchers are well-intentioned non-racists and they have a right to defend their research. But they need to address misrepresentations of their work not only by anti-racists but also by racists. With one or two exceptions, your criticisms of Saini’s book seem fervently and disproportionately focused on the former with little recognition of the seriousness of the latter.
Response by Mr. A:
You’re incorrect that the source I was citing contradicts what I said about educational attainment acting as a proxy for intelligence. The paper that I was citing (Okbay et al.) says, “we find genetic covariance between increased educational attainment and increased cognitive performance”. The term “genetic covariance” specifically refers to a shared genetic cause. You may disagree with the paper about this, but it is not misrepresenting the paper to say that this is what it concluded.
In your earlier comment, what you said wasn’t that my point was contradicted by the source I was citing; you said to “see their FAQ page”, by which I assume you mean the FAQ for the Social Science Genetic Association Consortium [https://www.thessgac.org/faqs]. Question 2.8 in their FAQ is, “Are the genetic variants associated with higher educational attainment in your study also associated with other outcomes?” And the answer is, “Yes. We found that genetic variants associated with increased educational attainment are negatively associated with grade retention (i.e., having to repeat a grade) and positively associated with grade point average (GPA), cognitive performance, and self-reported math ability. This suggests that genetic variants ‘predict’ (see FAQ 1.4) educational attainment at least in part through their correlation with cognitive development and academic performance.” In case it’s unclear what “cognitive performance” means in this context, the FAQ states that cognitive performance was “measured by scores on cognitive tests” (that is, IQ or similar tests). And the FAQ also is explicit that it is making a statement about causality: “We also found that many of the genetic variants that affect educational attainment also affect cognitive performance and math abilities.”
To summarize, the paper I was citing says that there are shared genetic causes between educational attainment and cognitive ability, and the research group’s FAQ states that these genetic variants are associated with educational attainment in part because of their correlation with cognitive performance, as well as that these variants are not merely correlated with these life outcomes, but actually have an effect on them. Rather than “specifically warning against this interpretation” (your words), both the paper itself and the research group’s FAQ are consistent with it. The Willoughby et al. paper is even more direct on this point, and actually uses the word “proxy”: “A polygenic score (PGS) constructed from this large sample is now able to predict 11–13% of the variance in educational attainment and 7–10% of the variance in cognitive ability, exceeding all previous benchmarks, and providing further evidence that educational attainment is a viable proxy for cognitive ability.”
Your response to Mr. B contains a mixture of criticism of traditional twin studies and of GWAS / GCTA studies, but it doesn’t work to mix and match criticism between these two types of study the way you’re doing, because these studies use different approaches. The distinction between correlation and causation is a popular criticism of twin studies, but in studies about the relation between genetic variants and life outcomes the causality can only point in one direction, because our life experiences don’t alter what’s in our genomes. (This point is mentioned in the Nature Reviews Genetics article.) It’s true that with many of these genetic variants, we don’t know the mechanism by which they affect these outcomes, and in some cases the mechanism may be an indirect one, as described in the Willoughby paper. But even that criticism does not apply in all cases, because in many cases we also know the developmental pathways by which the genetic variants affect brain structure. This is discussed in the Okbay et al. paper, and discussed in greater detail in the supplementary information of Lee et al. 2018 (“Gene discovery and polygenic prediction from a genome-wide association study of educational attainment in 1.1 million individuals”, published in volume 50 issue 8 of Nature Genetics). The Lee et al. supplement includes an in-depth discussion of the known biological mechanisms by which these genetic variants, known as “causal SNPs”, affect brain structure and educational attainment. I’ve mentioned this line of data several times, but you’ve never acknowledged it in any of your comments, maybe because the critics of behavioral genetics that you like to cite generally don’t acknowledge it either.
Obviously I agree with you that when research is misrepresented by racists, that deserves to be pointed out also, but there is an important difference here. There is an entire genre of books devoted to debunking race research, so that researchers who present arguments for the hereditarian hypothesis about group differences–or even those who just take it seriously as a scientific hypothesis, without directly supporting it–inevitably face a large amount of criticism and scrutiny. On the other hand, with a few exceptions, research and ideas that are politically convenient are not scrutinized as closely. The uneven standards to which these ideas are held are discussed at length in pages 201-208 of Nevan Sesardic’s book Making Sense of Heritability, in the section titled “double standards”. The most extreme example given by Sesardic is the experience of the behavioral geneticist Sandra Scarr, who is quoted describing how she was warned by student activists to “get out of town, or you will be killed”, and how she eventually made plans to flee the United States if one of her studies produced a politically inconvenient result. Looking at the current year, compare the reception of Saini’s book outside of Amazon to the fate of James Flynn’s book In Defense of Free Speech (which takes the hereditarian hypothesis seriously, but argues against it), or to Noah Carl’s experience at Cambridge. Books by anti-racist authors such as Saini generally are positively received regardless of their quality, but in our present society, can anyone seriously doubt that if an author misrepresents research in a racist manner, they’ll be held to account?
The reason I think it’s important to point out the misrepresentations made by authors such as Saini is not that I think this is inherently more important than pointing out misrepresentations made by racists, but that individuals who are willing to present criticism of anti-racists are in much shorter supply. I feel that I’m the wrong person to have this job in Saini’s case, and I was not expecting my review here to become this book’s top-rated review, but it’s better for me to do this than for it to not be done at all.
My response to Mr. A:
Dear Mr. A: Critiques of anti-racism are in “short supply”?? Anyone who has even glanced at the internet knows that discussions of race are dominated by an online community whose members rush to support anything that they interpret as affirming nonwhite inferiority. This alone should give you pause for thought and may help to account for the tsunami of upvotes your review received, which seems to have surprised even you. But in your account, the anti-racist community is the bigger threat because its members are bullying courageous researchers who are pursuing the truth. The problem with this argument is that the facts don’t support it. Critiquing research is not bullying, and the examples you give of actual bullying involve, almost without exception, people from outside the research community. That bullying is a symptom of increasing anger and polarization across the entire political spectrum in the U.S.—and it is more characteristic of the right than of the left. For example, ideological suppression is directed more often toward left-leaning than right-leaning academics, and white supremacists are responsible for the great majority of extremist murders in the U.S. (see Jeffrey Adam Sachs, “There is no campus free speech crisis,” Niskanen Center, 4/27/18, and ADL Annual Report, Jan 2019; see also Coleen Flaherty, “Old criticisms, New Threats,” Inside Higher Education, 6/26/17). Within the research community, “race science” is not taken seriously because it has not made its case. However, one would never know this from the distortions of genetic research by “race science” advocates in the public sphere. The same forces that promote racist bullying in the public sphere have also promoted a whole cottage industry of pseudo-intellectuals to support their ideology. These commentators draw unrepresentatively from the scientific literature, misinterpret results, cite seriously deficient studies, and fetishize non-peer reviewed popular sources. This has led a number of genetics researchers to express concern about the distortion of actual science by racists (Statement by American Society of Human Genetics, 10/19/18; see also the interviews with other genetics researchers in the NYT by Amy Harmon, “Why White Supremacists are Chugging Milk and Why Geneticists are Alarmed,” 10/17/18).
The exaggeration of hereditarian findings feeds this dynamic in a very troubling way. Saini’s book gives many examples of how distortions of science (repeat—distortions, not actual scientific findings) have been exploited to justify discrimination, abuse, and violence against racially marginalized groups. “Anti-racists,” as you call them, are simply no longer willing to let this go unchallenged.
Regarding your other points, I’m sure you know that Okbay and the other members of the SSGAC work cooperatively and interchangeably. So the “source” I referred to is that group. The “warning” I referred to (FAQ page [https://www.thessgac.org/faqs], second bullet point near bottom [What is Cognitive Persormance?”]) is from Mr. B’s friend Rietveld (evidently), who explains why he and the other members of the group use the term “cognitive performance” (a behavior) rather than “intelligence” (an ability)–specifically, because “it does not prejudge the sources or causes of differences among people, which could be explained by genetic factors, environmental factors, their interactions, or all of these.” My point exactly. And this is very relevant to the problems with your argument. Your comment (rightly) invokes “covariance,” “association,” “correlation,” “consistency with,” and “prediction” in describing the findings of these studies, but then ties itself in knots trying (wrongly) to make compelling inferences to specific genetic causes. The critiques of correlation and causation are “popular” (to use your term) because subtle but serious confusion about these concepts permeates the understanding of both twin and genomic studies, not only by the public but also by some researchers, particularly those with strong hereditarian leanings.
Allow me to elaborate. There are three different kinds of terms that are used to describe the findings of genetic research. One set of terms is correlational (the ones I commend you for, above). A second set is ambiguous—the terms in this group can refer either to correlation or to causation, depending on the context (this includes terms like “explains,” which can refer to either causal explanation or variance explained, and “effect,” which can refer either to a causal effect or an effect size). And a third set of terms refers only to causation—e.g. “causes,” “affects,” “produces,” “influences,” etc. People who discuss this research often use terms from all three of these categories, but don’t always distinguish them adequately, and hence may slip into over-interpretation. Furthermore, the third set of terms has another ambiguity, illustrated by your interpretation of the quote by Lee: “We also found that many of the genetic variants that affect educational attainment also affect cognitive performance and [self-reported] math abilities.” There are two possible senses of the word “affect” here. The first is that the genetic variants have some effects, large or small, directly or indirectly, on the dependent variable, though we don’t know what they are. But this interpretation is essentially empty, since all behaviors have some connection with genes, and the fourth law of behavior genetics indicates that the determination of complex behaviors involves so many genes with such small effects that no gene or genes can be reasonably ruled out—so we pretty much knew this anyway. The second sense is that the genetic variants have an identifiable causal pathway that is sufficiently specifiable and independent of interaction that it can reasonably be called a cause-and-effect relationship, as causation is ordinarily understood in science (i.e., nomological causation). Of course, there are probably going to be some interaction effects too—but to the degree that these interactions are multiple and poorly understood, then the meaning of the word “affect” begins to slide back into the first (vacant) sense. The problem for you and other hereditarians is that we know almost nothing about effects in the second sense, though discussions in the literature frequently imply that we do, especially when compounded by conflation with the first sense, and by slippage between the three types of correlational/causal terms. The Lee group is actually pretty good about keeping such distinctions clear (but see my bracketed correction to their quote above). However, those who discuss their work routinely fall into these traps. To give just one example, the Willoughby quote you gave in your most recent comment misstates Lee’s findings in terms of “cognitive ability” rather than “cognitive performance,” and in so doing makes an unwarranted inferential leap from correlation to causation, exactly what Rietveld cautions against.
Now, what about Plomin’s claim that we do know, at least, the direction of causation from genes to traits? Plomin is wrong here, and this is what I mean by his tendency to overstate things. The alleged causal factor is not the material structure of the gene but the function of the gene. It is well-known that gene function can be affected by environment in a number of different ways, and advancing genetic research has uncovered more and more interactions of these and other kinds. So the influence can run in both directions, and horizontally as well, all along the biochemical and organismic pathways. This is also relevant to your comments about Lee’s supplement. Yes, the authors cite a number of interesting, even fascinating, findings suggestive of possible causal pathways. But the most direct evidence they have comes from laboratory studies, many on nonhuman subjects, and many involving neural correlates, rather than established causes and effects (the term “correlates” is standard terminology in neuroscience and is not accidental: we don’t know the direction of causation for most neural events; just as genes affect the brain, so, too, does environment, often profoundly). The best we can say is that some genetic-biochemical links may be encompassed by some of this research. Most of these findings involve one or a small number of genes, without knowledge of how other genes affect the same link, or the ways that these may interact with each other, or with other biological processes, or how this entire system is altered by environmental factors. Furthermore, even if all this were nailed down, it only gets us to brain tissue and a few basic processes—not to brain organization, and the way that human behavior is mediated by that organization, and, again, it says nothing at all about how the environment acts on that organization and that behavior. In short, we know very little about causation in the second (and important) sense for biological processes, and even less about how these processes interact with the social and cultural context. As I mentioned before, this is not some set of minor details. It lies at the heart of all the fallacies of hereditarianism, which systematically overestimate genetic influences and systematically underestimate or ignore environmental ones. This is especially true for behavioral traits, not only because of their complexity and their sensitivity to the environment, but also because we have a great deal of evidence for the power of social conditions in shaping and limiting behavior. Unlike the claims of hereditarians, the evidence supporting environmental influence is well-established and replicable, not only by correlational and quasi-experimental research, but also by true experiments using randomized controlled trials and therefore permitting causal inference. Since the social conditions of racial groups differ from each other in well-known ways, often dramatically so, these facts are seriously delegitimizing to racists who fight to maintain beliefs in genetically determined stereotypes of racial inferiority and who eagerly seek out pseudo-intellectual sources peddling hereditarian overreach to aid them in this project. Hence, the critiques
******************************
NOTE: Mr. A never responded to my final comment, above. A few months later, I discovered that the review that initiated this conversation had disappeared. While it is possible that Amazon removed the review, there was nothing in the review that would have merited removal (Mr. A’s arguments were hereditarian but they were neither racist nor offensive). It seems likely, therefore, that Mr. A chose to withdraw the review himself. Perhaps he reconsidered his position, but this can only be speculation.
Psyche and Sense 