Behavioral economics has so far largely avoided discussing the psychological origins of preferences, as well as their relation to needs. This has not only restricted interdisciplinary exchange, but also significantly limits the predictive capabilities of models. For example, the revealed preference approach can only reliably predict repeating choices, while needing large amounts of observations for calibration. In this paper, I show how unifying preferences with the psychological concept of needs strengthens economic models, by developing a decision-making framework for well-being assessment and choice prediction. To present the direct merit of this approach, I show how this framework yields a systematic approximation scheme, which is able to solve limitations of current approaches by describing new alternatives, non-repeating choices, or otherwise unobservable desires. Meanwhile, the approximation scheme requires less observations on an individual level than current approaches. I achieve this by constructing a hierarchical dependency between human motivations and preferences through the language of needs. I show the basic feasibility of the approximation scheme through simulations on random populations. In practice, the framework is applicable in situations where individuals exert choices only once and measuring preferences is expensive, like evaluating policy proposals or predicting decisions under technological change
