The Psychology of Fake News

Recent evidence contradicts the common narrative that partisanship and politically motivated reasoning explain why people fall for 'fake news'.Poor truth discernment is linked to a lack of careful relevant knowledge, as well use familiarity source heuristics.There also large disconnect between what believe they will share on social media, this largely driven by inattention rather than purposeful sharing misinformation.Effective interventions can nudge media users think about accuracy, leverage crowdsourced veracity ratings improve ranking algorithms. We synthesize burgeoning literature investigating false or highly misleading news online. Contrary whereby politics drives susceptibility fake news, are ‘better’ at discerning from falsehood (despite greater overall belief) when evaluating concordant news. Instead, poor associated with heuristics such familiarity. Furthermore, there substantial media. This dissociation inattention, more so misinformation. Thus, successfully focus accuracy. Crowdsourced be leveraged Fabricated nothing new. For example, in 1835 The Sun newspaper New York published six articles purported life moon which came known 'Great Moon Hoax'. During 2016 US Presidential Election UK Brexit Referendum, however, different form (see Glossary) rose prominence (Box 1): political 'news' stories, primarily originating [1.Lazer D. et al.The science news.Science. 2018; 359: 1094-1096Crossref PubMed Scopus (727) Google Scholar]. Concern was redoubled 2020 face widespread misinformation disinformation [2.Wardle C. Information Disorder: Essential Glossary, Shorenstein Center Media, Politics, Public Policy. Harvard Kennedy School, 2018Google Scholar] coronavirus disease 2019 (COVID-19) pandemic [3.Loomba S. al.Measuring impact COVID-19 vaccine vaccination intent USA.Nat. Hum. Behav. 2021; (Published online February 5, 2021. https://doi.org/10.1038/s41562-021-01056-1)Crossref (0) [4.Pennycook G. Rand D.G. Examining beliefs voter fraud wake Election.Harvard Sch. Misinformation Rev. 2: 1-19Google Misleading hyperpartisan yellow journalism [5.Kaplan R.L. Yellow journalism.in: Donsbach W. International Encyclopedia Communication. John Wiley & Sons, 2008Crossref Scholar], related forms problematic content likely sources polarization [6.Faris R.M. al.Partisanship, Propaganda, Disinformation: Online Media U.S. Election. Berkman Klein Internet Society, 2017Google What it human psychology – its interaction [7.Lewandowsky al.Technology Democracy. Understanding Influence Technologies Political Behaviour Decision-Making, EU Science Hub2020Google Scholar,8.Kozyreva A. al.Citizens versus internet: confronting digital challenges cognitive tools.Psychol. Sci. Interest. 2020; 21: 103-156Crossref (2) explains failure distinguish accurate inaccurate online? Apart being theoretical interest, question has practical consequences: developing effective against depends understanding underlying psychology.Box 1Prevalence Fake NewsVarious analyses web browsing data have been used an attempt determine prevalence often using data, archives fact-checking websites, survey, Allcott Gentzkow [19.Allcott H. M. Social election.J. Econ. Perspect. 2017; 31: 211-236Crossref (1218) estimated particular set stories were shared Facebook least 38 million times 3 months leading up election (30 favoring Donald Trump). estimate represents lower bound since only reflects specific news.Other focused publishers (i.e., websites) individual articles. Based Twitter [117.Grinberg N. al.Fake twitter during election.Science. 2019; 363: 374-378Crossref (206) [77.Allen J. al.Scaling wisdom crowds.PsyArXiv. October 2, 2020. http://dx.doi.org/10.31234/osf.io/9qdza)Google Scholar,118.Guess A.M. al.Less you think: predictors dissemination Facebook.Sci. Adv. 5eaau4586Crossref (4) [89.Guess al.Exposure untrustworthy websites election.Nat. 4: 472-480Crossref (21) these studies concluded sites small proportion most people's diets, average user exposed little election.These important limitations, because available concern visit click through off-platform. But, course, vast majority time simply read post without clicking link actual website. so-called news' one category misinformation, much larger diets Scholar,119.Bradshaw al.Sourcing automation information over United States, 2016–2018.Polit. Commun. 37: 173-193Crossref (6) on-platform exposure remains open [120.Rogers R. scale Facebook's problem upon how classified.Harvard 1: 1-15Google feel premature conclude rates minimal, thus not (also Scholar]). especially true looking beyond new threats claims Scholar,44.Pennycook al.Fighting media: experimental scalable accuracy intervention.Psychol. 770-780Crossref (73) gained traction amplification (mostly Republican) elites.Accordingly, (and broadly) equally distributed across all users. In particular, conservatives older adults far Scholar,89.Guess Scholar,117.Grinberg Studies found associations conservatism belief USA [20.Pennycook Lazy, biased: partisan better explained reasoning.Cognition. 188: 39-50Crossref (177) Chile [121.Halpern al.From conspiracy theories trust others: factors influence exposure, believing news.in: Meiselwitz Computing Media. Design, Human Behavior Analytics. HCII 2019. Lecture Notes Computer Science. vol 11578. Springer, Cham2019: 217-232Crossref (5) Germany [122.Zimmermann F. Kohring Mistrust, disinforming vote choice: panel survey origins consequences 2017 German Parliamentary Election.Polit. 215-237Crossref (8) but Hungary [24.Faragó L. al.We we doctored ourselves: connection news.Soc. Psychol. (Gott). 51: 77-90Crossref who engage less lower-quality [71.Mosleh al.Cognitive reflection correlates behavior Twitter.Nat. 12: 1-10Crossref even if exposures substantially higher subpopulations may particularly vulnerable content. Finally, originates sometimes transitions audiences picked traditional outlets either via direct repetition debunking (which result inadvertent amplification). Various Other election. These elites. Accordingly, here presented However, come many forms, several literatures clearly related, outside scope our review (although draw some connections throughout). include work [9.Sunstein C.R. Vermeule Conspiracy theories: causes cures.J. Polit. Philos. 2009; 17: 202-227Crossref (126) superstition [10.Lindeman Aarnio K. Superstitious, magical, paranormal beliefs: An integrative model.J. Res. Pers. 2007; 41: 731-744Crossref (118) rumors [11.Berinsky A.J. Rumors health care reform: experiments misinformation.Br. 47: 241-262Crossref (171) bullshit receptivity [12.Pennycook al.On reception detection pseudo-profound bullshit.Judgm. Decis. Mak. 2015; 10: 549-563Google misperceptions [13.Amazeen M.A. al.Correcting consumer misperceptions: effectiveness effects rating contextual correction formats.J. Mass Q. 95: 28-48Google among others. examples organized campaigns (e.g., Russian Research Agency, relating global warming Election). When considering believe, essential two fundamentally ways conceptualize One approach ‘discernment’, extent believed ‘relative’ Discernment, typically calculated minus (akin 'sensitivity' d' signal theory [14.Wickens T. Elementary Signal Detection Theory. Oxford University Press, 2002Google Scholar]) captures ‘overall’ one's gives insight into failures ('falling news'). Another belief, regardless (calculated sum akin calculating 'bias' Critically, alter need ability tell [15.Batailler, al. A identification (in press)Google Scholar]: increasing decreasing headlines equivalent no effect does affect discernment). popular discern rooted motivations. argued consumers (mis)information [16.Kahan D.M. Misconceptions, logic identity-protective cognition.in: SSRN Electron. Cultural Cognition Project Working Paper Series No. 164, Yale Law 605, Economics 575. 2017Crossref 'identity-protective cognition' faced valenced content, leads them overly consistent their identity skeptical inconsistent [17.Kahan Ideology, reasoning, reflection.Judgm. 2013; 8: 407-424Google argues place loyalty identities above fail favor ideologically [18.Van Bavel J.J. Pereira brain: Identity-based model belief.Trends Cogn. 22: 213-224Abstract Full Text PDF (68) accounts contend strong causal motivation dominant factor explaining It belief: People (versus discordant) Scholar, 20.Pennycook 21.Pennycook al.Shifting attention reduce online.Nature. https://doi.org/10.1038/s41586-021-03344-2Crossref 22.Pereira al.Identity concerns drive news.PsyArXiv. September 18, 2018. http://dx.doi.org/10.31234/OSF.IO/7VC5D)Google 23.Vegetti Mancosu sophistication misinformation.Polit. 678-695Crossref (1) 24.Faragó 25.Drummond al.Limited climate change.Environ. 081003Crossref (Figure 1B ). note, concordance smaller Scholar,21.Pennycook Scholar,26.Bago B. fast slow: deliberation reduces (but true) headlines.J. Exp. Gen. 149: 1608-1613Crossref (22) other words, discordant trump truth. necessarily indicate reasoning. Such differences could arise unbiased rational Bayesian) inference built prior factual differ party lines owing environments) [27.Tappin B.M. al.Thinking inferences reasoning: paradigmatic study designs undermine inference.Curr. Opin. 34: 81-87Crossref (7) 28.Tappin al.Rethinking reasoning.J. 29, http://dx.doi.org/10.31234/OSF.IO/YUZFJ)Crossref 29.Tappin al.Bayesian biased? Analytic thinking updating.Cognition. 204: 1-12Crossref 30.Baron Jost J.T. False equivalence: liberals States biased?.Perspect. 14: 292-303Crossref (34) 31.Gerber Green Misperceptions perceptual bias.Annu. 1999; 189-210Crossref (185) 32.Leeper T.J. Slothuus parties, public opinion formation.Polit. 2014; 35: 129-156Crossref 33.Friedman Motivated skepticism inevitable conviction? Dogmatism politics.Crit. 2012; 24: 131-155Crossref (10) 2 details).Box 2Challenges Identifying Politically ReasoningThe observation ideology/partisanship ideology/partisanship) taken [22.Pereira Scholar,123.Ditto P.H. al.At bias bipartisan: meta-analytic comparison conservatives.Perspect. 273-291Crossref (78) Scholar,124.Clark C.J. Winegard Tribalism war peace: nature evolution ideological epistemology significance modern science.Psychol. Inq. 1-22Crossref pattern actually provide clear politicall