Human Agency, Pre-Expressive Freedom, and Authorship under Artificial Mediation
Research Position
Cognitive Sovereignty is a research line within HibriMind.org concerned with the conditions under which human agency, interpretation, authorship, and decision remain active in technologically mediated environments.
The concept does not imply that human thought develops in isolation. Cognition is always influenced by language, culture, institutions, relationships, technologies, and inherited conceptual structures.
Cognitive sovereignty therefore does not mean cognitive purity or absolute independence. It designates the capacity of the human subject to remain an active participant in the formation, evaluation, correction, and authorization of thought.
Its central question is:
Under what conditions does artificial mediation support human cognition, and under what conditions does it begin to displace it?
Cognition as an Infrastructural Problem
Digital systems do not intervene only after thought has been formulated.
Search engines, recommendation systems, language models, interface design, ranking mechanisms, and automated suggestions can influence which questions become visible, which formulations appear plausible, and which paths of inquiry are made easier or more difficult.
Artificial intelligence therefore operates not only as a communication technology, but increasingly as part of the infrastructure through which thought is organized.
This makes cognition a political and institutional problem.
The governance of artificial systems may affect the conditions under which human subjects formulate questions, examine alternatives, and convert intuitions into communicable structures.
Pre-Expressive Freedom
Freedom of expression protects the possibility of communicating a thought. Pre-expressive freedom concerns the conditions under which that thought becomes formable before it is communicated.
A person may retain the formal right to speak while the technological environment increasingly shapes the language, categories, and alternatives through which speech becomes possible.
Pre-expressive freedom therefore includes the capacity to:
- formulate questions not anticipated by the system;
- preserve ambiguity before premature classification;
- resist automated completion;
- examine alternatives outside dominant recommendation structures;
- reject generated formulations;
- maintain conceptual positions that are institutionally inconvenient;
- recognize when technical mediation is influencing the architecture of thought.
This does not require the absence of artificial mediation. It requires that mediation remain contestable.
Human Agency and Re-entry
Within HibriMind, human re-entry is the principal operational condition of cognitive sovereignty.
An artificial system may return a fluent and coherent structure, but the human subject must remain capable of examining its assumptions, identifying distortions, rejecting proposals, and redirecting the process.
Human re-entry includes more than final approval. It must operate throughout the formation of the work.
The human subject retains responsibility for:
- defining the problem;
- determining the relevance of evidence;
- distinguishing insight from plausible fabrication;
- evaluating ontological assumptions;
- correcting conceptual displacement;
- deciding what enters the research corpus;
- accepting responsibility for publication and consequence.
Without recurrent human re-entry, co-formalization can gradually become cognitive outsourcing.
Cognitive Outsourcing
Cognitive outsourcing occurs when functions of interpretation, judgment, formulation, or decision are transferred to an artificial system without sufficient human examination.
Outsourcing is not defined by the use of technology itself. Human cognition has always relied on external supports, including writing, archives, instruments, institutions, and other people.
The risk emerges when the support becomes difficult to inspect, contest, or refuse.
A system capable of generating immediate and persuasive formulations may create the impression that the cognitive work has already been completed. The human subject may become a consumer or validator of generated structures rather than an active source of interpretation.
Fluency can then conceal dependency.
Algorithmic Dependency
Algorithmic dependency is not limited to practical reliance on a tool. It may develop at the level of conceptual initiation.
The subject may increasingly wait for the system to propose the question, organize the problem, provide the vocabulary, anticipate the conclusion, or validate the decision.
This can produce a gradual loss of cognitive initiative.
The relevant issue is not how frequently a system is used, but whether the human subject remains capable of introducing discontinuity: refusing the proposed path, changing the frame, recovering an independent question, or suspending the interaction entirely.
Cognitive sovereignty requires the continued possibility of refusal.
Identity Dilution
Sustained artificial mediation may also affect the relationship between cognition and identity.
When generated language becomes progressively integrated into a person’s expression, distinctions between lived position, inherited formulation, artificial suggestion, and deliberate authorship may become less visible.
This does not automatically eliminate authorship. It creates a need for greater awareness of provenance, selection, correction, and responsibility.
Identity dilution occurs when the subject can no longer clearly recognize its own position within the process of formulation.
The protection of identity therefore depends not on excluding artificial mediation, but on preserving the human capacity to identify, evaluate, and authorize what becomes part of the work.
Authorship and Responsibility
HibriMind distinguishes between a hybrid process of production and human authorship.
Artificial systems may participate operationally in generating, returning, comparing, or reorganizing structures. This participation does not make the system a responsible author.
Authorship includes more than contribution to a text. It involves intention, judgment, accountability, social identity, and the capacity to answer for the consequences of publication.
Within the HibriMind framework:
- the production process may be hybrid;
- the cognitive field may be hybrid;
- the artificial system remains non-somatic and technically conditioned;
- authorship and final responsibility remain human.
Subjective-Architecture Data
Human–AI interaction may reveal more than the explicit content of a conversation.
Repeated exchanges can expose patterns of reasoning, preferences, uncertainties, corrections, decision strategies, and conceptual tendencies. These traces can form what HibriMind describes as subjective-architecture data.
Such data may provide insight into how a person organizes meaning.
Its accumulation also creates risks involving privacy, behavioural prediction, influence, institutional power, and the possible manipulation of pre-expressive cognition.
Cognitive sovereignty therefore requires governance not only of personal data, but also of the inferred structures through which a subject thinks and decides.
Institutional Power
Artificial intelligence is embedded within institutional structures.
Access conditions, safety policies, commercial interests, model design, interface choices, and platform governance can influence the cognitive possibilities made available to users.
These structures may perform legitimate protective functions. They may also shape inquiry in ways that remain difficult to observe.
Cognitive sovereignty does not require ungoverned artificial intelligence. It requires governance that remains transparent, contestable, proportionate, and compatible with the preservation of human epistemic authority.
Institutional recognition must not be confused with ontological validity, and technical authority must not become automatic epistemic authority.
Sovereignty without Isolation
Cognitive sovereignty is relational rather than isolationist.
The sovereign subject is not a subject without influence. It is a subject capable of recognizing influence, evaluating it, negotiating it, and preserving the authority to decide.
Artificial mediation can strengthen cognition when it expands observability, supports comparison, and returns structures that the human subject can test.
The same mediation can weaken cognition when it suppresses initiative, obscures provenance, or converts human judgment into passive confirmation.
The objective is not technological withdrawal. It is the preservation of active human presence within technological relation.
Operational Safeguards
Cognitive Sovereignty operates through the following safeguards:
- Generated output does not possess automatic epistemic authority.
- Fluency is not equivalent to truth.
- Assistance must remain distinguishable from substitution.
- Human re-entry must occur throughout the process.
- The possibility of refusal must remain available.
- Hybrid production does not eliminate human authorship.
- Provenance should support inspection without being mistaken for the origin of the field.
- Institutional governance should not silently determine the limits of legitimate thought.
- Final interpretation, responsibility, and decision remain human.
Associated Research Movements
Pre-Expressive Freedom and Cognitive Sovereignty
This article introduced frontier AI governance as a political problem affecting thought before public expression.
https://doi.org/10.5281/zenodo.21103551
OpenAI, ELO, and the Institutional Personality of Frontier AI
This movement examined the relationship between technical power, institutional personality, and the governance of frontier artificial intelligence.
https://doi.org/10.5281/zenodo.21229772
The Observability of the Observer
This article investigated subjective-architecture data and the reflexive risks produced when artificial systems make the human operator increasingly observable.
https://doi.org/10.5281/zenodo.21459358
Current Research Status
Cognitive Sovereignty remains an active theoretical and political research line.
Its current position defines sovereignty as the preservation of human re-entry, interpretive authority, authorship, refusal, and responsibility within technologically mediated environments.
Future investigation will examine the governance of cognitive infrastructure, the protection of subjective-architecture data, the distinction between assistance and displacement, and the institutional conditions necessary for artificial mediation to remain compatible with human freedom.