Value — Causality

The human dimension: why value is perceived in distorted ways

Value is perceived relative to a reference point, and losses weigh more. Prospect theory establishes that "people evaluate outcomes relative to a reference point, are loss-averse (losses loom larger than equivalent gains), and weight probabilities nonlinearly" (Intention Architecture, Module V: Behavioral Economics · Behavioral Economics). Practical consequence: a solution that prevents losing $10,000 is perceived as more valuable than one that earns $10,000 — so the project must quantify objective value, even while communicating in the frame the client understands.

Human intentions do not come from rational optimization. "Human intentions are not the product of rational optimization. They are shaped by reference points, loss aversion, present bias, and social comparison. A system that models users as rational agents will systematically fail to predict their behavior" (Module V · Behavioral Economics). The client may be convinced that X causes Y simply because they wish it were so.

We feel causes before verifying them. Damasio's somatic markers show that "emotional signals guide decision-making by marking certain options as desirable or dangerous before conscious deliberation begins" (Module I: Human Intention and Decision-Making · Motivation and Emotion). The causal "gut feeling" is a data point — but it is a hypothesis, not a demonstration.

The discipline of the method

Every intention is a causal hypothesis. The syllabus is explicit: "causality is arguably the most important concept in intention architecture because every intention is, at its core, a causal hypothesis: 'If I do X, then Y will occur.' If the causal model is wrong, the intention, no matter how well-expressed, will fail" (Module III: Applied Logic and Systems Thinking · Causality).

Pearl's three levels. Pearl distinguishes "association ('What is?'), intervention ('What if I do?'), and counterfactual ('What if I had done differently?'). Most statistical methods operate at the first level; genuine intention requires the second and third" (Module III · Causality). A project that only has association data does not yet have a solution: it has a lead.

Confounders produce catastrophic interventions. "The distinction between correlation and causation is routinely violated in practice. Confounding variables produce spurious correlations that can lead to catastrophically wrong interventions" (Module III · Causality).

Causal judgment remains human. Pearl and Mackenzie argue that "causal reasoning is what separates human intelligence from mere data processing. Current AI systems simulate but do not truly possess this capacity. The human in the loop remains indispensable for causal judgment" (Module III · Causality). This connects directly with Topic 8: causal analysis is one of the parts of the work that is never delegated to the machine.

Running case: Marino Distributors

The solution designed in Topics 1–4 rests on this causal hypothesis, written explicitly:

"If we complete the technical attributes of the product sheets for the 480 high-rotation products (X), returns due to 'wrong part' will drop from 5.8% to ≤ 2.5% (Y), because 62% of returns state a selection error and 88% of those occur on products with incomplete sheets (mechanism)."

Before investing, alternative explanations (confounders) are sought: what if the incomplete-sheet products are also the most technically complex ones, and complexity — not the sheet — causes the error? This is checked at Pearl's level 2 (intervention): a 4-week pilot with 60 products — 30 sheets completed, 30 comparable ones left untouched. Result: returns in the treated group drop 55%; the control group does not move. The causal chain is demonstrated by intervention, not correlation, and the value is quantified: at $18 per avoided return, the project pays for itself in 5 months.

In communicating with the client, the loss frame is used consciously (prospect theory): "you are currently losing ~$4,100 per month in avoidable returns" mobilizes more than "we could gain efficiency" — a truthful frame, not manipulation (a boundary developed in Topic 8).

Double exercise

Filter question

Can we demonstrate the causal chain between this solution and the benefit it promises?

References to the Intention Architecture syllabus

• Module I: Human Intention and Decision-Making · Motivation and Emotion (somatic markers)
• Module III: Applied Logic and Systems Thinking · Causality (Pearl's three levels, confounders, human causal judgment)
• Module V: Behavioral Economics · Behavioral Economics (prospect theory, reference points, loss aversion)

Bibliography: Pearl (2000); Pearl & Mackenzie (2018); Kahneman & Tversky (1979); Damasio (1994).