This book is entertaining, highly readable, and ultimately extremely frustrating. Gilbert summarizes scores of psychological and sociological studies on "happiness," and concludes that we humans are spectacularly bad at predicting what will make us happy. But the evidence he relies on seems, for the most part, one-dimensional and even trite, and his unquestioning reliance on this research makes his own theory of happiness seem shallow.

The biggest problem is his failure to address what he or the research subjects mean by "happiness." The same word is used throughout the book to refer to, say, the momentary pleasure one gets from a bite of ice cream, as well as to a more profound and lasting sense of contentment and meaning over time.

Although he acknowleges the definitional problem in the first chapter, he fails to conduct any systematic inquiry into what research subjects might mean when they say they are "happy" in response to a particular research question. It seems obvious subjects might apply a different definition of happiness when asked to predict their future happiness level than when asked to rate their mood at a particular moment in time. Gilbert's failure to consider this possibility -- or to explain how the research controls for it -- undermines the overall persuasiveness of his argument, and leads one to suspect many of his conclusions would be contradicted by more precisely tailored research (or a more rigorous analysis of the results).

For example, Gilbert says that while most prospective mothers predict that having children will enhance their happiness, the research shows that those predictions are wrong. He reaches this conclusion by relying on studies in which mothers were asked how they were feeling at particular moments throughout the day. The responses indicated, for the most part, that the mothers were "less happy when taking care of their children than when eating, exercising, shopping, napping or watching television." These studies, he claims, show not only that prospective mothers are wrong when they predict that having children will increase their happiness, but also belie the mothers' claims later in life that having children made them happy.

One problem with this is, of course, that he never tells us what he or the research subjects mean by "happiness." In the particular example described above, I strongly suspect that if the subjects were given a chance to fully articulate their feelings, their predictions would be fairly close to reality. Most would-be parents would probably say they expect that having children will add depth and meaning to their lives, and will give rise both to moments of great joy and to hours of tedium and frustration. This prediction, for the most part, will probably turn out to be true. Moreover, most people with this belief would probably answer "yes" to a survey question asking whether they believed that having children would increase their happiness. But these very same people, when asked how they felt at a particular moment, might well respond "frustrated," "bored," "overwhelmed," or even "miserable" -- especially if the question were asked at the end of a long afternoon with a cranky toddler. Gilbert does not seem to consider that short-term displeasure can be entirely consistent with long-term satisfaction -- or that a meaning and satisfaction -- and ultimately happiness -- often emerge directly from -- not despite of -- a struggle.

To give another example, Gilbert says that most people would probably predict that being jilted at the altar would cause them tremendous unhappiness, but that most people who actually have been jilted probably think that it was "the best thing that ever happened to them." In truth, the actual experience is probably both of those things, and if allowed to elaborate on their feelings, most people would probably imagine they would feel close to how they would actually feel. The feelings would probably be something along the following lines: "it was a tremendously humiliating experience that caused me great embarassment and pain, and in fact, still causes me great embarassment and pain, but ultimately I'm glad I discovered what a selfish jerk my fiance was before I married him." Such a bride might accurately state that being jilted at the altar was the best and worst thing that ever happened to her.

Gilbert does not seem to acknowledge that such apparently contradictory responses can, in fact, be entirely consistent. The most painful experiences might ultimately have the most meaning precisely because of their intensity, and an experience that is accurately predicted to cause great suffering might utlimately become a catalyst for positive change. To put it more bluntly, the cancer that caused a shift in your world view might be the best thing that ever happened to you, but you might lay down your own life before you would allow such suffering to be inflicted on the ones you love. The apparent contradictions between predicted and actual feelings that are the focus of Gilbert's book may well reflect more the inadequacies of social science surveys than any deep-seated delusions about what ultimately will give us a sense of meaning, satisfaction and contentment in our lives.

All and all, this is a book with a lot of fun vignettes, but without the depth that would make this a truly satisfying read.