Friday, August 11, 2017

Let's Talk About Sex, Baby

I spend a good deal of time on the Internet.  I'm involved in a variety of online communities, I'm of the age group that was on the cutting edge of Facebook and still haven't moved on to anything cooler, I have an embarrassing familiarity with memes, and my gif game is on point.  Being an "Internet person" means that sometimes you find incredibly amazing content, and sometimes you find absolute trash.  Some of the absolute trash I've accidentally stumbled into has been from communities that rally around men who use biology, psychology, and "biotruths" to justify various forms of misogyny.  When a memo by a Google employee arguing against Google's diversity initiatives was published recently,1 I realized that a lot of people, even people I know, love, and respect, either hold similar beliefs or can be convinced to believe them simply by slapping some peer-reviewed sources and scientific labels on them.  The problem is that a lot of the sources that these arguments cite are often bad science, not supported by the scientific community or by other studies on the topic, or are being misused and misrepresented.

And so, today I'm delving into the land of sexual dimorphism and evolutionary psychology in an attempt to debunk some of these ideas that are spread around.  As a note, this post acknowledges trans* and intersex people, but since most of the research is on male-female differences, I'm going to be focusing on that.  With a deep breath and the knowledge that I will probably invoke the ire of at least one person on Twitter, let's begin.  Full disclosure:  I am a woman.  I'm sure you're shocked.

We know there are many differences between men and women.  The arguments that are often made, that were a central tenet of the Google memo, try to focus on nature as opposed to nurture- ignoring culturally ingrained constructs and socially appropriate behaviors in deference to biology and trans-cultural phenomena.  I'm going to partially address two of the focuses of these debates today: first, sexual dimorphism.  Sexual dimorphism is just the scientific term for the differences between sexes beyond their primary sex organs.  There are definitely physical differences between men and women.  Men are, on average, 9% taller than women.2  On average, women have 40-60% of the upper body strength of men and 70-75% of the lower body strength (I'll blame this for the fact that I only bench press 65 pounds).3  There are average differences in lung capacity,4 pain tolerance,5 and likelihood of catching and severity of illness.6  You'll notice that none of the things I've listed have anything to do with the brain, psychology, or behavior.

So what differences are there in the brain?  Men's brains are, again on average, 10-15% larger than women's.7  There are a few important caveats here: 1) while brain size may be positively correlated with intelligence, it by no means accounts for intelligence, and there are a variety of third variables (nutrition, socioeconomic status, education level) that may influence both,8 and 2) men also tend to weigh 10-15% more than women, so when body weight is corrected for, there actually is no difference.9  A meta-analysis (statistics performed across a number of studies) in 2014 indicated significant differences in volume of the amygdalae, hippocampus, left insula, putamen cerebellum and  anterior cingulate cortex (men larger), as well as the right insula, frontal gyri, and Heschel's gyrus (women larger).10  Many of these structures are part of the limbic system, which is involved with emotion, learning, and memory.  Interestingly, the men had larger volumes in the "emotion areas" that women have traditionally been associated with.

A study of over 5,000 individuals using the UK Biobank study found that men tend to display greater variance across brain region volume, and that the effect sizes of volume differences were small.11  Essentially, while they're may be statistically significant differences, they're not very big difference, mainly due to the amount of variance men's brains displayed.  There was some evidence that the connections between brain areas were more complex in women, but as of yet, we don't really know what that might mean, if anything.12

So here we get into two of the major flaws of the people who make MRI and volumetric based arguments for differences between men and women.  The first is that the majority of these studies are in "mature" populations- over 18.  This makes it very difficult to tell whether these morphometry differences are due to inherent, genetically determined blueprints, or if they come about because of experience and exposure.  I'll compare it to studies that have found that people with PTSD tend to have smaller hippocampi: does a lower hippocampus volume predispose you to getting PTSD, or does something about having PTSD shrink your hippocampus?  Cross-sectional study designs make it nearly impossible to tell, and most of the sources that people tend to cite on these points are cross sectional.

The second, and most major, problem with this line of argument is the problem of reverse inference.  Reverse inference happens all the time in MRI research, and it's a thing to be on the lookout for when evaluating studies.  The basic idea is that we show people pictures of fearful faces.  Their amygdala shows increased activity. "Awesome", we say.  "They amygdala is involved with fear responses.  We have done science!" Then we show people pictures of their dog, and their amygdala shows increased activity.  "Ah ha!", we say. "People are afraid of their dogs!".  It's one of the first logical fallacy most people learn: Red heads have no soul.  Sam Winchester has no soul.  Therefore, Sam Winchester is a red head.
Wrong.  This only applies if red heads are the only people that don't have souls, or if the only the that causes increased amygdala activation is fear.  Claiming that men and women display structural or functional differences in the limbic system does not in any way mean that they are more or less emotional, process emotion differently, or any of the other things that people may claim.  Researchers do this all the time in their papers, and it's a huge problem with functional imaging papers.  Don't let other people get away with applying this crappy science too.

After establishing that there are measurable physiological differences, people tend to turn to personality differences. The Psychology of Sex Differences, published in 1974, considered 2,000 different studies on gender differences in pretty much every domain imaginable: social behavior, abilities, personality traits, memory, etc.  It reported differences in only four domains: verbal memory, mathematical ability, visual-spatial ability, and aggression.13  A meta-analysis published in 2005 looked at the effect size of sex differences across 128 different behavioral variables- academic skills, speech, facial emotion processing, aggression, leadership, helping behaviors, self-esteem, motor behaviors, Neuroticism, Agreeableness, Openness, moral reasoning, and cheating behavior.  It found a small to zero effect size in 78% of the variables.  Large differences were found in physical abilities (not really relevant to women in the workplace) and sexuality (hopefully not relevant to women in the workplace).  The effect size in physical aggression was moderate (still hoping not relevant in the workplace, unless you're an MMA fighter), and effect size of relational aggression, which has gotten a lot of play in the news, was unclear.14

Let's move beyond that into some of the more specific claims- ones that Mr. Google cited coming from the Big Five Personality Scales.  These include Openness to Experience, Agreeableness, Neuroticism, Conscientiousness, and Extraversion.  All of these traits exist on a continuum, where every stage has what we would culturally consider positive or negative attributes to any given point, which means that regardless of findings, you don't have to think more than about thirty seconds to come up with a way to make them support your opinion.  There is a fair body of evidence that indicates that women score higher on Neuroticsm scales across cultures.15  This finding doesn't represent effect size at all though, and small average differences on a personality scale aren't going to create the major disparities we see in the workplace and relationships.  This finding could also be used to argue that women are less well equipped to handle stress, emotionally unstable, often in a bad mood.  On the flip side, you could argue that men are less inspired or unconcerned with things, less dynamic and engaged.  The same is true of any of the Big Five Personality measures.  They're somewhat like (fairly consistently validated) horoscopes: whatever you want to read into it, you can.  If you want to prove that women are (on average) less oriented toward or capable of performing in certain roles, you can find a way to interpret personality measures to do that.

The other field that certain types of people like to use as arguments is evolutionary psychology.  This puts forward the idea that psychological behaviors are adaptive, and have adapted differently for men and women.  It looks for patterns across cultures and species that can be explained as a byproduct of adaptation as opposed to random variation.  The field as a whole is extremely controversial in science.  To start off with, evolutionary psychology is often scientifically sketchy at best.  There are a few problems here: first, evolutionary psychology theories are often not testable, which makes them bad science.  Given how different modern life is from that of our ancestors, it seems likely that personality traits would have evolved for different reasons than we now apply them.  We can't really test that, though, because it's difficult to determine personality from human remains.  It's difficult to tell when and how psychological behaviors developed in the timeline of human evolution, and impossible to determine what was psychologically happening with these individuals.  Even if we do have evidence of behaviors, there is no way to know the underlying reasonings and internal happenings, which is where the important stuff for these purposes is.  If you try to draw across species comparisons, you can't very well ask animals their thoughts and motivations either.  The idea of observing a trait or behavior and saying "That's interesting, it could be because of the fact that women take nine months to reproduce", without the ability to manipulate the gestation period in women, or have a clear analog for the behavior in species that don't have a nine month gestation period is just that- ideas, not science, and doesn't "prove" anything, no matter how much these defenders want it to.

There are assumptions made by many evolutionary psychologists that are just not true.  A major example is specialization of brain areas.  While yes, there is specialization to some extent, those are by no means simple or static.  Look at any of the hundreds and hundreds of studies on neuroplasticity (NOT the kind that supposedly happens with brain training apps) to see how malleable these areas are.  Look at the Wikipedia page for a given chunk of cortex and for most of them, multiple roles or suspected roles will be listed.  We have amazing computational power that wouldn't really be possible if we limited ourselves to one task or ability per chunk.  This goes back to the problem of reverse inference.  It would be nice, but unfortunately doesn't work that way.

In the end, the Google memo, and other manifestos like it, are a bunch of hogwash, or whatever your preferred colloquial form of "bullshit" may be.  Any differences in psychological or personality traits that may be relevant to a work place (or romantic relationships, for the people who use these tactics in those settings) are either not consistently found or have a small effect size, and likely don't really matter in the settings they're applying them.  The "biological differences" theories put out there often ignore the interaction of genes and environment or are misinterpreted and misapplied (possibly out of naivity, possibly intentionally).  The use of evolutionary psychology to bolster a point is a red flag in general, and the reader should proceed with caution.  They're often rife with reverse inferences, cherry-picking data that supports their points (and there usually IS data that supports their point, even when there's a whole lot more that proves it wrong), and coming up with a theory and then attempting to back it up, as opposed to letting the data guide the theory.  Most importantly, slapping the label of "science" on something does not inherently make it true, productive, or worthy of our consideration.  "Peer reviewed" doesn't necessarily mean right, it just means methodologically sound.  "Biological" doesn't mean applied correctly.  Capitalizing on the common misunderstandings of the word "theory" as it is used incorrectly by those arguing against the "theory" of evolution or the "theory" of relativity, doesn't mean that anything with the word theory in it is commonly accepted by scientists. Using "science" to bolster misogynistic, racist, homophobic views (and for those defending this particular memo, please remember that the exact same types of arguments have historically been made to prove that black individuals are less intelligent and capable, more violent, etc.) doesn't make them more worthy of consideration, it just makes them more worthy of using actual science and statistics to smack it down.

You can find me on Twitter @reviewer3blog for all of the arguments and rebuttals I'm sure will inevitably follow this post.


  1. Yeah! This knowledge was very important for me. Thanks. Actually, gender selection natural is really amazing. You can have your desired gender baby with this.

  2. Pretty! It was really a wonderful blog. Thanks for the provided information.
    Escorts Brasil