Three Ways of Analyzing Decisions

These three ways of analyzing information help us understand different types of statements. Descriptive statements tell us what actually is or happens (like 'people often check their phones first thing in the morning'). Normative statements express values about what should be (like 'we should spend less time on devices'). Prescriptive statements give specific instructions for action (like 'turn off your phone during dinner'). Notice this: Recognizing which type of statement you're hearing or making helps you understand what kind of reasoning you're using.

These three analytical approaches help us separate facts, values, and recommendations. Descriptive thinking focuses on observing reality and explaining patterns without judgment. It answers 'what is' questions by documenting actual behavior, outcomes, or relationships. Normative thinking deals with values, ethics, and ideals, answering 'what should be' questions based on principles, standards, or goals. It establishes criteria for evaluating whether something is good, right, or desirable.

Prescriptive thinking builds on both to recommend specific actions, answering 'what should we do' questions. It provides clear guidance by connecting our values to our actions. For example, a doctor might observe that patients who exercise regularly recover faster (descriptive), believe that good health is important (normative), and then recommend specific exercises to a patient (prescriptive).

In business, a manager might notice that remote workers report higher satisfaction (descriptive), value both flexibility and collaboration (normative), and then create a hybrid work schedule (prescriptive). This framework helps avoid confusing observations with recommendations and makes our reasoning more transparent and effective.

The distinction between normative and descriptive statements has deep philosophical roots, particularly in the fact-value distinction articulated by David Hume and later by philosophers of science. This separation is fundamental to clear thinking across disciplines, from economics to ethics to artificial intelligence.

In economics, the normative-descriptive distinction separates positive economics (describing how economies work) from welfare economics (making value judgments about how economies should work) [1]. In ethics, the is-ought problem highlighted by Hume shows that descriptive statements about what 'is' cannot logically determine normative statements about what 'ought' to be without bridging premises [2].

AI researchers use these distinctions to categorize systems: descriptive AI predicts outcomes, normative AI evaluates options, and prescriptive AI recommends actions [3].

Hume (1739): Demonstrated that facts about the world cannot logically determine moral values without additional normative premises [1].

Simon (1997): Showed how decision-making requires both descriptive models of how people actually choose and normative models of how they should choose [2].

Russell & Norvig (2020): Explain how AI systems separate perception (descriptive), evaluation (normative), and action (prescriptive) capabilities [3].

The distinction between these three modes can blur in practice. What appears descriptive often carries hidden normative assumptions through word choice or framing. Prescriptive recommendations sometimes masquerade as descriptive observations to seem more objective. Additionally, people frequently switch between modes unconsciously, leading to confusion in communication. Cultural differences affect what people treat as facts versus values, and individual perspectives on controversial topics might differ dramatically on which statements are descriptive versus normative.