The current issue of The Atlantic carries a fascinating summary of “The Confidence Code” by Katty Kay and Claire Shipman. The essay runs through the evidence suggesting that women tend to have too little self-confidence. When asked how well they did on tests, women tend to estimate that they got fewer answers correct than they actually did. In one British study, a business school professor asked students how much they would deserve to earn five years after graduation. The women’s estimates were 20 percent lower than the men’s.

It’s interesting to read the evidence as a guy, especially if you’re a self-aggrandizing pundit who covers politics and public life. I almost never see problems caused by under-confidence, but I see (and create) problems related to overconfidence every day.

Much of the recent psychological research also suggests that overconfidence is our main cognitive problem, not the reverse. Daniel Kahneman’s book “Thinking, Fast and Slow” describes an exhaustive collection of experiments demonstrating how often people come to conclusions confidently and wrongly. When asked to estimate if more murders happen in Detroit or in Michigan, most people give higher estimates for Detroit even though every murder in Detroit also happens in Michigan.

Dan Ariely’s work shows how consistently we overpraise our virtues and rationalize our faults so we can think too highly of ourselves. Most of us call ourselves honest. But, in fact, most people regularly cheat in small ways, when the situation is right.

So my first reaction when reading of female under-confidence is not simply that this is a problem. It’s to ask, how can we inject more of this self-doubt and self-policing into the wider culture. How can each of us get a better mixture of “female” self-doubt and “male” self-assertion?

But my second reaction is to notice that people are phenomenally terrible at estimating their own self-worth. Some Americans seem to value themselves ridiculously too little while others value themselves ridiculously too highly.

The self-help books try to boost the “confidence” part of self-confidence, but the real problem is the “self” part. The self, as writers have noticed for centuries, is an unstable, fickle, vain and variable thing. Hundreds of years ago, David Hume noticed that when he tried to enter into what he called his most intimate self, he always stumbled on some particular perception or another. He never could catch himself without a perception of something else, and he never could see himself, only the perception.

When you try to come up with a feeling for self-confidence, you are trying to peer into a myriad of ever-changing mental systems, most of them below the level of awareness. Instead of coming up with a real thing, which can reliably be called self-confidence, you’re just conjuring an abstraction. In the very act of trying to think about self-confidence, your vanity is creating this ego that is unstable and ethereal, and is thus painfully fragile, defensive, boasting and sensitive to sleights.

If you want to talk about something real, it’s probably a mistake to use a suspect concept like self-confidence, which is self-oriented. It’s probably a better idea to think about competence, which is task-oriented. If you ask, “Am I competent?” at least you are measuring yourself according to the standards of a specific domain.

The person with the self-confidence mind-set starts thinking about his own intrinsic state. The person who sees herself as the instrument for performing a task thinks about some external thing that needs doing. The person with the confidence mind-set is like the painfully self-conscious person at a dinner party who asks, “How am I coming across?” The person with an instrumentalist mind-set is serving a craft and asks “What does this specific job require?” The person with a confidence mind-set is told “Believe in yourself.” This arouses all sorts of historical prejudices and social stereotypes. The person with an instrumentalist mind-set is told “Look accurately at what you have done.”

One of the hard things in life is learning to ask questions that you can actually answer. For example, if you are thinking about taking a job, it’s probably foolish to ask, “What future opportunities will this lead to?” You can’t know. It’s probably better to ask, “Will going to this workplace be rewarding day to day?” which is more concrete. If you are getting married, it’s probably foolish to ask an unknowable question like, “Will this person make me happy for 50 years?” It’s probably smarter to ask, “Is this person admirable enough that I want to live my life as an offering to them?” You can at least glimpse another’s habits here and now.

Similarly, if you start thinking about your self-confidence, you will just be inventing a self-referential story. It’s probably easier to go through life focusing on what specifically needs doing, rooted in a set of external obligations and criteria and thus quieting the self.