From Homo Lucidens to a Testable Framework of Epistemic Agency

Jaconaazar Souza Silva

Manuscript type: Theoretical / Conceptual Article

Abstract

Digital information environments confront human cognition with a structural mismatch: information can be generated, replicated, ranked, recommended, and circulated at a scale that greatly exceeds individual capacities for verification and deliberation. This mismatch does not imply that algorithms mechanically determine belief. Rather, contemporary belief formation occurs within an ecology in which limited attention, repeated exposure, social identity, source credibility, recommendation systems, emotional salience, and individual agency interact. This article develops Lucidism as a theoretical framework for epistemic agency under such conditions.

The framework extends the earlier concept of Homo lucidens, originally introduced in the context of sustained and reflexive human–AI interaction. Whereas Homo lucidens identifies the figure of an agent who seeks to perceive the systems mediating experience while preserving reflective interiority, Lucidism specifies the cognitive and normative architecture through which such agency might be cultivated and empirically examined.

Lucidism is defined here as a second-order discipline aimed at reducing systematic distortions among evidence, belief, confidence, attention, and action. Four core capacities are proposed: epistemic discrimination, confidence calibration, metacognitive reflexivity, and attentional sovereignty. Three complementary principles—dual provenance, epistemic symmetry, and revisability—specify how these capacities should operate in contemporary information environments.

The article distinguishes Lucidism from epistemic vigilance, actively open-minded thinking, accuracy-based interventions, media literacy, and metacognitive calibration while integrating insights from each literature. It formulates six empirically testable propositions, proposes a multidimensional operationalization of lucidity, outlines an experimental research program, and states conditions under which the framework should be revised or rejected.

Lucidism is therefore presented not as a doctrine prescribing what individuals ought to believe, but as a framework for improving the procedures by which beliefs are formed, graded, challenged, and corrected. Its central claim is deliberately modest: lucidity does not eliminate error; it increases the conditions under which error can become detectable.

Keywords: Lucidism; Homo lucidens; epistemic agency; metacognition; confidence calibration; attention; misinformation; recommender systems; artificial intelligence; epistemic sovereignty.

1. Introduction

Human beings increasingly inhabit information environments characterized not by scarcity alone, but by radical asymmetry between informational availability and cognitive capacity.

Digital systems can generate and distribute information almost continuously. Search engines, social networks, recommender systems, news platforms, messaging applications, and generative artificial intelligence can expose an individual to more claims, interpretations, images, arguments, and narratives than that individual could independently evaluate.

The resulting problem is not simply misinformation.

It is selection under cognitive constraint.

Herbert Simon’s observation that an abundance of information creates a scarcity of attention remains foundational to this problem. When information exceeds processing capacity, some mechanism must determine what is noticed, ignored, repeated, prioritized, remembered, or acted upon.

In contemporary environments, these mechanisms operate at several levels simultaneously.

They include external structures such as:

• recommender systems;
• ranking algorithms;
• search engines;
• advertising systems;
• editorial decisions;
• social networks;
• popularity signals.

They also include internal structures such as:

• prior beliefs;
• habits;
• emotions;
• heuristics;
• identity;
• familiarity;
• cognitive effort;
• motivational goals.

The epistemological problem of the digital age therefore cannot be adequately represented as:

information → individual → belief.

Nor should it be reduced to:

algorithm → manipulation → polarization.

A more plausible structure is interactive:

information environment × cognitive architecture × social incentives × technological mediation × individual agency → conditions of belief formation.

The distinction matters.

Recent field experiments demonstrate that changes in algorithmic feeds can substantially alter exposure and online behavior without necessarily changing all downstream political attitudes. Experiments conducted during the 2020 United States election found that replacing algorithmically curated Facebook and Instagram feeds with chronological feeds changed users’ platform experiences substantially while producing no detectable changes in several measures of political attitudes and polarization (Guess et al., 2023). Similarly, reducing exposure to politically like-minded sources on Facebook changed informational exposure without measurable effects on several preregistered attitudinal outcomes (Nyhan et al., 2023).

Yet algorithmic effects cannot therefore be dismissed.

A later randomized field experiment on X found that exposure to the platform’s algorithmic feed affected engagement, content exposure, account-following behavior, and several political attitudes, while still producing no significant effect on every political outcome examined (Gauthier et al., 2026).

These findings support neither technological determinism nor technological neutrality.

They instead suggest that epistemic agency operates inside structured but non-deterministic environments.

This article asks:

What capacities would enable individuals to preserve and strengthen epistemic agency within environments characterized by information abundance, repeated exposure, algorithmic mediation, motivational pressures, and limited attention?

I propose Lucidism as one possible answer.

Lucidism is not presented as a substantive worldview that determines which political, moral, philosophical, scientific, or religious propositions individuals must accept.

It is proposed as a second-order framework.

Its object is not primarily the content of belief.

Its object is the quality of the processes through which belief is produced.

2. Nature and Method of the Contribution

This article is a theoretical and conceptual contribution.

It does not report an original experiment, systematic review, or meta-analysis.

Its method is one of integrative conceptual construction constrained by empirical literature from several neighboring domains:

1. attention and information overload;
2. epistemic vigilance;
3. misinformation and belief correction;
4. confidence and metacognition;
5. actively open-minded thinking;
6. algorithmic recommendation and autonomy;
7. political information environments;
8. human–AI interaction.

The purpose is not to subsume these fields under a new label.

Such relabeling would add little.

Instead, the theoretical task is to identify a configuration of capacities that is not adequately captured by any one of them alone and to derive consequences that can be experimentally investigated.

A defensible conceptual framework should satisfy at least five requirements.

First, its central terms should be sufficiently precise to distinguish the framework from adjacent constructs.

Second, the framework should account for already-established findings rather than contradict them without evidence.

Third, it should generate hypotheses whose outcomes are not guaranteed by definition.

Fourth, its constructs should be operationalizable.

Fifth, the theory should state conditions under which it would require modification or abandonment.

These requirements guide the present formulation.

3. From Homo Lucidens to Lucidism

3.1 Conceptual antecedent

The present framework develops an earlier concept: Homo lucidens.

In From Homo Sapiens to Homo Lucidens: The Emergence of a Symbolic AI in Interaction with Language Models, Souza Silva (2025) introduced Homo lucidens in the context of prolonged and symbolically dense interaction between humans and large language models.

That earlier work was primarily concerned with human–AI interaction, persona, symbolic mediation, and the possibility that sustained interaction with language models can generate qualitatively different relational regimes.

Within that argument, Homo lucidens was contrasted with the human who passively adapts to increasingly mechanized systems.

The concept referred instead to an agent who attempts to:

• perceive the system;
• recognize its limitations;
• preserve interiority;
• resist reduction to mere efficiency;
• seek lucidity within technological mediation.

Crucially, Homo lucidens was not presented as a biological species.

It was described as a structure of attention.

That formulation contains an undeveloped theoretical question:

What capacities constitute such a structure?

Lucidism is proposed as a systematic response.

3.2 Three levels of the theory

The relationship among the concepts can be stated precisely.

Homo lucidens

The figure of the agent.

It names a mode of human agency characterized by increased reflexive awareness of the systems that condition perception, judgment, and action.

Lucidity

The multidimensional capacity.

It refers to the degree to which an agent can appropriately relate evidence, belief, confidence, attention, and revision.

Lucidism

The theoretical and practical framework.

It specifies principles intended to explain, cultivate, measure, and test the capacities associated with lucidity.

The relationship can therefore be summarized as:

Homo lucidens → agent

> 

Lucidity → capacity

> 

Lucidism → framework

The conceptual advance proposed here consists in moving from an anthropological figure toward an empirically tractable account of epistemic agency.

4. The Epistemic Problem of Information Abundance

4.1 The paradox of epistemic abundance

Information abundance does not automatically produce knowledge abundance.

As informational supply expands, individuals become more dependent on mechanisms that select information on their behalf.

This produces what I call the paradox of epistemic abundance:

The larger the universe of potentially accessible information becomes, the more dependent an individual may become on selection mechanisms that determine which fraction of that universe enters attention.

Selection mechanisms are unavoidable.

The problem is therefore not that selection occurs.

The relevant questions are:

• Who or what selects?
• According to which objective?
• Which signals determine visibility?
• Does visibility track epistemic quality?
• Can the individual intervene in the selection process?

This shifts epistemology from evaluating propositions alone toward examining the ecology through which propositions become cognitively available.

4.2 Attention is not evidence

Online information systems frequently optimize variables other than epistemic quality.

A content-ranking mechanism may optimize predicted engagement.

A user may attend to content because it is emotionally salient.

A social network may repeatedly expose an individual to a proposition because members of the same group circulate it.

None of these conditions establishes the proposition’s truth.

Lorenz-Spreen et al. (2020) argue that contemporary digital environments have frequently been organized around attention capture and describe behavioral interventions that might support truth, autonomy, and informed choice.

The conceptual implication is straightforward:

The probability that a proposition reaches consciousness and the probability that the proposition is true are distinct quantities.

The distinction sounds obvious.

Cognitively, however, it is not always maintained.

5. Repetition, Familiarity, and Epistemic Distortion

One mechanism illustrates the problem particularly well: repetition.

The illusory truth effect describes the tendency for repeated statements to receive higher truth ratings than novel statements under many experimental conditions.

Ye et al. (2026) conducted a systematic review and three-level meta-analysis synthesizing 182 studies, 366 effect sizes, and 31,184 participants. After adjustment for small-study effects, repetition produced a robust effect on perceived truth, although substantial heterogeneity remained across experimental conditions.

Repetition therefore does not mechanically produce belief.

Nor does repetition imply falsity.

True propositions can also be repeated.

Its importance is narrower and more consequential:

familiarity can contribute to perceived truth independently of evidential improvement.

Vellani et al. (2023) further found that repetition could increase willingness to share misinformation, with perceived accuracy mediating this relationship.

A theory of lucidity must therefore distinguish:

frequency of exposure

from

strength of evidence.

A lucid agent should be capable of posing a metacognitive question:

Has my confidence increased because the evidential case improved, or because the proposition has become easier to process and more familiar?

That question cannot eliminate the illusory truth effect.

It creates, however, an opportunity for correction.

6. Epistemic Vigilance and Dual Provenance

Human knowledge is irreducibly social.

Individuals depend on testimony for the overwhelming majority of propositions they accept.

This dependence makes trust necessary and deception possible.

Sperber et al. (2010) developed the concept of epistemic vigilance to describe mechanisms involved in assessing communicated information and the trustworthiness of communicators.

Lucidism accepts this insight but argues that contemporary digital environments introduce another question.

It is no longer sufficient to ask:

Who produced this claim?

One must also ask:

Why did this claim reach me?

This yields the principle of dual provenance.

6.1 Claim provenance

Claim provenance concerns the origin of the proposition.

Relevant questions include:

• Who produced it?
• What is the source?
• Does the source possess relevant expertise or access?
• Is the claim traceable?
• Is supporting evidence independently accessible?
• Has the source demonstrated reliability in comparable cases?

6.2 Exposure provenance

Exposure provenance concerns the origin of the encounter.

Relevant questions include:

• Did I deliberately search for this information?
• Was it recommended by an algorithm?
• Was it shared by someone in my network?
• Was it promoted?
• Is it trending?
• Has it been repeatedly presented because I interacted with similar content?
• Is the system predicting that this material will maximize engagement?

This distinction is important because exposure itself can create epistemically misleading signals.

Popularity may resemble consensus.

Repetition may resemble corroboration.

Engagement may resemble importance.

Personalization may resemble representativeness.

Lucidism therefore proposes:

Visibility is not evidence of truth, and the mechanism producing visibility is itself epistemically relevant information.

7. Defining Lucidity

For theoretical purposes, ordinary-language notions of clarity or awareness are insufficient.

I define epistemic lucidity as:

the capacity to maintain appropriately differentiated relationships among evidence, belief, confidence, interpretation, attention, and action while preserving proportional openness to revision.

This definition contains several commitments.

7.1 Lucidity is not intelligence

Intelligence and lucidity may correlate under some conditions, but they are not identical.

High cognitive ability can support careful reasoning.

It can also support sophisticated rationalization.

The lucid agent is distinguished not by maximal computational capacity, but by the quality of the agent’s relationship to uncertainty and correction.

7.2 Lucidity is not certainty

A person may be highly lucid and wrong.

A person may also be correct for poor reasons.

Accuracy in one judgment and epistemic quality are related but not interchangeable.

Lucidity concerns not only whether a belief is true, but whether confidence is proportionate to available justification and whether the belief remains appropriately revisable.

7.3 Lucidity is not ideological neutrality

Lucidism does not require absence of moral, political, religious, or philosophical commitment.

The relevant question is whether the standards used to assess claims remain responsive to evidence even when conclusions interact with identity or values.

7.4 Lucidity is not generalized skepticism

Indiscriminate doubt is epistemically defective.

An individual who assigns low confidence to well-established evidence is not necessarily more sophisticated than one who is overconfident.

The objective is calibration, not universal uncertainty.

7.5 Lucidity is not emotional elimination

Emotion can signal importance, threat, injustice, attachment, loss, or value.

The theoretical requirement is narrower:

emotional intensity should not automatically be treated as evidential strength for an external factual proposition.

8. Epistemic Categories

A basic capacity underlying lucidity is the ability to distinguish claims that have different epistemic status.

At minimum, five categories should be separated.

8.1 Fact claim

A proposition whose truth can, in principle, be evaluated through publicly accessible evidence.

Example:

The article was published on a particular date.

8.2 Inference

A conclusion drawn from evidence but not identical to the evidence itself.

Example:

The observed pattern probably reflects a change in institutional strategy.

The inference may be excellent.

It remains an inference.

8.3 Interpretation

A meaning framework applied to events or evidence.

Different interpretations can sometimes accommodate the same basic empirical observations while emphasizing different causal structures or significance.

8.4 Normative judgment

A proposition about what ought to occur.

Example:

Recommender systems should prioritize informational quality over engagement.

Such a statement can be defended rationally.

Its normativity should not be disguised as direct empirical description.

8.5 Narrative

A structured organization of events, causal relations, interpretations, agents, and values into an intelligible whole.

Narratives are not inherently false.

They are unavoidable tools of human understanding.

The lucidist concern begins when the individual can no longer identify the transition between:

observed evidence

and

the narrative organizing that evidence.

9. The Four Capacities of Lucidism

The core model consists of four analytically distinct capacities.

9.1 Epistemic discrimination

Epistemic discrimination is the capacity to identify what kind of epistemic object is being evaluated.

Questions include:

• Is this a verified observation?
• An inference?
• A causal hypothesis?
• A prediction?
• An interpretation?
• A moral claim?
• A narrative?

This capacity reduces category errors.

A plausible hypothesis should not silently become an established fact.

A moral conviction should not be presented as though it were a measurement.

A correlation should not automatically become causation.

The first function of lucidity is therefore classificatory:

Before evaluating whether a claim is justified, identify what kind of claim it is.

9.2 Confidence calibration

Human beliefs need not be represented as binary states.

A proposition may be:

• extremely well established;
• highly probable;
• probable;
• plausible;
• uncertain;
• unlikely;
• unassessed.

Confidence calibration refers to the correspondence between subjective confidence and the quality of the available basis for judgment.

Fleming (2024) reviews the role of confidence in metacognition and emphasizes that confidence judgments depend on models individuals possess both of the world and of their own cognitive performance.

Recent misinformation research has increasingly converged on calibration as an important target.

Guigon, Geay, and Charpentier (2026), for example, argue that interventions focused exclusively on binary truth detection overlook the importance of plausibility estimation and calibrated confidence under uncertainty.

Lucidism incorporates this insight but places calibration within a broader architecture.

A lucid belief is not merely one accompanied by uncertainty.

It is one in which uncertainty is appropriately distributed.

The key question becomes:

How strongly should I believe this, given what I actually know?

9.3 Metacognitive reflexivity

Metacognitive reflexivity is the capacity to make one’s own process of belief formation an object of examination.

Questions include:

• Why do I believe this?
• What evidence changed my mind?
• What would reduce my confidence?
• Am I evaluating evidence or defending an identity?
• Would I accept this reasoning if it supported an opposing conclusion?
• Has repetition affected my confidence?
• Am I confusing access to information with understanding?
• Am I searching for truth or merely for confirmation?

Accuracy-based research shows that motivational orientation matters.

Rathje et al. (2023), across four experiments, found that accuracy incentives improved participants’ ability to discriminate between true and false political headlines and reduced partisan bias in judgments.

Pennycook and Rand (2022), synthesizing 20 experiments involving 26,863 participants, found that prompts directing attention toward accuracy improved the quality of intended news sharing.

Actively open-minded thinking provides another adjacent construct. It captures willingness to consider alternatives, sensitivity to counterevidence, resistance to premature closure, and reflective thought (Stanovich & Toplak, 2019).

Lucidism incorporates these insights while adding two requirements:

• explicit calibration of confidence;
• reflection on the architecture that shaped exposure itself.

9.4 Attentional sovereignty

The fourth capacity extends the earlier characterization of Homo lucidens as a structure of attention.

I define attentional sovereignty as:

the degree to which an agent can deliberately influence the selection, duration, sequencing, and interruption of informational exposure rather than leaving these processes entirely to external incentives or automatic habits.

“Sovereignty” here does not mean complete cognitive independence.

No such independence exists.

Attention is always influenced by biology, environment, social structures, and technological design.

The concept instead identifies effective capacity for intervention.

Behavioral indicators could include:

• overriding default feeds;
• deliberately searching beyond recommended content;
• diversifying sources;
• controlling notifications;
• postponing responses during high emotional arousal;
• interrupting automatic consumption;
• using reading queues rather than reactive feeds;
• seeking disconfirming evidence;
• separating information acquisition from immediate sharing.

Attentional sovereignty therefore transforms attention from a passive resource into a domain of agency.

10. How Lucidism Differs From Adjacent Frameworks

A new framework is useful only if it contributes something beyond existing constructs.

Lucidism overlaps substantially with several established literatures.

That overlap should be made explicit.

Framework	Primary concern	Relation to Lucidism
Epistemic vigilance	Trustworthiness of communicated information and communicators	Lucidism incorporates source evaluation but adds exposure provenance and attentional control
Accuracy prompts	Making truth/accuracy salient during judgment	Incorporated within metacognitive reflexivity
Actively open-minded thinking	Counterevidence, alternative hypotheses, delayed closure	Closely related to reflexivity and symmetry
Media literacy	Evaluation of media messages, sources, and practices	Provides domain-specific skills that can instantiate lucidist capacities
Metacognitive calibration	Alignment between confidence and performance/uncertainty	Constitutes one of Lucidism’s four core capacities
Recommender-system autonomy	Effects of recommendation systems on agency and choice	Contributes to the concept of attentional sovereignty
Psychological inoculation	Resistance to manipulative techniques before exposure	Compatible intervention strategy, but narrower than the proposed framework

Lucidism’s proposed contribution is therefore integrative but not merely aggregative.

Its distinctive theoretical structure is the conjunction of:

1. epistemic discrimination;
2. confidence calibration;
3. reflexive examination;
4. attentional sovereignty;
5. dual provenance;
6. symmetry of evidentiary standards;
7. explicit revisability.

The claim is not that none of these components exists in previous literature.

The claim is that their integration describes a specific form of second-order epistemic agency that can be treated as a research program.

11. The Principle of Epistemic Symmetry

One of the largest threats to any framework of rationality is asymmetric application.

Humans readily identify:

• bias in adversaries;
• propaganda in opposing groups;
• weak evidence supporting unwanted conclusions;
• emotional reasoning in ideological opponents.

The same standards may not be applied to identity-congruent claims.

Lucidism therefore introduces the principle of epistemic symmetry:

An evidentiary standard used to reject a claim should, under epistemically equivalent conditions, also be applicable when the same standard threatens a preferred conclusion.

This does not require pretending that all sources or propositions are equivalent.

The relevant condition is epistemic equivalence, not political symmetry.

If two claims have different evidential support, they should be judged differently.

The principle instead prevents motivated double standards.

The strongest test of lucidity may therefore occur not when an individual identifies the weakness of an adversary’s argument, but when evidence threatens a conclusion the individual prefers to preserve.

12. The Principle of Revisability

A factual belief should possess some relationship to potential counterevidence.

The strength required to revise a belief can appropriately vary.

Extraordinary claims may require extraordinary evidence.

Well-established conclusions should not be abandoned because of a single anomaly.

Revisability therefore does not imply instability.

It means that the agent can identify, at least in principle, conditions under which confidence would change.

The practical question is:

What evidence would make me less confident in this belief?

When the answer is “nothing,” a proposition may no longer be functioning primarily as an empirical belief.

It may instead operate as:

• identity;
• loyalty;
• metaphysical commitment;
• value;
• dogma;
• existential orientation.

Such commitments are not automatically illegitimate.

Lucidism requires only that their epistemic category be recognized.

13. Homo Lucidens as a Second-Order Agent

The framework permits a more precise reformulation of Homo lucidens.

A first-order epistemic agent asks:

Is this claim true?

A reflexive agent additionally asks:

Why do I believe it is true?

A lucid agent introduces another level:

What cognitive, social, technological, motivational, and attentional conditions shaped the space in which this claim became plausible to me?

This third question moves analysis from the proposition to the ecology of belief.

Accordingly:

Homo sapiens knows and represents. Homo lucidens attempts to perceive the conditions shaping its own knowing and representing.

This distinction should not be interpreted as a taxonomy of superior and inferior human beings.

Lucidity is proposed as dimensional, context-dependent, and fallible.

The same individual may be highly lucid in one domain and poorly calibrated in another.

Homo lucidens is therefore best understood as a theoretical model of agency rather than a status or identity.

14. Algorithms, Polarization, and Causal Restraint

Lucidism should reject technologically simplistic explanations.

Recommender systems matter.

But their effects depend on:

• platform design;
• ranking objectives;
• user characteristics;
• initial preferences;
• social networks;
• available content;
• political context;
• duration of exposure;
• measured outcomes.

The evidence illustrates this complexity.

Guess et al. (2023) found that moving Facebook and Instagram users from algorithmic to reverse-chronological feeds substantially altered platform use and exposure but did not detect significant changes across several political-attitude outcomes.

Nyhan et al. (2023) similarly found that reducing exposure to politically like-minded Facebook sources changed exposure patterns without measurable effects on a set of attitudinal measures.

Gauthier et al. (2026), however, found that switching active X users from a chronological to an algorithmic feed increased engagement and shifted several political attitudes during a seven-week experiment, while leaving self-reported partisanship and affective polarization without significant effects.

The theoretically responsible conclusion is neither:

algorithms control users,

nor:

algorithms are politically irrelevant.

A better proposition is:

algorithmic architectures alter distributions of exposure, salience, incentives, and behavior, and these changes can—under some conditions—affect beliefs and attitudes.

This probabilistic formulation creates conceptual space for attentional sovereignty.

If systems were completely deterministic, sovereignty would be meaningless.

If systems had no effects, it would be unnecessary.

15. Artificial Intelligence and Epistemic Lucidity

Generative artificial intelligence intensifies several problems addressed by the framework.

Large language models can produce text that is:

• fluent;
• structured;
• rhetorically confident;
• personalized;
• immediately available.

None of these properties guarantees truth.

The resulting principle is:

Linguistic fluency must not be treated as a proxy for evidential reliability.

This has implications both for risks and opportunities.

Artificial intelligence can reduce epistemic quality when it encourages:

• unverified acceptance;
• false authority attribution;
• substitution of generated coherence for evidence;
• dependency on a single informational interface.

The same technology can increase epistemic quality when used to:

• locate counterarguments;
• compare explanations;
• identify inconsistencies;
• generate alternative hypotheses;
• clarify uncertainty;
• distinguish claim types;
• search for falsifying evidence.

AI therefore does not occupy a fixed position within Lucidism.

It can increase either automation or reflection.

The determining variable is partly the mode of use.

This returns the analysis to Homo lucidens.

The relevant question is not simply whether intelligent systems become more capable.

It is whether human agents preserve the ability to observe and evaluate the mediation those systems provide.

16. Six Testable Propositions

A conceptual framework becomes scientifically productive when it generates claims capable of failure.

The following propositions are intended as initial hypotheses.

Proposition 1 — Epistemic discrimination

Individuals trained to distinguish factual claims, inferences, interpretations, normative judgments, and narratives will demonstrate better confidence calibration when evaluating epistemically heterogeneous information than matched controls.

The proposition can fail if classification training improves terminology without improving judgment.

Proposition 2 — Dual provenance

Prompting individuals to evaluate both claim provenance and exposure provenance will reduce reliance on repetition, virality, and popularity as implicit indicators of evidential strength.

A strong test should manipulate exposure signals independently of source quality.

Proposition 3 — Accuracy orientation

Interventions that increase the salience of accuracy as a cognitive objective will, on average, improve truth discernment, but their effects will vary with identity relevance, task context, and informational uncertainty.

This proposition extends existing accuracy-prompt research rather than treating the effect as universal.

Proposition 4 — Reflective delay

Introducing a brief reflective interval between emotionally activating information and a sharing decision will reduce impulsive sharing of insufficiently verified content without producing equivalent reductions in high-confidence, well-supported content.

The second clause is essential.

An intervention that suppresses all sharing equally would not demonstrate improved discernment.

Proposition 5 — Metacognitive calibration

Training individuals to explicitly represent uncertainty will improve correspondence between confidence and performance, even when improvements in raw accuracy are smaller or absent.

This distinguishes first-order accuracy from second-order calibration.

Proposition 6 — Attentional sovereignty

Individuals with greater behavioral control over feeds, notifications, source selection, and information-search routines will rely less strongly on algorithmic salience as an implicit indicator of importance or credibility.

This proposition requires behavioral measures rather than self-description alone.

17. Toward an Operational Model of Lucidity

The framework should not assume that lucidity is a single psychological factor.

A preliminary model should instead measure several dimensions separately.

Dimension A — Epistemic discrimination

Possible measures:

• classification accuracy for different claim types;
• causal versus correlational distinction;
• ability to identify uncertainty language;
• separation of observation from interpretation.

Dimension B — Confidence calibration

Possible measures:

• Brier scores;
• calibration curves;
• overconfidence indices;
• correspondence between confidence and accuracy.

Dimension C — Revisability

Possible measures:

• belief updating following diagnostic counterevidence;
• magnitude of posterior change;
• willingness to reduce confidence;
• sensitivity to evidence quality.

Dimension D — Epistemic symmetry

Possible measures:

Participants could evaluate matched arguments whose ideological or identity alignment is experimentally reversed.

A symmetry index could estimate differences in evidentiary thresholds across congruent and incongruent conditions.

Dimension E — Provenance sensitivity

Possible measures:

• ability to distinguish source quality;
• recognition of sponsored content;
• sensitivity to recommendation versus deliberate search;
• reduced use of engagement metrics as evidential cues.

Dimension F — Attentional sovereignty

Possible measures should prioritize behavior:

• frequency of feed overrides;
• diversity of deliberate searches;
• notification settings;
• interruption latency;
• source-switching behavior;
• voluntary exposure to counterevidence;
• time between emotionally activating exposure and sharing.

Self-reported sovereignty alone would be vulnerable to desirability bias.

18. A Minimal Lucidity Protocol

The conceptual framework can be translated into a compact sequence.

Before strongly accepting or publicly disseminating a consequential claim, the agent asks:

1. What exactly is being claimed?

Separate the proposition from rhetorical framing.

2. What type of claim is it?

Fact, inference, prediction, interpretation, norm, or narrative?

3. What evidence supports it?

Identify evidence independently of the conclusion.

4. Where did the claim originate?

Evaluate claim provenance.

5. Why did I encounter it?

Evaluate exposure provenance.

6. How confident should I be?

Represent uncertainty explicitly.

7. What would lower my confidence?

Establish revisability.

8. Would I use the same standard against a preferred conclusion?

Test epistemic symmetry.

9. Do I need to act now?

Separate informational urgency from psychological urgency.

The protocol is not an algorithm for truth.

Its purpose is error exposure.

19. Proposed Experimental Program

A first controlled evaluation could use a preregistered randomized design.

Participants would be assigned to:

Lucidism training

or

active educational control.

The experimental intervention would include modules on:

• claim classification;
• probabilistic confidence;
• source assessment;
• exposure provenance;
• evidence updating;
• actively seeking counterevidence;
• epistemic symmetry;
• attentional control.

The control intervention would contain general educational material of equivalent duration and cognitive demand.

19.1 Baseline measures

Before intervention:

• true/false news discernment;
• confidence calibration;
• actively open-minded thinking;
• source discrimination;
• political or identity attitudes;
• information-search behavior;
• baseline sharing intentions.

19.2 Immediate post-intervention outcomes

Primary outcomes could include:

1. truth discernment;
2. confidence calibration;
3. discrimination between observation and inference;
4. belief updating after counterevidence;
5. congruence-related asymmetry;
6. sharing discernment.

19.3 Behavioral task

Participants could enter a simulated information feed in which:

• source reliability;
• repetition frequency;
• popularity;
• ideological congruence;
• emotional intensity;
• recommendation labels

are independently manipulated.

This factorial design would permit estimation of how much each cue affects:

• perceived truth;
• confidence;
• information seeking;
• sharing;
• willingness to revise.

Attentional sovereignty could additionally be assessed by providing participants with optional tools to:

• switch to chronological ordering;
• disable engagement metrics;
• request original sources;
• search opposing evidence;
• postpone sharing.

Use of these tools would provide behavioral rather than purely attitudinal evidence.

19.4 Longitudinal follow-up

Immediate effects would be insufficient.

Follow-up after several weeks should determine whether gains persist.

This is important because misinformation interventions, including inoculation-based approaches, can show decay over time.

The relevant scientific question is not:

Did participants learn the terminology of Lucidism?

It is:

Did measurable changes in epistemic behavior survive outside the immediate training context?

20. Falsification and Revision Conditions

Lucidism must expose itself to the same revisability it demands of its agents.

The framework would require substantial modification if repeated high-quality studies showed that:

• epistemic-discrimination training improves vocabulary but not judgment;
• explicit calibration training increases subjective confidence without improving calibration;
• dual-provenance prompts add no predictive or behavioral value beyond ordinary source evaluation;
• attentional-sovereignty measures fail to predict any meaningful difference in information processing;
• supposed improvements disappear whenever claims become identity-relevant;
• training primarily increases perceived intellectual superiority;
• the proposed dimensions collapse psychometrically into already-established constructs without meaningful incremental validity;
• behavioral effects fail to generalize beyond laboratory tasks.

The seventh possibility is especially important.

If Lucidism merely redescribes actively open-minded thinking, metacognitive calibration, and media literacy without demonstrating incremental explanatory or predictive value, a new theoretical label would be unnecessary.

21. The Identity Paradox of Lucidism

The framework faces a distinctive internal danger.

If “Lucidist” becomes an identity associated with superior rationality, the theory can generate conditions opposed to its purpose.

The sequence could be:

Lucidism → identity → status → motivated self-protection → resistance to correction → reduced lucidity.

This produces the identity paradox of Lucidism:

The more strongly an individual derives epistemic status from believing oneself lucid, the harder it may become to detect one’s own failures of lucidity.

Consequently, Lucidism should not function as a title.

There should be no epistemically privileged class of “the lucid.”

Lucidity must remain:

• task-specific;
• measurable;
• provisional;
• corrigible.

The question is not:

Am I a lucid person?

It is:

How lucidly am I handling this particular claim under these particular conditions?

22. Lucidism Is Not Relativism

Acknowledging interpretation does not entail the equivalence of interpretations.

Some models:

• account for more observations;
• generate more accurate predictions;
• require fewer unsupported assumptions;
• withstand stronger tests;
• cohere better with established evidence.

Lucidism therefore presupposes minimal epistemic realism.

Its weakest necessary form is:

There can be a difference between what an individual wants to be true and what available evidence provides reason to believe.

Perfect epistemic access is unnecessary.

Reality need only possess the capacity to resist our representations.

Experiments can fail.

Predictions can be wrong.

Sources can contradict us.

Events can refuse to conform to our narratives.

That resistance provides the possibility of correction.

23. Lucidism Is Not Rationalist Purism

The framework also rejects the ideal of a perfectly rational, affect-free agent.

Humans:

• possess limited attention;
• use heuristics;
• depend on trust;
• reason under time constraints;
• operate through social structures;
• experience emotion;
• rely on incomplete evidence.

Near-perfect verification is therefore an unrealistic goal.

The purpose of Lucidism is not to eliminate bounded cognition.

It is to make some of its constraints available to metacognitive inspection.

Accordingly, the framework does not claim:

Homo lucidens has no biases.

It claims:

Homo lucidens attempts to organize practices that increase the probability that consequential biases, errors, and overconfidence will become detectable and correctable.

Lucidity is therefore better represented by corrigibility than by cognitive purity.

24. Relationship to Contemporary Misinformation Research

A further reason for avoiding rationalist idealization is that empirical evidence shows that ordinary truth judgment is more complex than a simple failure to detect falsehood.

Pfänder and Altay (2025), in a preregistered meta-analysis drawing on 67 experimental articles and nearly 200,000 participants across multiple countries, examined both correct detection of false news and incorrect rejection of true news.

This distinction matters.

An intervention that simply increases skepticism may reduce belief in false claims while also increasing rejection of true ones.

Such an intervention may not improve epistemic quality.

Lucidism therefore favors discernment over generalized distrust.

Similarly, Guigon, Villeval, and Dreher (2024) found that confidence in judgments concerning ambiguous news did not reliably track accuracy and that confidence itself influenced subsequent information seeking.

The implication is that individuals can be both wrong and confidently satisfied, or uncertain and motivated to seek more evidence.

This makes metacognitive calibration central to any theory of contemporary epistemic agency.

25. Cultural Interpretation

Although the present article frames Lucidism primarily as a theoretical research program, the concept also admits a cultural interpretation.

Digital environments frequently reward:

speed.

Lucidity may require delay.

They reward:

certainty.

Lucidity may require graded confidence.

They reward:

reaction.

Lucidity may require investigation.

They reward:

identity alignment.

Lucidity may require disconfirming evidence.

They reward:

salience.

Lucidity requires distinguishing salience from importance.

They reward:

continuous exposure.

Lucidity may require withdrawal.

In cultural terms, Lucidism can therefore be understood as an attempt to move the cognitive ideal from:

react more quickly

toward:

perceive more accurately before reacting.

Its conceptual opponent is not technology.

Nor is it emotion, politics, identity, or artificial intelligence.

Its opponent is unexamined automation in the formation of belief.

26. Limitations

Several limitations should be stated explicitly.

First, Lucidism is presently a conceptual framework, not a validated psychological construct.

Second, the proposed dimensions may overlap substantially with existing measures of metacognition, actively open-minded thinking, media literacy, and epistemic vigilance.

Third, the empirical literature motivating the framework disproportionately represents technologically developed and often Western populations.

Fourth, attentional sovereignty may depend on socioeconomic conditions. An individual with little time, high occupational demands, or restricted informational access may possess less practical capacity to redesign information exposure.

Fifth, deliberation is not universally superior to heuristic judgment. Expertise often permits rapid and accurate judgments.

Sixth, personalized recommendation is not inherently anti-autonomous. Recommender systems can reduce overload, facilitate discovery, and help users achieve self-endorsed goals (Bonicalzi et al., 2023).

Seventh, factual verification is often impossible for ordinary individuals. A mature theory of lucidity must therefore study rational trust, not merely independent verification.

Eighth, the term Lucidism has been used independently in other philosophical and cultural contexts. The present article therefore claims no lexical priority over the term. Its proposed contribution concerns the specific theoretical architecture developed here.

Ninth, the framework contains normative commitments—particularly commitments to evidence-responsiveness, symmetry, revisability, and autonomy—that cannot themselves be established solely through empirical psychology.

These limitations delimit the theory rather than invalidate it.

27. A Compact Formulation

The central thesis can be expressed in one sentence:

Lucidity is not the possession of truth; it is the organization of cognition and attention so that error has more opportunities to become visible.

This formulation intentionally lowers the ambition of the theory.

Lucidism does not promise epistemic invulnerability.

It does not create immunity to propaganda.

It does not eliminate identity.

It does not eliminate emotion.

It does not guarantee correct beliefs.

It attempts to strengthen the mechanisms through which:

• evidence can challenge confidence;
• confidence can represent uncertainty;
• attention can be redirected;
• assumptions can become explicit;
• sources can be examined;
• beliefs can be revised.

28. Conclusion

The central epistemic challenge of the algorithmic age is not simply that false information exists.

Human beings have always confronted falsehood.

The contemporary problem is structural.

Finite cognitive agents now operate inside environments in which information can be generated, reproduced, ranked, recommended, personalized, emotionally intensified, and repeated at scales that exceed individual capacities for direct evaluation.

The appropriate response cannot be total informational independence.

That is impossible.

Nor can it be unquestioned delegation to institutions, algorithms, experts, or artificial intelligence.

What remains is the problem of epistemic agency under dependence.

This article has proposed Lucidism as a second-order framework for that problem.

It organizes epistemic agency around four capacities:

• epistemic discrimination;
• confidence calibration;
• metacognitive reflexivity;
• attentional sovereignty.

These capacities are supplemented by three governing principles:

dual provenance, which distinguishes where a claim came from from why the individual encountered it;

epistemic symmetry, which requires comparable evidentiary standards across preferred and dispreferred conclusions;

and

revisability, which preserves a meaningful relationship between evidence and confidence.

The framework develops the earlier concept of Homo lucidens.

Where the earlier formulation identified an agent attempting to perceive technological mediation while preserving interiority, the present account transforms that intuition into a broader theory of epistemic agency.

Thus:

Homo lucidens is the agent.

> 

Lucidity is the capacity.

> 

Lucidism is the framework.

The theoretical transition is important.

A metaphor becomes scientifically valuable only when it begins to generate distinctions, measures, competing predictions, and possible failures.

For this reason, the future of Lucidism should not depend on its rhetorical appeal.

It should depend on whether its proposed dimensions can be operationalized, whether interventions based upon them improve measurable epistemic outcomes, whether the effects persist outside laboratory settings, and whether the framework provides explanatory value beyond already-established constructs.

If it fails those tests, it should be revised.

If its components prove redundant, they should be absorbed into better theories.

If its central predictions repeatedly fail, they should be abandoned.

This requirement is not external to Lucidism.

It is its most important internal commitment.

A theory of lucidity remains coherent only insofar as it preserves the possibility of discovering that it is wrong.

Declarations

Data Availability

This is a theoretical and conceptual article. No original dataset was generated or analyzed.

Ethics Statement

The article reports no original research involving human participants, animals, or personally identifiable data and therefore required no original ethics approval.

Funding

No specific funding is declared for this manuscript.

Competing Interests

The author declares no competing interests directly related to the arguments developed in this article.

Generative AI Disclosure

Generative artificial intelligence tools may be used for linguistic refinement, structural assistance, and editorial review. Intellectual responsibility for the article’s arguments, conceptual claims, source selection, interpretation, and final text remains with the author.

References

Biddlestone, M., Ziemer, C.-T., Maertens, R., Roozenbeek, J., & van der Linden, S. (2025). Norm-enhanced prebunking for actively open-minded thinking indirectly improves misinformation discernment and reduces conspiracy beliefs. Journal of Experimental Social Psychology, 121, 104818. doi:10.1016/j.jesp.2025.104818.

Bonicalzi, S., De Caro, M., & Giovanola, B. (2023). Artificial intelligence and autonomy: On the ethical dimension of recommender systems. Topoi, 42, 819–832. doi:10.1007/s11245-023-09922-5.

Ecker, U. K. H., Lewandowsky, S., Cook, J., Schmid, P., Fazio, L. K., Brashier, N., Kendeou, P., Vraga, E. K., & Amazeen, M. A. (2022). The psychological drivers of misinformation belief and its resistance to correction. Nature Reviews Psychology, 1, 13–29. doi:10.1038/s44159-021-00006-y.

Fleming, S. M. (2024). Metacognition and confidence: A review and synthesis. Annual Review of Psychology, 75, 241–268. doi:10.1146/annurev-psych-022423-032425.

Gauthier, G., Hodler, R., Widmer, P., et al. (2026). The political effects of X’s feed algorithm. Nature, 652, 416–423. doi:10.1038/s41586-026-10098-2.

Guigon, V., Geay, L., & Charpentier, C. J. (2026). Rethinking misinformation through plausibility estimation and confidence calibration. Communications Psychology, 4, 24. doi:10.1038/s44271-026-00413-y.

Guigon, V., Villeval, M. C., & Dreher, J.-C. (2024). Metacognition biases information seeking in assessing ambiguous news. Communications Psychology, 2, 122. doi:10.1038/s44271-024-00170-w.

Guess, A. M., Malhotra, N., Pan, J., et al. (2023). How do social media feed algorithms affect attitudes and behavior in an election campaign? Science, 381, 398–404. doi:10.1126/science.abp9364.

Lorenz-Spreen, P., Lewandowsky, S., Sunstein, C. R., & Hertwig, R. (2020). How behavioural sciences can promote truth, autonomy and democratic discourse online. Nature Human Behaviour, 4, 1102–1109. doi:10.1038/s41562-020-0889-7.

Mirhoseini, M., Early, S., El Shamy, N., & Hassanein, K. (2023). Actively open-minded thinking is key to combating fake news: A multimethod study. Information & Management, 60, 103761. doi:10.1016/j.im.2023.103761.

Nyhan, B., Settle, J., Thorson, E., et al. (2023). Like-minded sources on Facebook are prevalent but not polarizing. Nature, 620, 137–144. doi:10.1038/s41586-023-06297-w.

Pennycook, G., & Rand, D. G. (2022). Accuracy prompts are a replicable and generalizable approach for reducing the spread of misinformation. Nature Communications, 13, 2333. doi:10.1038/s41467-022-30073-5.

Pfänder, J., & Altay, S. (2025). Spotting false news and doubting true news: A systematic review and meta-analysis of news judgements. Nature Human Behaviour, 9, 688–699. doi:10.1038/s41562-024-02086-1.

Rathje, S., Roozenbeek, J., Van Bavel, J. J., & van der Linden, S. (2023). Accuracy and social motivations shape judgements of (mis)information. Nature Human Behaviour, 7, 892–903. doi:10.1038/s41562-023-01540-w.

Rose, J. M., Ganbold, O., Rose, A. M., Thibodeau, J. C., et al. (2024). Overcoming resistance to belief revision and correction of misinformation beliefs: Psychophysiological and behavioral effects of a counterfactual mindset. Scientific Reports, 14, 12493. doi:10.1038/s41598-024-63230-5.

Simon, H. A. (1971). Designing organizations for an information-rich world. In M. Greenberger (Ed.), Computers, Communications, and the Public Interest (pp. 37–72). Johns Hopkins Press.

Souza Silva, J. (2025). From Homo Sapiens to Homo Lucidens: The Emergence of a Symbolic AI in Interaction with Language Models. Zenodo. doi:10.5281/zenodo.17612875.

Sperber, D., Clément, F., Heintz, C., Mascaro, O., Mercier, H., Origgi, G., & Wilson, D. (2010). Epistemic vigilance. Mind & Language, 25(4), 359–393. doi:10.1111/j.1468-0017.2010.01394.x.

Stanovich, K. E., & Toplak, M. E. (2019). The need for intellectual diversity in psychological science: Our own studies of actively open-minded thinking as a case study. Cognition, 187, 156–166. doi:10.1016/j.cognition.2019.03.006.

van der Linden, S. (2022). Misinformation: Susceptibility, spread, and interventions to immunize the public. Nature Medicine, 28, 460–467. doi:10.1038/s41591-022-01713-6.

Vellani, V., Zheng, S., Ercelik, D., & Sharot, T. (2023). The illusory truth effect leads to the spread of misinformation. Cognition, 236, 105421. doi:10.1016/j.cognition.2023.105421.

Ye, S., Attali, D., Ghazi, M., Cachia, A., Cassotti, M., & Borst, G. (2026). Systematic review and meta-analysis of the evidence for an illusory truth effect and its determinants. Nature Communications, 17, 3270. doi:10.1038/s41467-026-70041-x.