Politics, Social Science, US Election

Were Trump Voters Irrational?

In September 2016, in collaboration with my colleagues Richard West and Maggie Toplak, I published a book titled The Rationality Quotient. In it, we described our attempt to create the first comprehensive test of rational thinking. The book is very much an academic volume, full of statistics and technical details. We had expected our academic peers to engage with the statistics and technical details, and they did begin to do just that after its publication.

But then the November 8, 2016 United States presidential election intervened.

The nature of my email suddenly changed. I began to receive many communications containing gallows humor, like “Wow, you’ll sure have a lot to study now” or “We sure need your test now, don’t we?” Many of these emails had the implication that I now had the perfect group to study—Trump voters—who were obviously irrational in the eyes of my email correspondents.

Subsequent to the election, I also received many invitations to speak. Several of these invitations came with the subtle (or sometimes not-so-subtle) implication that I surely would want to comment—after first giving my technical talk, of course—on the flawed rational thinking of the voters who had done this terrible thing to the nation. One European conference that solicited my participation had as its theme trying to understand the obviously defective thinking not only of Trump voters, but of Brexit voters as well. The wordy conference prospectus clearly presumed that every educated person would view any opposition to increased globalization as obviously irrational. I—the author of a rational thinking test—was seen as the ideal candidate to give the imprimatur of science to this conclusion. No less insistent have been friends and relatives who assume that I am the perfect person to affirm their view that a substantial number of people who cast ballots for Trump were irrational in their thinking.

Rather than reply to these emails, or give all the talks requested, or affirm all of the leading questions addressed to me, I thought it would be more efficient to write this single essay and disappoint all of my progressive correspondents at once. I am very sorry, progressives, to have to tell you that the Trump voters were, in fact, not irrational—or at least no less rational than the Clinton voters. Several kinds of analyses in cognitive science support this conclusion.

Rationality in Cognitive Science

Rationality is a tortured term in intellectual discourse because it has a multitude of definitions. The term is claimed by many disciplines—philosophy, economics, decision theory, psychology—and defined slightly differently by each. In The Rationality Quotient, we grounded our test of rational thinking in definitions from the interdisciplinary field of cognitive science.

Cognitive scientists recognize two types of rationality: instrumental and epistemic. Instrumental rationality is achieved when we act with optimal efficiency to achieve our goals. Epistemic rationality concerns how well beliefs map onto the actual structure of the world—that is, whether our beliefs are accurate, or true. A quick and memorable way to differentiate the two is to say that they concern what to do (instrumental rationality) and what is true (epistemic rationality). Of course, the two are related. In order to take actions that fulfill our goals, we need to base those actions on beliefs that are properly calibrated to the world. In order to understand the rationality (or irrationality) of the Trump voters, I will focus first on instrumental rationality and then turn to epistemic rationality.

Utility is not only about hedonism and pleasure, but also desirability and worth

Instrumental rationality—the optimization of the individual’s goal fulfillment–means behaving in the world so that you get what you most want, given the resources (physical and mental) available to you. More technically, the model of rational judgment used by decision scientists is one in which a person chooses options based on which option has the largest expected utility. The term utility is a slippery word, however, and it is actually used by decision scientists in ways that do not map exactly onto anything in general discourse. The term as used in cognitive science does not refer to its primary dictionary definition of “usefulness.” Instead, in decision theory, utility means something closer to “goodness.” It is important to realize that utility is not the same as pleasure. Instead, utility refers to the good that accrues when people achieve their goals—and a person’s goal is not always to maximize pleasure. Utility is thus more closely related to the notions of worth or desirability than it is to pleasure or hedonism.

More important for discussions of voter rationality, however, is that utility does not just mean monetary value. The value of winning a kayaking trip in a raffle is measured in its consumption utility, not its monetary value. For someone who has water phobias, its utility might be less than zero. Utility can be assigned to things even less concrete than a kayaking trip. For instance, people gain utility from holding and expressing specific beliefs and values. Failing to realize this is the source of much misunderstanding about voting behavior.

When what a person acts, wants, and desires are expressed as preferences. Decision theory is actually neutral on what a want or desire can be. It is the public, not economists, who tend to emphasize money or material wealth. Decision theorists are perfectly happy to call the non-material goal of seeking social prestige a desire with a utility value. Utility theory also does not dictate that every goal has to reflect strict self-interest in a narrow sense. We can have as our goal that other people achieve their goals and that can have utility value for us. Thus, neither material goals nor self-interested goals are necessary according to utility theory. Many goals that motivate people are neither self-interested nor material, such as preserving the environment for posterity.

Failure to appreciate these nuances in rational choice theory is behind the charge that the Trump voters were irrational. A common complaint about them among Democratic critics is that they were voting against their own interests. A decade ago, this was the theme of Thomas Frank’s popular book What’s the Matter with Kansas? and it has recurred frequently since. The idea is that lower income people who vote Republican (not necessarily for Trump—most of these critiques predate the 2016 election) are voting against their interests because they would receive more government benefits if they voted Democratic. Many of these critiques contain the presumptions that, to be rational, preferences must be self-interested and that people’s primary desires are monetary. I have just discussed how rational choice theory contains no such presumptions, so on that basis alone the claim that such voters are irrational is unfounded.

In addition to being misplaced, leftists never seem to see how insulting this critique of Republican voters is. Their failure to see the insult illustrates precisely what they get wrong in evaluating the rationality of the Trump voters. Consider that these What’s the Matter with Kansas? critiques are written by highly educated left-wing pundits, professors, and advocates. Perhaps we should ask one of them whether their own vote is purely self-interested and for their own monetary benefit. They will say no, of course. And they will deny as well that their vote is irrational. Progressives will say that they often vote against their own monetary interests in order to do good for other people. Or they will say that their vote reflects their values and worldview—that they are concerned about the larger issues that are encompassed by that worldview (abortion legislation or climate change or gun restriction). Leftists seem unable to see that Republican voters—even lower income ones—may be just as attached to their own values and worldviews. The stance of the educated progressive making the What’s the Matter with Kansas? argument seems to be that: “no one else should vote against their monetary interests, but it’s not irrational for me to do so, because I am enlightened.”

The implicit insult in the Kansas argument often goes unrecognized, and, if I may use some cognitive science jargon here, it is a form of ‘myside’ bias. For example, leftists who work for nonprofit organizations are often choosing their values over monetary reward. And likewise, conservatives joining the military are often also choosing their values over monetary reward. The What’s the Matter with Kansas? argument seems to ignore or deny this symmetry. Many Republican voters with modest incomes cast a vote to help others rather than for their own monetary interests—precisely as do the progressive Democrats who find such Republican behavior puzzling. So no, neither the Kansas voters in Frank’s book, nor the Trump voters are voting against their interests, broadly—and correctly—defined. Even if part of the Kansas critique is correct (they are voting against their purely economic interests), these voters are not necessarily irrational because they may be sacrificing monetary gain in order to express their values or worldview.

But what about temperament, character, and fitness for office? Surely it was irrational to vote for Trump if temperament is relevant, Democrats might say. But this argument is not a slam-dunk from the standpoint of rationality. It is simply not self-evident how people should trade off temperament versus worldview in their voting choices. This is especially true in the 2016 presidential election, where the candidates were unusually differentiated in their worldviews. In that election, Clinton represented what I will term the Global and Groups perspective (GG) and Trump represented the Country and Citizen perspective (CC). Clinton signalled to the electorate that she represented the GG perspective by emphasizing global concerns (climate change and global climate agreements; increasing US refugee intake; rights and protections for noncitizens) and continually addressing groups in her speeches (the groups of Democratic identity politics: LGBT, African-Americans, Hispanics, etc.). Trump signalled to the electorate that he represented the CC perspective by continually emphasizing country in his speeches (“make America great again”) and addressing his audiences as citizens with nation-level interests rather than group interests (trade deals that disadvantaged American workers; securing the country’s borders; etc.).

Hillary Clinton during the third and final presidential debate

These two candidates (Clinton and Trump) more sharply differentiated these worldviews than any other combination of candidates in 2016. Bernie Sanders would have watered down the GG perspective, because his criticism of some trade deals made him less of a globalist than Clinton, and Sanders placed less emphasis on the Democratic identity groups. Similarly, Jeb Bush or Marco Rubio as Republican candidates would have watered down the CC perspective by being sympathetic to global trade deals and pursuing voters as groups in the manner that Democrats do (Hispanic voters in particular). Clinton and Trump represented the GG and CC worldviews in much purer form. The issue for a Republican voter or an independent voter with a CC worldview was thus how to weight the temperament issue against worldview (for simplicity, we will stipulate here that the temperament issue resides with Trump). Because there is no way to ascertain what weighting of these factors (temperament versus worldview) is optimal for a given person, it cannot be said that a voter who chooses worldview over temperament is irrational.

To my Democratic friends who demur from my conclusion here, I pose a thought experiment. Imagine that the candidates for a presidential election were Ted Cruz on the Republican side and Al Sharpton on the Democratic side. Now it is the candidate with the GG worldview who has the character and fitness-for-office issues. Who would you vote for?

When I am successful in forcing Democrats to give a response to this imaginary election, a substantial number admit that they would vote for Sharpton. They justify their choice by citing things that are very rational, given their worldview: they worry about appointments to the Supreme Court, abortion, and gun legislation. The Democrats justify their choice in much the same way as the Trump voters did when they eschewed disqualifying him on the basis of temperament. The Trump voters worried about open borders and encouraging cities to defy federal immigration law, etc.—they worried about threats to their CC worldview in the same manner that the hypothetical Sharpton voters worry about threats to their GG worldview. The calculus of decision theory is not precise enough to dictate a particular weighting of temperament and worldview in something as abstract and multidimensional as a presidential voting choice. After choosing Sharpton over Cruz, few Democrats would consider themselves irrational. Cognitive science would agree with them. But, in the same manner, when those with the opposite worldview vote for Trump over Clinton, they are being no less rational. On instrumental grounds, neither the voters choosing Sharpton over Cruz nor the voters choosing Trump over Clinton can be deemed irrational.

The Epistemic Rationality of the Trump Voters

If you are particularly ill-disposed toward Trump voters, at this point you may still be feeling that, deep down, there is something else wrong with the Trump supporters that was not covered in my discussion of instrumental rationality. You might feel that something in the domain of knowledge is wrong with the Trump voters: they don’t know enough, or they seem to be misinformed, or they don’t seem to listen to evidence. You would be right that there is something else that is worth assessing—another aspect of rationality that covers these additional concerns: epistemic rationality.

Concern with Trump voters in the epistemic domain is, however, not unique because this is a charge (the charge of epistemic irrationality) that Democrats have made about Republicans for some time now. Liberal Democrats have become accustomed, as we all have, to media presentations that are critical of conservative Republicans who do not accept the conclusions of climate science, or of evolutionary biology. These media presentations are correct, of course. The role of human activity in climate change is established science, and evolution is a biological fact. Thus, the denial of climate science or of evolutionary science clearly has a negative connotation, and rightly so.

However, there is a trap lying in wait for progressives here. It is very tempting for them to say: Well, the Democrats get climate science right, and Republicans get it wrong; the Democrats get evolution right, and conservative Republicans get it wrong; so therefore we liberal Democrats are getting everything factually right about all of the other charged topics that figure in political disputes—crime, immigration, poverty, parenting, sexuality, and so on. Such an argument is essentially the claim that Democrats are epistemically more rational than Republicans.

This type of thinking is what some years ago prompted the Democratic Party to declare itself the “party of science” and to label the Republican Party as the science deniers. That stance spawned a series of books with titles like Mooney’s The Republican War on Science (2005). As a political strategy, this “party of science” labelling might be effective, but epistemic superiority cannot simply be declared on the basis of a few examples. A cognitive scientist is forced to be pedantic here and rain on the progressive parade. In fact, any trained social scientist would be quick to point out the obvious selection effects that are operating. The issues in question (climate science and creationism/evolution) are cherry-picked for reasons of politics and media interest. In order to correctly call one party the party of science and the other the party of science deniers, one would of course have to have a representative sampling of scientific issues to see whether members of one party are more likely to accept scientific consensus.

In fact, it is not difficult at all to find scientific issues on which it is liberal Democrats who fail to accept the scientific consensus. Leftists become the “science deniers” in these cases. In fact, and ironically, there are enough examples to produce a book parallel to the Mooney volume cited above titled Science Left Behind: Feel-Good Fallacies and the Rise of the Anti-Scientific Left (2012). To mention an example from my own field, psychology: liberals tend to deny the overwhelming consensus in psychological science that intelligence is moderately heritable.

This isn’t the only instance of left-wing science denial, though. In the area of economics, progressives are very reluctant to accept the consensus view that when proper controls for occupational choice and work history are made, women do not make more than 20 per cent less than men for doing the same work.

Progressives tend to deny or obfuscate (just as conservatives obfuscate the research on global warming) the data indicating that single-parent households lead to more behavioral problems among children. Overwhelmingly progressive university schools of education deny the strong scientific consensus that phonics-based reading instruction facilitates most readers, especially those struggling the most. Many progressives find it hard to believe that there is no bias at all in the initial hiring of women for tenure-track university positions in STEM disciplines. Progressives tend to deny the consensus view that genetically modified organisms are safe to consume. Gender feminists routinely deny biological facts about sex differences. Largely Democratic cities and university towns are at the forefront of the anti-vaccine movement which denies a scientific consensus. In the same cities and towns, people find it hard to believe that there is a strong consensus among economists that rent control causes housing shortages and a diminution in the quality of housing. [Research citations for all the above are available from the author here.]

I will stop here because the point is made. There is plenty of science denial on the Democratic side to balance the anti-scientific attitudes of Republicans toward climate change and evolutionary theory. Neither political party is the party of science, and neither party exclusively contains the science deniers. Each side of the ideological divide accepts or denies scientific consensus depending upon the issue in question. Each side finds it hard to accept scientific evidence that undermines its own ideological beliefs and policies.

More formal studies have indicated that there are few differences in factual knowledge of the world between Republicans and Democrats. The Pew Research Center reported one of its News IQ surveys in 2015 (What the Public Knows, April 28, 2015) and found very few partisan differences. People in the sample answered 12 questions about current events (identifying the route of the Keystone XL pipeline; knowledge of how many Supreme Court justices are women; etc.) and the Republicans outperformed the Democrats on 7 of the 12 items. Democrats outperformed the Republicans on 5 of the items. On average, the Republicans in the sample answered 8.3 items correctly, the Democrats answered 7.9 items correctly, and the independents answered 8.0 items correctly.

The 2013 News IQ survey from the Pew Center (What the Public Knows, September 5, 2013) showed the same thing. People in the sample answered 13 questions about world events (identifying Egypt on a map of the Middle East, etc.). The Republicans outperformed the Democrats on 5 items, the Democrats outperformed the Republicans on 7 of the items, and there was no difference on one item. On average, the Republicans in the sample answered 6.5 items correctly, the Democrats answered 6.4 items correctly, and the independents answered 6.6 items correctly. In summary, the Pew surveys find few partisan differences in current events knowledge. Even if the Trump voters had come disproportionately from independent voters compared with previous Republican nominees, there would not have been a knowledge deficit among Trump voters.

Similar findings are obtained in specific areas of knowledge related to voting such as economics. George Mason University economist Daniel Klein and colleague Zeljka Buturovic (Econ Journal Watch, May 2011, 157-173) gave a 17-item questionnaire on knowledge of economics to over 2000 online respondents. They found that individuals labeling themselves libertarian or very conservative scored higher than individuals labeling themselves as liberal or progressive. Importantly, their major conclusion was not that conservatives were more economically knowledgeable than leftists. Instead, they stressed as one of their major findings how such surveys are tilted by the selection of questions. For example, the item “rent-control laws lead to housing shortages” (correct answer: true) is more difficult for progressives because it challenges their ideology; whereas the item “a dollar means more to a poor person than it does to a rich person” (correct answer: true) is more difficult for conservatives because it challenges their ideology.

Measures of so-called “knowledge” in such a domain are easily skewed in a partisan manner by selection effects. This is a version of the “party of science” problem discussed previously. Whether the Democrats or the Republicans are the “party of science” depends entirely on how the issue in question is selected. The 17-item measure used by Klein was relatively balanced (8 items biased against leftists and 9 items biased against conservatives). With all the caveats in place about the difficulty of item matching, the weak conclusion that can be drawn is that existing research provides no evidence for the view that conservatives are deficient in the domain of economic knowledge—a domain critical for rational voting behavior.

Trump supporter

Similar sampling problems plague studies of conspiracy beliefs. These are important to study because perhaps the problem with the Trump voters is not that they have acquired too little knowledge but that they have acquired too much misinformation. The early research literature on the relation between ideology and conspiracy belief seemed to suggest that conspiratorial thinking was, in fact, more strongly associated with the political right. However, more recent research has suggested that this finding was simply a function of the distribution of specific conspiracy beliefs that were studied. Research using more balanced items has suggested that conspiracy beliefs are equally prevalent on the political right and left. We have confirmed this latter trend in the research literature in our own studies of our rational thinking measure, the Comprehensive Assessment of Rational Thinking (CART), which contains a subtest measuring the tendency to believe in conspiracy theories.

Our subtest covered a wide range of conspiratorial beliefs. Most importantly, our measure includes both right-wing and left-wing conspiracy items as well as a good number of items that spanned the political divide. Unlike some previous measures, it was not just a proxy for right-wing political attitudes. Some of the commonly studied conspiracies that we assessed were: the assassination of President John F. Kennedy, the 9/11 attacks, fluoridation, the moon landing, pharmaceutical industry plots, the spread of AIDS, oil industry plots, and Federal Reserve conspiracies. The results from our study were consistent with the more recent work on this issue. There was not a significant correlation between political ideology and the score on the conspiracy beliefs subtest of the CART. Overall then, there is no evidence from social science that Trump voters (overwhelmingly Republicans and independents) are epistemically less rational because they are less knowledgeable, or because they have acquired more misinformation.

Although there is no strong evidence that there are differences in the amount of knowledge that progressive and conservative voters have accumulated, it might be that the problem with conservatives (and Trump voters) is in the process of knowledge accumulation (in belief forming mechanisms). There are right and wrong ways to acquire knowledge. A person can acquire a true fact in the wrong manner. One problematic process in knowledge acquisition was mentioned above and has been extensively studied by cognitive psychologists—myside bias. If a person acquired a true political fact by a process of searching exclusively for things that support their political position, they may well be acquiring knowledge in the technical sense, but the knowledge base will be skewed and selective. It will have been acquired in the wrong way.

What then is the evidence that either Trump voters in particular, or Republicans, or conservatives in general, are more likely to display myside bias? The evidence, to the contrary, is that there is little relation between political beliefs and myside bias. Psychologists have a variety of paradigms for studying myside bias, but they all have the following general logic. Subjects might read an essay, or evaluate a purported experiment, or be presented with actual numerical data from an experiment (the results vary little across these different stimulus presentations). Sometimes the arguments and/or data are presented as supporting their political position (myside supporting) and other times it is presented as opposing (otherside supporting) their political position. Whether rating an essay or evaluating the quality of an experiment, subjects reliably rate myside supporting evidence higher than otherside supporting evidence. This is a manifestation of myside bias. Likewise, when evaluating actual numerical data, subjects view it as more convincing data when it is myside supporting than when it is otherside supporting.

Myside bias is ubiquitous, and it has been known for some time that it is displayed across the political spectrum, so the extreme argument that Republicans are characterized by myside bias and that Democrats are unbiased in how they view evidence was falsified years and years ago. But a weaker form of the hypothesis has remained in psychology and in general discourse for some time—that Republicans are more mysided in their thinking than Democrats (another, related, way of making the old “party of science” argument).

Recently, psychologist Peter Ditto of the University of California, Irvine and colleagues have greatly clarified the evidence on this weaker form of the argument. They meta-analyzed 41 experimental studies of partisan differences in myside bias that involved over 12,000 subjects. After amalgamating all of these studies and comparing an overall metric of myside bias, Ditto and colleagues concluded that the degree of partisan bias in these studies was quite similar for progressives and conservatives.

Thus, the lack of partisan differences found in actual acquired knowledge discussed previously is mirrored by a lack of partisan differences in the biasing process of myside thinking. These findings have not stopped the overwhelmingly liberal psychology world (see the writings of Jonathan Haidt, Lee Jussim and others on ideological bias in psychology) from seeking other process differences. In fact, there is a whole subspecialty area in social psychology devoted to showing that negative traits such as prejudice and stereotyping and unfairness are associated with the conservative temperament. There is even a theory—the “intrinsic thesis”—that hypothesizes that the increasing political polarization surrounding scientific issues is due to the “psychological deficiencies among conservatives as compared to liberals” (p. 36, Nisbet et al., Annals of the American Academy of Political and Social Science, 2015). Recently, there has been a flurry of psychological studies purporting to prove that conservatives are more prejudiced, less open-minded, and indeed less intelligent than liberals. The problem is that many of these studies have not been replicable, were poorly designed, or were designed and interpreted in a biased manner. [For additional research citations, see here.]

There’s Got to be Something Else Wrong Then: Promethean Claims of Broad Rationality Advocates

In summary, both in terms of the knowledge acquisition process and knowledge content, there is no strong evidence that the Trump voters were more epistemically irrational than the Clinton voters. In terms of both components of rationality—instrumental and epistemic—there is no support in the empirical literature for attributing a unique problem of rationality to Trump voters. Those who do not find this conclusion palatable might feel prone to object that the analysis so far seems somehow too narrow—that it leaves out some larger factors like, for instance, which overall goals and worldviews it is rational to have. Such skepticism would be right, in a sense. I have in fact adopted, deliberately, a somewhat narrow view of rationality because it is the view that is most easily tested, and most easily related to data that social scientists can produce. There are indeed broader versions of rationality, but they bring with them a host of problems—not the least of which is that they render claims that “X is irrational” untestable. Thus they provide no solace to the “Trump voters were irrational” sentiment because they hold out no hope of proving that the claim might be true.

In fact, with the Ted Cruz/Al Sharpton thought experiment, I was trying to illustrate the difficulty of evaluating goals and/or worldviews. The thought experiment, from the standpoint of a person with the GG world view, was a choice between a candidate with their worldview but temperamentally poorly suited to the presidency (Sharpton) and a candidate with an unpalatable worldview but much better suited to the presidency (Cruz). The point—to someone with the GG worldview—was not to show that one or the other choice was correct, but simply to illustrate the difficulty of this type of tradeoff and to provoke some associated recognition of the fact that the CC voter was presented with a similarly difficult tradeoff when faced with Trump/Clinton choice. It was to show that a Democrat feeling the attraction of Sharpton over Cruz should similarly understand the attraction of Trump over Clinton for Republicans. Even a Democrat feeling that they would actually pull the lever for Cruz on the basis of temperament/fitness for office would certainly understand the decision of a fellow Democrat to vote for Sharpton and would never call a fellow Democrat irrational for this choice.

Democrats who choose Sharpton in the thought experiment and who think that the Trump voters were irrational are showing a strong myside bias. They are signalling that they think they can do what philosophers cannot—discern which goals are irrational to have. Similarly, Republicans who voted for Trump and who think that the Democrats who pick Sharpton in the thought experiment are irrational are also showing a strong myside bias. They are signalling that they think they can do what philosophers cannot—discern which goals are irrational to have. Both groups are essentially saying that they are absolute arbiters of the correct worldview.

Needless to say, cognitive science does not endorse either of these myside responses. Both myside responders claiming irrationality on the other side are actually showing an extreme form of irrationality themselves. So claims that the Trump voters are irrational on this basis actually impeach the reasoning of the person making the claim. Or—another way to put it—it is itself a form of irrationality to think that one can discern that one’s own worldview is inherently superior to others (at least within the restricted set of viable Western political philosophies).

Critiques of Expressive Rationality Don’t Work Either (As Strategies for Calling the Trump Voters Irrational)

The GG and CC worldviews might be said to be carriers of utility values. The policies and actions they lead to have concrete effects on the country, resulting in economic and civic changes that people experience directly. But like many religious worldviews, people might be drawn to the GG and CC worldviews (and to voting for them), not because of the consumption utility they bring directly, but simply to express something about their values. Here we have entered the domain of some slightly different concepts of rationality that cognitive scientists have explored.

Psychologists and philosophers have both emphasized how a person’s decisions do more than convey utility to themselves but also send meaningful signals to other actors and symbolically reinforce the self-concept of the decision maker. The late philosopher Robert Nozick has discussed how symbolic actions that help maintain a valued concept of personhood are not irrational despite their lack of a causal connection to experienced utility. For many of us, the act of voting serves just this symbolic function. Many of us are aware that the direct utility we derive from the influence of our vote on the political system (a weight of one millionth or one hundred thousandth depending on the election) is less than the effort that it takes to vote, yet all the same we would never miss an election!

Voting has symbolic utility for us. It represents who we are. We are “the type of person” who takes voting seriously. We are expressing a value by voting. Nozick notes that we are apt to view a concern for symbolic utility as irrational. This is especially the case when the lack of a causal link between the symbolic action and the actual outcome has become manifestly obvious yet the symbolic action continues to be performed. Nozick mentions various anti-drug measures as possibly falling in this category. In some cases, evidence has accumulated to indicate that an anti-drug program does not have the causal effect of reducing actual drug use, but the program is continued because it has become the symbol of our concern for stopping drug use.

Although it would be easy to classify many instances of acts carried out because of symbolic utility as irrational because of a lack of causal connection to the outcome actually bearing the utility, Nozick warns that we need to be cautious and selective in removing symbolic actions from our lives. No one wants to live a life without any symbolic meaning. Nozick’s notion of symbolic utility has affinities with many similar concepts in social science. Political economist Shaun Hargreaves Heap argues for distinguishing what he terms expressive rationality from instrumental rationality. When engaged in expressively rational actions, people are not optimizing anything but are instead attempting to articulate and explore their values. Yale political scientist Dan Kahan has written extensively about identity-protective cognition, whereby people process information not to ascertain the truth or to maximize consumption, but to protect one’s status in an affinity group or to maintain a cultural identification.

Choices based on such values may actually lower the personal welfare (in terms of direct consumption utility) of the individual, as when we vote for a political candidate who will act against our material interests but who will express other societal values that we treasure. The concept of ethical preferences in economics has the same function of severing the link between observed choice and consumption utility. The boycott of nonunion grapes in the 1970s, the boycott of South African products in the 1980s, and the interest in fair-trade products that emerged in the 1990s are examples of ethical preferences affecting people’s choices.

These expressive choices, symbolic choices, and ethical preferences—are they rational? It is rarely easy to say. For example, it is easy to see how expressive choices can run amok in the escalating and reverberating circuits of “meaning making” in the modern world. As Nozick noted years ago:

[C]onflicts may quickly come to involve symbolic meanings that, by escalating the importance of the issues, induce violence. The dangers to be specially avoided concern situations where the causal consequences of an action are extremely negative yet the positive symbolic meaning is so great that the action is done nevertheless.

This quote is from a 1993 book. In the wake of 9/11, Nozick’s warning is chilling. But calling to mind 9/11 is stacking the deck. It is rarely so easy to label a symbolic act dysfunctional. And the difficulty of evaluating an expressive action is just my point, because voting is highly symbolic. It is expressive of one’s identity and cultural commitments. And in the eyes of many voters, it is an act with ethical implications—not just voting itself, but voting a particular way. At the level of elections, there is no rational way of assessing the tradeoff between the worth of an expressive signal and its negative consequences. Voting for Trump to signal opposition to the agenda of the Democrats even if a voter did not like Trump cannot be labelled rational or irrational. In fact, it cannot even be labelled any less rational than voting for Clinton to signal opposition to the agenda of the Republicans even if a voter did not like Clinton. In fact, I would go further. Thinking that one can evaluate the rationality of someone else’s expressive act is itself a form of irrationality. Actually, it probably reflects a famous effect studied by psychologists: the bias blind spot.

The bias blind spot is the label for the finding that it is relatively easy for people to recognize bias in the decisions of others, but it is difficult to detect bias in their own judgments. This is probably just what is happening in the political domain when people feel that they can judge exactly when having expressive goals is rational. The reason is that judgments of another partisan’s seeking of symbolic utility or expressive rationality are invariably saturated with myside bias. Why your own side would choose to signal a value at a utility cost seems perfectly obvious, yet when your political opponents do it, it seems utterly irrational. Republicans can clearly see the irrationality of Democratic city councils divesting city funds in corporations disliked on the left (often at a cost in real return on city-invested dollars). Democrats likewise denigrate the enthusiasm of Republicans for “just say no” campaigns surrounding drugs and sex and point out the irrationality of the Republicans not caring if the programs work or not. Such judgments are overwhelmingly determined by myside bias. The other side is judged deeply irrational when they abandon cost-benefit analysis to signal a value choice, but when my own side sacrifices utility, money, or outcome goals in order to signal a value, that is OK because our values are right (seems to be the reasoning!).

The Bottom Line

I am afraid that my Democratic friends are just going to have to reconcile themselves to the conclusion that the cognitive science of rationality does not support their judgment of the Trump voters. You can say whatever you want about the rationality or irrationality of Trump himself, but cognitive science does not support the claim that his voters were irrational—or, more specifically, that they were any less rational than the Clinton voters. Politics is not the place to look for objective rightness or wrongness—and that is what judgments about the rationality of voting entail. Our judgments in this domain are uniquely susceptible to myside bias.

Many of our most contentious political issues hinge on values and culture rather than facts. That may be a good thing. It could be signalling that our society has already handled the easiest issues—those that can be solved by educating everyone to accept the same facts and then implementing the obvious solution that follows from these facts. We may have achieved a social structure that is so optimized that the remaining disputes revolve largely around values and cultural choices. Rather than calling the Trump voters irrational, it might be a better idea to engage with their Country and Citizen cultural concerns and treat them as equally valid and rational as the Global and Groups cultural concerns that largely drove the Clinton voters.