The Promise and Limits of Rationality

Steven Pinker is the greatest Whig theorist of our day. In book after book, he argues that the world is improving through the continual application of our rational intelligence. In The Better Angels of Our Nature, he showed that violence—whether in the form of crime or wars—has been on the decline for thousands of years. In Enlightenment Now, he argued that humankind has never been better off, despite persistent claims that things are getting worse. In both cases, he sees human reason as the engine of social amelioration, both in the technology it creates and the cooperation it commends. All we need for more progress is better science and more peaceful collaboration.

In his latest, Rationality: What It Is, Why It Seems Scarce and Why It Matters, Pinker explores human rationality, this dynamo of improvement, in more detail. In his view, the limits to reason are what hold us back. His overall thesis is that man is a pretty rational creature. Whether in a hunter-gatherer society or in the modern world, people make inferences from the evidence to achieve their goals. But human nature also restricts reason’s domain and effectiveness.

The book is best when articulating the nature and importance of rationality. It is weaker in discussing reason’s frailties, particularly in group decision-making. For instance, the book never mentions the disturbing political reality that it is rational to be ignorant of democratic policy, because one’s votes are so unlikely to make a difference. As a result, rationality is likely not enough to sustain a good polity and generate social improvement. For society to flourish, it is necessary for citizens to be imbued with civic virtue beyond any rational calculus. 

Pinker does nicely explain a number of inherent weaknesses in individual rationality, however. First, our minds have not developed as unbiased calculators because bias helped us survive in past eras of our evolution. The context that led to accurate shortcuts the past few hundred thousand years may distort our assessments in modernity. For instance, most of us have an unreasonably strong aversion to losses. From a strictly rational perspective, one should be willing to endure the same risk either to make a gain or to avoid a loss of the same amount. But people are usually more willing to take the similar risk to avoid the loss. The reason is that in the evolutionary era where people lived on the margin of existence, even small losses could lead to disastrous outcomes, even death, and it was worth taking greater risks to avoid them.

Second, people tend to embrace arguments and evidence that are in their own narrow interest rather than those that are objectively stronger. Indeed many avoid even going down a train of reasoning for fear that they will not like where it takes them. Again, the explanation for such bias and motivated reasoning may be evolutionary. As Pinker says, we evolved not “as intuitive scientists but as intuitive lawyers.” There were greater returns to winning an argument than to objective truth.

Third, determining the truth in social affairs is harder than in natural science. We cannot run the social conditions of the world over again, changing its conditions to isolate the causes of a social phenomenon. Causation is ultimately about counterfactuals. If A causes B, it follows that if A does not happen, neither will B given otherwise exactly similar initial conditions. But precisely defined counterfactual social worlds live only in our imagination. Thus, motivated reasoning inevitably dominates social science more than natural science. Not only are the real-world stakes in social disputes generally more immediate and personal (what will be the effect of higher taxes on me) than in purely scientific ones (does this gene cause this disease), but the effects of policy are genuinely hard to pin down.

Much of the book consists of Pinker’s descriptions of different forms of rationality and the extent and reasons for our departure from its dictates. He is an excellent expositor, whether discussing pure logic, Bayesian inference, or the heuristic flaws in human reasoning. For instance, he gives a lucid explanation of the famous Monty Hall problem that has bedeviled even great mathematicians. In the problem, a game show host (Monty Hall) promises a lucrative prize behind one of three doors but leaves only goats behind the other two. He then invites you to choose a door. After you do, Monty opens one of the other two doors and shows you the goat behind it. He then invites you to switch your door, if you wish. Should you do so? The answer is yes, although most people don’t bother to change their initial choice.

The reason is that when Monty opens a door, it provides actionable new information. Monty could have opened either of two doors, but he always chooses the one with the goat. That makes it more likely that the one he did not open has the prize than the one you chose first. Monty knows where the prize is hidden when he opens the door, and you should act on his knowledge.

But as Pinker also explains, our mistake is not surprising. We do not generally enjoy the company of an omniscient being like Monty to give us clues about what to do. Our thinking is primed for the ordinary situation where the probability of finding a prize behind the first door is not changed by a random opening of one of the other two.

Pinker also outlines common problems in our rationality that have substantial political effects. One is what is called availability bias. We base our views of how common a phenomenon is on how often we hear about it, not on the hard data of likelihood. As a result, people wildly overestimate the dangers of plane crashes and shark attacks that gain plenty of coverage in the news while underestimating the underreported dangers of household accidents and driving. And this bias can drive politics. Three Mile Island and Chernobyl were top news stories, so much that they remain bywords for nuclear power plant disasters today. But as Pinker notes, such power plants provide safe and clean energy compared to the alternatives. The pollution created by oil and gas kills thousands of people every year. But these deaths go unreported and make no difference to policy. The same is true of the deaths caused by police. Such deaths understandably get enormous attention. But the hugely greater number of people killed by criminals generally becomes a mere background statistic.

What Pinker wants above all is evidence-based opinions and policy. Experts and citizens alike should update their prior beliefs based on a dispassionate weighing of new evidence.

Experts are not as subject to availability bias. They are paid to rely on data, not anecdotes. But democracy means that people driven by anecdotes, not experts driven by data, often make more policy. And experts are no panacea either, because they can be infected by political prejudice, what Pinker calls “myside bias.” Likely because of evolutionary reasons, people are eager to be part of a team and with the decline of religion and regionalism, the left and right have become the dominant axes around which tribes are formed. For a Whig such as Pinker, decisions driven either by a totalizing ideology or by anecdote are equally anathema.

I was surprised that Pinker did not discuss the use of structural mechanisms to discipline ideology and anecdote. One with potential is prediction markets where citizens and experts alike put their money where their mouth is. That gives people incentives to leave ideology and misleading anecdotes aside at least in their beliefs about future occurrences. If prediction markets are widely enough disseminated, they militate against availability bias, because people will become more likely to base their judgments on rational predictions of the future rather than scatter shots from the past.

What Pinker wants above all is evidence-based opinions and policy. Experts and citizens alike should update their prior beliefs based on a dispassionate weighing of new evidence. Thus, even if people have different priors, the injection of empirical evidence to social debates should ideally lead to more convergence in views.

Unfortunately, that meeting of the minds often does not happen, particularly in politics. And the signal weakness of Pinker’s book is that he does not fully appreciate the obstacles to the regular updating of citizens’ views, nor does he consider the institutional solutions for overcoming them.

Pinker does recognize the tragedy of the commons that is inherent in politics. Because motivated reasoning may bring us strategic benefits, “each of us has motive to prefer our truth” rather than “the truth” as when we spin false tales about how some subsidy that benefits only people like us is really in the public interest. Even if we would all be better off in the long run if we followed true beliefs, there is no mechanism for enforcing a true, collective understanding. All too true.

But even if people were wholly well-motivated, they would lack incentives to update their beliefs about the collective goods that are at the core of democracy. One’s vote is extremely unlikely to make the difference between the victory and defeat of the candidate with the better evidence-based policy. Therefore, embarrassingly for a devotee of rationality, it is in one sense rational to be ignorant of information about most political disputes and thus be impervious to updating. The payoff of spending time on one’s own enterprises is far greater. Personal returns may be uncertain, but the returns from grazing on policy debates are infinitesimal.

The reality of this kind of “rational ignorance” (a well-known concept in public choice literature distinct from Pinker’s own use of the same term for avoiding information that would help overcome bias and make better choices) creates two kinds of problems for his analysis. First, rationality is not enough to solve the greatest collective action problem of politics. A citizen needs a sense of duty to become informed. It cannot simply be derived from updating beliefs about factual truths, even the truth that democracy would work better if citizens update. For most citizens, such duties are more likely to come from patriotism or virtues that are not, strictly speaking, rational. Thus, there is a paradox about social rationality: it needs emotional attachments to flourish.

Second, rational ignorance suggests that the central issue for any society that wants to have more belief-updating in politics is institutional design, not injunctions to be more rational. Pinker is a psychologist, not a political scientist, but here he writes from the perspective of general social science, not from a single discipline. It would have been useful for him to canvass the rich literature on how to use political structure to improve rational decision-making in a democracy. For instance, limiting the activities of the national government helps people focus on relatively few types of policies, when they evaluate national officials, making it more likely that they will update on relevant information. Federalism too creates incentives for updating. An individual can capture the benefits of learning about good policies in other states. Free movement in turn gives officials more incentive to update so they can attract people and businesses from other states, thereby raising their status.

Despite some missed opportunities, Rationality remains a fine survey of the benefits and heuristic limits of all-too-human reason. It is also a model of writing for a popular audience. Pinker’s prose is in many cases memorable and pithy, as when he writes: “The press is an availability machine. It serves up anecdotes which feed our impressions in a way that is guaranteed to mislead.” As an apostle of reason, Pinker sometimes ignores the need for virtue to make its application more effective, but no one has written a clearer case for its salience to the flourishing of mankind.