Decision scientists often distinguish between normative and descriptive models of decision making, and less often add the additional distinction between normative and prescriptive models.
Normative Models
Normative models are models or theories about processes in judgment and decision making that putatively result in making the best choice (optimizing). While the models describe a process, it is not always considered important that decision makers use the same process, so long as their decisions correspond to those that would have been made if the normative process had been used, in recognition of the adage that there is (often) more than one way to skin a cat. “Putatively” is italicized above to emphasize the fact that that normative models and theories are just that, models and theories. They may in fact not be normative. That leads to a sometimes confusing state of affairs when a decision scientist wants to argue that a normative process is not normative or that a bias is in fact normative. That confusion aside, there remains debate among decision scientists about the normative status of most normative models, although the dissenting voices are sometimes few and far between.
Descriptive Models
Descriptive models refer to processes people actually use, independent of whether or not those processes are normative. Traditional economic theory has a long history of assuming the choices people actually make (descriptive processes) are in fact normative, and working backwards from that assumption to derive the subjective value (utilityIn decision science, "utility" refers to a quantitative measure representing an individual's preference or satisfaction with respect to a particular outcome. In this framework, utility...) that different choices must have. Psychologists studying judgment and decision making, including behavioral economists, on the other hand, have a long history of developing creative ways to demonstrate that descriptive models are often non-normative, with the most famous examples coming out of the heuristics and biasesIn decision science, bias refers to systematic and predictable deviations from accepted models of rationality. In scientific language, "systematic" and "predictable" are not meant to... tradition that has led to more than one Nobel Prize in economics.
Prescriptive Models
The additional nod to prescriptive models is in recognition of the fact that people cannot realistically be expected to use some normative models given limitations in cognitive capacity or available time or other resources (see Bounded RationalityThe concept of bounded rationality refers to the idea that human rationality is constrained by (1) available information, (2) cognitive / information processing limitations, and...). Prescriptive models refer to processes that can be recommended to decision makers given the reality of these real-world constraints or bounds, processes that may not be optimal if such constraints were removed but that are a reasonably close approximation of optimality, and may in fact be optimal given the boundary conditions of real-world decisions. Basic strategy and card countingBasic Strategy The basic strategy, or just basic, is the best way to play each hand without keeping track of cards removed from play during... are good examples of prescriptive models. No known human is capable of calculating optimal blackjackBlackjack, also referred to as 21, is a popular casino card game where players attempt to have a hand value closer to 21 than the... More play without the help of a computer—and prior to the 1950s even with the help of a computer. Basic strategy and card counting reliably approximate these calculations given real-world constraints, seeking to get blackjack players as close to optimal as possible given the casino context. There remains extensive disagreement as to whether and when prescriptive models should be considered normative or whether un-enactable normative models are a reasonable standard of rationality. That will be a recurring theme throughout the Substack posts.