Structural Counterfactuals: A Brief Introduction

Judea Pearl

doi:10.1111/cogs.12065

Abstract

Recent advances in causal reasoning have given rise to a computational model that emulates the process by which humans generate, evaluate, and distinguish counterfactual sentences. Contrasted with the "possible worlds" account of counterfactuals, this "structural" model enjoys the advantages of representational economy, algorithmic simplicity, and conceptual clarity. This introduction traces the emergence of the structural model and gives a panoramic view of several applications where counterfactual reasoning has benefited problem areas in the empirical sciences.

Highlights

One of the most striking phenomena in the study of conditionals is the ease and uniformity with which people evaluate counterfactuals
Not many students of conditionals asked the question: How do we, humans, reach such consensus? More concretely, what mental representation permits such consensus to emerge from the little knowledge we have about Oswald, Kennedy and 1960’s Texas, and what algorithms would need to be postulated to account for the swiftness, comfort and confidence with which such judgments are issued
Before describing specific applications of the structural theory, it will be useful to summarize its implications in the form of two “principles.” The entire set of tools needed for solving causal and counterfactuals problems are based on only these two: Principle 1: “The law of structural counterfactuals.”

Summary

Introduction

One of the most striking phenomena in the study of conditionals is the ease and uniformity with which people evaluate counterfactuals. These canonical examples (attributed to Ernst Adams, (1975)) represent a compelling proof of the ubiquity of the indicative/subjunctive distinction, and of the amazing capacity of humans to process, evaluate and form consensus about counterfactuals. Taking Kennedy’s assassination as a working example, the distinction is as follows: To evaluate the indicative conditional S1 (“If Oswald didn’t kill Kennedy, someone else did”) we start by assigning truth values to variables that are known (or believed) to be true in the story. The evaluation of the subjunctive conditional S2 (“If Oswald hadn’t killed Kennedy, someone else would have”) demands a different procedure.

An outline of the structural theory

The two principles of causal inference

Summary of applications

Conclusions