Are we willing to share what we believe is true? Factors influencing susceptibility to fake news

Michał Piksa, Karolina Noworyta, Aleksander B. Gundersen, Jonas R. Kunst, Mikołaj Morzy, Jan Piasecki, Rafał RygułaFrontiers in Psychiatry 2023 · DOI

TL;DR

While we are surprisingly good at identifying whether a news story is objectively true or false, our decision to "like" or "share" it has almost nothing to do with its veracity. Instead, social media engagement is driven primarily by cognitive utility—how well the information reinforces our existing worldview—suggesting that fact-checking alone is an insufficient weapon against the spread of misinformation.

What Problem Does This Paper Address?

The "infodemic" accompanying the COVID-19 pandemic highlighted a frustrating paradox: even when high-quality fact-checking is available, fake news continues to go viral. Most research focuses on why people believe lies, but this paper tackles the gap between veracity judgment (is this true?) and behavioral engagement (will I share this?).

The authors frame this through the lens of Friston’s "free-energy principle," suggesting that humans are "entropy-minimizing" machines. We don't just look for truth; we look for information that reduces the gap between our internal mental models and the outside world. If a piece of fake news helps "make sense" of the world according to our biases, we are biologically incentivized to engage with it.

Methods

The researchers employed a 2x2x2 within-subjects factorial design to test how three specific factors influence our reaction to news:

  1. Truthfulness: Is the news objectively true (based on WHO guidelines) or fake (invented/verified false)?
  2. Cognitive Utility: Is the news congruent or incongruent with the participant's pre-existing views on COVID-19?
  3. Presentation Style: Is the news "sober" (factual, neutral) or "populistic" (sensationalized, high-contrast images, emotional language)?

They recruited 201 American adults via Prolific, filtering for high-quality respondents. Participants were categorized as "Acceptors" (viewed COVID-19 as a real threat) or "Denialists" (viewed it as a hoax) to determine the "congruence" of the news items shown to them. Participants rated 80 news items on a 6-point Likert scale for both veracity and likelihood of engagement (liking/sharing).

Datasets

The study utilized a custom-designed dataset of 80 news items formatted to mimic Facebook posts (headline, image, subtext, and source).

  • True items: Sourced from official WHO guidelines.
  • Fake items: Invented by the research team and verified as false.
  • Congruence: Items were balanced so that half aligned with "Acceptor" views and half with "Denialist" views.

The authors have made this dataset publicly available via the Jagiellonian University Repository to encourage further research.

Key Insights and Ideas

  • The Veracity-Engagement Decoupling: The most striking "aha moment" is that the cognitive processes for judging truth and deciding to share are distinct. People are actually quite decent at spotting fake news, but they often choose to share it anyway if it supports their "team."
  • The "Trojan Horse" Hypothesis: The authors find strong support for the idea that misinformation is most effective when it acts as a "carrier." If a post contains information that is highly congruent with a user's beliefs, it "drags" the false information along with it into the user's social circle.
  • The Failure of Sensationalism: Interestingly, "sober" presentation styles were often more effective at garnering both belief and engagement than "populistic" or sensationalized styles. This suggests that the "shouting" match of social media might be less effective than fake news that mimics the calm tone of traditional journalism.
  • Entropy over Accuracy: Sharing behavior is less about spreading "truth" and more about reducing the "sensory entropy" of one's mental model. We share things that make our world feel more predictable and consistent.

Key Results

  • Veracity Judgment: Objective truth was the strongest predictor of whether someone thought a story was true ($F = 408.87, p < 0.001$).
  • Engagement Paradox: Behavioral engagement was significantly higher for fake news than for true news ($F = 64.10, p < 0.001$).
  • Congruence is King: The probability of liking or sharing was much more closely tied to whether the news fit the participant's preconceived beliefs ($F = 118.30, p < 0.001$) than whether it was actually true.
  • Interaction Effects: The "most engaging" items were those that were true, congruent, and presented in a sober manner. However, if a story was fake but congruent, it still received high engagement, whereas true but incongruent stories were often ignored.

Limitations and Future Work

The primary limitation is the sample imbalance: only 13 participants identified as "Denialists" compared to 186 "Acceptors." While the statistical model (ANOVA) is robust to this, it limits the generalizability of findings regarding those with fringe views.

Furthermore, the study was conducted in a simulated environment. Real-world social media involves social pressure, algorithmic curation, and "echo chamber" effects that aren't fully captured in a survey.

Future Work: The authors suggest moving toward real-world observational data (e.g., analyzing Twitter/X or Facebook sharing histories) and using machine learning to develop psycholinguistic models that can predict which "Trojan Horse" stories are most likely to go viral. They also advocate for public health campaigns that focus on "compassion and understanding of individual worldviews" rather than just debunking facts.

Summary updated

Abstract

We conclude a common notion that the spread of fake news can be limited by fact-checking and educating people might not be entirely true, as people will share fake information as long as it reduces the entropy of their mental models of the world. We also find support for the Trojan Horse hypothesis of fake news dissemination.
BibTeX
@inproceedings{piksa2023are,
  author = {Michał Piksa and Karolina Noworyta and Aleksander B. Gundersen and Jonas R. Kunst and Mikołaj Morzy and Jan Piasecki and Rafał Ryguła},
  title = {{Are we willing to share what we believe is true? Factors influencing susceptibility to fake news}},
  booktitle = {Frontiers in Psychiatry},
  year = {2023},
  doi = {10.3389/fpsyt.2023.1165103},
}
Figure