Psychological Foundations of Illusory Truth

The core concept: processing fluency

Cognitive fluency describes the subjective ease with which our brain handles information. The easier it is, the more pleasant it feels — and the more we mistake it for "true," "beautiful," or "good."

graph TD
    A[Information] --> B{Fluent?}
    B -->|High| C[Positive feeling]
    B -->|Low| D[Negative feeling]
    C --> E[Attribution: true / familiar / right]
    D --> F[Attribution: false / suspicious / strange]
    style C fill:#c8e6c9
    style D fill:#ffcdd2

Fluency increases with:

A message can therefore gain "perceived truth" without changing its content — just by becoming more fluent.

The 3 cognitive pillars of the effect

1. Familiarity confused with truth

This is the historic core mechanism (Hasher, Goldstein & Toppino, 1977). System 1 uses familiarity as a truth heuristic: "If I've heard it before, it's probably because it's true."

This heuristic evolved for good reasons: in an ancestral environment, what's heard repeatedly is on average more reliable. In an environment saturated with information (media, ads, social networks, LLMs), it becomes a trap.

2. The memory trace

Each exposure strengthens a memory trace. The brain doesn't reliably distinguish the source ("I read it on a trusted site" vs. "I scrolled past it on Twitter"). It often stores the content but not its context.

This is the source confusion effect: with time, we remember the message but forget where we heard it. Repetition alone has been enough to etch the trace.

Key study — Begg, Anas & Farinacci (1992): 36 statements were presented, some labelled "this is true" and others "this is false." A week later, subjects judged as true both the genuinely-true ones and the false-but-repeated ones (presented once with the "false" tag, then forgotten). Repetition beats the warning label.

3. The illusion of prior knowledge

The brain interprets familiarity as evidence that "I already knew this." This illusion boosts confidence, which in turn boosts belief.

graph LR
    A[Repeated exposure] --> B[Familiarity]
    B --> C['I already knew this']
    C --> D[Subjective confidence]
    D --> E[Information credibility]
    E --> F[More sharing / defence]
    F --> A
    style A fill:#e1f5fe
    style F fill:#fff3e0

It's a self-reinforcing loop: the more convinced we are, the more we share, the more the environment reinforces familiarity.

Three studies that change your view

Fazio, Brashier, Payne & Marsh (2015) — knowledge doesn't immunise

Subjects read claims like "A sari is the traditional outfit of women in Scotland." Some know the right answer (kilt). Yet after a single repetition, the false claim is judged more credible — even by subjects who know it's false.

Business implication: don't assume your expert prospects will resist repeated claims. They'll resist a bit better, but not enough to cancel the effect.

Pennycook, Cannon & Rand (2018) — fake news

A single exposure to a false headline ("Trump wants to send Americans to Mars by 2024") boosts its perceived credibility one week later. The "Disputed by 3rd party fact-checkers" warning reduces — but does not eliminate — the effect.

Implication: a fact-check helps but always arrives after exposure. Repetition keeps a head start.

Lin et al. (2022) — TruthfulQA and LLMs

LLMs fine-tuned on the web reproduce and amplify common falsehoods ("humans only use 10% of their brain"). The bigger the model, the more fluent it is on these claims — and the more credible it makes them.

Implication: LLMs are fluency machines. Without guardrails, they're natural amplifiers of illusory truth.

Limits and conditions of the effect

The effect is robust but not universal. It is reduced when:

The effect is strengthened when:

• The claim is plausible (not absurd)
• The content is complex (System 2 too costly)
• The subject is under cognitive load (mobile, multitasking, fatigue)
• Exposures are spaced in time (spacing effect)
• The channel varies (audio + text + image)

Connection to other biases

graph TB
    A[Illusory truth] -.-> B[Mere exposure effect]
    A -.-> C[Availability heuristic]
    A -.-> D[Halo effect]
    A -.-> E[Confirmation bias]
    A -.-> F[LLM hallucinations]
    B --> G[Familiar = liked]
    C --> H[Easy to recall = frequent]
    D --> I[Familiar = competent]
    E --> J[We seek confirming info]
    F --> K[Fluency generated by model]
    style A fill:#fff9c4,stroke:#f57f17

Understanding these close cousins lets you dose your interventions:

• Want to plant an idea? Illusory truth (repetition).
• Want to plant a positive feeling? Mere exposure.
• Want to plant a conviction? Combine both + authority + social proof.

Synthetic diagram

graph LR
    A[Stimulus] --> B[Cognitive processing]
    B --> C{Fluent?}
    C -->|Yes| D[High fluency]
    C -->|No| E[Low fluency]
    D --> F[Felt familiarity]
    F --> G['True' heuristic]
    G --> H[Reinforced belief]
    E --> I[Effort, suspicion]
    style D fill:#c8e6c9
    style H fill:#a5d6a7
    style E fill:#ffcdd2
    style I fill:#ef9a9a

Key takeaways

• Processing fluency drives the bias: easy = true.
• Three pillars: familiarity, memory trace, illusion of prior knowledge.
• The effect persists even when the audience knows the right answer (Fazio et al., 2015).
• LLMs naturally amplify illusory truth — a stack risk worth knowing.
• The effect shrinks but doesn't disappear with expertise, motivation and warnings.

→ Next chapter: foundations quiz, then sales applications.