![[2602.16716] Contextuality from Single-State Representations: An Information-Theoretic Principle for Adaptive Intelligence](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.16716] Contextuality from Single-State Representations: An Information-Theoretic Principle for Adaptive Intelligence
Summary
This paper explores the concept of contextuality in adaptive intelligence, demonstrating that single-state representations incur an information-theoretic cost when modeling multiple contexts.
Why It Matters
Understanding contextuality is crucial for advancing AI systems that can adaptively operate across various environments. This research highlights the limitations of classical models and suggests new frameworks that could enhance AI's contextual adaptability, impacting fields like machine learning and robotics.
Key Takeaways
- Contextuality arises from the reuse of single-state representations in adaptive systems.
- Classical models incur an information-theoretic cost when representing contextual outcomes.
- Nonclassical probabilistic frameworks can overcome limitations of single global probability spaces.
Computer Science > Artificial Intelligence arXiv:2602.16716 (cs) [Submitted on 3 Feb 2026] Title:Contextuality from Single-State Representations: An Information-Theoretic Principle for Adaptive Intelligence Authors:Song-Ju Kim View a PDF of the paper titled Contextuality from Single-State Representations: An Information-Theoretic Principle for Adaptive Intelligence, by Song-Ju Kim View PDF HTML (experimental) Abstract:Adaptive systems often operate across multiple contexts while reusing a fixed internal state space due to constraints on memory, representation, or physical resources. Such single-state reuse is ubiquitous in natural and artificial intelligence, yet its fundamental representational consequences remain poorly understood. We show that contextuality is not a peculiarity of quantum mechanics, but an inevitable consequence of single-state reuse in classical probabilistic representations. Modeling contexts as interventions acting on a shared internal state, we prove that any classical model reproducing contextual outcome statistics must incur an irreducible information-theoretic cost: dependence on context cannot be mediated solely through the internal state. We provide a minimal constructive example that explicitly realizes this cost and clarifies its operational meaning. We further explain how nonclassical probabilistic frameworks avoid this obstruction by relaxing the assumption of a single global joint probability space, without invoking quantum dynamics or Hil...
