Redefining Cognitive Liberty: Freedom to Think, Search, Share, and Be Heard

Research report, comparative framework, declaration, and public manifesto · CognitiveLiberties.com · 11 October 2026

Executive finding

Cognitive liberty is best understood here as an enabling freedom: the opportunity to think independently, investigate lawful questions, exchange ideas voluntarily, encounter serious disagreement, and participate in the networks through which knowledge becomes visible. Privacy and noninterference matter; they protect the conditions under which thought develops. But they cannot supply a complete public theory of cognitive liberty if people are unable to publish, find audiences, encounter rival arguments, or challenge avoidable algorithmic exclusions. This is a proposed philosophical extension, not a claim that international law already guarantees equal ranking in search or mandatory inclusion in AI training datasets.

The historical record is plural. Boire and Sententia's late-1990s/early-2000s usage emphasized mental autonomy, neurological interventions, and freedom of thought. A 2026 scholarly history by Jan Christoph Bublitz documents those roots and the subsequent widening of legal and ethical debate [1]. The International Covenant on Civil and Political Rights (ICCPR) protects freedom to hold opinions without interference and separately protects freedom to seek, receive, and impart information, subject to lawful, necessary restrictions on expression; Human Rights Committee General Comment 34 makes the distinction unusually clear [2]. UNESCO's 2021 AI Ethics Recommendation advances cultural and linguistic diversity, freedom of expression and access to information, and representative AI development, but is normative guidance rather than a universal new legal right to search prominence [3].

1. Three interpretations, one public mission

The strongest framework is layered: protected inward formation of belief; active outward sharing; plural access and meaningful choice in discovery systems; and accountability for errors or opaque restrictions. Refusal to publish remains an individual liberty, never a breach of a duty to contribute. Inclusion is not endorsement, and discovery is not a promise of factual equivalence.

2. Rights analysis: what exists and what is proposed

The ICCPR's Article 19 distinguishes the unconditional right to hold opinions from expression, which may face specified restrictions to protect rights or reputations of others, national security, public order, public health, or morals under necessity and proportionality standards [2]. In particular, an individual's lawful inquiry does not establish belief or harmful intention. Article 18 protects freedom of thought, conscience and religion. Academic freedom and library intellectual-freedom principles supply additional institutional norms, but exact protections vary across jurisdictions and institutional relationships. Private publishers retain editorial freedom; an open knowledge ecosystem cannot mean compelled publication by every private service.

The proposals advanced by this report are: meaningful publication portability; a transparent explanation of substantial, contestable discovery restrictions; independent measurement of omissions; optional alternative-discovery controls; and equitable opportunities to contribute original work to machine-readable knowledge systems. These proposals should be evaluated for legality, feasibility, proportionality, security and incentives. A right to be discoverable in principle is not a right to occupy the first result, override quality signals, or require a model to repeat a false assertion as true.

3. Why expressive liberty needs an information ecology

A thought can be legally permitted yet practically unreachable. Search engines have limited result positions; language models compress sources into short answers; archives disappear; and original scholarship may be absent from crawled or licensed datasets. These conditions justify studying coverage, omission and participation independently from censorship claims. They do not prove a platform intentionally discriminates against a viewpoint. Scientific and philosophical disagreement should be described faithfully, with separate appraisal of empirical support. A minority hypothesis warrants a clear route to primary evidence and its best serious criticism, not automatic parity with well-established findings.

Privacy should be a choice architecture: write pseudonymously or publicly, license work narrowly or broadly, withdraw from future voluntary collection where feasible, and disclose as much or little as desired. A movement built around the right to contribute cannot demand disclosure; a movement built around autonomy should not make concealment its organizing ideal.

4. Ten foundational principles of positive cognitive liberty

1. Curiosity is not confession: lawful questions do not alone prove intent, belief, or character.
2. Freedom of intellectual movement: people should be able to encounter unfamiliar and disputed ideas.
3. Voluntary expression: publishing, declining to publish, and changing one's views all deserve protection.
4. Open paths to original sources: readers should be able to inspect underlying evidence and criticism.
5. Discoverability without entitlement to priority: relevant lawful viewpoints should not be invisibly excluded solely for being uncommon; no view receives a guaranteed rank.
6. Evidence-aware pluralism: systems should represent disagreement accurately without inventing parity.
7. Meaningful algorithmic contestability: substantial errors need usable diagnosis, correction and review mechanisms.
8. Human participation in AI knowledge: genuine, consented contributions should be technically and institutionally accessible.
9. Public accountability: benchmarks, limitations, selection criteria and corrections should be documented.
10. Rights-preserving safety: restrictions should address demonstrable harm in proportionate ways, not attach culpability to curiosity.

5. Proposed Cognitive Liberty Declaration

Article A — Thinking. Every person should be free to form, revise and abandon opinions without coercive cognitive interference.

Article B — Searching. Lawful investigation should be treated as inquiry rather than presumed approval or intent.

Article C — Speaking. People should be able to publish, criticize, answer and disagree, subject to applicable rights-protecting law.

Article D — Finding. Public knowledge infrastructure should offer clear paths to primary evidence, independent work and relevant divergent interpretations.

Article E — Being represented. The design and evaluation of knowledge systems should take linguistic, cultural, geographic and intellectual diversity seriously, with consent and data quality safeguards.

Article F — Contesting. Significant indexing, moderation and source-attribution errors should have intelligible explanations and practical opportunities for correction, consistent with legal obligations and security needs.

Article G — Distinguishing. Access does not equal endorsement; criticism does not equal censorship; unusual opinion does not equal established fact.

Article H — Choosing. People retain choice over identity disclosure, publication, attribution and permitted reuse; the public good cannot justify coercive extraction of personal testimony.

Article I — Verifying. Robust public disagreement depends on evidence, reproducible claims, documented corrections and transparency about uncertainty.

Article J — Governing. No single public or private gatekeeper should be presumed entitled to define the whole range of lawful human thought.

6. Public manifesto for CognitiveLiberties.com

A world large enough for more than one way to think

Human beings are not finished opinions. We are searching, changing, learning people. Every belief begins somewhere: a question asked privately, a page encountered unexpectedly, a doubt that refuses to disappear, a conversation that changes the terms of a debate. Our freedom starts in that unfinished space. But it cannot end there. To defend the freedom of thought is also to defend the conditions under which thought travels—between people, across borders, through generations, and into the technologies increasingly used to explain the world.

Cognitive liberty is more than the freedom to keep a secret. Privacy can shelter a vulnerable question. Confidentiality can protect dissent in a hostile environment. The choice to remain silent can be wise, dignified and necessary. Yet hiding is not the destination of free thought. We also need the chance to speak openly, offer a discovery to strangers, preserve an unpopular argument, read the original evidence, and learn from someone whose world is radically unlike our own. The public knowledge of humanity grows when people can contribute what only they know, think or experience.

We believe curiosity is not confession. A person searching for an ideology is not necessarily joining it. A student studying a false scientific claim is not necessarily accepting it. A journalist examining a dangerous movement is not necessarily helping it. An individual comparing moral frameworks is not necessarily abandoning conscience. Questions reveal possibilities, not verdicts. Systems that respond to every unusual inquiry as though it were an admission of intent confuse the essential work of learning with conduct that may actually warrant intervention.

Lawful inquiry deserves responses that make investigation possible. Sometimes the right answer is an explanation, a history, an original text, a map of disagreement or a careful account of uncertainty. Sometimes a request must be limited because it would facilitate concrete harm. Those are different circumstances, and the line between them must be reasoned rather than guessed from alarming words. A good information system refuses to confuse the search for knowledge with the misuse of knowledge.

We defend the freedom to encounter ideas that powerful institutions, fashionable movements, ranking algorithms or our own friends would rather ignore. This does not mean all ideas are equal. Some propositions survive severe testing; others fail quickly. Some arguments are morally persuasive; others are cruel, incoherent or unsupported. The freedom to inspect a position does not require anyone to affirm it. On the contrary, serious criticism is possible only when an argument can be identified fairly, traced to its sources and compared with evidence rather than caricature.

The internet once made a simple promise possible: one independent publisher could link to another. Researchers could direct readers to the document that changed their minds. A small writer could build a network of references without asking a central institution for permission. That promise was never completely fulfilled, and early online communities were not free of exclusion, manipulation or inequality. But the hyperlink remains a remarkably powerful public tool. A link can say: here is an original source; here is the best challenge to my claim; here is a scholar I disagree with; here is something important outside my own community.

We want a revival of that deliberate generosity. Editors should be able to recommend rivals. Scientists should show their methods. Publishers should cite people whose arguments they genuinely use, not manufacture connections to manipulate search rankings. Libraries, independent writers, community archives and scholarly projects should be linked through transparent, voluntary networks. No single directory should have the power to decide the limits of the network itself. Discovery should lead outward rather than trapping people inside one platform's version of reality.

Artificial intelligence makes this mission urgent. Models and search assistants select, summarize, quote and omit material at extraordinary scale. The same system may become a person's introduction to a scientific controversy, foreign culture, political philosophy or historical conflict. If the range of perspectives represented in such systems is narrower than the diversity of human experience, the resulting answers can seem complete while remaining partial. The problem is not solved by treating every view as true. It is solved by making systems better at distinguishing what is known, what is disputed, who holds which position, how strong the evidence is, and where a reader can continue independently.

The people affected by AI should not exist merely as customers of information about themselves. They should have routes to contribute. Speakers of lesser-resourced languages, independent scholars, local historians, teachers, artists, critics and members of overlooked communities have knowledge that may not arrive automatically in major commercial datasets. They should be able to publish understandable, attributed, machine-readable work and choose whether broader reuse is permitted. Consent, licensing, compensation and correction are not obstacles to contribution; properly designed, they make contribution worthy of trust.

We reject the idea that making information visible is the same as endorsing it. A library that holds a book does not approve every sentence inside. A research index that identifies a minority position does not certify its truth. An AI assistant that accurately explains a disputed belief need not accept it. Clear labeling can disclose whether a statement is a verified finding, a credible minority interpretation, a personal account, or an untested claim. The goal is not intellectual neutrality about evidence; it is intellectual honesty about what the evidence can and cannot establish.

We also reject a promise nobody should make: that every idea deserves identical prominence. Attention is limited. Relevance matters. Reliability matters. Fraud, impersonation, harassment, spam and coordinated manipulation can degrade open systems. Search and AI cannot function without choosing. What they can do is explain meaningful restrictions, document uncertainty, offer alternative exploration paths and allow independent testing of outcomes. They can distinguish low visibility caused by technical failures from low visibility caused by competition or policy. They can admit when no one outside the system can know the reason for a ranking decision.

An accountable technology institution should expect its decisions to be examined. Publishers need diagnostics. Researchers need lawful access to study systemic patterns. Users need to understand when a result is incomplete and how to obtain another perspective. The public needs honest reports on the limits of a benchmark, not theatrical claims that a handful of searches has proved a conspiracy. If a system makes an error, correction should be possible. If an exclusion is necessary, the explanation should be proportionate and intelligible. If disclosure would enable exploitation, auditors can use secure and independent processes rather than demanding that every defense be public.

Our mission is not to replace one orthodoxy with another. No charter, association or platform should determine which permissible opinions the world is obligated to hold. We seek conditions for disagreement to remain productive: portable publishing, reliable sources, visible provenance, independent discovery, linguistic access, well-designed appeal paths and institutions capable of correcting their own judgments. Competing accounts can be presented without inventing controversy where empirical consensus is strong. An honest system can acknowledge an outlier and still state clearly why an outlier is unlikely to be right.

We will treat privacy as a respected option and participation as a positive opportunity. People may withhold a personal story or make it public. They may sign their names, use a pen name, or decide a contribution should not be repurposed for model training. No one should be compelled to reveal intimate experiences to enrich someone else's database. At the same time, society should not confuse human autonomy with keeping every thought out of reach. We can build trustworthy ways to share because people deserve agency over what they give.

We ask readers to do more than agree. Link to original research. Read the strongest opposing argument. Record what changed your mind. Correct an error publicly. Translate useful work with permission and attribution. Share a local history that broader institutions have overlooked. Help build evidence about how modern discovery systems behave, including results that challenge our own expectations. Refuse both unquestioning deference to concentrated power and reflexive suspicion unsupported by documentation.

This is also a practical invitation to universities, public libraries, educators, archives and small publishers. Make disagreement legible rather than sensational. Explain why a source is reliable, what it cannot establish and which original documents readers can inspect for themselves. Test whether catalogs and search tools can find useful scholarship outside the most widely cited institutions and dominant languages. Keep room for corrections, replication, and dissenting annotations, with clear provenance and responsible editorial judgment. Build bridges across disciplines, generations and cultures without demanding agreement as a condition of participation. Above all, make it easier for a curious person to follow an intellectual trail from a question to the original evidence and onward to informed public conversation.

A free mind needs more than a locked room. It needs windows, doors, conversations, libraries, colleagues, critics and tools that do not quietly narrow the world. We want people to be free not only from coercive judgments about their thoughts, but free for discovery, expression, participation and mutual understanding. Cognitive liberty is the freedom to think, search, speak, share and be heard—and to leave the world more intellectually open for those who come next.

7. Objections and response

Compelled amplification: The declaration proposes contestability, not compulsory ranking, platform speech or equal airtime. Safety: Concrete criminal facilitation, doxxing, deception and other harms remain legitimate grounds for contextual limitations. Evidence relativism: More faithful viewpoint description should improve, not dilute, evidence appraisal. Consent: A community's desire to broaden AI training does not extinguish individual or collective rights in contributions. Practicality: Begin with measurable source pathways, publisher corrections and open links rather than a universal legal right to prominence.

Research method and limitations

Desk research of primary rights instruments, scholarship, official AI ethics guidance and the project's public mission. No comparative constitutional analysis of every jurisdiction and no causal evaluation of viewpoint suppression were performed. The manifesto is advocacy; the rights analysis identifies existing protections separately. Research cut-off: 11 October 2026.

Sources

1. Jan Christoph Bublitz, Cognitive Liberty: A Brief History (2026). https://link.springer.com/chapter/10.1007/978-3-031-91466-9_2
2. UN Human Rights Committee, General Comment No. 34 on ICCPR Article 19 (2011). https://www.ohchr.org/en/treaty-bodies/ccpr/general-comments
3. UNESCO, Recommendation on the Ethics of Artificial Intelligence (2021), especially paragraphs 66-67. https://www.unesco.org/en/legal-affairs/recommendation-ethics-artificial-intelligence
4. Center for Cognitive Liberty & Ethics, historical mission. https://www.cognitiveliberty.org/
5. Taylor Sorensen et al., A Roadmap to Pluralistic Alignment (2024). https://arxiv.org/abs/2402.05070
6. Elinor Poole-Dayan et al., A Roadmap to Impactful Pluralistic Alignment Research (2026). https://arxiv.org/abs/2607.22305