Belief, Evidence, and Elites in the Age of Mass Information¶

Speaker: Kevin O'Connor, UC Irvine
Venue: Rotman Institute of Philosophy, University of Toronto, 2026
Video: YouTube

TL;DR¶

False beliefs persist and spread through social factors—conformity, institutional curation of evidence, and network structure—rather than individual cognitive biases alone. O'Connor examines historical cases (variolation in 18th-century England, tobacco industry strategy) and mathematical models to show how unbiased actors (journalists, industry-funded scientists) can systematically promote false beliefs without fraud by selectively sharing evidence or funding specific research methods.

Main Arguments¶

1. Belief is fundamentally social. Humans learn most beliefs from other people, not from direct experience. While cognitive biases matter, they explain only part of why false beliefs persist. The social structure of belief transmission—who we trust, how information flows through networks, conformity pressures—is equally or more important.

2. Conformity bias prevents belief change. The Asch conformity experiments (1950s) showed ~33% of subjects publicly agree with obviously false statements to avoid contradicting a group. Even people who privately know the truth will withhold it if doing so would violate social norms. In tight-knit communities with weak external links, this creates stable polarization: good beliefs can't spread inward, bad beliefs can't spread outward.

3. Industrial influence doesn't require fraud. The tobacco industry used two subtle strategies: - Selective sharing: Distributing only those real studies that happen (by chance variation) not to find a tobacco–cancer link, while suppressing the much larger body of positive evidence. - Industrial selection: Funding scientists already using methods favorable to industry findings. The researchers do honest work; the funding just selects which questions get asked and which methods receive resources. Example: funding anti-arrhythmia drug efficacy studies (they work at stopping arrhythmias) while avoiding mortality studies (the drugs kill people). Result: ~100,000 deaths over a decade.

4. "Fair reporting" can systematically promote false beliefs. Journalists following the fairness doctrine—giving equal airtime to "both sides"—overweight minority false beliefs when evidence heavily favors one position. Mathematical models show this always produces worse population beliefs than reporting proportional to the actual body of evidence. In scientific contexts, fairness harms accuracy.

5. Other journalistic norms backfire. Single-study reporting (treating one noisy study as definitive), novelty bias (surprising findings often mean false positives), and reporting on false claims (spreading them even while debunking them) all exacerbate misinformation spread.

6. Social media amplifies these effects. Platforms enable conformist clustering, allow propagandists to impersonate trusted sources and target vulnerable communities, and algorithmically reward novelty and emotional content over accuracy.

Historical Cases¶

Variolation (1700s). Lady Mary Wortley Montagu encountered inoculation against smallpox in Turkey—a lifesaving practice with ~5% mortality vs. 50% for natural infection. Despite knowing it worked and having a physician (Maitland) who'd performed it successfully, English physicians refused to adopt it; even Maitland wouldn't do it in England under peer scrutiny. Conformity bias silenced the evidence. The practice eventually spread only when Princess Caroline of Ansbach (the King's wife) publicly chose variolation for her children—exploiting conformity in the opposite direction.

Tobacco and cancer (1950s–1980s). After Consumer Digest published overwhelming evidence linking smoking to cancer, tobacco company sales dropped for the first time in decades. The industry hired a PR firm to "fight science with science," creating the tobacco industry research committee (a propaganda body, not a real research institute). They distributed pamphlets claiming "9 studies find no link" while omitting the 50+ that found a link. This strategy was so effective it became a template for climate denial, ozone depletion denial, and acid rain denial—all orchestrated by overlapping networks of scientists and PR firms.

Anti-arrhythmic drugs (1970s–1980s). Heart arrhythmias precede heart attacks. Question: if we stop arrhythmias, do we prevent death? Scientists split: some studied "do anti-arrhythmia drugs stop arrhythmias" (yes), others studied "do they prevent death" (no). Pharmaceutical companies funded only the former. Based on that biased evidence base, companies manufactured and sold the drugs widely. It took ~10 years to identify that the drugs cause heart attacks, killing an estimated 100,000+ people.

Implications for Online Misinformation¶

Conformity effects intensify on platforms with curated social networks (echo chambers, algorithmic clustering).
Propagandists can directly impersonate trusted sources and target tight-knit communities (e.g., Russian operations targeting Bernie Sanders supporters before 2016 to suppress voter turnout).
Journalistic incentive problems worsen: with readership driven by clicks/likes, pressure for catchy, novel, emotional content intensifies.
Novel tactics keep evolving. "Fake news" (2016) was familiar by 2018, displaced by deepfakes and memes.

Key Takeaways¶

Misinformation can be real, non-fraudulent data, selectively shared.
Institutions operating under standard practices can still systematically harm belief accuracy.
Trust and expertise matter. Science, despite flaws, is more reliable than following ungrounded individual "experts."
Simply providing more information doesn't change polarized minds. Building bridges through shared identity and trust is more effective.
We cannot assume we know what misinformation looks like online; subtlety and adaptation mean ongoing vigilance.

Connections¶

Notes¶

A lucid, research-grounded talk that balances historical depth with mathematical modeling. O'Connor successfully argues that individual-level cognitive explanations are incomplete; the social and institutional structures of belief transmission deserve equal attention. Her emphasis on how unbiased actors (honest scientists, responsible journalists, well-meaning institutions) can systematically produce false beliefs is a valuable corrective to cynical "it's all lies" framings and oversimplified "just educate people" solutions.

The variolation case and the contrast with modern anti-vaccine clusters is particularly striking—same conformity mechanisms, different era. The critique of "fairness" in science journalism is important and under-discussed in media criticism.