In introductory treatments of economics, one often encounters a distinction drawn between what is called positive economics and what is called normative economics. In these names — and in typical discussion — there are problems.

The meaning of positive here is restricted to fact, as opposed to speculation. Now, on the one hand, supposedly positive economics, like all attempts by human beings to understand the world, is permeated by speculations, which in scientific effort are hypotheses. (The philosophic movement called positivism arose with incompetent aspirations.) On the other hand, contrasting the normative with something called positive entails an implication, insinuation, or declaration that the normative cannot be placed on as solid a foundation as the rest of our understanding. Sometimes a lack of present agreement is treated as if proof that there is no objective ethical truth; sometimes the question is just begged. In any case, the distinction is irrational.

Instead using the terms descriptive and prescriptive steps away from the worst aspect of using positive, though it would be less corrosive to refer to non-prescriptive economics as, well, non-prescriptive or as non-normative.

However, in behavioral science, elements drawn from prescriptive theory are often useful non-prescriptively, either as approximations or as bounding cases. Economic rationality and expected-utility maximization (the latter sometimes conflated with the former) are such elements.

Some economists would not even recognize economic rationality or expected-utility maximization as prescriptive in any case, because they are meta-preferential — they express a preference for structures of preference that have ordering properties such as transitivity and acyclicity, but say nothing about ultimate objectives and thus, in themselves, say nothing about whether one should prefer tomatoes to apples or life over death.

The prescriptive arguments for economic rationality and for expected-utility maximization are to the effect that those who conform realize more of their objectives — regardless of what those objectives might be — than those who do not, with it usually treated as tautologic that one desires such maximization.[1]

The non-prescriptive arguments for economic rationality and for expected-utility maximization as approximations note that these are relatively tractable models of behavior for which evolutionary dynamics will select. Because the models are taken from prescriptive work, some people mistake or misrepresent any use of them as necessarily prescriptive, but the claim is neither that social or other biologic evolution ought to select for something approximated by such behavior nor that agents ought to engage in the behavior for which evolution selects. (If anything, what is illuminated is that evolution selects for a propensity to such prescriptions!)

I endorse use of these models as tractable approximations in many cases, but I also embrace use of a weaker notion of economic rationality as a bounding case. A boundary of economic outcomes is given by considering what those outcomes would be were agents economically rational.

Behavioral economics concerns itself with when-and-how people actually behave, and especially with failures of the aforementioned models. Although this research is not what I do, I acknowledge its value. However, a great deal of what passes for behavioral economics involves an inferential leap from identifying a real or apparent deviation of behavior from one of these models to a conclusion that this-or-that result could be obtained by state intervention, with the researcher looking away from any proper examination of the behavior of agents determining practices of the state. Behavioral economics is thus used as the motte for a statist bailey. Additionally, even behavioral researchers with no apparent statist agenda often fail to recognize when behavior that seems at odds with these models is or may be instead at odds with some presumption of the researcher.[2]

[1] The main-stream of economic theory treats completeness of preferences as a feature of economic rationality but I've never seen a prescriptive argument even attempted for this feature. The prescriptive cases for transitivity and for acyclicity seem to presume an absence of conflicting, prior meta-preferences. The prescriptive argument for expected-utility maximization is especially problematic.

[2] While I have problems with some of the work and with much of the rhetoric of Gerd Gigerenzer, he has ably identified important cases of such failure on the part of researchers.