- The paper demonstrates that purely quantum systems cannot support robust agency because the no-cloning theorem and linearity of quantum dynamics prevent reliable state copying and simulation.
- It employs a decision-theoretic framework to analyze quantum circuits, revealing that even optimal designs achieve only marginally better fidelity than random selection.
- The findings imply that classical resources, like decoherence and preferred bases, are essential for agency, challenging quantum theories of consciousness and free will.
Agency and the Limits of Purely Quantum Systems
Introduction
This paper rigorously examines the physical prerequisites for agency, specifically interrogating whether a purely quantum system—one evolving unitarily in a coherent regime, absent decoherence or collapse—can satisfy three minimal conditions for agency: (1) constructing a world-model, (2) evaluating consequences of alternative actions, and (3) reliably performing the action that maximizes expected utility. The analysis demonstrates that these conditions are fundamentally incompatible with the constraints imposed by quantum mechanics, notably the no-cloning theorem and the linearity of quantum dynamics. The implications are substantial, delineating the necessity of classical resources for agency and challenging the viability of quantum theories of agency, free will, and consciousness.
The paper adopts a decision-theoretic framework for agency, requiring agents to internally represent their environment (world-model), simulate the outcomes of alternative actions, and select the action with maximal expected utility. This model-based deliberation is supported by empirical and theoretical work in philosophy, neuroscience, and AI, and is distinguished from looser notions of agency (e.g., self-maintenance or simple reflex agents) by its reliance on internal modeling and evaluation of alternatives.
The deliberative process presupposes the ability to copy information: agents must duplicate their world-model to simulate multiple actions. In classical systems, copying is trivial; in quantum systems, it is fundamentally restricted. The paper emphasizes that this copying requirement is not merely a technicality but a necessary condition for robust, generalizable agency in complex environments.
Quantum Constraints: No-Cloning and Linearity
No-Cloning Theorem
The no-cloning theorem prohibits the perfect copying of arbitrary unknown quantum states. This restriction directly impedes both the construction of a world-model (copying environmental information) and the deliberation process (duplicating the world-model for simulating actions). Approximate cloning strategies—state-dependent, probabilistic, and deterministic-imperfect cloning—are analyzed, but none provide sufficient fidelity or generality for reliable agency in the purely quantum regime.
Linearity of Quantum Dynamics
Even if an agent has access to multiple identical copies of the relevant quantum state, the linearity of quantum operations precludes the implementation of a selection mechanism that deterministically applies the best action based on simulated outcomes. The paper provides explicit constructions and calculations showing that controlled unitaries cannot, in general, select and enact the optimal action without entangling the control and target registers, resulting in mixed states and degraded fidelity.
The authors construct and evaluate several quantum agency circuits, varying the number of copies and the set of deliberation unitaries. Performance metrics include fidelity to the target state, Bloch vector length, and angular error. The results show that, even with optimal circuit design and access to multiple copies, the average fidelity achieved is approximately $2/3$, only marginally better than random guessing ($1/2$). The inability to achieve high-fidelity selection of the best action underscores the fundamental limitations imposed by quantum mechanics.
In the classical limit—where the agent knows the basis of environmental states and can design unitaries accordingly—perfect agency is achievable. This regime corresponds to the emergence of classicality via decoherence, where preferred bases allow for reliable copying and selection.
Implications for Theories of Agency, Consciousness, and Free Will
Constraints on Quantum Agency
The results place principled constraints on the physical realization of agency, ruling out purely quantum agents in the absence of classical resources. This challenges models that posit fully quantum agents (QQ) or quantum agents in classical environments (QC), as both require access to preferred bases or classical reference frames.
Quantum Theories of Consciousness
Quantum theories of consciousness, such as Orch OR and quantum integrated information theory, are critically examined. The necessity of classical resources for agency implies that any quantum theory of consciousness must explicitly account for the classical mechanisms enabling deliberation and memory. The subjective privacy of experience, sometimes attributed to unclonability, is shown to be problematic: if conscious states are uncloneable quantum states, their information cannot be used for deliberation or memory formation.
Theories of Free Will
Libertarian accounts of free will that invoke quantum indeterminacy or collapse are constrained by the need for classical copying and selection mechanisms. The paper highlights unresolved challenges in modeling deliberation and choice within a purely quantum framework, particularly regarding the generation and evaluation of alternative actions.
Emergence of Classical Agency
The analysis clarifies how classical agents emerge within a quantum universe: agency becomes physically possible only when classical features such as preferred bases and reliable copying are present. Decoherence plays a central role in this transition, enabling the stable, copyable states required for model-based deliberation and action selection.
Technological Implications
Quantum computers, despite their quantum substrate, rely on classical control and preferred computational bases for input preparation and output interpretation. The inability to simulate agential behavior in a purely quantum regime without significant classical components imposes a fundamental limitation on quantum AI architectures.
Conclusion
The paper establishes that agency, as formalized by model-based deliberation and utility-maximizing action selection, cannot be realized in a purely quantum system due to the no-cloning theorem and the linearity of quantum dynamics. Classical resources—preferred bases, decoherence, and reliable copying—are indispensable for agency. These findings have far-reaching implications, constraining the physical basis of agency, informing the emergence of classical agents in quantum universes, and challenging quantum theories of consciousness and free will. Future research must address the integration of classical and quantum resources in models of agency and cognition, and further elucidate the mechanisms by which classicality emerges from quantum substrates.