The Empirical-Mythological Inverse Feedback Modelنموذج التغذية الراجعة العكسية بين البيانات والأسطورة

Data does not become useless when reality is manipulated. It becomes insufficient.

That is the reason for the empirical-mythological inverse feedback model. The name is awkward because the problem is awkward. In a distorted information environment the empirical signals still matter: trust surveys, bot activity, polling shifts, platform behavior, speech patterns, voting data, institutional confidence, market reactions. But those signals describe the surface of a process driven by symbols, fear, memory, status, humiliation, and collective narrative. A model that reads only the surface mistakes the smoke for the fire.

“Mythological” here does not mean false. It means the shared symbolic structures through which a group makes suffering, power, danger, belonging, and destiny intelligible. Jung called them archetypes. Plato, when he came to found his city, reached for a story rather than an argument: the myth of the metals in Book III of the Republic, a fiction the rulers tell so that citizens will accept the order they live under. Modern social science has its own vocabularies for the same difficulty: cultural cognition, the narrative policy framework, political myth, social identity, moral foundations, collective memory. The words differ. The problem is the same. Human beings do not process public life as isolated fact machines.

The model has two movements.

The first movement runs from empirical data to symbolic claim. Start with the measurable layer. Do not begin with a myth and go hunting for proof. Begin with observed behavior: trust collapse after a scandal, sudden support for a strongman, refusal to accept correction, fear clustered around migration, technology, crime, sect, class, or national humiliation. Then ask which symbolic structures appear to organize that behavior.

The symbolic claim has to risk failure. “The savior archetype is active” is not enough. What would count as evidence? Increased support for personalist leadership during institutional distrust? Language that frames politics as rescue rather than representation? Greater tolerance for emergency powers? Reduced interest in procedural constraint? If those indicators do not appear, the claim is discarded or revised.

The second movement runs back from symbolic claim to empirical interpretation. Once a symbolic structure survives testing, it can help read new data. A spike in authoritarian preference may not be only a policy response. It may be the activation of rescue politics under humiliation. A conspiracy wave may not be only ignorance. It may be a way to restore agency when institutions have become illegible. A disinformation campaign may spread not because its claims are credible, but because it names an injury the audience already carries.

That is the inverse move. Data filters myth. Myth, once filtered, explains patterns in data that data alone leaves flat.

The danger is obvious. Without discipline the model becomes pseudoscience: it names archetypes with theatrical confidence and stops testing anything. That version deserves to be rejected.

So the method needs rules.

First, every symbolic claim must be operationalized. If loss of agency is doing the explanatory work, specify observable proxies: language of powerlessness, demand for punitive leadership, withdrawal from civic participation, attraction to simplified enemies, survey responses about control.

Second, every symbolic claim must compete against non-symbolic explanations. Material insecurity, party identity, media exposure, education, age, religion, geography, and institutional performance may account for the pattern better. The symbolic layer earns its place only if it adds explanatory power once those are in the model.

Third, the method must document what failed. A serious research program records the narratives it discarded. If every myth survives, the filter is fake.

Fourth, the model must stay local before it becomes general. The savior figure in one society is not automatically the strong leader in another. Symbols travel, but they change shape through language, religion, class, colonial memory, and state form.

A concrete case makes the demand clearer. Take vaccine refusal. A data-only model can track partisan identity, exposure to false claims, local infection rates, education, and trust in medical institutions. A symbolic layer asks which deeper structures give refusal its meaning: bodily sovereignty, distrust of state intrusion, purity, religious boundary, racialized medical memory, anti-elite identity, fear of dependency. The empirical phase tests which of those predicts behavior once the ordinary variables are held constant. The inverse phase then asks the harder question, and the existing evidence poses it sharply.

In a randomized trial of 1,759 American parents, Brendan Nyhan, Jason Reifler, Sean Richey, and Gary Freed found that information refuting the claimed link between the measles vaccine and autism did reduce belief in that link. None of the four messages they tested raised the stated intention to vaccinate, and the refutation lowered it among the parents whose attitudes toward vaccines were least favorable to begin with. The factual layer moved. The behavioral layer did not, and in the most resistant group it moved the wrong way.

That finding has to be read carefully, because the literature surrounding it is widely misreported. Earlier work by Nyhan and Reifler described a backfire effect, in which a correction left the targeted group more committed to the misperception than before. The effect turned out to be rare. Wood and Porter tested fifty-two issues across more than ten thousand subjects and found no correction capable of triggering it. Nyhan has since written that corrective information usually produces modest but real gains in accuracy, and that the persistence of misperception is better explained by the decay of those gains, and by contrary cues from elites and media, than by any recoil in the correction itself.

So the claim is not that facts fail. The claim is narrower and harder to answer. Facts often land on the factual layer and stop there. The correction settles the medical proposition while leaving untouched whatever the refusal was carrying: a judgment about who may enter the body, who has lied before, who is being asked to trust whom. The gap between corrected belief and unchanged conduct is exactly where a symbolic layer would have to earn its place, and it is measurable.

The point is not to replace empirics with myth. The point is to stop assuming that empirics can read symbolic life without a theory of symbols.

Sovereignty raises the stakes. Borderless information operations do not only spread false claims. The RAND account of the contemporary Russian model describes something closer to saturation: high volume across many channels, rapid, continuous and repetitive, with no commitment to objective reality and none to internal consistency. A campaign built on those lines is not trying to win an argument. That such campaigns work by locating symbolic weak points is a hypothesis rather than a finding, and it is the hypothesis this model exists to test. The candidate weak points are humiliation, sectarian fear, lost greatness, corrupted elites, foreign puppetry, demographic panic, religious violation. A state defending only the factual layer may win the corrections and lose the public. A state defending only the mythic layer becomes propaganda itself. The difficult task is to defend shared meaning without falsifying reality.

That is why the model belongs inside the larger Project of Synthesis. It is not anti-empirical. It is model pluralism under context discipline. It lets different methods work where each is strongest, then forces them to answer to each other.

The model is not proven. It is a research program, and it has a first test. Choose several cases in which correction moved belief without moving conduct, code both the empirical and the symbolic variables, and ask whether the symbolic layer improves prediction after the ordinary variables are included. If it does not, the model shrinks. If it does, it earns more room.

That is the only way this idea stays alive without becoming mysticism with footnotes.

Sources

• Carl Jung, The Archetypes and the Collective Unconscious, Collected Works vol. 9 part 1, 2nd ed., Princeton University Press, 1969.
• Plato, Republic, Book III, 414b-415d.
• Dan M. Kahan, “Cultural Cognition as a Conception of the Cultural Theory of Risk,” in Handbook of Risk Theory: Epistemology, Decision Theory, Ethics, and Social Implications of Risk, Springer Netherlands, 2012, 725-759. https://doi.org/10.1007/978-94-007-1433-5_28
• Deborah Stone, Policy Paradox: The Art of Political Decision Making, 3rd ed., W.W. Norton, 2012.
• Emery Roe, Narrative Policy Analysis: Theory and Practice, Duke University Press, 1994.
• Chiara Bottici, A Philosophy of Political Myth, Cambridge University Press, 2007.
• Jonathan Haidt, The Righteous Mind: Why Good People Are Divided by Politics and Religion, Pantheon, 2012.
• Brendan Nyhan and Jason Reifler, “When Corrections Fail: The Persistence of Political Misperceptions,” Political Behavior 32, no. 2 (2010), 303-330. https://doi.org/10.1007/s11109-010-9112-2
• Brendan Nyhan, Jason Reifler, Sean Richey, and Gary L. Freed, “Effective Messages in Vaccine Promotion: A Randomized Trial,” Pediatrics 133, no. 4 (2014), e835-e842. https://doi.org/10.1542/peds.2013-2365
• Thomas Wood and Ethan Porter, “The Elusive Backfire Effect: Mass Attitudes’ Steadfast Factual Adherence,” Political Behavior 41, no. 1 (2019), 135-163. https://doi.org/10.1007/s11109-018-9443-y
• Brendan Nyhan, “Why the Backfire Effect Does Not Explain the Durability of Political Misperceptions,” Proceedings of the National Academy of Sciences 118, no. 15 (2021), e1912440117. https://doi.org/10.1073/pnas.1912440117
• Christopher Paul and Miriam Matthews, “The Russian ‘Firehose of Falsehood’ Propaganda Model: Why It Might Work and Options to Counter It,” RAND Corporation, PE-198-OSD, 2016. https://www.rand.org/pubs/perspectives/PE198.html