Tackling Pinker's defense of evolutionary psychology

I previously addressed the criticisms of my criticisms of evolutionary psychology by Jerry Coyne; Now I turn to the criticisms of my criticisms he solicited from Steven Pinker. This is getting a bit convoluted, so let me first state the basics.

I dislike evolutionary psychology. Pinker is an advocate for evolutionary psychology. What brought on this back-and-forth was that I was a member of a panel at a science fiction convention that discussed evo psych; I made a few brief comments on my blog that were capsule summaries of my discussion there. In the section below, the paragraphs preceded by an "M:" and in italics are my words excerpted from those comments; the parts preceded by a "P:" are Pinker's commentary. All clear?

M: Fundamental assumptions of evo psych: That you can infer an adaptive history from the distribution of current traits — that they are adaptations at all is an assumption usually not founded in evidence (this is not to deny that that there are features that are clearly the product of selection, but that you can’t pick an arbitrary attribute and draw elaborate scenarios for its origins). . .

P: Of course “arbitrary” and “elaborate” are the straw-man giveaways here. What about carefully selected attributes, and minimal assumptions about phylogeny with a focus on function, as we do for other organs? You can ask what the spleen is for – and it would be perverse to do physiology without asking such a question – without “drawing elaborate scenarios for its origins.”

Whoa, whoa, whoa — that skips right over the really important word: "adaptive". Start there. That's my primary objection, the habit of evolutionary psychologists of taking every property of human behavior, assuming that it is the result of selection, building scenarios for their evolution, and then testing them poorly.

We already know that that is impossible. The repertoire of human behavior is so complex and rich, and relatively recently evolved, that to argue that every behavior is the product of specific selection imposes an untenable genetic load. The bulk of the genetic foundation of our psychology (and I agree that there must be one!) must be byproducts and accidents. The null hypothesis of evolutionary psychology should be that a behavior is non-adaptive, yet for some reason all I ever see is adaptive hypotheses.

The spleen is an interesting example. There are components of the spleen that are definitely functional and almost certainly adaptive: its functions as a blood reservoir, as an element of the immune system, as part of the erythrocyte cycling mechanism. You can examine the evolution of those functions phylogenetically; for instance, some teleosts lack the erythropeotic functions of the spleen, while the majority use it as a blood reservoir. You can begin to dissect its history comparatively, by looking at what has a clear functional role and looking at the pattern of emergence of those properties.

What you can't do is pick any particular property of the spleen and invent functions for it, which is what I mean by arbitrary and elaborate. For instance, the spleen is located in most people in the upper left quadrant of the abdomen; are you going to make an adaptive case for why it's on the left rather than the right? The actual reason almost certainly has nothing to do with adaptation or selection, and everything to do with historical and developmental mechanisms that are neutral with respect to selection.

M:. . . That behavioral features that have been selected for in our history are represented by modular components in the brain – again with rare exceptions, you can’t simply assign a behavioral role to a specific spot in the brain, just as you can’t assign a behavior to a gene.

P: No one in Ev Psych points to specific spots in the brain – that’s cognitive neuroscience, not evolutionary psychology. The only assumption is that there are functional circuits, in the same way that a program can be fragmented across your hard drive.

Now this is one of my peeves with evolutionary psychology. The evo psych literature is thick with papers emphasizing "modularity"; that evolutionary psychology FAQ I referenced before makes it clear that it's an important concept in the field (and also ties it to concepts in computer science). Yet it is meaningless. Sometimes there's the implication that the "module" is a discrete element in the brain, but it's never clear whether they're talking about a genetic module (an epistatic network of genes) or a neural module (an interconnected network of neurons), and when pressed, they retreat, as Pinker does here, to an admission that it could be just about anything scattered anywhere in the brain.

So my question is…why talk about "modules" at all, other than to reify an abstraction into something misleadingly concrete? Evolutionary psychologists don't do neurobiology, and they don't do genetic dissections, and they don't do molecular genetics, so why do they insist on modularity? It's premature and a violation of Occam's razor to throw the term around, and also completely unnecessary — a behavior could be a product of diffuse general phenomena in the brain without diminishing its importance at all.

M: . . . That the human brain is adapted to a particular environment, specifically the African savannah, and that we can ignore as negligible any evolutionary events in the last 10,000 years, that we can ignore the complexity of an environment most of the evo psych people have never seriously studied, and that that environment can dictate one narrow range of outcomes rather than permit millions of different possibilities.

P: The savannah is a red herring – that’s just a convenient dichotomization of the relevant continuum, which is evolutionary history. A minimal commitment to “pre-modern” gives you the same conclusions. By saying that the brain could not have been biologically adapted to stable government, police, literacy, medicine, science, reliable statistics, prevalence of high-calorie food, etc., you don’t need to go back to the savannah; you just need to say that these were all relevantly recent in most people’s evolutionary history. The savannah is just a synechdoche.

Ah, a syncechdoche. This is the evolutionary psychology version of the religious argument that it's "just a metaphor."

Again, this is a peeve I have with the field. I agree with the general principle that of course the brain is a product of our evolutionary history, and that there is almost certainly a foundation of genetically defined, general psychological properties of the mind…and that a great many specific psychological properties are not biologically adapted. Pinker is writing good common sense here.

But over and over, you see evolutionary psychologists falling into this trap of examining a behavior and then fitting it to some prior specific environment. They talk of a Savannah Mind or they generalize it to the Environment of Evolutionary Adaptedness. It's another reification of the unknown. You don't like "savannah"? Change it to "Pleistocene". It's just as broad and meaningless. It's an attempt to reduce the complex and diverse to a too simple unit.

M: I’d also add that most evo psych studies assume a one-to-one mapping of hypothetical genes to behaviors. . .

P: Completely untrue – this was Gould’s claim in the 1970s, which confused a “gene for x” (indispensable in any evolutionary thinking, given segregation) in the sense of “increases the probability of X, averaging over environments and other genes” with “a gene for X” in the sense of “necessary and sufficient for X.” Every honest biologist invokes “gene for X” in the former sense; evolution would be impossible if there were no additive effects of genes. No one believes the latter – it’s pure straw.

By one-to-one, I mean the assumption that a behavior trait can be mapped to a contribution from a gene that was subject to selection for that trait; that it might be an additive property of a pleiotropic gene will be nominally noted, as Pinker does here, but operationally ignored. Remember, the issue is not whether genes contribute to our psychology, a point I totally agree with, but whether we can assign a selective origin to a behavior. That is a much, much harder problem.

M [continuation of previous sentence]:. . . and never actually look at genes and for that matter, ignore most human diversity to focus on a naive typological simplicity that allows them to use undergraduate psych majors at Western universities as proxies for all of humanity”

P: It’s psychologists, not evolutionary psychologists, who focus on Western undergrads –field research and citations of anthropology are vastly more common in ev psych than in non-ev-psych. PZ is engaging in prosecution here, not analysis – he’s clearly ignorant of the sociology of the fields.

As for diversity – is he arguing for genetic differences among human groups, a la Herrnstein & Murray?

First, this has already been addressed by Stephanie Zvan: when you look in the evolutionary psychology journals at papers identified as evolutionary psychology, you find…a focus on Western undergrads. I throw up my hands in exasperation. Look at the actual work done in your field, not the abstract ideal you hold in your head. I get my vision of evolutionary psychology by reading the papers.

Secondly, what a weirdly off-target attempt at ad hominem. Once again, my criticisms are being addressed by imagining motives; in Jerry Coyne's critique, I'm an uber-liberal offended at the consequences a genetic component to behavior might have on my egalitarian biases; now Pinker takes a swipe by tarring me with the likes of Herrnstein & Murray. Make up your minds!

For the record, of course there are genetic differences in human populations! It's an open question whether any of them make significant contributions to human psychology, however. I'm open to evidence either way.

But my remark was about cultural diversity (which also, by the way, exists). Setting aside the notion of a genetic component for now, we know that culture creates different minds. How can you analyze the causes of a behavior if your work focuses on a relatively uniform sample?

M: Developmental plasticity vitiates most of the claims of evo psych. Without denying that some behaviors certainly have a strong biological basis, the differences in human behaviors are more likely to be a product of plasticity than of genetic differences. . .

P: Plasticity is just learning at the neural level, and learning is not an alternative to innate motives and learning mechanisms. Plasticity became an all-purpose fudge factor in the 1990s (just like “epigenetics” is today). But the idea that the brain is a piece of plastic molded by the environment is bad neuroscience. I reviewed neural plasticity in the chapter “The Slate’s Last Stand” in The Blank Slate, with the help of many colleagues in neuroscience, and noted that the plasticity that allows feedback during development and learning during ontogeny is superimposed on an innate matrix of neural organization. For example if you silence *all* synaptic activity in the brain of a developing mouse with knock-outs, the brain is pretty much normal.

Speaking of straw men…I found The Blank Slate entirely unreadable, unlike most of Pinker's books, because of the gigantic straw man erected in the title. This flailing against me is a product of this weird idea that I reject the contribution of our genes to our minds, but just as there are no evolutionary psychologists who believe everything in our brains is genetically predetermined, there is no such thing in serious science as a "blank slater".

There is a continuum, and we're arguing about degrees. For example, take a child of French parents and raise them in the United States, they'll grow up speaking fluent English (or Spanish, depending on the household), and vice versa — an American child raised in France will speak French like a native. There is no genetic component to the details of language. Yet when you compare diverse languages you can start to pick out commonalities, and when you look at the neural substrates of language you do see shared anatomy and physiology — I do not hesitate to accept that there is an evolved component of human language. The differences between speakers are learned, the universals may well be biological.

Which means that when evolutionary psychologists try to parse out variations between different groups, racial or sexual, I suspect it's most likely that they are seeing cultural variations, so trying to peg them to an adaptive explanation is an exercise in futility. When evolutionary psychologists try to drill down and identify the shared components, I'm much more willing to see their efforts as interesting.

That last sentence by Pinker is a lovely example of nonsensical denial of the importance of plasticity. "Pretty much normal" means that on broad, superficial inspection the various components of the brain are present — hindbrain, midbrain, forebrain, various nuclei and pathways, they're all there. I've seen the same thing in zebrafish: the peripheral motor nerves I studied as a graduate student form perfectly normally even if you knock out all the acetylcholine receptors, so that the muscles are totally unresponsive to physiological inputs.

This does not surprise me. Most of the patterning of the brain is set up in the embryo before neuronal connectivity is established; the clock-like activity of mitotic rate genes defines the size of various bits of the brain; adhesive and repulsive cell surface interactions lay out the major pathways. Does Pinker think someone trained in developmental neurobiology would expect that the brain would collapse into a formless blob in the absence of action potentials and synaptic transmission?

But it is still absurd to call the deprived brain "pretty much normal". When you look deeper, you find subtle and important differences. The clearest examples are found in experiments with visually deprived cats: sew one eyelid shut, or both, or alternate, in a young kitten, and you can find all kinds of changes in visual processing, detectable at both the physiological and anatomical levels. The visual cortex forms, projections from the lateral geniculate terminate in roughly the right place, but they absolutely depend on visual input to fine tune their connections. Human children born with visual deficits in one eye will also have lifelong deficits in visual processing, even if the original problem is corrected.

Try raising a child without contact with other humans. I guarantee you that their brains, when physically examined, would look "pretty much normal"…but does anyone really believe that psychologically, on the level evolutionary psychologists study brains, that they'd be "pretty much normal"?

This is "pretty much normal" behavior from evolutionary psychologists, though. Point out that that their inferences about neuronal circuitry are bogus, they tell you that they don't study neurons anyway; tell them that the behaviors they study are awfully plastic and flexible, and presto, hey, look, brains and neurons are patterned by genetic elements. The sleight of hand is impressive, except when you realize that science shouldn't be about magic tricks.