Causality Without Causal Models

2025-11-26

AI Summary

The Halpern–Pearl definition of causality is overly dependent on structural equation models (SEMs), limiting its applicability to broader semantic frameworks. Method: We propose an abstraction of causality and explanation that is model-agnostic, formally reconstructing the core semantic features—counterfactual dependence, actuality, and minimality of intervention—within a formal counterfactual logic, thereby eliminating reliance on SEM-specific syntax or structure. Contribution/Results: This framework enables causal reasoning over complex propositions involving disjunctions, negation, belief operators, and nested counterfactuals—the first such treatment in the literature. It supports retrospective counterfactual inference and unifies causal and explanatory reasoning across all model classes admitting a counterfactual semantics. By decoupling causality from SEMs, our approach significantly enhances expressive power, domain generality, and foundational depth of causal reasoning.

Technology Category

• Reasoning under Uncertainty: Causality
• Knowledge Representation and Reasoning: Action, Change, and Causality
• Machine Learning: Causal Learning

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Perhaps the most prominent current definition of (actual) causality is due to Halpern and Pearl. It is defined using causal models (also known as structural equations models). We abstract the definition, extracting its key features, so that it can be applied to any other model where counterfactuals are defined. By abstracting the definition, we gain a number of benefits. Not only can we apply the definition in a wider range of models, including ones that allow, for example, backtracking, but we can apply the definition to determine if A is a cause of B even if A and B are formulas involving disjunctions, negations, beliefs, and nested counterfactuals (none of which can be handled by the Halpern-Pearl definition). Moreover, we can extend the ideas to getting an abstract definition of explanation that can be applied beyond causal models. Finally, we gain a deeper understanding of features of the definition even in causal models.

Problem

• Abstracting causality beyond causal models
• Extending counterfactual reasoning to complex formulas
• Generalizing causal definitions for broader applications

Innovation

• Abstracting causality from causal models
• Enabling counterfactual analysis in broader models
• Extending causal definitions to complex logical formulas