Laurie Paul’s fascinating paper on the rationality of choosing to have children has already received a great deal of attention in the blogosphere. Perhaps everything worth saying has already been said. But I wanted to point to some evidence that we ought not place the kind of weight on people’s experiences, in the context of assessing how their choices have gone, that Paul suggests we should.

Paul argues that the decision whether to have children cannot be made rationally. More specifically, the assessment of whether having children will be deeply fulfilling or deeply frustrating cannot be made rationally, because having a child is a transformative experience: one can’t know what it will be like to have a child until one does, and therefore one can’t know what the experience will be like for you. Now clearly there is a trivial sense in which this is true, and just as clearly Paul doesn’t mean this trivial sense. Trivially, you can’t fully know what it is like to do something until you have done it. For instance, you can’t know what it is like to spin around until you are so dizzy you fall over while listening to death metal in a room that smells of boiled cabbage, until you have tried it. But it’s not that trivial sense that Paul is concerned with. Rather, she claims that the experience of having a child is transformative, in a deep way. One’s values change, one very sense of self changes. And that fact entails that we can’t rationally choose whether to have children, on the grounds of what it will or would be like to have them.

In making a rational decision, we use some kind of rough measure of the expected utility of the options. The expected utility of an option is the degree of satisfaction choosing it brings, multiplied by the probability of achieving that satisfaction. So the expected utility of my choosing whether to change jobs is some complicated function of the overall satisfaction I think I will get from the new job, adjusted for the probability that I might be wrong or things might go wrong, compared to the overall satisfaction of other options similarly adjusted. The problem with transformative experiences is that we can’t even begin to make these kinds of calculations. I can’t know what (dis)satisfaction that choice might bring me if the choice will change not just my situation but, in ways I can’t anticipate, what counts as satisfying for me. So the choice is very much a leap into the dark (interestingly, Martha Nussbaum has argued on similar grounds that it is actually irrational to choose transformative experiences. She claims that Odysseus acted rationally in turning down Calypso’s offer of immortality, because he could not know what it would be like to be immortal).

The worry I have about this line of argument is this: there is a great deal of evidence concerning people’s judgments of how satisfied they are with choices, and that evidence suggests that we ought to place relatively little weight on these judgments. 50 years of research on cognitive dissonance shows that under a variety of conditions, people adjust their beliefs to accord with their actions, when they do not receive sufficient satisfaction from their actions without the adjustment. So, for instance, in the classic essay writing task, subjects who are induced to write an essay defending a view that they are unlikely to agree with (eg, college students induced to write essays defending rises in tuition fees for their college) will later tend to agree that fees should rise, when, and only when, they are not paid enough by the experimenters to explain their agreeing to write the essay. Cognitive dissonance produced by an awareness of the attractions of forgone opportunities produces a phenomenon known as the spreading of alternatives; the positive qualities of the option chosen are magnified and those of forgone options are depreciated. So we can expect people’s self-rated satisfaction with the choice of having children to be higher than the rewards they receive from it, on average. Those who find it rewarding will rate it as rewarding because they find it rewarding, whereas others will rate it as rewarding because they don’t find it rewarding. Because the choice is irrevocable and momentous, the effects will be pronounced.

One might respond that the process whereby people come to believe that they find something rewarding doesn’t matter. The fact is, they do find it rewarding. But it is not at all clear that people induced by cognitive dissonance to alter their perception of the value of a choice really find it rewarding. Perhaps they only think they do. This claim would explain the dissociation between people’s expressed satisfaction with the choice to have children and the fact that parents have lower rates of life satisfaction, on average, than the childless, and that parents rate the time they spend in childcare as among the least satisfying in their lives.