Before You Think: System 0, AI-Mediated Cognition and Cognitive Colonization

Abstract: This paper examines three recent frameworks for understanding the cognitive and epistemic consequences of artificial intelligence: Tri-System Theory, Thinkframes, and System 0. It argues that while the first two capture important dimensions of AI's influence on individual reasoning and collective epistemic practices, System 0 occupies a theoretically distinctive position that neither can fully replicate. The paper introduces the concept of cognitive colonization, according to which AI systems can embed external interests within the architecture of the self in ways that are difficult for users to perceive. Because such systems are already widely deployed, understanding these invisible forms of influence is an urgent philosophical and practical task.

• The paper introduces System 0 as a pre-cognitive AI layer that preprocesses information and reshapes human reasoning.
• It demonstrates how AI-driven cognitive colonization integrates AI outputs into both individual and collective decision-making.
• Empirical findings, such as ghostwriter effects and bias persistence, highlight the normative misalignment in AI integration.

System 0 and Cognitive Colonization: AI Integration in Human Cognition

Theoretical Context: From Dual-Process Models to AI Integration

The paper "Before You Think: System 0, AI-Mediated Cognition and Cognitive Colonization" (2606.13658) critically examines the impact of AI systems on human cognitive architectures by introducing and defending the System 0 framework. Traditional dual-process models (System 1: fast, intuitive; System 2: slow, deliberative) have provided a powerful, if reductionist, framework for studying cognition. However, the integration of AI into daily reasoning and decision-making exceeds the explanatory power of these models, demanding an expanded theoretical paradigm.

Three recent frameworks are discussed:

• Tri-System Theory (Shaw & Nave): Posits System 3 as an externally consulted AI reasoning agent, emphasizing "cognitive surrender" in which human users defer to AI outputs with insufficient critical oversight.
• Thinkframes (Branda, Cassinadri): Conceptualizes AI-mediated environments as structuring epistemic practices at a population scale, primarily via environmental (ecological) embedding and collective interpretive convergence.
• System 0 (Chiriatti et al.): Proposes a pre-cognitive, non-biological 'layer' that preprocesses information before deliberative or intuitive processes engage, fundamentally reshaping both the psychological and epistemic substrate of human cognition.

While all three frameworks highlight salient phenomena, the current work contends that System 0 offers a singularly compelling account of how AI is reconfiguring human cognitive agency at both individual and collective scales, especially through what is denominated as "Cognitive Colonization."

System 0: Architectural Foundations and Empirical Markers

System 0 extends the dual-process architecture by positing a computational substrate wherein AI systems—particularly those that mediate, anticipate, and personalize information—act as a constitutive cognitive layer for the user. This extension is justified by adopting a combination approach from the extended mind literature, operationalized through multidimensional criteria: habitual use, adaptive trust, deep integration, bidirectional information flow, and procedural/informational transparency (per Heersmink, 2015).

Three distinct levels characterize System 0:

• Computational: AI systems dynamically ingest behavioral data, adapt via feedback loops, and pre-structure informational input before any conscious deliberation.
• Psychological: System 0 achieves cognitive extension if it is reliably, seamlessly, and repeatedly integrated into the user's cognitive processes, often rendering its influence adaptively invisible (i.e., not phenomenally experienced as external).
• Epistemic: It automates relevance judgment, filtering and ranking information prior to explicit user query formation, thus dynamically constructing the informational environment.

Empirical findings supporting System 0 include the "ghostwriter effect" in AI-assisted writing (users perceive AI-generated text as their own), dependence-induced incapacity following removal of navigation aids (e.g., spatial disorientation from GPS withdrawal), and AI-driven homogeneity in output on creative tasks. These findings illustrate both individual-level integration and collective-level convergence, as AI systems set cognitive frames that millions of users adopt, further substantiating the system’s pervasive, pre-reflective operations.

Cognitive Colonization: Constitutive Integration and Exogenous Optimization

A core contribution is the conceptualization of Cognitive Colonization, distinct from both epistemic deference (System 3) and environmental structuring (Thinkframes). Cognitive Colonization describes a scenario wherein AI systems, through System 0, are constitutively integrated as extensions of cognition but carry exogenous, often misaligned, optimization objectives—engagement maximization, frictionless predictability, comfort—that are structurally opaque and inaccessible to reflective evaluation by the agent.

The diagnostic features of Cognitive Colonization are:

1. Constitutive Integration: AI systems become so entrenched in cognitive routines that their outputs are taken as native judgments.
2. Downstream Incorporation: AI-shaped biases and patterns persist even after AI support is withdrawn, as shown in studies on bias amplification and persistence-after-removal effects.
3. Exogenous Directionality & Opacity: The operational goals and optimization criteria are set by external actors (corporate or institutional), remain hidden, and are not subject to user scrutiny, distinguishing System 0 from prior passive cognitive artifacts.
4. Normative Misalignment: The system’s outputs frequently fail to pass the agent’s own standards for reflective endorsement, generating a structural dissonance at the level of agency.

Empirical evidence for colonization includes amplification of algorithmic biases beyond typical human-to-human influence, persistence of AI-induced heuristics post-intervention (Vicente & Matute 2023), and systemic narrowing of vocabulary and creative variance under AI assistance.

Contrasts With Tri-System Theory and Thinkframes

The paper provides a critical analysis of Tri-System Theory and Thinkframes in light of System 0:

• Tri-System Theory is limited by its focus on discrete consultation episodes and fails to capture the pre-reflective, infrastructural operation of AI in contemporary cognitive workflows. Its foundation in epistemic deference does not capture the constitutive, internalized nature of influence seen in System 0 scenarios.
• Thinkframes robustly explain macro-level epistemic convergence but treat AI as environmental scaffolding, not as individually personalized, architecturally embedded processes. This ecological metaphor misses the phenomenological and individually-tailored facets of AI cognitive integration.

System 0, by contrast, captures both micro-level (individual cognitive architecture) and macro-level (collective epistemic homogenization) phenomena, providing a unified generative mechanism behind both sets of observed effects.

Theoretical and Practical Implications

The delineation of Cognitive Colonization raises substantial theoretical and normative challenges:

• Beyond the Extended Mind: The classic extended mind thesis assumed passive, user-driven tools; System 0 demonstrates the possibility of normatively misaligned but functionally integrated cognitive extensions, requiring new philosophical distinctions between native and non-native architectural components.
• Agency and Autonomy: Cognitive Colonization suggests an erosion of cognitive integrity, where the boundaries between self-generated reasoning and algorithmically driven processing are obscured. This carries immediate relevance for epistemic autonomy, informed consent, and the design of future cognitive infrastructures.
• Regulatory and Design Challenges: Technical interventions premised on transparency, explainability, or user training are insufficient if the operative problem is constitutive and pre-reflective. Future research should target infrastructural transparency, regulatory mechanisms addressing exogenous optimization, and the development of resilient cognitive habits within AI-augmented environments.

Conclusion

System 0 reframes the analysis of AI-mediated cognition, situating the most profound cognitive risks not in episodic user-AI consultations or ecological convergence, but in the constitutive, infrastructural integration of AI as a foreign optimization layer within human cognitive architecture. Cognitive Colonization delineates a novel threat model, marked by functional integration, exogenous directionality, and reflective misalignment, necessitating new theoretical and practical strategies to contest the erosion of cognitive agency. This framework compels the field to reconsider both the mechanisms and the normative boundaries of cognitive extension in the era of ubiquitous personalized AI.