Empowerment Gain and Causal Model Construction: Children and adults are sensitive to controllability and variability in their causal interventions

2025-12-08

AI Summary

This study investigates how “empowerment”—an intrinsic reward mechanism grounded in perceived controllability and outcome variability—guides causal structure learning and intervention design in humans (children and adults). Method: We formally operationalize empowerment as an information-gain measure within a causal Bayesian network framework and integrate it into a reinforcement learning model of causal intervention selection—the first such formulation. Contributions/Results: (1) We establish empowerment as a theoretical bridge linking Bayesian causal inference with goal-directed behavior; (2) we demonstrate that children’s efficient causal learning arises from an innate preference for high-empowerment interventions, independent of statistical cue reliance; (3) we introduce a computationally tractable, cognitively plausible paradigm for causal modeling. Empirical results show participants robustly prefer high-empowerment interventions, and this preference significantly predicts causal inference accuracy—providing a unifying theoretical foundation for developmental cognitive science and robust AI-based causal learning.

Technology Category

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

Application Category

• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Learning about the causal structure of the world is a fundamental problem for human cognition. Causal models and especially causal learning have proved to be difficult for large pretrained models using standard techniques of deep learning. In contrast, cognitive scientists have applied advances in our formal understanding of causation in computer science, particularly within the Causal Bayes Net formalism, to understand human causal learning. In the very different tradition of reinforcement learning, researchers have described an intrinsic reward signal called "empowerment" which maximizes mutual information between actions and their outcomes. "Empowerment" may be an important bridge between classical Bayesian causal learning and reinforcement learning and may help to characterize causal learning in humans and enable it in machines. If an agent learns an accurate causal world model, they will necessarily increase their empowerment, and increasing empowerment will lead to a more accurate causal world model. Empowerment may also explain distinctive features of childrens causal learning, as well as providing a more tractable computational account of how that learning is possible. In an empirical study, we systematically test how children and adults use cues to empowerment to infer causal relations, and design effective causal interventions.

Problem

• Investigates how empowerment cues influence causal inference in humans
• Explores empowerment as a bridge between Bayesian learning and reinforcement learning
• Examines children's and adults' use of controllability and variability in interventions

Innovation

• Causal Bayes Net formalism for human learning
• Empowerment reward signal in reinforcement learning
• Empowerment cues for causal inference and interventions