Know-it-all neuroscientist explains Trump’s election victory to the rest of us

Alex Gamma points us to the latest atrocity of pseudo-science in the popular press:

There’s a horrible piece in “Scientific American” arguing that knowing some neuroscience explains the results of the election, in particuar why people voted for Trump (https://blogs.scientificamerican.com/mind-guest-blog/trump-s-victory-and-the-neuroscience-of-rage/)

It says things like “To understand this election you must understand the brain’s threat detection mechanism”, or “This neuroscience perspective explains the seemingly incomprehensible situation of a privileged billionaire becoming the champion of working class men and women…”, or “Whether these real divisions in society will explode into factionalism or unite us will be determined by the same neural circuitry in our prefrontal cortex that separates “us from them.””

It’s bad on so many levels.

On Facebook, I pointed out that the explanatory force in this piece is located entirely on the psychological/social level: people feel angry and threatened etc by social disruption and alienation etc and then make a decision based on their emotions and on not reasoning etc.

The neural apparatus that is invoked doesn’t do any explanatory work. You can test that by omitting, in turn, the neural apparatus and the psychosocial story and see which one does the explaining. So it’s another instance of the overhyping of neuroscience by people who should know better.

Also, insofar as the psychosocial story rests on evolutionary reasoning (“us” vs “them”, tribalism…) it is pure speculation as we don’t have reliable evidence about the minds and social behaviors of our ancestors.

But that’s only the beginning of the problems with this piece. Others are

– it assumes, without evidence, to know why Trump voters voted like they did
– it assumes, without evidence, that *all* Trump voters voted for the very same reason

So it brutally ignores the variability in voter motivations and all the nuances of the social, economic, ethnic and personality backgrounds that play into them. It ignores the role of the media in shaping the perception of reality and in creating mass behaviors. It compresses all this into a reductionistic non-explanation of “you did this because your prefrontal cortex did X and your limbic system did Y”.

Finally, there’s some irony in people now fearing what will happen to science under the new science-challenged president. The fact is if science is on the level of this piece (and much, if not most of it, is), then it does deserve to go down the drain.

I agree with Alex. The article in question begins:

Pollsters, politicians, much of the press and public are dismayed by Donald Trump’s surprising victory in the presidential election, but not neuroscientists.

So . . . neuroscientists all predicted that Hillary Clinton would get 50% of the two-party vote, but not 52%? Pretty cool, huh?

Also this bit:

Whom to marry, where to live, or even what entrée to select from a dinner menu, are decisions we make not by reason, but rather by how we “feel.”

Wow—science! Jeez, and before reading this I thought my choice of steak instead of chicken had been driven by hard, rational calculation. It’s a good think we have neuroscience to tell us that we rely on emotion to decide who we marry. Who’d a thunk it?

As a political scientist, though, what really irks me is the bit about “The pollsters got it wrong because . . .” Off by 2%, dude! I agree that this 2% was a problem—and it was more than 2% in some key states—but, shoot, man, what kind of accuracy are you expecting here? Suppose the polls had been off by 1%? Then would that be ok with you?

As a small-d democrat, I’m also repulsed by this guy’s characterization of Republican voters (yes, they voted approx 50% Republican for the House and Senate, not just for president, also they gave Mitt Romney 48% of the two-party vote in 2012, etc.) as “an appeal to the brain’s limbic system.”

Also bizarre is this bit: “It is impossible to feel love or hate for someone you have not yet met.” Millions of people feel hate for Donald Trump or Hillary Clinton (or both!) without having met either of them.

This particular Hari Seldon concludes with a statement about “the electorate has concluded.” Kind of amazing how the electoral college is part of neuroscience too.

What I’m wondering is, if this was all so damn obvious, why didn’t he publish it before the election? That would’ve been much cooler.

The funny thing is, I know of only one neuroscientist who made a public prediction of the 2016 election. He gave Hillary Clinton a 99% chance of winning the election. Afterward he wrote that “the polls were off, massively.” Actually the polling errors were not “massive”; they were in line with historical polling errors. This neuroscientist wrote, “The entire polling industry . . . ended up with data that missed tonight’s results by a very large margin.” Actually the margin wasn’t so large. But it was consequential. The neuroscientist who predicted the election wrong ahead of time called the result a “giant surprise,” but the other neuroscientist who wrote after the election wasn’t bewildered at all.

Maybe the two neuroscientists should get together and work this one out.

28 thoughts on “Know-it-all neuroscientist explains Trump’s election victory to the rest of us

  1. If the numbers guessed at now are correct and black turnout was off by perhaps 1% of the total electorate, then aren’t all the pieces about the white vote kind of stupid too? It’s somewhat fascinating that almost no effort is being spent on what may have been actual salient point of the election. Is that an unwillingness to face the truth in favor of absurdities like negative brain waves and whatever narrative one imagines (dragons?)? I sometimes think it’s yet another form of racism: the denial that black participation was important and that it’s lack was crucial. But then maybe that’s the neurons doing something funny.
    .

  2. It’s the Dems in denial. They can’t face up to the fact they lost the election and will not analyse why their politics went wrong. I’m afraid science is just a casualty of politics here, as “pop science” overrides this OP’s evidence belief mechanism to confirm what he already believed. I still get the same rubbish in my FB feed too: “Oh, look at those evil racist Republicans.” As if endless “info-memes” are supposed to motivate me to vote Hillary. More like “disinfo-memes”. Little discussion from Dems on how they managed to lose an election when they had their “Buggins’ turn” candidate in place. They faced their dream opponent (the 1 in 16 they wanted to run against). They had all their identity politics artillery in place. Funding. Their perfect demographics, or whatever, their color-book Machiavellian “analysts” and “strategists” call it. It’s tragic and heart-breaking looking at Dems in denial. Nothing convinces one so much of human stupidity as watching it in action.

    • Mark:

      What is so ridiculous is that if Clinton had received the exact same share of the popular vote but with a slightly different distribution so that she won the electoral college, I’m pretty sure the neuroscientist would’ve offered a confident explanation of why that happened, too!

  3. Just to make it clear. This essay in Sci. Am. is best understood in terms of Dems telling each other what they want to think, and what the Party’s Machiavellian “analysts” and “strategists” told them to think. It is virtue signalling: “I’m a scientist and I can toe the Party line”.

    • AG: “this article is bad because the neuroscientist is using a single simplistic idea to explain the motivations of an entire group without evidence”
      Commenter: yes, and I have a simplistic idea to explain why the neuroscientist and all democrats do this!

  4. “The fact is if science is on the level of this piece (and much, if not most of it, is), then it does deserve to go down the drain.”

    This is very important. The “progressive” impulse to settle political questions with science is as dangerous as the “conservative” impulse to settle them with moral sentiment. Scientific “explanations” of political decisions get in the way of understanding them. We sit in judgment on our political opponents, rather than engaging with them.

    I’ve been reading Wyndham Lewis’s The Art of Being Ruled a lot lately. “Social reform today [1926],” he begins, “is a very fluid and mercurial science,” and adds that “politics and science are today commutative.” Much later in the book, in a brilliant chapter called “Fascism as an Alternative,” he writes about the situation in Mussolini’s Italy:

    “In ten years a state will have been built in which at last no trace of European ‘liberalism’ or its accompanying democratic ‘liberty’ exists. […] In such a state it is difficult to see how ‘politics’ could exist. ‘Economics’ will simply disappear. All the boring and wasteful sham-sciences that have sprung up in support of the great pretenses of democracy, and in deference to notions of democratic freedom, will die from one day to the next: for they are the most barren of luxuries, and no one would be interested in keeping them alive for their own sakes (in the way the arts are sometimes kept alive) for an hour.”

    I have highlighted the sentence that Alex’s statement reminded me of. The fascist approach to ‘politics’ is essentially to do away with it (in any recognizably democratic sense) and to replace it with “decision making” by a powerful person. What’s interesting is how the “science” of decision-making simulates this attitude, albeit with a pretense of democracy. Before each of Ezra Pound’s broadcasts from Rome during the second world war, Italian radio provided the disclaimer that Pound was being given the airwaves “in accordance with the Fascist policy of the free expression of opinion by those qualified to hold it.” If one didn’t know better, one would think they were trying to be funny.

    But hasn’t “the age of Obama” been precisely a time when one felt unqualified to hold one’s own political opinions? I’m thinking here, for example, of the “science” of “nudging” people towards policy goals by “priming” their “implicit biases”. One has felt disqualified by dint of one’s limbic system, let’s say. The neuro-scientific end run around what people think and believe about their polity is in many ways like telling people that 97% of all scientists believe in climate change. They may be right, of course. But I think the effort to “qualify” political opinions with science has finally undermined the basis of science’s real authority. Scientific American must shoulder much of the blame. It will no doubt draw its excuses from science too.

  5. I’m a reductionist and believe that in some distant future, the physicists will explain everything because everything is physics. Whom to marry, where to live, or even what entrée to select from a dinner menu, are decisions we make not by reason, but just something that emerges from the matter we consist of and the matter of our surroundings obeying the laws of physics.

    But physics isn’t quite there yet, and neither is neuroscience. Social science that explains things at the preferences/beliefs level retains some usefulness, for now.

    • This is actually never going to happen and this is beyond any doubt. Your proposition implies full determinism and lack of free will. While we can’t reject the hypothesis of determinism (nor the hypothesis of free will! But then we know we can’t even reject solipsism!) we are in position to assert we can’t know with perfect precision the position and speed of any single particle of the universe. This is due to the Heisenberg’s uncertainty principle! This is sufficient to disprove your assertion. And note that it is not even necessary! We could have come to the same conclusion even without this principle (e.g. the complexity of interactions between the particles! or the existence of consiousness). Reductionism can’t work and this is particularly true for large social systems or interacting agents.

      • Firstly, I agree with you that physics will not be able to “explain” in human-interpretable terms the macro-social goings-on of the world because I don’t regard a system of billions upon billions of interdependent partial differential equations as an “explanation” even if it represents a useful model. I say this as a reductionist; there will always be a place for macro analysis of social phenomena (and if there isn’t, I don’t think I’d like to live in that world).

        That aside, I disagree with you regarding determinism. Even today, we are able to describe mathematically/reductively the macroscopic properties of matter based on underlying quantum mechanical phenomena which are represented by probabilistic models (and, to our knowledge, are probabilistic in nature). In short, reductionism doesn’t necessarily imply hard determinism.

  6. As a neuroscientist dismayed by the Trump election, I resent this guy speaking for all neuroscientists. That article is the lowest form of neuroscience perversion that claims some vague notion we have about the brain translates directly to the moral and the social. We understand so little about how the brain works. I completely agree with all the criticisms Alex Gamma levied. It’s disgusting.

  7. “The pollsters got it wrong” is just wrong; the people’s interpretation of the polls was wrong. 538 gave Mrs Clinton a 2/3 chance of winning and Mr Trump a 1/3 chance. People looked at that and thought Mrs Clinton’s chances were 100%. Mr Trump’s 1/3 chance was way higher than 0, and this was just that 1 chance in 3 that happened to come out a winner.

      • Was that a joke? I’ll admit that 1/100 events are less probable than 1/3 ones, but still, 1/100 events happen. At the risk of sounding like a frequentist, don’t we need a sequence of predictions to get a sense of calibration? I know that we can make a single observation and still think the 1/100 prediction is worse than a 1/3 prediction when the event occurs, but we also know the errors are pretty big.

  8. Typical “post hoc ergo propter hoc”, which is especially convenient if the projections made in other fields of research appeared to have turned out “wrong” (i.e., the 15 minutes of fame for pseudo-scientific explanations).
    Personally, I was really surprised how prevalent that way of thinking is even in the data-savvy sciences. It doesn’t matter how much pollsters talk about accounting for uncertainty, and about the huge margin of error this election had; as soon as you dare settle on a number with a percentage sign in the end, people go wild.

    • >”data-savvy sciences”

      Your average neuroscientist is not at all data-savvy. What gave you that impression?

      Try this,

      1) Go pick a random neuroscience paper.
      2) Check if it says “mechanism X leads to behavioral outcome Y”. If not, go to step 1.
      3) See if they plot the molecular/histological data regarding X vs the behavioral data regarding Y.

      It is very rare to see simple scatter plots of this nature from neuroscientists.

      • Totally true about the publishing process. Partly tongue in cheek… Try going to a clinical conference and NOT hearing someone link their particular research to some neuroscientific finding. Neuroscience is what makes psychology and other inferential processes valid amongst the insecure. Small N + Large Imaging device = “better” science than the rest of us. A more realistic view is that neuroscience is like putting a computer in an MRI and hoping the large EMF doesn’t impact the readings too much. To infer from the output that the computer is processing Bayesian Analysis using STAN would be impressive. To also infer that all computers showing that pattern are therefore using STAN and even have it installed would be silly. Neuroscience runs the risk of swapping self-report for imaging and potentially then making the same leaps of faith.

        It is analogous to physicists, when they reach a certain age, attempting to bestow their philosophies of life on the masses. All of which is much more to do with social behaviour, trend following and being human than science.

  9. The distillation of the likelihood of a binary election result into a single probability like 65% or 99% is popular due to short attention spans and the betting markets – where it does have meaning-, but this has become generally accepted as a popular ‘scientific’ communication. This numerical result is contingent on a specific set of assumptions that should be communicated as an essential component of the calculation. If the 65% was communicated with an assumption of “No October or November ‘surprise'” along with other specific assumptions, we might be talking (or maybe that should be hearing) less about the failure of the statisticians and even more about the failure of the FBI. Not to overplay the “where I come from” card, but an actuary only communicates numerical results along with the applicable set of assumptions which are part and parcel of the results. A numerical result delivered without the assumptions that were used in the calculation is considered meaningless.

    • Both 538 and the overconfident neuroscientist in question have published quite a lot about their assumptions. Indeed neuroscientist has published his computer code in its entirety. Neuroscientist seems to have fallen a bit too much in love with his code, particularly the fact that if you just overlook the correlation of polling error among states, you can compute the probability mass function of electoral college results in quadratic time. I wrote elsewhere that neuroscientist’s model is going to leave him very embarrassed after some election, but I hope it’s not this one. Then, guess what.

Leave a Reply to econostatistician Cancel reply

Your email address will not be published. Required fields are marked *