On Sciences and Humanities: Reflections on Coyne and Konnikova

This post is seemingly off-topic on this blog. In reality, it goes to the core of the problems I’m trying to address. It just takes a philosophical and methodological road to get there.

On Scientific American’s Literally Psyched blog, Maria Konnikova, a graduate student in Psychology at Columbia University and a fiction writer, has a piece with a somewhat sophomoric title “Humanities aren’t a science. Stop treating them like one.” I read it with a smile but quite sympathetically. (And not simply because Anna and I are Russians by origin and therefore have a knack for melodramas.) As an anthropologist, I’m used to reading this kind of prose and believe that it reflects some real controversies in the way knowledge is constructed in Western societies. Konnikova’s text is not particularly sophisticated, does not rely on too many case studies and is not well-referenced. But it’s very well-written. Her main thesis is most dispassionately communicated in the very last paragraph:

“The tools of mathematical and statistical and scientific analysis are invaluable. But their quantifiable certainty is all too easy to see as the only “real” way of doing things when really, it is but one tool and one approach—and not one that is translatable or applicable to all matters of qualitative phenomena.”

At the end of the day an aspiring writer and a psychologist quite appropriately writes for the Literally Psyched blog.

But then a rather odd piece followed from Jerry Coyne. Jerry Coyne is Professor in the Department of Ecology and Evolution at the University of Chicago and a member of both the Committee on Genetics and the Committee on Evolutionary Biology. He’s the author of a book with a corny title Why Evolution Is True and a rather childish cover design featuring dinosaurs. His unexpected response to Konnikova came from his blog with the same corny title that he uses to sell his book. A fearsome crusader against creationists, Coyne apparently has enough time on his hands to police the unruly thoughts of a Literally Psyched blogger. The essence of his rebuttal is well captured in the last paragraph:

“It may be uncharitable of me, but I suspect Konnikova’s trying to tout the humanities and social sciences as “other ways of knowing.” But that won’t work. As the social sciences and humanities mature, they come to realize that their criteria for finding truth are the same as those used in biology and physics. Indeed, they even become more mathematical.  It may be harder to suss out what’s true about humans than about, say, ants, but that reflects our more complex culture, not different ways of knowing about different organisms.”

Just like Francis of Assisi imitated Christ, Coyne walks in Charles Darwin’s footsteps by trying to understand “the origin of species: the evolutionary process that produces discrete groups in nature.” And he professes a similar belief to “Jesus is the only way.” Anthropologists have long realized that whether science, religion or nationalism, these domains of culture are constructed in similar ways and, although different on the surface, they may converge on the level of deep affinities. Just like for Christians or Moslems, there’s only one God, for Coyne there’s only one way of knowing and it’s Science.

But every now and then you do expect some level of reflexivity from a scientist that will set science apart from religious faith. Not in Coyne’s case. He acknowledges that “he doesn’t know anything about” folkloristics or forensic linguistics (the examples Konnikova uses to show the failures of quantitative methodologies). Coyne apparently feels much more confident about other branches of linguistics when he says that “linguists do indeed use quantitative analysis, and reconstruct the history of languages in ways similar to those used by biologists to reconstruct the history of life.” To his disappointment, if what he has in mind are Quentin Atkinson and Russell Gray’s researches, these scholars are psychologists applying phylogenetic methods to language data. They do publish their models in such “scientific” journals as Nature, Science and Proceedings of the Royal Society of London (the journals that Coyne reads), but not because they arrived at a superior knowledge of Indo-European languages by applying superior methods but because such field-specific venues as The Journal of Indo-European Studies (which Coyne does not read) do not consider their methods appropriate for the linguistic material and their knowledge of Indo-European language data deep enough. Linguists have criticized them on multiple occasions. Scholars who apply formal methods to language evolution and who do get published in The Journal of Indo-European Studies prefer other quantitative methods that have nothing to do with biological systematics.

Nothing indicates that Coyne knows anything about psychology or literature either. His self-description circumscribes his level of expertise pretty narrowly (from an overall sciences and humanities viewpoint):

“[H]e uses a variety of genetic analyses to locate and identify the genes that produce reproductive barriers between distinct species of the fruit fly Drosophila: barriers like hybrid sterility, ecological differentiation, and mate discrimination. Through finding patterns in the location and action of such genes, he hopes to work out the evolutionary processes that originally produced genetic change, and to determine whether different pairs of species may show similar genetic patterns, implying similar routes to speciation.”

But he nevertheless has a very strong point of view on what Konnikova wrote. And again her piece was published on the Literally Psyched blog, not on a Drosophila blog. What Coyne endorses and Konnikova implicitly (she could have been more specific at that) rejects is the sweeping application of field-neutral (with some remote ancestry in biology) “scientific” methodologies, which usually come in the form of statistical models because quantitative methods are easier to scale across disciplines than the more field-specific qualitative methods, without a proper regard for the nature of data in psychology, history, sociology, anthropology, political science, etc. The most immediate example that comes to my mind is the phenomenon of the brilliant mathematician, Nick Patterson. He began as a cryptographer for the national defense departments in the U.K. and the U.S., then worked for many years on Wall Street predicting markets and as of late is Senior Computation Biologist in Medical and Population Genetics at Harvard working with David Reich on decoding Neandertal and other archaic as well as modern genomes. This is how scalable quantitative methods and skills are! Now it’s up to Patterson to decide whether we admixed with Neandertals or not. But Patterson is a horizontal, not a vertical expert. He’s not an archaeologist, anthropologist or linguist. How relevant are his statistical methods to what is essentially a problem of history? How reliable are his estimates and interpretations? As reliable as were market growth projections in the early 2008?

Coyne also takes for granted that social sciences have no other way to analyze their material than by applying a scientific test:

“Well, how does one go about finding out whether self-reported experience is influenced by culture, class, experience, and a particular research situation? You do a scientific test!”

Coyne does not realize that he is not just misapplying the scientific method but actually advising people to begin wasting their time. There is no need to apply a scientific test to determine whether self-reported experience is influenced by culture, etc. It unquestionably and absolutely is. That culture, etc. influences self-reported experience is truer than evolution. Some unfortunate scientists would have saved themselves and their readers a lot of time if they looked for the causes of high IQ in modern Ashkenazi Jewish individuals in the long-term tradition of Talmudic scholarship, not in Ashkenazi Jewish genes. But since “scientists” tend to think of themselves as somehow free from cultural constraints and scientific knowledge as independent of the deictic source of its production, they assume that everybody else now has to prove, using scientific methods, that culture matters.

It is this endorsement by Coyne of Science detached from its precise field of practice plus his distorted reality field in which a social minority believes that it somehow transcends the rules of society and can substitute its own “laws of nature” for these rules, that led to the writing of a self-infatuated and didactic blog post by a person who has no necessary reason, experience or expertise to blog about psychology or the humanities.

Coyne’s piece attracted 100 comments. Like it usually happens on “science blogs,” swarms of unidentified “scientists” displayed allegiance to Coyne’s God of Science and launched verbal attacks at Konnikova. However, there was one rare defender of Konnikova who corroborated Konnikova’s examples with her own:

“The example this brings to mind for me, in the field of psychology, is working with autistic children. I often see a clash between therapists with great form and those with great content. Therapists with great form are often fond of selecting an extraordinarily measurable goal, like having a child say “blue” when a blue card is held up. Data is collected. Charts are made. Percentages, pre and post teaching accuracy, and rate of acquisition is recorded with care. Usually a big fuss is made over the superiority of being “data driven”. Three months later, there is, from a data driven perspective, an absolutely beautiful example of progress to behold – on paper. In reality, the kid can say “blue” when you hold up one, very specific, blue card.”

A real scientist would start seeing a pattern behind the experience of Konnikova and the experience of one of Coyne’s readers. But, for some reason, the great defender of Science in front of creationists, Jerry Coyne, fails to see this pattern. Most likely this is because it’s not his field of study. Autistic children are not Drosophila. But it never occurs to him that he has stepped way outside of his area of expertise.

Outside of a few supporters of Konnikova, Coyne and his commenters seem to have converged on three points 1) criteria for truth and the fundamental scientific method of ascertaining facts are the same across sciences, social sciences and humanities; 2) quantitative methods are not synonymous with or exhaustive of the scientific method (Darwin and Einstein formulated their theories without doing calculations); 3) there are no better ways of knowing than science, and that humanities and social sciences will eventually mature to become more precise, measurable and replicable; 4) Konnikova’s position is disguised postmodernism and as such is a pesky but temporary roadblock on the way to the complete triumph of science.

Now, I’m fine with points 1-2. I can’t predict the future, so I’m not sure about point 3, but I suspect that different fields of study will have their own field-specific quantitative repertoires. And by becoming field-specific, these quantitative repertoires will become qualitative at the core. But I would definitely supplement point 4 with the following one: reflective, critical, post-modernist, iconoclastic, interpretative and other similar approaches are consistent with points 1-2 and represent a necessary step toward the progress of science. Going back to the questionable point 3, science will not progress if it’s devoid of what is listed in my point 4. My issue with post-modernism (and let’s label Konnikova “neo-post-modernist” for the sake of the argument) has always been not its reflexivity, which I fully embrace, but its inability to practice what it preaches. Postmodernism takes what is essentially a scientific stance out of “circulation” and fetishizes it. Once a postmodernist sees its own reflexivity in action, it recoils from it because now it looks too much like science. Notably, Konnikova in her piece did not present a single example of what constitutes a positive “advance in understanding” in the humanities, which at the same time is not “science.” All of her examples were of misapplications of the scientific method in history and linguistics. The mutual animosity that exists between “scientists” and “postmodernists” derives from the inability of scientists to abstract themselves from their day-to-day scientific routine of “knowing the world better than anything else can” and from the inability of postmodernists to roll up their sleeves and actually prove that the world is socially constructed. Scientists like Coyne remind me of Don Quixote who is busy following an ideal role model (Amadis de Gaula in the case of Don Quixote, Charles Darwin in the case of Coyne) in an effort to remake the world, which ends up being a product of his imagination. Postmodernists are, on the other hand, more like Hamlet who endlessly delays avenging his father’s death.

Update: 08.17.12. I just discovered that Razib “covered” Konnikova’s article, as part of a 5-post-a-day blogging mill. His readership (6 comments) was not nearly as excited about the opportunity to bash someone Razib called “19th century romantic” as Coyne’s readers were (100 comments). There is nothing wrong with being a romantic. What is questionable is romanticizing professional science, which is typically what the amateur Razib Khan does on his blog. “Question science, and it may surprise you with what it has discovered!” preaches Khan. When it comes to actual scientific knowledge, Razib underwhelms. He praises Basyesian phylogenetics as holding superiority in quality and economy over knowledge generated by domain specialists. Not too long ago, Bayesian phylogenetics was used to argue that Out-of-Africa with Complete Replacement has higher probability than Out-of-Africa with Admixture. Alan Templeton (“Statistical Hypothesis Testing in Intraspecific Phylogeography: Nested Clade Phylogeographical Analysis vs. Approximate Bayesian Computation,” Mol Ecol 18 (2009): 319-331; “Coherent and Incoherent Inference in Phylogeography and Human Evolution,” Proceedings of the National Academy of Sciences 2010) was the first one to alert the scientific community that this inference was incoherent because a special theory (Out-of-Africa with Complete Replacement) cannot have a higher probability than a general theory (Out-of-Africa with Admixture). History proved him right when geneticists discovered evidence for archaic admixture within and outside of Africa. Another failed application of Bayesian statistics comes from my own field of kinship studies. Fiona Jordan (“A Phylogenetic Analysis of the Evolution of Austronesian Sibling Terminologies,” Human Biology 83 (2), 2011) published a reconstruction of proto-Austronesian sibling terminology using Bayesian analysis. But since she misunderstood the complex native semantics of sibling terms in her primary dataset, the method led to an erroneous conclusion. I corresponded with Fiona on this and she admitted the mistake but the article was too far in the publishing process to make the corrections.

Razib parades science’s ability to democratize knowledge (and it often does!) but this shouldn’t come at the expense of deteriorating quality of scholarship. There maybe a self-serving interest here for Razib – the more all-purpose, wandering “scientific” methods are out there, the easier it is for someone with a background in biochemistry to blog about everything from religion to human origins. But this is fantasy science, not actual science, and the difference is as real as between fantasy football and real football. There can be absolutely no real science without what Razib dismisses so ignorantly as “individual erudition” and “powerful intuition about specific subjects through means esoteric or ascetic.”