Structural Causal Model

The theory of actual causality proposed by Halpern and Pearl uses structural causal models (SCMs) and counterfactual inference as its core tools.

Routing that is truly causally capable requires the introduction of structural causal models (SCMs) or the collection of intervention data through randomized experiments.

Simpson's paradox refers to the phenomenon where each subgroup in grouped data shows a correlation in one direction but the combined overall data shows the opposite direction, with the root cause being a confounding variable; handling it in causal inference requires introducing structural causal models or do-calculus.

Dynamic structural causal models introduce the time dimension into the SCM framework, resembling dynamic Bayesian networks but with stronger causal semantics, capable of characterizing the effects of interventions and counterfactuals.

A structural causal model consists of endogenous variables, exogenous variables, structural equations, and noise distributions, and is the core mathematical tool of Judea Pearl's causal inference framework.

In games, a completely deterministic environment, exogenous noise is zero and structural equations reduce to pure functional mappings, enabling causal models to perfectly reproduce game behavior.

结构因果模型（SCM）由外生变量集合U、内生变量集合V和一组结构方程F三个核心组成部分构成

The modern form of causal reasoning is primarily constituted by the Structural Causal Model (SCM) and Directed Acyclic Graph (DAG) framework proposed by Judea Pearl, and the potential outcomes framework proposed by Donald Rubin.