David Brooks / The New York Times

If you want to deter crime, it seems that you’d want to lengthen prison sentences so that criminals would face steeper costs. In fact, a mountain of research shows that increases in prison terms have done nothing to deter crime. Criminals, like the rest of us, aren’t much influenced by things they might have to experience far in the future.

If a police officer witnesses the death of his partner, you’d want to quickly send in a grief counselor. In fact, immediate counseling freezes and fortifies memories, making the aftershocks more damaging.

If you want to get people to vote more, you’d want to tell them what a problem low turnout is. In fact, if you want people to vote, tell them everybody else is already voting and they should join the club. Voting is mostly about social membership.

These are three examples of policies and practices that are based on bad psychology. The list of examples could go on and fill this page. That’s because we spend trillions of dollars putting policies and practices into place, but most of these efforts are based on the crudest possible psychological guesswork.

Fortunately, people in the behavioral sciences are putting policies to the test.

One of the things we know is that seemingly trivial changes can have big effects. People who are presented with a wide variety of choices of, say, yogurt, will eat more than people who are presented with a small array of choices or no choice. People who were randomly given a short, wide 22-ounce glass, poured 88 percent more juice or soda into it than people who were offered a tall, narrow 22-ounce glass, but they believed they only poured in half as much as they actually did.

Sometimes the behavioral research leads us to completely change how we think about an issue. For example, many of our anti-discrimination policies focus on finding the bad apples who are explicitly prejudiced. In fact, the serious discrimination is implicit, subtle and nearly universal. Both blacks and whites subtly try to get a white partner when asked to team up to do an intellectually difficult task. In computer shooting simulations, both black and white participants were more likely to think black figures were armed. In emergency rooms, whites are pervasively given stronger painkillers than blacks or Hispanics. Clearly, we should spend more effort rigging situations to reduce universal, unconscious racism.

The research is also leading to new policy approaches. The most famous involve default settings. Roughly 98 percent of people take part in organ donor programs in European countries where you have to check a box to opt out. Only 10 percent or 20 percent take part in neighboring countries where you have to check a box to opt in.

In one clever program, dieters were told to phone in their weight to a nurse daily. Every day they called, they got an encouraging text and a lottery ticket, with a chance of winning a small amount. These dieters lost three times more weight than people who didn’t get tickets.

My problem with these efforts is that they are still so modest. What about the big problems? How do we get people to restrain government now so that debt down the road won’t be so ruinous? How do we calculate the multiplier effects of tax cuts or spending increases among different subgroups, or under different emotional conditions? How do we rig budget negotiations so participants can actually come to a deal?

These are the big questions, and most of our policies rely on crude folk psychology from a few politicians. But there’s hope. As Brian Wansink notes in Eldar Shafir’s “The Behavioral Foundations of Public Policy,” the 20th century saw great gains in sanitation and public health. The 21st century could be a great period for behavior change.