Neoclassical economic theory rules out systematic errors in consumption choice. According to the basic view, individuals know what they choose. They are able to predict how much utility an activity or a good produces for them now and in the future and they can maximize their utility. This implies that behavior reveals consistent preferences. This approach makes it impossible to detect and understand sub-optimal consumption decisions, due to problems of self-control and the misprediction of utility. We propose the economics of happiness as a methodological approach to study these phenomena. Based on proxy measures for experienced utility, it is, in principle, possible to directly address whether some observed behavior is sub-optimal and is therefore reducing a person's well-being. We discuss recent evidence on smoking and eating habits, TV viewing and commuting choice.