research news theory

Routines and Practices: Studying the Making of News

• First Online: 24 September 2023

Cite this chapter

• Julie Firmstone 2  

223 Accesses

This chapter explores established and emerging routines and practices of news production to consider how they shape news content and to ask what these practices mean for who has power over what becomes news. Formative studies and concepts such as gatekeeping and news values are introduced. Several newly evolving routines and practices of digital journalism—including aggregation, curation, and the use of audience analytics and algorithms—are explored to consider the role of digital technology in shaping news production. Findings from a range of studies are synthesised to evaluate the extent to which these practices replace or supplement existing practices, what distinguishes the routines and practices of today from the past, and what implications these distinctions have for news quality now and in the future.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Subscribe and save.

• Get 10 units per month
• Download Article/Chapter or eBook
• 1 Unit = 1 Article or 1 Chapter
• Cancel anytime
• Available as EPUB and PDF
• Read on any device
• Instant download
• Own it forever
• Compact, lightweight edition
• Dispatched in 3 to 5 business days
• Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

Zamith differentiates between “non-purposive forms of audience feedback gathered by sophisticated audience information systems (audience analytics), and distilled into real-time, quantified measures of user-behaviours (audience metrics)” ( 2018 : 1118). Zamith, R. 2018. On Metrics-Driven Homepages. Journalism Studies, 19 , 1116–1137.

Allern, S. 2002. Journalism and commercial news values. Nordicom Review, 23, 137–152.

Google Scholar  

Anderson, C. W. 2010. Journalistic networks and the diffusion of local news: the brief, happy news life of the “Francisville Four”. Political Communication, 27, 289–309.

Article   Google Scholar  

Anderson, C. W. 2011. Between creative and quantified audiences: web metrics and changing patterns of newswork in local US newsrooms. Journalism, 12, 550–566.

Anderson, C. W. 2013a. Rebuilding the News: Metropolitan Journalism in the Digital Age, Philadelphia, Temple University Press.

Anderson, C. W. 2013b. What aggregators do: towards a networked concept of journalistic expertise in the digital age. Journalism, 14, 1008–1023.

Baraniuk, C. 2018. Would You Care If This Feature had Been Written by a Robot? UK: BBC Online. 30/1/2018. https://www.bbc.co.uk/news/business-42858174

Bardoel, J. & Deuze, M. 2001. Network journalism: converging competences of media professionals and professionalism. Australian Journalism Review., 23, 91–103.

Bednarek, M. & Caple, H. 2017. The Discourse of News Values: How News Organisations Create Newsworthiness, Oxford, Oxford University Press.

Book   Google Scholar  

Bell, A. 1991. The Language of the News Media, Oxford, Blackwell.

Bettag, T. 2000. Evolving definitions of news. The Harvard International Journal of Press/Politics, 5, 105–107.

Boczkowski, P. J. 2004. Digitizing the News: Innovation in Online Newspapers, London; Cambridge, Mass, MIT Press.

Boczkowski, P. J. & De Santos, M. 2007. When more media equals less news: patterns of content homogenization in Argentina’s leading print and online newspapers. Political Communication, 24, 167–180.

Breed, W. 1955. Social control in the newsroom: a functional analysis. Social Forces, 32, 326–335.

Brighton, P. & Foy, D. 2007. News Values, London, Sage.

Bro, P. & Wallberg, F. 2015. Gatekeeping in a digital era: principles, practices and technological platforms. Journalism Practice, 9, 92–105.

Bruns, A. 2005. Gatewatching: Collaborative Online News Production, Oxford, Peter Lang.

Caple, H. & Bednarek, M. 2013. Delving Into the Discourse: Approaches to News Values in Journalism Studies and Beyond. Reuters Institute for the Study of Journalism. Oxford, University of Oxford.

Carlson, M. 2017. Journalistic Authority: Legitimating News in the Digital Era, New York, Columbia University Press.

Carlson, M. 2018. Automating judgment? Algorithmic judgment, news knowledge, and journalistic professionalism. New Media & Society, 20, 1755–1772.

Carlson, M. & Lewis, S. C. 2015. Boundaries of Journalism: Professionalism, Practices and Participation, London, Routledge.

Chadha, K. & Wells, R. 2016. Journalistic responses to technological innovation in newsrooms. Digital Journalism, 4, 1020–1035.

Chyi, H. I., Lewis, S. C. & Zheng, N. 2016. Parasite or partner? Coverage of Google News in an era of news aggregation. Journalism & Mass Communication Quarterly, 93, 789–815.

Coddington, M. 2018. Aggregation and Journalism. Oxford Encyclopedia of Journalism Studies. Oxford, Oxford University Press.

Coleman, S., Thumim, N., Birchall, C., Firmstone, J., Moss, G., Parry, K., Stamper, J. & Blumler, J. G. 2016. The Mediated City: The News in a Post-Industrial Context, London, UK, Zed Books Ltd.

Cottle, S. E. 2003. Media Organisation and Production, London, Sage

Davies, N. 2008. Flat Earth News: An Award-Winning Reporter Exposes Falsehood, Distortion and Propaganda in the Global Media, London, Chatto & Windus.

Deuze, M. 2005. What is journalism?: Professional identity and ideology of journalists reconsidered. Journalism, 6, 442–464.

Dickinson, R. 2007. Accomplishing journalism: towards a revived sociology of a media occupation. Cultural Sociology, 1, 189–208.

Duffy, A., Tandoc, E. C. & Ling, R. 2018. Frankenstein journalism. Information, Communication & Society, 21, 1354–1368.

Fanta, A. 2017. Putting Europe’s Robots on the Map: Automated Journalism in News Agencies Oxford: Reuters Institute for the Study of Journalism.

Firmstone, J. 2008. The editorial production process and editorial values as influences on the opinions of the British press towards Europe. Journalism Practice , 2, 221–229.

Firmstone, J. & Coleman, S. 2015. Public engagement in local government: the voice and influence of citizens in online communicative spaces. Information, Communication & Society, 18, 680–695.

Firmstone, J. & Whittington, R. 2021. Local Political Journalism: Systematic Pressures on the Normative Functions of Local News. In: Morrison, J., Birks, J. & Berry, M. (eds.) Routledge Companion to Political Journalism (pp. 84–93). London, Routledge.

Galbraith, J. 2008. A squatter on the fourth estate: Google News. Journal of Library Administration, 46, 191–206.

Galtung, J. & Ruge, M. H. 1965. The structure of foreign news. Journal of International Peace Research, 1, 64–90.

Gandy, O. J. 1982. Beyond Agenda Setting: Information Subsidies and Public Policy, Michigan, Ablex Publishing Company.

Gandy, O. H. J. 2000. Race, ethnicity and the segmentation of media markets. In J. Curran & M. Gurevitch (eds.), Mass media and society , 3rd Ed (pp. 44–69). London, Arnold.

Gans, H. J. 1979. Deciding What’s News, New York, Pantheon.

Gillespie, T. 2014. The Relevance of Algorithms. In: Gillespie, T., Boczkowski, P. J. & Foot, K. (eds.) Media Technologies. Cambridge, MA, MIT Press.

Golan, G. 2006. Inter-media agenda setting and global news coverage. Journalism Studies, 7, 323–333.

Golding, P. & Elliot, P. 1979. Making the News, London, Longman.

Goode, L. 2009. Social news, citizen journalism and democracy. New Media & Society, 11, 1287–1305.

Hall, S. 1973. The Determinations of News Photographs. In: Cohen, S. & Young, J. (eds.) The Manufacture of News: Deviance, Social Problems and the Mass Media. London, Constable.

Hanusch, F. 2017. Web analytics and the functional differentiation of journalism cultures: individual, organizational and platform-specific influences on newswork. Information, Communication & Society, 20, 1571–1586.

Harcup, T. & O’Neill, D. 2001. What is news? Galtung and Ruge revisited. Journalism Studies, 2, 261–280.

Harcup, T. & O’Neill, D. 2017. What is news?: News values revisited (again). Journalism Studies, 18, 1470–1488.

Hartley, J. 1988. Understanding News, London, Routledge.

Hurley, R. J. & Tewksbury, D. 2012. News aggregation and content differences in online cancer news. Journal of Broadcasting & Electronic Media, 56, 132–149.

Jackson, D. & Moloney, K. 2016. Inside churnalism. Journalism Studies, 17, 763–780.

Lewis, J., Williams, A. & Franklin, B. 2008. A compromised fourth estate? Journalism Studies, 9, 1-20.

Lewis, S. C. 2012. The tension between professional control and open participation. Information, Communication & Society, 15, 836–866.

Linden, C.-G. 2017. Algorithms for journalism: the future of news work. The Journal of Media Innovations 4, 60–76.

Lowry, W. 2014. News Routines. In: Donsbach, W. (ed.) The International Encyclopedia of Communication. Wiley Online.

McChesney, R. W. 2003. The problem of journalism: a political economic contribution to an explanation of the crisis in contemporary US journalism. Journalism Studies, 4, 299–329.

McCombs, M. 2004. Setting the Agenda: The Mass Media and Public Opinion, Oxford, Polity.

McCombs, M. 2005. A look at agenda-setting: past, present and future. Journalism Studies, 6, 543–557.

McManus, J. H. 1992. What kind of commodity is news. Communication Research, 19, 787–805.

Nechushtai, E. & Lewis, S. C. 2018. What kind of news gatekeepers do we want machines to be? Filter bubbles, fragmentation, and the normative dimensions of algorithmic recommendations. Computers in Human Behavior , 90, 298–307.

Newman, N., Fletcher, R., Schulz, A., Simge, A., Robertson, C. T. & Nielsen, R. K. 2021. Digital News Report 2021. 10th ed. Oxford, Reuters Institute for the Study of Journalism, Oxford University.

Niblock, S. & Machin, D. 2007. News values for consumer groups: the case of Independent Radio News, London, UK. Journalism, 8, 184–204.

O’Neill, D. & Harcup, T. 2009. News Values and Selectivity. In: Wahl-Jorgensen, K. & Hanitzsch, T. (eds.) The Handbook of Journalism Studies (pp. 161–174). New York, Routledge.

Palmer, J. 2000. Spinning Into Control: News Values and Source Strategies, London, Leicester University Press.

Petre, C. 2015. The Traffic Factories: Metrics at Chartbeat, Gawker Media, and The New York Times. New York, Tow Centre for Digital Journalism.

Petre, C. 2021. All the News That’s Fit to Click: How Metrics are Transforming the Work of Journalists, Princeton, Princeton University Press.

Phillips, A. 2010. Faster and Shallower: Homogenisation, Cannibalisation and the Death of Reporting. In: Lee-Wright, P., Phillips, A. & Witschge, T. (eds.) Changing Journalism (pp. 81–98). Abingdon, Routledge.

Phillips, A. 2015. Futures of Journalists: Low Paid Piece Workers or Global Brands?. In: Vos, T. P. & Heinderyckx, F. O. (eds.) Gatekeeping in Transition (pp. 61–82). New York, Routledge.

Rogstad, I. 2016. Is Twitter just rehashing? Intermedia agenda setting between Twitter and mainstream media. Journal of Information Technology & Politics, 13, 142–158.

Schlesinger, P. 1978. Putting Reality Together, London, Constable.

Schudson, M. 1995. The Power of News, London; Cambridge, Mass;, Harvard University Press.

Schudson, M. 1999. What Public Journalism Knows About Journalism But Doesn’t Know About “Public”. In: Glasser, T., L (ed.) The Idea of Public Journalism. New York, Guildford.

Schudson, M. 2000. The Sociology of News Production Revisited (Again). In: Curran, J. & Gurevitch, M. (eds.) Mass Media and Society. 3rd ed. London, Arnold.

Schultz, I. 2007. The journalistic gut feeling: Journalistic doxa, news habitus and orthodox news values. Journalism Practice, 1, 190–207.

Shoemaker, P. J. & Reese, S. D. 2014. Mediating the Message in the 21st Century: A Media Sociology Perspective, New York, Routledge/Taylor & Francis Group.

Shoemaker, P. J., Vos, T. P. & Reese, S. D. 2009. Journalists as Gatekeepers. In: Wahl-Jorgensen, K. & Hanitzsch, T. (eds.) The Handbook of Journalism Studies (pp. 73–87). New York, Routledge.

Sjøvaag, H., Stavelin, E., Karlsson, M. & Kammer, A. 2019. The hyperlinked Scandinavian news ecology: the unequal terms forged by the structural properties of digitalisation. Digital Journalism , 7(4), 507–531.

Staab, J. F. 1990. The role of news factors in news selection: a theoretical reconsideration. European Journal of Communication, 5, 423–443.

Strömbäck, J., Karlsson, M. & Hopmann, D. N. 2012. Determinants of news content: comparing journalists’ perceptions of the normative and actual impact of different event properties when deciding what’s news. Journalism Studies, 13, 718.

Tandoc, E. C. 2014. Journalism is twerking? How web analytics is changing the process of gatekeeping. New Media & Society, 16, 559–575.

Tuchman, G. 2002. The Production of News. In: Jensen, K. B. (ed.) A Handbook of Media and Communication Research (pp. 78–90). London, Routledge.

Usher, N. 2016. The Constancy of Immediacy: From Printing Press to Digital Age. In: Alexander, J. C., Breese, E. B. & Luengo, M. A. (eds.) The Crisis of Journalism Reconsidered: Democratic Culture, Professional Codes, Digital Future (pp. 170–189). New York, Cambridge University Press.

Van Dalen, A. 2012. The algorithms behind the headlines: how machine-written news redefines the core skills of human journalists. Journalism Practice, 6, 648–658.

Van Hout, T. & Van Leuven, S. 2016. Investigating ‘Churnalism’ in Real-Time News. In: Franklin, B. & Eldridge, S. (eds.) The Routledge Companion to Digital Journalism Studies . London, Routledge.

Vliegenthart, R. & Walgrave, S. 2008. The contingency of intermedia agenda setting: a longitudinal study in Belgium. Journalism & Mass Communication Quarterly, 85, 860–877.

Vos, T. P. 2015. Revisiting Gatekeeping Theory During a Time of Transition. In: Vos, T. P. & Heinderyckx, F. O. (eds.) Gatekeeping in Transition (pp. 3–24). New York, Routledge.

Chapter   Google Scholar  

Vos, T. P. & Heinderyckx, F. O. 2015. Gatekeeping in Transition, New York, Routledge.

Vu, H. T. 2014. The online audience as gatekeeper: the influence of reader metrics on news editorial selection. Journalism, 15, 1094–1110.

Wanta, W. 2009. Gatekeeping. In: Donsbach, W. (ed.) The International Encyclopedia of Communication. London, Blackwell.

Welbers, K., Van Atteveldt, W., Kleinnijenhuis, J., Ruigrok, N. & Schaper, J. 2016. News selection criteria in the digital age: professional norms versus online audience metrics. Journalism, 17, 1037–1053.

White, D. M. 1950. The “Gate Keeper”: A Case Study in the Selection of News. Journalism Quarterly , 27(4), 383–390.

Zamith, R. 2018. On metrics-driven homepages. Journalism Studies, 19, 1116–1137.

Download references

Author information

Authors and affiliations.

University of Leeds, Leeds, UK

Julie Firmstone

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Julie Firmstone .

Rights and permissions

Reprints and permissions

Copyright information

© 2024 The Author(s), under exclusive license to Springer Nature Switzerland AG

About this chapter

Firmstone, J. (2024). Routines and Practices: Studying the Making of News. In: The Shaping of News. Palgrave Macmillan, Cham. https://doi.org/10.1007/978-3-031-21963-4_4

Download citation

DOI : https://doi.org/10.1007/978-3-031-21963-4_4

Published : 24 September 2023

Publisher Name : Palgrave Macmillan, Cham

Print ISBN : 978-3-031-21965-8

Online ISBN : 978-3-031-21963-4

eBook Packages : Literature, Cultural and Media Studies Literature, Cultural and Media Studies (R0)

Share this chapter

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Publish with us

Policies and ethics

• Find a journal
• Track your research

• Tools and Resources
• Customer Services
• Communication and Culture
• Communication and Social Change
• Communication and Technology
• Communication Theory
• Critical/Cultural Studies
• Gender (Gay, Lesbian, Bisexual and Transgender Studies)
• Health and Risk Communication
• Intergroup Communication
• International/Global Communication
• Interpersonal Communication
• Journalism Studies
• Language and Social Interaction
• Mass Communication
• Media and Communication Policy
• Organizational Communication
• Political Communication
• Rhetorical Theory
• Share Facebook LinkedIn Twitter

Article contents

Digital journalism and epistemologies of news production.

• Rodrigo Zamith Rodrigo Zamith Journalism Department, University of Massachusetts Amherst
• , and  Oscar Westlund Oscar Westlund Department of Journalism and Media Studies, Oslo Metropolitan University
• https://doi.org/10.1093/acrefore/9780190228613.013.84
• Published online: 18 July 2022

News is the result of news production, a set of epistemic processes for developing knowledge about current events or issues that draw upon a range of newsgathering techniques and formatting choices with the objective of yielding a publishable and distributable product designed to inform others. That process, however, has changed considerably over time and in parallel to broader economic, political, professional, social, and technological changes. For example, during the past two decades alone, there has been greater audience fragmentation and an emphasis on audience measurement, new forms of strategic exploitation of information channels and digital surveillance of journalists, greater aggregation of news and more avenues for professional convergence, a media environment awash in user-generated content and challenges to traditional outlets’ epistemic authority, and more opportunities for interactivity and miniaturized mobilities. In concert, these and other forces have transformed news production processes that have become increasingly digital—from who the actors are to the actants that are available to them, the activities they may engage in, and the audiences they can interact with.

Such impacts have required scholars to revisit different theories that help explain how news is produced and with what consequences. Whereas the field of journalism studies draws on a rich history of multidisciplinary theorizing, epistemologies of journalism have received increased attention in recent years. There is a close link between news production and epistemology because the production of news inherently involves developing news information into one form of knowledge. As such, an epistemological lens allows scholars to examine the production, articulation, justification, and use of knowledge within the social context of digital journalism. An analytic matrix of 10 dimensions—the epistemologies of journalism matrix—helps scholars examine different forms of journalism through an epistemological lens. The matrix focuses on identifying the key (a) social actors, (b) technological actants, and (c) audiences within a space of journalism; examining their articulation or justification of (d) knowledge claims and their distinct (e) practices, norms, routines, and roles; differentiating between the (f) forms of knowledge they typically convey; and evaluating the similarities and dissimilarities in their typical (g) narrative structure, (h) temporality, (i) authorial stance, and (j) status of text.

By applying that matrix to four emerging forms of journalism (participatory journalism, live blogging, data journalism, and automated journalism), it can be seen that digital journalism and news production are becoming even more heterogeneous in terms of their implicated entities, cultures and methods, and positionality in relation to matters of knowledge and authority. First, contemporary news production is deeply influenced by myriad technological actants, which are reshaping how knowledge about current events is being created, evaluated, and disseminated. Second, professional journalists are losing epistemic authority over the news as key activities are delegated to algorithms created by non-journalists and to citizens who have become more present in news production. Third, the outputs of news production are becoming more diverse both in form and in content, further challenging long-standing norms about what is and is not “journalism.” In short, history has shown that news production will continue to evolve, and an epistemological lens affords scholars a useful and adaptable approach for understanding the implications of those changes to the production of knowledge about news.

• digital journalism
• news production
• epistemology
• data journalism
• participatory journalism
• live blogging
• automated journalism
• social media

Introduction

News , or “public knowledge claiming to report on current events in the world” ( Westlund & Ekström, 2018 , p. 3), is more pervasive in citizens’ lives today than ever before. It may be accessed around the clock and in a multitude of ways, including through typical reading, watching, viewing, and listening activities as well as newer “snacking” and monitorial activities such as scrolling through headlines while waiting in the elevator ( Costera Meijer & Groot Kormelink, 2015 ). Those activities may be performed actively via acts such as searching or passively via exposure coordinated by algorithmic recommendation systems. News itself may be accessed through a wider range of media and digital platforms and from a larger multitude of sources. These include legacy news media and digital news start-ups working in and for local, regional, national, or international contexts ( Ali et al., 2019 ; Chua & Duffy, 2019 ; Heft & Dogruel, 2019 ), as well as citizen journalists ( Kim & Lowrey, 2015 ) and alternative news media ( Figenschou & Ihlebæk, 2019 ; Holt et al., 2019 ).

News is neither a “given” nor a necessarily stable object, however. It is the result of news production , defined here as the epistemic processes for developing knowledge about current events or issues that draw upon a range of newsgathering techniques and formatting choices with the objective of yielding a publishable and distributable product designed to inform others. This definition highlights the intrinsic link between news and epistemology, as news can be distilled into different forms of knowledge about the world ( Ekström & Westlund, 2019a ; Ekström et al., 2021 ; Nielsen, 2017 ; Zelizer, 1993 ). It also underscores that news is necessarily shaped by activities such as sourcing and filtering information ( Domingo et al., 2008 ), which may be produced by human actors or technological actants ( Lewis & Westlund, 2015a ) and further formatted with particular platforms ( Hågvar, 2019 ; Westlund, 2013 ) and audiences in mind ( Weischenberg & Matuschek, 2008 ). Finally, it recognizes the close link between news production and distribution―which, indeed, may sometimes occur simultaneously as in the case of broadcasting, live tweeting, and producing newsletters―while acknowledging that the latter is most often examined as a separate, subsequent step (for a detailed examination of news distribution, see Braun, 2019 ; see also Hermida, 2020 ; Wallace, 2018 ).

Adopting an epistemological lens allows scholars to recognize that news is often contested and that much of the contestation occurs implicitly―and sometimes explicitly―along epistemic lines, as with critiques about the veracity of a given news account and allegations of bias ( Carlson, 2017 ; Compton & Benedetti, 2010 ). This lens also allows scholars to be mindful of the fact that news varies in substance and form between genres, across platforms, and depending on epistemic processes and formats for publishing ( Ekström & Westlund, 2019b ). In short, news is the result of a dynamic and heterogeneous process.

This article aims to capture that dynamism in order to illustrate the evolution of digital news production, particularly since the turn of the 21st century and mostly in relation to Western journalistic practices. The article therefore does not review the emergence and growth of some important research into news production from the mid-20th century, including the influential work produced by the likes of Herbert Gans, Gaye Tuchman, and Pamela Shoemaker and Stephen Reese. Such work is aptly reviewed by Hanusch and Maares (2021) , who describe it as part of a wave of scholarship that illustrate the importance of news routines; the role of intra-, inter-, and extra-organizational relationships; and strategic rituals in shaping news production processes and, consequently, news products (see also Westlund & Ekström, 2019 ). However, for expediency, this article instead focuses on epistemologies of digital news production, recognizing that present ideas about journalistic knowledge production are shaped by past work.

The article begins with a synthesis of the significant economic, political, professional, social, and technological developments that have played a structuring role in the developments of news production in the field, such as the proliferation of mobile devices and organized disinformation campaigns. Then, it describes some of the key theories that have been used to study news production, centering on an epistemological lens that emphasizes its rhetorical, practical, and evaluative elements. Next, the article systematically examines four emerging forms of journalistic news production―which is characterized by “ambitions toward the publishing of truthful accounts of current events in the world” ( Westlund & Ekström, 2018 , p. 3)―through an epistemological lens. That examination focuses on participatory journalism, live blogging, data journalism, and automated journalism because they are not only marked by some novelty and represent rapidly evolving forms of journalism but also associated with a significant amount of recent scholarship that merits synthesis. The article concludes with a discussion in which it is argued that scholars can only go so far in understanding news and journalism by focusing on who does journalism or what the news materials produced are, and that the examination of epistemic practices proves a worthwhile addition to that endeavor.

Key Shifts in the 21st Century

The news production process has changed considerably over time and in parallel to broader economic, political, professional, social, and technological changes ( Barnhurst, 2011 ; Braun, 2015 ; Bruns, 2008 ; Fenton, 2011 ; Hanusch & Maares, 2021 ; Napoli, 2011 ; Westlund & Quinn, 2018 ; Zelizer, 2019 ). Indeed, as scholars have observed, entities typically regarded as being outside the space of journalism can play a major role at particular points of its development. For example, the U.S. Postal Service played a crucial role in creating the distribution infrastructure for newspapers during the early U.S. republic and, in turn, not only helped shape U.S. news processes but also created a sense of national identity and belonging among the citizens of the emerging nation ( John, 1995 ). Although chronicling all the changes that have impacted news production is not possible within a single article, 20 particularly consequential shifts since the turn of the century are highlighted here to illustrate how news production has been transformed alongside changing forces. These forces are grouped for illustrative purposes, recognizing that some transcend simple categories—that is, they may be simultaneously economic and technological, and so on.

Economically, today’s news environment is characterized by greater audience fragmentation , which refers to the process through which (or the phenomenon in which) mass audiences are split into more diffuse and specialized groups in their media consumption ( Neuman, 1991 ). That has promoted specialization in the production of news and the creation of niche outlets to satisfy new and narrower segments ( Napoli, 2011 ). Similarly, there is now greater emphasis on audience measurement , or the process of quantifying, analyzing, and synthesizing information about individuals’ content preferences and how they interact with that content ( Napoli, 2011 ; Tandoc, 2019 ). The current emphasis on measurement is enabled largely by audience analytics, which provides more (and more detailed) data about audience behaviors and facilitates keying journalistic products to audience demands ( Zamith, 2018 ). The nature of commodification , or the transformation of a good or service into a product that can be sold for profit within a market ( Hamilton, 2004 ), has also changed since the turn of the 21st century due to the unbundling of news products, rise of non-journalistic platforms, and increase in competition from platform companies ( Steensen & Westlund, 2021 ) as well as alternative news media ( Holt et al., 2019 ). This has resulted in pressures for journalists to do more with less and a renewed emphasis on subscription-based and nonprofit economic models ( Pickard, 2020 ). Economic conditions have also resulted in greater occupational precarity , or deteriorating professional conditions that lead to insecure labor conditions ( Örnebring, 2018 ). This situation is characterized by a growing dependence on unpaid labor and outsourced workers, less full-time work, and a more general fear of indiscriminate layoffs, constraining journalists’ ability to adhere to journalistic ideals and remain autonomous ( Örnebring, 2018 ).

Politically, journalists must contend with greater amounts of disinformation , or information that is deliberately false or misleading ( Jack, 2017 ). A range of actors―including state-sponsored groups―have sought to sow disinformation by strategically exploiting trustworthy information channels and outright impersonating trusted news brands, forcing journalists to rethink how they verify information within a speed-oriented craft while also further complicating eroding trust in journalism ( Marwick & Lewis, 2017 ). News media have been subjected to changes in regulations , or rules, laws, and codes prescribed by some authority, typically a government ( Flew & Swift, 2013 ). Scholars have observed that Western countries have generally moved toward overall deregulation, resulting in greater corporate ownership and emphasis on consumer-oriented journalism ( Fenton, 2011 ). However, they have also observed substantially different approaches taken within European countries, where journalists and news organizations sometimes have access to direct, government-sponsored grants and indirect subsidies, and where news audiences often have access to robust public service broadcasters ( Murschetz, 2020 ). Such approaches are markedly different from those in the United States, where the primary government support mechanism is frequently just a general tax break for nonprofit entities and donors ( Pickard, 2020 ). In addition, many countries still operate under strict information control regimes that limit what journalists can publish ( Xu, 2015 ). There is also now greater digital surveillance of journalists, which enables an actor, such as a government, to use digital tools to continuously monitor the activities of another actor ( Ataman & Çoban, 2018 ). Indeed, journalists―and investigative journalists in particular―increasingly report serious concerns about being tracked, leading to some self-censorship and increased difficulty getting confidential sources to share information ( Lashmar, 2017 ). There have been changes in the amounts and types of state subsidies for news media, or the direct aid provided by governments to support the activities of independent organizations ( Kreiss & Ananny, 2013 ). Although public support for such media remains high, their subsidies have been repeatedly reduced or threatened in recent years ( Fenton, 2011 ), and the lack of subsidies in some countries has resulted in the development of “news deserts” as commercial models have faltered ( Pickard, 2020 ).

Professionally, there is now greater aggregation of news, or the practice of manually or algorithmically bringing together information from different products into a single one, typically based on some curation criteria ( Bakker, 2012 ). This has resulted in the proliferation of news aggregator sites such as Google News and apps such as Apple News that do not originate news but serve as competitors and key audience brokers by virtue of their strategic position within the contemporary news landscape ( Coddington, 2019 ). Such aggregators tend to promote freely accessible content, making shifts away from ad-supported news more challenging. Similarly, there are new forms of convergence , or the integration of previously distinct media components and technologies to create new organizational forms and processes ( Pavlik, 2004 ). This has promoted internal collaboration across a news organization’s departments ( Nielsen, 2012 ) as well as external collaboration with non-media partners (e.g., Hacks/Hackers; see Usher, 2014 ). It has also promoted a digital-first ethos, where newsworkers are expected to quickly produce content for online environments and engage through social media platforms in ways that challenge traditional journalistic ethics ( Singer, 2012 ). This shift exacerbated the continuous deadline pressures introduced by live broadcasts, accelerated by 24-hour news, and taken to a new level with the “death of the deadline” in online news, resulting in time-obsessed and stepped-up news cycles that emphasize temporal competition and improvisatory practice ( Barnhurst, 2011 ). Recent research on breaking news has also shown how journalists take timing into consideration, and are mindful of when to release their stream of online news ( Ekström et al., 2021 ).

Socially, journalists now operate in a media environment filled with user-generated content , or non-journalistic content created by active audience members that is typically published online and accessible at negligible cost ( Jönsson & Örnebring, 2010 ). This has enabled outsiders to enter journalism, provided new content subsidies for news organizations, created new competitors within a competitive attention economy, and challenged news professionals’ gatekeeping powers ( Bruns, 2008 ). Similarly, there are now more opportunities for dark participation , or antisocial forms of online participation that include harassment, trolling, and “doxxing,” which refers to the practice of publicly revealing private, and often sensitive, information about an individual or organization ( Quandt, 2018 ). Such participation induces some journalists to self-censor, withdraw from public spaces, or quit the profession altogether—and disproportionately affects journalists from historically marginalized communities ( Lewis et al., 2020 ; Stahel & Schoen, 2020 ). These developments have paralleled (and driven) challenges to traditional epistemic authority , or an entity’s socially accepted “power to define, describe, and explain bounded domains of reality” ( Gieryn, 1999 , p. 1). Journalists in many areas of the world must now cope with low and/or declining levels of trust in media, as well as eroding control over information ( Fletcher & Park, 2017 ). News production is also conditioned by placeification , or the shaping of an artifact by the places in which it is produced, practiced, and consumed ( Gutsche & Hess, 2020 ). In many countries, news production now occurs primarily in large, urban centers as a result of broader societal place-based realignments, with consequences for trust in non-urban centers and for journalists to witness certain events firsthand ( Radcliffe & Ali, 2017 ). Indeed, as Schmitz Weiss (2015 , p. 127) contends, “location plays a significant role in how communities function and how they see themselves,” and scholars have argued that structural inequalities and political polarization in places such as the United States have taken on a place-based dimension as a result of broader social, economic, technological, and professional shifts (e.g., Usher, 2019 ).

Technologically, journalists work within an environment characterized by greater interactivity , which refers to the technological attributes of mediated environments that allow users to connect with and through technology ( Bucy, 2004 ). News consumers now expect to be able to interact with news content, whether through responsive websites or dynamic products such as interactive data visualizations ( Zamith, 2019b ). In addition, news organizations now routinely use external hyperlinks as reference tools, which in turn can promote transparency ( Sjøvaag et al., 2019 ) and contribute to heterogeneous news flows and inter-media connectivity ( Steensen & Eide, 2019 ). The current environment is also marked by miniaturized mobilities , or information and communication technologies designed to fit a mobile lifestyle, such as smartphones and smartwatches ( Elliott & Urry, 2010 ). These mobilities have enabled journalists to work outside the newsroom in more diverse and effective ways (see also Duffy et al., 2020 ; Westlund & Quinn, 2018 ). Social media , or platforms that allow users to traverse a network of contacts via contributions such as posts and tweets ( boyd & Ellison, 2007 ), have enabled journalists to adopt new practices such as ambient journalism to find novel stories and potentially draw upon a larger range of sources ( Hermida et al., 2014 ). They have also substantially altered how information spreads ( Swart et al., 2019 ). More broadly, however, the space of journalism is now characterized by an immense number of transparent intermediaries , or actors and actants that exert a structuring role in media production and distribution yet are unseen by most media consumers ( Braun, 2015 ). These include algorithmic recommendation tools that shape individuals’ exposure to content―both in terms of what journalists see and which of their work gets seen by news consumers―and promotes practices such as search engine optimization of headlines ( Gillespie, 2014 ). Notably, throughout the 2010s, many publishers aimed to build a presence on social media platforms ( Steensen & Westlund, 2021 ). However, amid growing concerns about their loss of power and revenue in the long term ( Nielsen & Ganter, 2018 ), some publishers have shifted toward platform counterbalancing ( Chua & Westlund, 2019 ).

In concert, these economic, political, professional, social, and technological forces have transformed multiple aspects of journalism and in particular have had material impact on news production—from who the actors are to the actants that are available to them, the activities they may engage in, and the audiences they can interact with (see Lewis & Westlund, 2015a ). Such impacts have required scholars to revisit different theories that help explain how news is produced and with what consequences.

Theorizing News Production

There is a long tradition of theorizing news production, much of which draws heavily on psychology, sociology, political economy, and cultural studies (see Ahva & Steensen, 2019 ; Hanusch & Maares, 2021 ; Steensen & Westlund, 2021 ). An early and enduring example is gatekeeping theory , which refers to the process through which actors or actants (gatekeepers) can include or exclude information before it reaches an audience—as with a newspaper editor who chooses which newswire stories to include and exclude ( White, 1950 ). Theorizing from this stream has since argued that such decisions are the product of professional socialization and structural constraints, including the inculcation of news values, practices, and norms ( Vos & Heinderyckx, 2015 ).

Similarly, scholars of journalism have drawn on institutional theory throughout the years to contend that institutions—typically defined as meso-level variables such as beliefs, norms, and formal rules—mediate the relationship between macrostructures such as journalism and the micro-actions of individuals or organizations ( Cook, 1998 ). This line of thinking has proved fruitful in explaining the uniformity in certain aspects of news production and the often cautious responses to disruption and uncertainty ( Lowrey, 2011 ). This theoretical perspective broadly shares key tenets with field theory , as proposed by Bourdieu (1993) , which has proven particularly influential in recent scholarship (e.g., Wu et al., 2019 ). That perspective imagines society as being composed of multiple “fields” (with journalism being one of them) that have field-specific norms, traditions, and practices that shape behavior but are themselves shaped through their intersection with other fields as well as broader cultural, economic, and political forces ( Bourdieu, 1993 ). Such theorizing has opened avenues for examining cultural resources that, for example, lend greater social legitimacy to certain news production actors and activities over others ( Benson, 2006 ).

These examples illustrate but one, primarily sociological, stream of theories that have been applied to the study of news production (for a broader collection, see Ahva & Steensen, 2019 ; Hanusch & Maares, 2021 ; Steensen & Westlund, 2021 ). However, they are also illustrative in that they have all been developed and occasionally recast in some manner in response to the aforementioned economic, political, professional, social, and technological developments. Indeed, as Wallace (2018) wrote while aiming to remodel gatekeeping theory, sociotechnical developments have “changed gatekeeping selection processes and news flow patterns. Accordingly, gatekeeping theory must also change” (p. 275).

This article is centered on a lens that has garnered increased attention in recent years: epistemologies of journalism ( Ekström & Westlund, 2019a ). There is a close link between news production and epistemology because the production of news inherently involves developing news information into one form of knowledge. Indeed, the very existence of journalistic authority is largely dependent on a public’s perception that journalism―or some entities within it―offers valuable and unique public knowledge ( Carlson, 2017 ). Moreover, scholars have long contended that journalists are members of interpretive communities that are united by shared meanings about news production and the practice of collectively interpreting key events ( Zelizer, 1993 ).

An epistemological lens focuses on understanding the production, articulation, justification, and use of knowledge within the social context of journalism ( Ekström, 2002 ). In other words, it helps scholars examine what newsworkers know, how they know it, and how they justify their accounts―“the news”―as a form of knowledge ( Ekström & Westlund, 2019a ). This has required scholars to revisit who produces journalism, what epistemic values and activities are accepted as being journalistic, and how those constellations produce distinct forms of journalism―each with sufficiently different epistemological processes and claims.

As Lewis and Westlund (2015a ) argue, digital journalism involves a larger and more heterogeneous set of social actors, technological actants, and audiences than ever before. The boundaries that help establish who is a journalist have blurred considerably, with individuals previously at journalism’s periphery now considered central to its enactment ( Belair-Gagnon & Holton, 2018 ). Some news production is already automated, raising questions about the nonhuman and nonjournalistic epistemic logics and processes imbued in the associated algorithms ( Diakopoulos, 2019 ). Audiences have also changed in terms of how they are imagined, constituted, distributed, and measured, complicating how journalists come to understand (and aim to service) what they perceive to be needs of diverse audiences ( Napoli, 2011 ; Tandoc, 2019 ; Zamith, 2018 ).

Ekström and Westlund (2019a ) observe that research on epistemic values and activities within journalism have centered on three interrelated aspects. The first focuses on how journalistic knowledge claims and epistemic authority are articulated in discourse and through texts. The second draws on journalists’ narrated reflections of their practices, norms, and routines to examine how they think about and enact different epistemic notions. The third evaluates how journalistic knowledge claims are justified in news products and the extent to which they are accepted, rejected, or remixed by those who consume them.

Although news is sometimes treated as homogeneous―especially in statistical modeling that reduces it to a single endogenous or exogenous variable―the scholarship clearly observes that it is instead quite heterogeneous as a result of distinct news production practices, objectives, and constellations. Nielsen (2017) helps illustrate one degree of epistemological divergence in outlining three different forms of knowledge that can be conveyed through digital journalism: news-as-impression, or decontextualized snippets of information as with brief news alerts; news-as-item, or typical-length news articles and video news reports about news episodes; and news-about-relations, or in-depth, explanatory, and durable news products that aim to show the bigger picture. Matheson and Wahl-Jorgensen (2020) also point to five key aspects for distinguishing between types of journalism: narrative structure, or the way in which information is organized; temporality, or how time is accounted for; journalistic role, or the responsibilities, values, and objectives of the news product; authorial stance, or the journalist’s perspective on conventions such as objectivity and balance; and status of text, or whether the product is treated as a finished or evolving product.

Drawing on this literature, this article attempts to explicate the epistemologies of news production through a matrix of 10 dimensions referred to as the epistemologies of journalism matrix . This matrix focuses on identifying the key (a) social actors , (b) technological actants , and (c) audiences within a space of journalism; examining their articulation or justification of (d) knowledge claims and their distinct (e) practices , norms , routines , and roles ; differentiating between the (f) forms of knowledge they typically convey; and evaluating the similarities and dissimilarities in their typical (g) narrative structure , (h) temporality , (i) authorial stance , and (j) status of text .

Epistemologies of News Production

To illustrate the heterogeneity of news production and the value of evaluating its epistemologies through the 10 aforementioned dimensions, the dimensions are applied to four emerging forms of journalism: participatory journalism, live blogging, data journalism, and automated journalism (Table 1 ; see also Ekström & Westlund, 2019a ). Scholars are encouraged to build upon the epistemologies of journalism matrix by incorporating additional forms of journalism.

Table 1. Epistemologies and Different Forms of Journalism

Note : The epistemologies of journalism matrix outlines the most dominant news production patterns for each of 10 dimensions. In this table, it is applied to distinct forms of predominantly digital journalism, as per the authors’ knowledge of the sectors and existing scholarly work. Exceptions to the dominant patterns can exist in different geographical contexts and among different sorts of news publishers.

Participatory Journalism

Participatory journalism is known as a form of digital journalism that promotes active and intentional engagement between newsworkers and individuals previously thought of as mostly passive audiences ( Singer et al., 2011 ). Although journalism has long offered audiences an opportunity to have a voice, whether through purposive sourcing or dedicated sections for letters to the editor, this more recent form aims to center citizens’ contributions in multiple stages of the news production process ( Lawrence et al., 2018 ). It may manifest itself both in perception (beliefs about the role of audiences) and in practice (affordances and efforts to involve audiences) and entail direct, indirect, and sustained exchanges designed to empower audiences ( Coddington et al., 2018 ). As Westlund and Ekström (2018) argue, scholarship on participatory journalism must now consider both proprietary and nonproprietary platforms. Importantly, proprietary platforms are those that belong to and are controlled by a specific company (with the inner workings often black-boxed) and which may be used by others through the purchase of a license or their participation in a monetization scheme. Some news organizations are proprietors of platforms and algorithms of their own. However, news companies also rely on platforms (e.g., Facebook and Chartbeat) that are not proprietary to them. Such third-party platforms, which include the likes of Twitter ( Hermida et al., 2014 ) and WhatsApp ( Kligler-Vilenchik & Tenenboim, 2020 ), are now deeply embedded in journalistic practice. That, in turn, has structured the affordances, possibilities, and expectations for acts of participatory journalism. Publishers are increasingly focusing on reducing their dependency on third parties and developing their own proprietary solutions, both for economic purposes and to introduce new affordances for participation. Moreover, as scholars have observed, not all participation is prosocial; a considerable amount involves harassment, bullying, and hate speech ( Lewis et al., 2020 ; Quandt, 2018 ).

News can be produced via participatory journalism by an extensive range of individual social actors that is typically led by journalists, social media editors, and audience engagement editors but may involve a range of previously passive actors such as citizen journalists ( Wall, 2017 ). Its production processes are still human-centric, although they draw upon proprietary and third-party technological actants such as social media platforms to facilitate participation at different stages of news production ( Westlund & Ekström, 2018 ). The audiences are not only diverse but also active, as nearly any member is theoretically able to engage in participatory journalism due to the low barrier to entry ( Coddington et al., 2018 ).

Participatory journalism involves practices, norms, routines, and roles oriented toward curation and requiring an openness to collaboration that has historically been a source of professional tension ( Lewis, 2012 ). It is driven by a logic that may be normatively characterized as democratically oriented and critically characterized as communicative capitalism ( Vujnovic et al., 2010 ; see also Zamith, 2018 ). The knowledge claims made within participatory journalism differ from traditional claims in that they assert themselves to be enhanced by public engagement―they are presumed to be actively vetted and informed by others’ observed and lived experiences―and thus purport to represent a collaborative form of knowledge production ( Anderson & Revers, 2018 ). As a form of knowledge, participatory journalism may take different shapes but is most commonly seen as typical news-as-items, wherein participants inform but do not revolutionize traditional journalistic products ( Borger et al., 2019 ).

The narrative structure of the products of participatory journalism are typically coherent and adhere to traditional structures, like the inverted pyramid for texts ( Engelke, 2020 ). Regarding temporality, products tend to adhere to an interpretive framework that draws upon more diverse sets of sources in an ordered manner ( Borger et al., 2019 ). The authorial stance differs in that it is more subjective and involves weaker professional control resulting from efforts to promote networked balance and co-presence ( Lawrence et al., 2018 ). The status of the text is typically implicitly conveyed as static and presumed to be finished, lest it involve a live or rapidly evolving news event ( Ekström & Westlund, 2019b ).

More generally, the tension between professional control and open participation ( Lewis, 2012 ) is associated with the epistemological authority of journalists in producing and defining news. Studies find that journalists remain in control of those processes or cede only a portion of their control (e.g., Engelke, 2019 ). Although scholars continue to see potential for greater participation in the news, the degree of “dark participation” has proven to be a significant barrier ( Quandt, 2018 )―evidenced, for instance, by the removal of user-commenting affordances on many leading news websites. Nevertheless, participatory journalism has in some cases substantively reshaped sourcing practices, yielding less elite and more diverse source networks ( Hermida et al., 2014 ) and ultimately producing more cautious knowledge claims. Moreover, whereas citizens’ direct participation in news production may be more limited than some scholars envisioned at the turn of the century, their indirect participation―by privately sharing news materials on tracked social media platforms, posting firsthand videos through semi-public accounts, and publicly discussing news and news coverage―has further reshaped journalism beyond this specific form ( Engelke, 2019 ).

Live Blogging

Live blogging is a form of digital journalism that focuses on ongoing, near real-time reporting of both planned and unexpected news events through brief and sequential posts on digital websites and platforms ( Matheson & Wahl-Jorgensen, 2020 ). This approach to journalism, which major news organizations have employed as far back as the early 2000s ( Thurman & Walters, 2013 ), is routinely used to cover sporting events, political speeches, and breaking news such as terror attacks ( Thorsen & Jackson, 2018 ). Although it is sometimes considered to be a text-based parallel to live broadcast news, it differs in the extent to which it typically engages with audiences and how it conveys its narrative. It is closely associated, and thus frequently interchanged, with the notion of live tweeting.

News can be produced through live blogging by an extensive range of individual social actors that include both staff journalists and citizens acting as journalists ( Thurman & Rodgers, 2014 ). This is made possible through the use of technological actants that are often not proprietary to news organizations, such as content management systems and blogging platforms, as well as through social media platforms ( Thorsen, 2013 ). Live blogging may be performed as one-way communication with general (and specialized) audiences, but it sometimes includes affordances for audience engagement―as in directly soliciting and answering questions during unfolding events or incorporating contextual information provided by members of the audience ( Bennett, 2016 ).

Its practices, norms, routines, and roles are characterized primarily by speed and curation, as actors must not only observe real-time events and break them as news but also quickly make sense of those events in order to distinguish their products from competitors’ while incorporating content created by other members of a social network ( Thurman & Walters, 2013 ). As such, its practitioners seemingly are more cautious with their knowledge claims because they explicitly recognize that emerging events can be confusing, even when observed firsthand, and the immediacy of their posts makes fact-checking difficult, if not unfeasible. They may, however, draw on the public to verify information for knowledge claims, such as by asking other users to confirm the physical address where a news event is taking place. As a form of knowledge, it is typically composed of a series of individual products (i.e., bullets or tweets) best characterized as news-as-impression but that add up to (and can be consumed as) news-as-item once the event is over.

The narrative structure of live blogging is fragmented and not organized by textual coherence but, rather, by reverse chronology, with the latest observation usually on top ( Matheson & Wahl-Jorgensen, 2020 ). Its temporality is characterized by overlapping moments in time, as previously reported developments are contextualized while new developments are reported, and new voices are occasionally interspersed via affordances such as retweets ( Matheson & Wahl-Jorgensen, 2020 ). The authorial stance is marked by networked balance and co-presence, rather than objectivity, because authors typically adopt a mix of their observations and opinions while inviting and including discrete moments shared by fellow journalists, sources, and audience members ( Matheson & Wahl-Jorgensen, 2020 ). The status of the text is explicitly conveyed as dynamic and temporary, with an understanding that updates are often open, incomplete, and unfinished ( Matheson & Wahl-Jorgensen, 2020 ).

The consequence of these attributes is that live blogging is more willing to cede some of its epistemological authority in defining news in large part because of how it produces news. Although journalism is often described as “a first draft of history,” live blogging is more accommodating of partial accounts, forgiving of corrections, and willing to include unverified claims. In other words, it recognizes itself as being particularly temporary within the ecosystem of journalism—a moment in time that will be replaced by fuller accounts. Moreover, live blogging is more distanced from objectivity norms and open to audiences, making knowledge production about “news” a more distributed endeavor. As Matheson and Wahl-Jorgensen (2020 , p. 313) state, “the live blog can be understood as a journalistic response to the logics of social media”—although, it is contended in this article, to a lesser degree than participatory journalism.

Data Journalism

Data journalism may be conceptualized as a hybrid form that is grounded in “data analysis and the presentation of such analysis” ( Coddington, 2015 , p. 334). It may also be delineated by its content, which

has a central thesis (or purpose) that is primarily attributed to (or fleshed out by) quantified information (e.g., statistics or raw sensor data); involves at least some original data analysis by the item’s author(s); and includes a visual representation of data ( Zamith, 2019b , p. 478).

The form is not itself new—it is an outgrowth of a longer tradition of precision journalism and computer-assisted reporting ( Houston, 1996 ; Splendore, 2016 )—but data journalism distinguishes itself by decoupling from investigative journalism and calling greater attention to best practices in data sharing and visualization ( Cairo, 2019 ; Coddington, 2015 ). It has also developed during a period when journalists have greater access to digital data and accessible tools, is now produced by major news organizations, and now has its own award bodies ( Zamith, 2019b ). However, journalists do struggle to get access to worthwhile and reliable data in many geographical contexts ( Lewis & Nashmi, 2019 ; Porlezza & Splendore, 2019 ).

The social actors involved in the production of data journalism typically have backgrounds in statistics, computer science, design, and/or journalism—and a new professional class has emerged that reflects a “cross-field hybridity” ( Coddington, 2015 , p. 337) by incorporating multiple of those backgrounds ( Hermida & Young, 2017 ). It is defined in part by the technological actants that enable it, including statistical analysis software and data visualization tools, as well as the premium that is placed on open-source solutions ( Splendore, 2016 ). Its audiences are typically passive but may take an active role in shaping news production—high-profile data journalism projects have involved audience participation, although participatory affordances are typically limited ( Zamith, 2019b )—and are usually given opportunities to interact with content.

The practices, norms, routines, and roles of this form focus on central tendencies rather than outliers ( Young & Hermida, 2015 ) and emphasize the ideals of transparency and sharing ( Coddington, 2015 ), yet they still legitimate themselves through the lens of some key traditional journalism principles ( Borges-Rey, 2020 ). Its knowledge claims are rooted in science and quantification, and further benefit from a mythology around the objectivity of quantified claims ( Lewis & Westlund, 2015b ). Its forms of knowledge involve news-as-item for many of its “everyday” variants ( Zamith, 2019b ) as well as the deeper analyses better characterized as news-about-relations ( Young & Hermida, 2015 ).

The narrative structure of data journalism is ordered and, in many ways, adheres to traditional structures ( Borges-Rey, 2020 ), but it is more interactive and modular to accommodate visualizations, which are inherent to the storytelling ( Cairo, 2019 ; Young & Hermida, 2015 ). Regarding temporality, it follows an ordered, interpretive framework that incorporates human sources but is most dependent on data sources ( Porlezza & Splendore, 2019 ; Zamith, 2019b ). Its authorial stance is objective, with some recognition that the author’s account is incomplete and thus open to further interpretation via the interactive features of data visualizations. The status of text is typically presumed to be finished or semi-finished, although texts may include visuals, models, and modules that automatically update as new data are entered.

Data journalism ultimately redefines epistemological authority as the result of data and scientific analyses that are further illustrated through anecdotal lived experience ( Young & Hermida, 2015 ). Indeed, as Cairo (2019) contends, “Numbers and charts look and feel objective, precise, and, as a consequence, seductive and convincing” (p. xi). Data journalism has mainstreamed hypothesis-testing and data-driven logics within journalism, although epistemological tensions still emerge when traditional journalists work alongside their more data-oriented counterparts ( Borges-Rey, 2020 ). However, although the production of data journalism marks an epistemological shift from traditional journalism, it is not a break. As Borges-Rey (2020) notes, data journalists routinely oscillate between “newshound” and “techie” approaches to news production. Furthermore, data journalists often legitimize their work as news production by referencing journalistic ideals and adopting its language ( Coddington, 2015 ).

Automated Journalism

Automated journalism refers to “algorithmic processes that convert data into narrative news texts with limited to no human intervention beyond the initial programming” ( Carlson, 2015 , p. 417). It is a more advanced form of computational journalism ( Coddington, 2015 ) that uses algorithms to largely automate the collection, writing, publication, and/or the distribution of news ( Diakopoulos, 2019 ). Although machine-driven forms of journalism also trace many of their roots to precision journalism and computer-assisted reporting—and similarly require some form of data to be executed—they rapidly gained social capital within journalistic spaces starting in the mid-2000s ( Zamith, 2019a ). There are now companies such as Automated Insights that are advancing the technical capabilities and professional use of algorithms for automating news production, and they count major news organizations such as the Associated Press as their clients ( Carlson, 2018 ). Perhaps most important, automated journalism has changed the scale at which journalism can be produced ( Diakopoulos, 2019 ). It has also introduced new ways of communicating journalism, as with chatbots ( Jones & Jones, 2019 ).

The social actors involved in automated journalism are mostly highly technical and include technologically oriented journalists, computational linguists, and vendors of proprietary algorithms ( Carlson, 2018 ; Diakopoulos, 2019 ). Its technological actants include mostly proprietary algorithms for natural language processing and natural language generation that give humans some degree of structured control (e.g., creating templates) but aim to require minimal human involvement ( Dörr, 2016 ). The audiences in automated journalism generally remain passive, although content may be personalized based on predictions from historical data and their active choices ( Zamith, 2019a ). Those recommendation systems can be designed to fit commercial purposes as well as distinct democratic models ( Helberger, 2019 ).

The practices, norms, routines, and roles of automated journalism are oriented toward abstraction, structuration, quantification, and personalization, with the objective of simultaneously breaking news down to granular, discrete elements while using those elements to create news products that are indistinguishable from their human-generated counterparts ( Coddington, 2015 ; Graefe et al., 2018 ). Its knowledge claims are derived from mechanical analyses of data that give them “algorithmic authority” by virtue of their presumed impartiality―even as those algorithms are themselves biased by the humans who create them ( Carlson, 2015 ). Its forms of knowledge include both news-as-item (e.g., automated news stories) and bite-size “structured information” ( Splendore, 2016 , p. 349) that can be used to power news-as-impression (e.g., chatbots and automated notifications).

The narrative structure of automated journalism is highly structured―indeed, its most common products are based on templates―and may be both coherent (as in the case of news stories; see Diakopoulos, 2019 ) and fragmented (as in the case of chatbots; see Jones & Jones, 2019 ). Regarding temporality, it typically follows an ordered, systematically interpreted framework that draws chiefly upon semistructured documents and structured data sets ( Dörr, 2016 ). Its authorial stance is objective, again drawing upon the purported impartiality of the algorithms that produced the news ( Broussard, 2018 ; Carlson, 2018 ). The status of text is typically presumed to be finished, although there is greater presumption of dynamism in response to the automated personalization of those texts ( Zamith, 2019a ).

As Carlson (2018) argues, automated journalism “represents a core departure from how journalism has been understood and cannot be contained as an extension of journalism’s professional logic” (p. 1765). Under this form, human judgment should play a limited (or unchanging) role in the production of knowledge about the news; instead, production should be guided by abstracted principles and enacted by algorithms ( Coddington, 2015 ). Furthermore, it shifts the idea of news as public, shared knowledge toward individual, personalized knowledge ( Splendore, 2016 ). It thus challenges traditional notions of journalistic epistemology even as it arguably serves as the apotheosis of one its key production values: objectivity ( Carlson, 2018 ). However, although this stream of journalism emphasizes the technical by its very nature, scholars have argued that the technological actants and activities involved in this space remain deeply influenced by human actors ( Broussard, 2018 ; Diakopoulos, 2019 ). Consequently, and in large part due to the current state of technology, the epistemological break in contemporary practice is more limited than theory would suggest—and this phenomenon is unlikely to change in the near future.

Discussion and Research Directions

Until recently, scholars have studied and described news production as a set of human-oriented activities that largely share a universal set of characteristics (see review in Westlund & Ekström, 2019 ). The authors of this article have deliberately sought to do otherwise, and instead called attention to recent arguments underscoring the growing role of technological actants in journalism and the heterogeneous nature of news production―which, in turn, have implications for how people come to understand “news.” This position primarily draws on three streams of research: the epistemologies of journalism (e.g., Ekström & Westlund, 2019a , 2019b ), sociotechnical approaches to understanding news work (e.g., Lewis & Westlund, 2015a ; Zamith, 2019a ), and systematic comparisons of diverging news production processes (e.g., Matheson & Wahl-Jorgensen, 2020 ). The authors contribute to those streams by proposing the epistemologies of journalism matrix, which provides scholars with an analytic framework for examining the heterogeneity of news production in terms of its implicated entities, its cultures and methods, and its positionality in relation to matters of knowledge and authority.

The utility of the matrix is illustrated through an examination of four forms of journalism: participatory journalism, live blogging, data journalism, and automated journalism. The analysis highlights three points. First, contemporary news production is deeply influenced by myriad technological actants, which are reshaping how knowledge about current events is being created, evaluated, and disseminated. Second, professional journalists are losing epistemic authority over the news as key activities are delegated to algorithms created by non-journalists and to citizens who have become more present in news production. Third, the outputs of news production are becoming more diverse both in form and in content, further challenging long-standing norms about what is and is not “journalism.” However, those are but four forms and hardly capture all of what journalism encompasses. The authors thus invite scholars to expand on the matrix by applying it to other forms of journalism—and, in the process, refine the matrix itself and advance its theoretical implications.

In addition, the authors believe it is important for any scholar studying news production to be mindful of three key developments in their future work. First, it is apparent that what is “news” to different people is quite different today from times past. The history of journalism has been marked by many significant changes as to what is considered news, how it is shaped, and who distributes it. However, digital devices and platforms have made news available 24/7, and the ease of producing and disseminating content these days has contributed to an explosion of news produced by a large and diverse array of actors. Moreover, that news is increasingly sought on just a few platforms (e.g., Google and Facebook) that often flatten traditional media hierarchies by placing news produced by professional journalistic outlets alongside content created by nonprofessionals. The consequence is that there are now more interlopers seeking to pass their content off as “news”—from individual trolls seeking to get a rise out of people ( Quandt, 2018 ) to actors hoping to monetize their content ( Braun & Eklund, 2019 ) and states seeking to gain political advantage ( Marwick & Lewis, 2017 )—which has further complicated a historically contested term. Moreover, the past decade has been marked by low or declining levels of trust in news media in many areas of the world ( Fletcher & Park, 2017 ), as well as sustained attacks on news media ( Carlson et al., 2021 ; Waisbord, 2020 ).

Second, the heterogeneity of “news” and “news production” requires scholars to think carefully about how they operationalize those variables in their work ( Mast et al., 2017 ; Waisbord, 2018 ). For example, there is a substantive and growing body of literature on news consumption and news avoidance that builds on quantitative data and analyses of media effects ( Skovsgaard & Andersen, 2020 ). Such studies often conceptualize and operationalize news and news production processes in ways that make them appear more homogeneous than they are in practice ( Mast et al., 2017 ). As such, differences in research findings may be due, in part, to distinct understandings of those concepts, in light of their heterogeneity. It is imperative, therefore, for scholars to both examine the evolution of these understandings and account for them in research by either offering more granular options or detailing their operationalizations.

Third, the power dependencies in news production have changed markedly in recent years ( Ekström & Westlund, 2019b ). It is now much more difficult for practitioners to adhere to the values typically associated with their occupational ideology or to resist changes instituted by superiors and consolidating ownership ( Coddington, 2019 ; Vos & Heinderyckx, 2015 ). News producers, once seen as gatekeepers, are now themselves gatekept by algorithms employed by platform companies ( Gillespie, 2014 ; Wallace, 2018 )—algorithms that producers often believe they must adjust to even as they recognize such actions only make them more dependent ( Nielsen & Ganter, 2018 ; Pickard, 2020 ). Their future is sometimes tied to technologies developed far from newsrooms ( Braun & Eklund, 2019 ; Diakopoulos, 2019 ; Tandoc, 2019 ). Thus, contemporary analyses of news production should account for power differences among institutional actors—recognizing that journalistic actors are now less likely to exert dominance.

At the same time, although this article has focused on change and on digital journalism, it is important to recognize that a non-negligible amount of what is commonly referred to as “journalism” has remained reasonably stable—and that much of the change is rooted in pre-digital expectations, practices, and capabilities ( Zelizer, 2019 ). Moreover, this article has focused on the mainstream applications of journalism in Western contexts, and it is important to recognize that the histories and legacies of other places impact the developmental trajectories—and epistemological notions—of digital journalism differently in those contexts ( Mellado, 2021 ).

Nevertheless, history has shown that news production will continue to evolve alongside broader economic, political, professional, social, and technological shifts—and in doing so spring new forms and assemblages. An epistemological lens affords scholars a useful and adaptable approach for understanding the implications of those changes to the production of knowledge about news. Nevertheless, it is apparent that future scholarship will demand further theoretical and methodological development in order to keep up with a rapidly changing ecosystem and information regime.

Acknowledgments

The work of Oscar Westlund was supported by Riksbankens Jubileumsfond [grant number RJ P16-0715].

Further Reading

• Ekström, M. , & Westlund, O. (2019). Epistemology and journalism . In Oxford research encyclopedia of communication . Oxford University Press.
• Pressman, M. (2018). On press: The liberal values that shaped the news . Harvard University Press.
• Ryfe, D. M. (2019). Journalism and the public . Polity.
• Steensen, S. , & Westlund, O. (2021). What is digital journalism studies? Routledge.
• Tandoc, E. C. (2019). Analyzing analytics: Disrupting journalism one click at a time . Routledge.
• Usher, N. (2021). News for the rich, white, and blue: How place and power distort American journalism . Columbia University Press.
• Ahva, L. , & Steensen, S. (2019). Journalism theory. In K. Wahl-Jorgensen & T. Hanitzsch (Eds.), The handbook of journalism studies (pp. 38–54). Routledge.
• Ali, C. , Schmidt, T. R. , Radcliffe, D. , & Donald, R. (2019). The digital life of small market newspapers . Digital Journalism , 7 (7), 886–909.
• Anderson, C. W. , & Revers, M. (2018). From counter-power to counter-pepe: The vagaries of participatory epistemology in a digital age . Media and Communication , 6 (4), 24.
• Ataman, B. , & Çoban, B. (2018). Counter-surveillance and alternative new media in Turkey . Information, Communication and Society , 21 (7), 1014–1029.
• Bakker, P. (2012). Aggregation, content farms, and Huffinization . Journalism Practice , 6 (5–6), 627–637.
• Barnhurst, K. G. (2011). The problem of modern time in American journalism . KronoScope , 11 (1/2), 98–123.
• Belair-Gagnon, V. , & Holton, A. E. (2018). Boundary work, interloper media, and analytics in newsrooms . Digital Journalism , 6 (4), 492–508.
• Bennett, D. (2016). Sourcing the BBC’s live online coverage of terror attacks . Digital Journalism , 4 (7), 861–874.
• Benson, R. (2006). News media as a “journalistic field”: What Bourdieu adds to new institutionalism, and vice versa . Political Communication , 23 (2), 187–202.
• Borger, M. , van Hoof, A. , & Sanders, J. (2019). Exploring participatory journalistic content: Objectivity and diversity in five examples of participatory journalism . Journalism , 20 (3), 444–466.
• Borges-Rey, E. (2020). Towards an epistemology of data journalism in the devolved nations of the United Kingdom: Changes and continuities in materiality, performativity and reflexivity . Journalism , 21 (7), 915–932.
• Bourdieu, P. (1993). The field of cultural production: Essays on art and literature . Columbia University Press.
• boyd, d. , & Ellison, N. B. (2007). Social network sites: Definition, history, and scholarship . Journal of Computer-Mediated Communication , 13 (1), 210–230.
• Braun, J. A. (2015). This program is brought to you by . . . . Yale University Press.
• Braun, J. A. (2019). News distribution . In Oxford research encyclopedia of communication . Oxford University Press.
• Braun, J. A. , & Eklund, J. L. (2019). Fake news, real money: Ad tech platforms, profit-driven hoaxes, and the business of journalism . Digital Journalism , 7 , 1–21.
• Broussard, M. (2018). Artificial unintelligence: How computers misunderstand the world . MIT Press.
• Bruns, A. (2008). The active audience: Transforming journalism from gatekeeping to gatewatching. In C. A. Paterson & D. Domingo (Eds.), Making online news (pp. 171–184). Lang.
• Bucy, E. P. (2004). Interactivity in society: Locating an elusive concept . The Information Society , 20 (5), 373–383.
• Cairo, A. (2019). How charts lie: Getting smarter about visual information . Norton.
• Carlson, M. (2015). The robotic reporter . Digital Journalism , 3 (3), 416–431.
• Carlson, M. (2017). Journalistic authority: Legitimating news in the digital era . Columbia University Press.
• Carlson, M. (2018). *Automating judgment? Algorithmic judgment, news knowledge, and journalistic professionalism[https://doi.org/10.1177/1461444817706684].* New Media & Society , 20 (5), 1755–1772.
• Carlson, M. , Robinson, S. , & Lewis, S. C. (2021). *Digital press criticism: The symbolic dimensions of Donald Trump’s assault on U.S. journalists as the “enemy of the people”[https://doi.org/10.1080/21670811.2020.1836981].* Digital Journalism , 9 (6), 737–754.
• Chua, S. , & Duffy, A. (2019). Friend, foe or frenemy? Traditional journalism actors’ changing attitudes towards peripheral players and their innovations . Media and Communication , 7 (4), 112–122.
• Chua, S. , & Westlund, O. (2019). Audience-centric engagement, collaboration culture and platform counterbalancing: A longitudinal study of ongoing sensemaking of emerging technologies . Media and Communication , 7 (1), 153–165.
• Coddington, M. (2015). Clarifying journalism’s quantitative turn . Digital Journalism , 3 (3), 331–348.
• Coddington, M. (2019). Aggregating the news: Secondhand knowledge and the erosion of journalistic authority . Columbia University Press.
• Coddington, M. , Lewis, S. C. , & Holton, A. E. (2018). Measuring and evaluating reciprocal journalism as a concept . Journalism Practice , 12 (8), 1039–1050.
• Compton, J. R. , & Benedetti, P. (2010). Labour, new media and the institutional restructuring of journalism . Journalism Studies , 11 (4), 487–499.
• Cook, T. E. (1998). Governing with the news: The news media as a political institution . University of Chicago Press.
• Costera Meijer, I. , & Groot Kormelink, T. (2015). Checking, sharing, clicking and linking . Digital Journalism , 3 (5), 664–679.
• Diakopoulos, N. (2019). Automating the news: How algorithms are rewriting the media . Harvard University Press.
• Domingo, D. , Quandt, T. , Heinonen, A. , Paulussen, S. , Singer, J. B. , & Vujnovic, M. (2008). Participatory journalism practices in the media and beyond . Journalism Practice , 2 (3), 326–342.
• Dörr, K. N. (2016). Mapping the field of algorithmic journalism . Digital Journalism , 4 (6), 700–722.
• Duffy, A. , Ling, R. , Kim, N. , Tandoc, E. , & Westlund, O. (2020). News: Mobiles, mobilities and their meeting points . Digital Journalism , 8 (1), 1–14.
• Ekström, M. (2002). Epistemologies of TV journalism: A theoretical framework . Journalism , 3 (3), 259–282.
• Ekström, M. , Ramsälv, A. , & Westlund, O. (2021). The epistemologies of breaking news . Journalism Studies , 22 (2), 174–192.
• Ekström, M. , & Westlund, O. (2019a). Epistemology and journalism . In Oxford research encyclopedia of communication . Oxford University Press.
• Ekström, M. , & Westlund, O. (2019b). The dislocation of news journalism: A conceptual framework for the study of epistemologies of digital journalism . Media and Communication , 7 (1), 259–270.
• Elliott, A. , & Urry, J. (2010). Mobile lives . Routledge.
• Engelke, K. M. (2019). Online participatory journalism: A systematic literature review . Media and Communication , 7 (4), 31–44.
• Engelke, K. M. (2020). Enriching the conversation: Audience perspectives on the deliberative nature and potential of user comments for news media . Digital Journalism , 8 (4), 447–466.
• Fenton, N. (2011). Deregulation or democracy? New media, news, neoliberalism and the public interest . Continuum , 25 (1), 63–72.
• Figenschou, T. U. , & Ihlebæk, K. A. (2019). Challenging journalistic authority . Journalism Studies , 20 (9), 1221–1237.
• Fletcher, R. , & Park, S. (2017). The impact of trust in the news media on online news consumption and participation . Digital Journalism , 5 (10), 1281–1299.
• Flew, T. , & Swift, A. (2013). Regulating journalists? The Finkelstein review, the convergence review and news media regulation in Australia . Journal of Applied Journalism & Media Studies , 2 (1), 181–199.
• Gieryn, T. F. (1999). Cultural boundaries of science: Credibility on the line . University of Chicago Press.
• Gillespie, T. (2014). The relevance of algorithms. In T. Gillespie , P. Boczkowski , & K. Foot (Eds.), Media technologies: Essays on communication, materiality, and society (pp. 167–194). MIT Press.
• Graefe, A. , Haim, M. , Haarmann, B. , & Brosius, H.‑B. (2018). Readers’ perception of computer-generated news: Credibility, expertise, and readability . Journalism , 19 (5), 595–610.
• Gutsche, R. E., Jr. , & Hess, K. (2020). Placeification: The transformation of digital news spaces into “places” of meaning . Digital Journalism , 8 (5), 586–595.
• Hågvar, Y. B. (2019). News media’s rhetoric on Facebook . Journalism Practice , 13 (7), 853–872.
• Hamilton, J. (2004). All the news that’s fit to sell: How the market transforms information into news . Princeton University Press.
• Hanusch, F. , & Maares, P. (2021). News production. In K. B. Jensen (Ed.), A handbook of media and communication research (3rd ed., pp. 93–111). Routledge.
• Heft, A. , & Dogruel, L. (2019). Searching for autonomy in digital news entrepreneurism projects . Digital Journalism , 7 (5), 678–697.
• Helberger, N. (2019). On the democratic role of news recommenders . Digital Journalism , 7 (8), 993–1012.
• Hermida, A. (2020). Post-publication gatekeeping: The interplay of publics, platforms, paraphernalia, and practices in the circulation of news . Journalism & Mass Communication Quarterly , 97 (2), 469–491.
• Hermida, A. , Lewis, S. C. , & Zamith, R. (2014). Sourcing the Arab spring: A case study of Andy Carvin’s sources on Twitter during the Tunisian and Egyptian revolutions . Journal of Computer-Mediated Communication , 19 (3), 479–499.
• Hermida, A. , & Young, M. L. (2017). Finding the data unicorn . Digital Journalism , 5 (2), 159–176.
• Holt, K. , Ustad Figenschou, T. , & Frischlich, L. (2019). *Key dimensions of alternative news media[https://doi.org/10.1080/21670811.2019.1625715]. Digital Journalism , 7 (7), 860–869.
• Houston, B. (1996). Computer-assisted reporting: A practical guide . St. Martin’s.
• Jack, C. (2017). Lexicon of lies . Data & Society.
• John, R. R. (1995). Spreading the news: The American postal system from Franklin to Morse . Harvard University Press.
• Jones, B. , & Jones, R. (2019). Public service chatbots: Automating conversation with BBC news . Digital Journalism , 7 (8), 1032–1053.
• Jönsson, A. M. , & Örnebring, H. (2010). User-generated content and the news: Empowerment of citizens or interactive illusion? Journalism Practice , 5 (2), 127–144.
• Kim, Y. , & Lowrey, W. (2015). Who are citizen journalists in the social media environment? Digital Journalism , 3 (2), 298–314.
• Kligler-Vilenchik, N. , & Tenenboim, O. (2020). Sustained journalist–audience reciprocity in a meso news-space: The case of a journalistic WhatsApp group . New Media & Society , 22 (2), 264–282.
• Kreiss, D. , & Ananny, M. (2013). Responsibilities of the state: Rethinking the case and possibilities for public support of journalism . First Monday , 18 (4).
• Lashmar, P. (2017). No more sources? Journalism Practice , 11 (6), 665–688.
• Lawrence, R. G. , Radcliffe, D. , & Schmidt, T. R. (2018). Practicing engagement . Journalism Practice , 12 (10), 1220–1240.
• Lewis, N. P. , & Nashmi, E. A. (2019). Data journalism in the Arab region: Role conflict exposed . Digital Journalism , 7 (9), 1200–1214.
• Lewis, S. C. (2012). The tension between professional control and open participation . Information, Communication and Society , 15 (6), 836–866.
• Lewis, S. C. , & Westlund, O. (2015a). Actors, actants, audiences, and activities in cross-media news work . Digital Journalism , 3 (1), 19–37.
• Lewis, S. C. , & Westlund, O. (2015b). Big data and journalism . Digital Journalism , 3 (3), 447–466.
• Lewis, S. C. , Zamith, R. , & Coddington, M. (2020). Online harassment and its implications for the journalist–audience relationship . Digital Journalism , 8 (8), 1047–1067.
• Lowrey, W. (2011). Institutionalism, news organizations and innovation . Journalism Studies , 12 (1), 64–79.
• Marwick, A. , & Lewis, R. (2017). Manipulation and disinformation online . Data & Society.
• Mast, J. , Coesemans, R. , & Temmerman, M. (2017). Hybridity and the news: Blending genres and interaction patterns in new forms of journalism . Journalism , 18 (1), 3–10.
• Matheson, D. , & Wahl-Jorgensen, K. (2020). The epistemology of live blogging . New Media & Society , 22 (2), 300–316.
• Mellado, C. (2021). Beyond journalistic norms: Role performance and news in comparative perspective . Routledge.
• Murschetz, P. C. (2020). State aid for independent news journalism in the public interest? A critical debate of government funding models and principles, the market failure paradigm, and policy efficacy . Digital Journalism , 8 (6), 720–739.
• Napoli, P. M. (2011). Audience evolution: New technologies and the transformation of media audiences . Columbia University Press.
• Neuman, W. R. (1991). The future of the mass audience . Cambridge University Press.
• Nielsen, R. K. (2012). How newspapers began to blog . Information, Communication and Society , 15 (6), 959–978.
• Nielsen, R. K. (2017). Digital news as forms of knowledge: A new chapter in the sociology of knowledge. In P. Boczkowski & C. W. Anderson (Eds.), Remaking the news: Essays on the future of journalism scholarship in the digital age (pp. 1–27). MIT Press.
• Nielsen, R. K. , & Ganter, S. A. (2018). Dealing with digital intermediaries: A case study of the relations between publishers and platforms . New Media & Society , 20 (4), 1600–1617.
• Örnebring, H. (2018). Journalists thinking about precarity: Making sense of the “new normal.” International Symposium on Online Journalism , 8 (1), 109–126.
• Pavlik, J. V. (2004). A sea-change in journalism: Convergence, journalists, their audiences and sources . Convergence , 10 (4), 21–29.
• Pickard, V. (2020). *Restructuring democratic infrastructures: A policy approach to the journalism crisis[https://doi.org/10.1080/21670811.2020.1733433]. Digital Journalism , 8 (6), 704–719.
• Porlezza, C. , & Splendore, S. (2019). From open journalism to closed data: Data journalism in Italy . Digital Journalism , 7 (9), 1230–1252.
• Quandt, T. (2018). Dark participation . Media and Communication , 6 (4), 36.
• Radcliffe, D. , & Ali, C. (2017). Small-market newspapers in the digital age . Tow Center for Digital Journalism.
• Schmitz Weiss, A. (2015). Place-based knowledge in the twenty-first century . Digital Journalism , 3 (1), 116–131.
• Singer, J. B. (2012). The ethics of social journalism. Australian Journalism Review , 34 (1), 3–16.
• Singer, J. B. , Domingo, D. , Heinonen, A. , Hermida, A. , Paulussen, S. , Quandt, T. , Reich, Z. , & Vujnovic, M. (2011). Participatory journalism: Guarding open gates at online newspapers . Wiley-Blackwell.
• Sjøvaag, H. , Stavelin, E. , Karlsson, M. , & Kammer, A. (2019). The hyperlinked Scandinavian news ecology . Digital Journalism , 7 (4), 507–531.
• Skovsgaard, M. , & Andersen, K. (2020). Conceptualizing news avoidance: Towards a shared understanding of different causes and potential solutions . Journalism Studies , 21 (4), 459–476.
• Splendore, S. (2016). Quantitatively oriented forms of journalism and their epistemology . Sociology Compass , 10 (5), 343–352.
• Stahel, L. , & Schoen, C. (2020). Female journalists under attack? Explaining gender differences in reactions to audiences’ attacks . New Media & Society , 22 (10), 1849–1867.
• Steensen, S. , & Eide, T. (2019). News flows, inter-media connectivity and societal resilience in times of crisis . Digital Journalism , 7 (7), 932–951.
• Swart, J. , Peters, C. , & Broersma, M. (2019). Sharing and discussing news in private social media groups . Digital Journalism , 7 (2), 187–205.
• Thorsen, E. (2013). Live blogging and social media curation: Challenges and opportunities for journalism. In K. Fowler-Watt & S. Allan (Eds.), Journalism: New challenges (pp. 123–145). Centre for Journalism & Communication Research.
• Thorsen, E. , & Jackson, D. (2018). Seven characteristics defining online news formats . Digital Journalism , 6 (7), 847–868.
• Thurman, N. , & Rodgers, J. (2014). Citizen journalism in real time: Live blogging and crisis events. In E. Thorsen & S. Allan (Eds.), Citizen journalism: Global perspectives (pp. 81–95). Lang.
• Thurman, N. , & Walters, A. (2013). Live blogging: Digital journalism’s pivotal platform? Digital Journalism , 1 (1), 82–101.
• Usher, N. (2014). Making news at The New York Times . University of Michigan Press.
• Usher, N. (2019). Putting “place” in the center of journalism research: A way forward to understand challenges to trust and knowledge in news . Journalism & Communication Monographs , 21 (2), 84–146.
• Vos, T. , & Heinderyckx, F. (2015). Gatekeeping in transition . Routledge.
• Vujnovic, M. , Singer, J. B. , Paulussen, S. , Heinonen, A. , Reich, Z. , Quandt, T. , Hermida, A. , & Domingo, D. (2010). Exploring the political–economic factors of participatory journalism . Journalism Practice , 4 (3), 285–296.
• Waisbord, S. (2018). Truth is what happens to news . Journalism Studies , 19 (13), 1866–1878.
• Waisbord, S. (2020). Mob censorship: Online harassment of US journalists in times of digital hate and populism . Digital Journalism , 8 (8), 1030–1046.
• Wall, M. (2017). Mapping citizen and participatory journalism . Journalism Practice , 11 (2–3), 134–141.
• Wallace, J. (2018). Modelling contemporary gatekeeping . Digital Journalism , 6 (3), 274–293.
• Weischenberg, S. , & Matuschek, C. (2008). News production and technology . In W. Donsbach (Ed.), The international encyclopedia of communication . Wiley.
• Westlund, O. (2013). Mobile news: A review and model of journalism in an age of mobile media . Digital Journalism , 1 (1), 6–26.
• Westlund, O. , & Ekström, M. (2018). News and participation through and beyond proprietary platforms in an age of social media . Media and Communication , 6 (4), 1.
• Westlund, O. , & Ekström, M. (2019). News organizations and routines. In K. Wahl-Jorgensen & T. Hanitzsch (Eds.), The handbook of journalism studies (2nd ed., pp. 73–88). Routledge.
• Westlund, O. , & Quinn, S. (2018). Mobile journalism and mojos . In Oxford research encyclopedia of communication . Oxford University Press.
• White, D. M. (1950). The “gate keeper”: A case study in the selection of news. The Journalism Quarterly , 27 (4), 383–391.
• Wu, S. , Tandoc, E. C. , & Salmon, C. T. (2019). A field analysis of journalism in the automation age: Understanding journalistic transformations and struggles through structure and agency . Digital Journalism , 7 (4), 428–446.
• Xu, D. (2015). Online censorship and journalists’ tactics . Journalism Practice , 9 (5), 704–720.
• Young, M. L. , & Hermida, A. (2015). From Mr. and Mrs. Outlier to central tendencies . Digital Journalism , 3 (3), 381–397.
• Zamith, R. (2018). Quantified audiences in news production . Digital Journalism , 6 (4), 418–435.
• Zamith, R. (2019a). Algorithms and journalism . In Oxford research encyclopedia of communication . Oxford University Press.
• Zamith, R. (2019b). Transparency, interactivity, diversity, and information provenance in everyday data journalism . Digital Journalism , 7 (4), 470–489.
• Zelizer, B. (1993). Journalists as interpretive communities . Critical Studies in Mass Communication , 10 (3), 219–237.
• Zelizer, B. (2019). Why journalism is about more than digital technology . Digital Journalism , 7 (3), 343–350.

Related Articles

• Algorithms and Journalism
• Mobile Applications and Journalistic Work
• Epistemology and Journalism

Printed from Oxford Research Encyclopedias, Communication. Under the terms of the licence agreement, an individual user may print out a single article for personal use (for details see Privacy Policy and Legal Notice).

date: 14 September 2024

• Cookie Policy
• Privacy Policy
• Legal Notice
• Accessibility
• [81.177.180.204]
• 81.177.180.204

Character limit 500 /500

• Original article
• Open access
• Published: 31 March 2021

The role of analytical reasoning and source credibility on the evaluation of real and fake full-length news articles

• Didem Pehlivanoglu   ORCID: orcid.org/0000-0002-9082-9976 1 ,
• Tian Lin 1 ,
• Farha Deceus 1 ,
• Amber Heemskerk 1 ,
• Natalie C. Ebner 1 , 2 , 3 , 4   na1 &
• Brian S. Cahill 1   na1  

Cognitive Research: Principles and Implications volume  6 , Article number:  24 ( 2021 ) Cite this article

13k Accesses

18 Citations

7 Altmetric

Metrics details

Previous research has focused on accuracy associated with real and fake news presented in the form of news headlines only, which does not capture the rich context news is frequently encountered in real life. Additionally, while previous studies on evaluation of real and fake news have mostly focused on characteristics of the evaluator (i.e., analytical reasoning), characteristics of the news stimuli (i.e., news source credibility) and the interplay between the two have been largely ignored. To address these research gaps, this project examined the role of analytical reasoning and news source credibility on evaluation of real and fake full-length news story articles. The project considered both accuracy and perceived credibility ratings as outcome variables, thus qualifying previous work focused solely on news detection accuracy.

We conducted two independent but parallel studies, with Study 2 as a direct replication of Study 1, employing the same design but in a larger sample (Study 1: N  = 292 vs. Study 2: N  = 357). In both studies, participants viewed 12 full-length news articles (6 real, 6 fake), followed by prompts to evaluate each article’s veracity and credibility. Participants were randomly assigned to view articles with a credible or non-credible source and completed the Cognitive Reflection Test as well as short demographic questions.

Consistent across both studies, higher analytical reasoning was associated with greater fake news accuracy, while analytical reasoning was not associated with real news accuracy. In addition, in both studies, higher analytical reasoning was associated with lower perceived credibility for fake news, while analytical reasoning was not associated with perceived credibility for real news. Furthermore, lower analytical reasoning was associated with greater accuracy for real (but not fake) news from credible compared to non-credible sources, with this effect only detected in Study 2.

Conclusions

The novel results generated in this research are discussed in light of classical vs. naturalistic accounts of decision-making as well as cognitive processes underlying news articles evaluation. The results extend previous findings that analytical reasoning contributes to fake news detection to full-length news articles. Furthermore, news-related cues such as the credibility of the news source systematically affected discrimination ability between real and fake news.

Introduction

Fake news refers to “fabricated information that mimics news media content in form but not in organizational process or intent” (Lazer et al., 2018 , p. 1094). While fake news is certainly not a new occurrence—e.g., tabloid magazines have been around for nearly a century (Murray, 2013 )—its prominence in and impact on our culture has been growing. This is also related to enhanced global connectedness and broader use of online media platforms in modern society which have drastically increased access to news but also increased distribution of misinformation via fake news. One study estimated that the average American encountered between one and three fake news articles during the month prior to the 2016 presidential election (Allcott & Gentzkow, 2017 ). Given the prevalence of fake news, the relevant question is, how good are people at detecting real and fake news? Recent polls indicate that a significant portion of Americans (47%) report having difficulty distinguishing between real and fake news (Associated Press, 2019 ). Analysis of Facebook activity of the top 20 fake and real news stories showed that user engagement was greater for fake compared to real news stories (Silverman et al., 2016 ). Further, in an analysis of 126,000 real and fake news stories tweeted by about 3 million Twitter users, fake compared to real news spread more than 4.5 million times faster and in a wider range (Vosoughi et al., 2018 ).

Thus, it is crucial to investigate the processes involved in the evaluation of real and fake news. Here, we will address the following understudied research questions: (1) Is current evidence regarding an impact of analytical reasoning on fake news detection robust to methodological change (i.e., by presenting full-length articles as opposed to headlines only)?; (2) Does systematically varying the credibility of the news source influence news article evaluation?; and (3) What can we learn from examining the perceived credibility of the news articles, beyond real and fake news detection accuracy?

A cognitive account of fake and real news detection

According to Dual-Process Theory, individuals engage in two modes of information processing: a quick, intuitive mode (called System 1 ) and a slow, deliberate mode (called System 2 ; De Neys, 2012 ; Ferreira et al., 2006 ; Kahneman, 2011 ; Stanovich, 2009 ). System 1 is associated with low analytical reasoning and reliance on cognitive heuristics when making decisions (i.e., mental shortcuts based on prior knowledge and beliefs; Evans, 2007 ; Kahneman et al., 1982 ). System 2, in contrast, is associated with high analytical reasoning and involves careful and systematic consideration of information, and therefore, is less error prone than System 1.

In line with Dual-Process Theory, individuals who scored higher on a measure of analytical reasoning (i.e., Cognitive Reflection Test [CRT]; Frederick, 2005 ) were better at detecting fake news than individuals who scored low on analytical reasoning, regardless of whether the news content aligned with their political beliefs (Pennycook & Rand, 2019b ; also see Bago et al., 2020 ; Pennycook & Rand, 2019a for evidence supporting the role of analytic reasoning over and above political ideology on fake news detection). Furthermore, engagement in analytic reasoning accounted for ~ 56% to 95% of the variance in accurate detection of fake news (Pennycook & Rand, 2020 ). Lastly, while delusion-prone individuals, dogmatic individuals, and religious fundamentalists were more likely to believe fake news, these relationships were partially or fully explained by lower levels of analytical reasoning (Bronstein et al., 2019 ). In sum, there appears to be consistent evidence that lower analytical reasoning is associated with poorer fake news detection.

Current study

From previous research we know that the prevalence of fake news is significant and that individuals are poor at detecting fake news, due to low engagement of analytical reasoning. Previous research, however, focused on real and fake news detection accuracy using news headlines only, which does not capture the rich context news is frequently encountered in real life. Additionally, while previous studies considered characteristics of the evaluator (i.e., analytical reasoning), characteristics of the news stimuli (i.e., news source credibility) and the interplay between the two have been largely ignored. This paper went beyond previous work by employing full-length news articles (involving full news story along with a headline) to determine the role of: (i) analytical reasoning on evaluation of real and fake full-length news articles; (ii) credibility of the news source on evaluation of news articles; and (iii) perceived credibility of news articles, in addition to detection accuracy. Next, we will discuss the theoretical background leading to these central research aims.

Impact of analytical reasoning on real and fake news evaluation for full-length articles

In a typical fake news paradigm, participants are presented with news headlines only that are either factually accurate (real news) or not (fake news). Following each headline, participants are asked to make a set of evaluations, including, but not limited to, veracity (i.e., real vs. fake), familiarity, and willingness to share. Given that in real life, people are not typically restricted to solely using the headline to evaluate a news article (i.e., people typically can go beyond browsing headlines and read the full article), we employed full-length news articles. Limited research has attempted to shift the research field by adopting more ecologically valid news evaluation methodology. Besides being more ecologically valid, full-length articles provide rather rich contextual information and a larger set of diagnostic cues to determine credibility of the news (e.g., coherence in story line, writing and grammatical style). These additional features of full-length news articles as opposed to news headlines only inform the news evaluation process. To our knowledge only Schaewitz et al. ( 2020 ) employed full articles and found that people with high compared to those with low need for cognition were less susceptible to misinformation via fake news. Their design, however, did not involve a systematic manipulation of news veracity as they only used fake news stories. Thus, systematic variation of news veracity within a relatively more naturalistic decision-making context that allows for full exploration of the entire article, as done in the present study, has potential to further understanding of the cognitive mechanisms underlying real and fake news evaluation.

According to the Naturalistic Decision Making framework (Klein, 2008 , 2015 ), in fast-paced complex settings, decision makers mostly rely on past experiences to find the first workable option rather than trying to find the best possible solution, which requires analytical reasoning and is resource-intensive. People in real life come across real news more frequently than fake news (Allen et al., 2020 ; Guess et al., 2019 ). It is therefore possible that detection of real news relies on relatively more naturalistic decision-making processes which do not require analytical reasoning to the same extent as those involved in (less frequently encountered) fake news stories (Gigerenzer, 2007 ). Detection of fake news, in contrast, may rely more on deliberative processes that require high analytical reasoning and careful scrutinization of potential deceptive cues; which full-length news articles may be more diagnostic of than (brief) headlines. Based on these considerations, we predicted that higher analytical reasoning would be associated with increased fake news accuracy, while there would be no relationship between analytical reasoning ability and real news detection accuracy (Hypothesis 1).

Effects of systematic variation of news source credibility on real and fake news evaluation

The Elaboration Likelihood Model put forth by Petty and Cacioppo ( 1986 ) is a dual-process model of persuasion. According to this model, information is processed via a central, systematic route when the decision maker is both motivated and has the necessary cognitive resources to do so. However, when the decision maker lacks either the necessary motivation or the cognitive resources, they will process information via a peripheral, heuristic route. Importantly, this model posits that heuristic cues such as the credibility of the source (in our case the news source of the article) will have a greater effect when the decision maker is processing the message via the peripheral route (Carpenter, 2015 ; Petty & Cacioppo, 1986 ; Ratneshwar & Chaiken, 1991 ). Thus, it is possible that news source credibility moderates real and fake news evaluation, especially when information is processed peripherally (i.e., involving lower analytical reasoning).

To our knowledge, there are no studies examining the impact of analytical reasoning on accuracy for both real and fake news under systematic variation of news source credibility. Given that individuals rely more on heuristics as cognitive resources decrease (Cacioppo et al., 1986 ; Petty & Cacioppo, 1986 ) and that low analytical reasoning is associated with reduced ability to detect fake news (Bronstein et al., 2019 ; Pennycook & Rand, 2019b ), we hypothesized that lower analytical reasoning would be associated with increased accuracy for real and particularly fake news paired with a credible compared to a non-credible news source ( Hypothesis 2 ).

Beyond accuracy, the role of perceived credibility on real and fake news evaluation

Most fake news studies have focused on accuracy as the primary outcome measure, while neglecting perceived credibility of real and fake news as relevant evaluation metric. Pennycook and Rand ( 2019a ) demonstrated that mainstream online news sources (e.g., cnn.com; npr.org) were perceived as more credible than online sources of partisan (e.g., breitbart.com; dailykos.com) or fake (e.g., thelastlineofdefense.org; now8news.com) news. This finding suggests that the source of a news item may be an important piece of information when evaluating the credibility of an article. Indeed, Luo et al. ( 2020 ) showed that perceived credibility of news headlines was greater when paired with more credible news sources (but see Schaewitz et al., 2020 for no effect of news source on perceived credibility of fake news).

Based on this evidence, we propose that perceived credibility may constitute a relevant, but currently understudied, construct involved in news evaluation. We hypothesized that higher analytical reasoning would be associated with less perceived credibility for fake news, while analytical reasoning ability would not affect perceived credibility of real news ( Hypothesis 3 ). Furthermore, we predicted that lower analytical reasoning would be associated with greater perceived credibility for real and particularly fake news paired with a credible compared to a non-credible news source ( Hypothesis 4 ).

To enhance scientific rigor and reproducibility (Open Science Collaboration, 2015 ), we adopted a two-study approach in this paper. In particular, we conducted two parallel but independent studies to systematically test in Study 1 and replicate with a large sample in Study 2 our research hypotheses.

Participants

Study 1 recruited 360 undergraduates from the Department of Psychology’s SONA system. A total of 68 participants were removed from the final analysis for the following reasons: 3 had average reading times 3 SD s greater than the group average, 41 had incomplete news evaluation data, and 24 failed the attention checks (e.g., Please answer 2 to this question ). The final analysis sample in Study 1 thus comprised 292 participants.

Study 2 used the same recruitment methods as Study 1; assuring through the SONA system that not the same participants were enrolled across the two studies. The initial sample consisted of 424 undergraduate students. A total of 67 participants were removed from the final analysis for the following reasons: 1 had average reading times 3 SD s greater than the group average, 42 had incomplete news evaluation data, and 24 failed the attention checks. The final analysis sample for Study 2 thus comprised 357 participants. Table 1 presents sample characteristics for participants in Study 1 and Study 2.

Both studies adopted a 2 (Veracity: real vs. fake; dichotomous; within-subjects) × 2 (Source: credible vs. non-credible; dichotomous; between-subjects) mixed design. Participants were randomly assigned to evaluate 6 real and 6 fake news articles either from credible ( N  = 138 in Study 1; N  = 171 in Study 2) or non-credible ( N  = 154 in Study 1; N  = 186 in Study 2) news sources (see below for more details).

Study materials were identical in Study 1 and 2.

News articles

To select fake news articles, we used the “junk news” archive maintained by the reputable fact-checking website Snopes.com (Junk News Archives, n.d.). For real news articles, we used the “true news” archive maintained by Snopes ( www.snopes.com/archive/ ) which involves news articles from reputable news organizations (e.g., Washington Post, NPR). From these archives, we selected 6 fake and 6 real news articles that varied by topic, including healthcare (e.g., doctors refusing care on religious grounds), religion (e.g., Mormonism and same-sex marriage, Pope Francis), education (e.g., California textbooks, guns on campuses), crime (e.g., prison escape, felony assault), and politics (e.g., the Black Lives Matter movement, gun confiscations). We conducted an independent pilot study with 98 college students from the Department of Psychology’s SONA system to assess the credibility of the selected 12 news articles (i.e., How credible was this news article? ; rated on a scale from 1 =  Not at all credible to 10 =  Completely credible ). Real news articles were rated as more credible ( M  = 5.90, SD  = 1.09) than fake news articles ( M  = 4.00, SD  = 1.39); t (97) = 13.40, p  < 0.001).

We conducted an additional independent pilot study with 161 college students from the Department of Psychology’s SONA system to determine the final set of news sources for use in our study paradigm. Participants were asked to indicate the level of credibility ( How credible is this news source? ) on a scale from 1 =  Not at all credible to 10 =  Completely credible for 10 commonly known news organizations (i.e., 5 credible sources: NPR, CNN, Washington Post, New York Times, BBC; 5 non-credible sources: True Pundit, Conservative Daily News, World News Daily Report, Liberty Writers News, Red State). The three sources with the highest averages (i.e., NY Times [ M  = 7.00, SD  = 2.30], Washington Post [ M  = 6.84, SD  = 2.23], and NPR [ M  = 6.80, SD  = 2.21]) were selected as “credible sources” and the three sources with the lowest averages (i.e., True Pundit [ M  = 4.30, SD  = 1.70], Red State [ M  = 4.34, SD  = 1.73], and Conservative Daily News [ M  = 4.55, SD  = 1.83]) were selected as “non-credible sources” for use in the study. Additional file 1 : Appendix A provides a full set of the news articles used in this project.

We created two experimental lists to control pairing of Veracity of the news article (real vs. fake; within-subjects) and Credibility of the news source (credible vs. non-credible; between-subjects). The two lists comprised the same 12 unique articles and were counterbalanced across participants. In List 1, the 6 real and the 6 fakes news articles were randomly paired with credible news sources (i.e., NY Times, Washington Post, NPR; credible condition). In List 2, the 6 real and the 6 fakes news articles were randomly paired with non-credible news sources (i.e., True Pundit, Red State, Conservative Daily News; non-credible condition). Presentation order within each list was pseudorandomized, with the constraint that the same type of news articles (real vs. fake) was not repeated more than two times in a row. For each list, (approximately) half of the participants received the reversed order to counter order effects.

• Cognitive reflection test

The CRT (Frederick, 2005 ) is a three-item task designed to measure the degree to which analytical reasoning is used when solving problems. For example, one item asks: “ A bat and a ball cost $1.10 in total. The bat costs $1.00 more than the ball. How much does the ball cost? ” Participants high in analytical reasoning would overcome the impulse to give the intuitive answer 10 cents and would instead give the correct answer of 5 cents. Thus, a greater CRT score reflects higher analytical reasoning.

Study procedures for Study 1 and Study 2 were identical unless noted otherwise. Participants accessed the study link through the SONA system website and completed the study remotely through Qualtrics ( https://www.qualtrics.com/ ). Prior to study enrollment, all participants consented electronically to participate.

During the News Evaluation Task , participants were presented with 12 news articles (6 real, 6 fake). Each article was presented on the screen for at least 60 s to ensure sufficient reading time, as determined in an internal pilot. Beyond the 60-s window, the task was self-paced. Footnote 1 After reading each article, participants were prompted with the following questions (in this order): accuracy ( Is this news article real or fake? ; response option: Real vs. Fake ), confidence ( How confident are you in your decision regarding the authenticity of this news article? ; response option: 1 ( Not at all confident ) to 10 ( Completely confident )), perceived credibility ( How credible do you find this news article? ; response option: 1 ( Not at all credible ) to 10 ( Completely credible )), media sharing ( Would you share this news article on social media? ; response option: Yes vs. No ), and familiarity ( Have you seen this article before? ; response option: Yes vs. No ). Participants were not informed about the number of articles presented to them to avoid response biases (e.g., 50/50 real vs. fake response rate).

After evaluating the news articles, participants completed the CRT and a short demographic questionnaire. Footnote 2 Study duration was about 1 h in each of the two studies.

Data analysis

We used multilevel random intercept models (Gelman & Hill, 2007 ; Hox, 2010 ) to accommodate for the nested data structure. Specifically, we conducted cross-random effects analyses with cross-classification of news articles and participants, and a nesting structure for repeated observations within participants. This approach allows evaluations made by the same participant to be correlated across different news articles, as well as accounts for dependencies of evaluations of the same news article made by different participants.

Our analyses included two separate models, one for accuracy Footnote 3 and one for perceived credibility. Complete datasets and analysis code can be found at https://osf.io/yrabp/ . For the binary outcome variable accuracy (0 = wrong, 1 = correct), we used mixed effects logistic regression; for the ordinal/continuous outcome variables perceived credibility we employed multilevel regression. Each model considered the fixed effect of veracity of the news article (0 = real, 1 = fake), credibility of the source (0 = credible, 1 = non-credible), and the CRT score of each participant (continuous variable) as predictors. We further estimated the interactions between these independent variables in each model. We also entered the random intercepts of evaluations for news articles and participants to estimate the variability of mean evaluations across news articles and participants, respectively. Reading time (beyond the fixed 60-s window), familiarity, gender, and presentation order were entered as covariates.

We applied maximum likelihood estimation for all model parameters and used the Wald tests to determine significance of the effects. For significant interactions, we compared (using z tests for pairwise comparisons) and plotted predicted marginal means (using a mean of 0 and ± 1 SD for interactions involving the continuous CRT variable) from the estimated model parameters to facilitate understanding of significant interactions. All analyses were performed in Stata 16.1 (StataCorp, 2019 ).

Consistent across both studies, the Veracity × CRT interaction on accuracy was significant [Study 1: χ 2 (1)  = 23.84, p  < 0.001; Study 2: χ 2 (1)  = 10.78, p  = 0.001]. As shown in Fig.  1 , real news accuracy did not change across levels of analytical reasoning (indexed by CRT scores) [Study 1/Panel A: z  = 1.37, p  = 0.339; Study 2/Panel B: z  = 0.5, p  = 0.619]. Accuracy for fake news, however, increased with higher analytical reasoning [Study 1/Panel A: z  = 4.53, p  < 0.001; Study 2/Panel B: z  = 4.13, p  < 0.001], thus supporting Hypothesis 1 . Furthermore, also consistent across both studies and depicted in Fig.  1 , higher analytical reasoning was associated with better detection of fake than real news [Study 1/Panel 1A: z  = 3.79, p  < 0.001; Study 2/Panel 1B: z  = 3.43, p  = 0.001].

Percent accuracy for real (gray line) and fake (black line) news articles across levels of analytical reasoning (continuous; indexed by Cognitive Reflection Test (CRT) scores) in Study 1 (Panel A) and Study 2 (Panel B). Error bars denote standard errors. The medium analytical reasoning level indicates the mean CRT score in the current sample while the low and high levels indicate 1 SD below and above the mean CRT score, respectively. The y-axis start point reflects the 50% chance level. Consistent across both studies, real news accuracy did not change across levels of analytical reasoning, while accuracy for fake news increased with higher analytical reasoning

The three-way interaction between Veracity × CRT × Source was not significant in Study 1 ( χ 2 (1)  = 0.42, p  = 0.517), but was significant in Study 2 ( χ 2 (1)  = 6.2, p  = 0.013). In particular, as shown in Fig.  2 (for Study 2), lower analytical reasoning was associated with greater accuracy for real news from credible compared to non-credible sources ( z  = 3.42, p  = 0.001). Furthermore as depicted in Fig.  2 , news source credibility did not influence accuracy for fake news across levels of analytical reasoning ( z s < 1, p s > 0.272); and higher analytical reasoning was associated with greater accuracy for fake news irrespective of news source credibility (all z s > 2.55, p s < 0.02). These findings partially supported Hypothesis 2 . Footnote 4

Veracity × CRT × Source interaction in Study 2; this 3-way interaction was not significant in Study 1. Percent accuracy for real (gray lines) and fake (black lines) news articles from credible (solid lines) and non-credible (dashed lines) news sources across levels of analytical reasoning (continuous; indexed by Cognitive Reflection Test (CRT) scores) in Study 2. Error bars denote standard errors. The medium analytical reasoning level indicates the mean CRT score in the current sample while the low and high levels indicate 1 SD below and above the mean CRT score, respectively. The y-axis start point reflects the 50% chance level. Lower analytical reasoning was associated with greater accuracy for real news paired with credible compared to non-credible sources, while news source did not influence accuracy for fake news across levels of analytical reasoning

• Perceived credibility

Consistent across both studies, the Veracity × CRT interaction was significant [Study 1: χ 2 (1)  = 14.28, p  < 0.001; Study 2: χ 2 (1)  = 24.57, p  < 0.001]. As depicted in Fig.  3 , perceived credibility for real news was overall higher than perceived credibility for fake news and was not influenced by levels of analytical reasoning [Study 1/Panel A: z  = 0.97, p  = 0.66; Study 2/Panel B: z  = 0.52, p  = 0.6]. In contrast, higher analytical reasoning was associated with less perceived credibility for fake news [Study 1/Panel A: z  = 4.22, p  < 0.001; Study 2/Panel B: z  = 3.55, p  < 0.001], in line with Hypothesis 3 .

Mean perceived credibility rating (1 =  Not at all credible to 10 =  Completely credible ) for real (gray line) and fake (black line) news articles across levels of analytical reasoning (continuous; indexed by Cognitive Reflection Test (CRT) scores) in Study 1 (Panel A) and Study 2 (Panel B). Error bars denote standard errors. The medium analytical reasoning level indicates the mean CRT score in the current sample while the low and high levels indicate 1 SD below and above the mean CRT score, respectively. Note that the y-axis ranges from 1 to 7 to reflect the actual range of responses given by participants. Consistent across both studies, perceived credibility for real news was not influenced by levels of analytical reasoning, while higher analytical reasoning was associated with less perceived credibility for fake news

The three-way interaction between Veracity × CRT × Source was not significant in either of the studies [Study 1: χ 2 (1)  = 1.49, p  = 0.222; Study 2: χ 2 (1)  = 0.67, p  = 0.413]. Thus, our data did not support Hypothesis 4 .

The present two-study project, with a built-in replication, is the first to examine evaluation of both real and fake news under consideration of cognitive factors (i.e., analytical reasoning), characteristics of the news stimuli (i.e., source credibility) as well as the interplay between the two using a novel, relatively more ecologically valid full-length article paradigm. In addition, our approach went beyond investigation of real and fake news evaluation accuracy in also determining effects on the perceived credibility of the articles. Consistent across both studies, higher analytical reasoning was associated with greater accuracy and reduced perceived credibility for fake news, while analytical reasoning ability did not moderate accuracy and perceived credibility of real news. Furthermore, in Study 2 (but not in Study 1) news source credibility influenced the relationship between analytical reasoning ability and news detection accuracy for real (but not fake) news. These novel findings have potential to advance theory and empirical understanding of cognitive processes underlying news evaluations, as discussed next.

Higher analytical reasoning improves fake news detection in full-length articles

Consistently across both studies and in line with our predictions, higher analytical reasoning was associated with more accurate detection of fake news articles. Thus, extending previous evidence from headlines-only studies (Bronstein et al., 2019 ; Pennycook & Rand, 2019a , 2020 ; Pennycook et al., 2015 ), by using full-length news articles the present study provides support for a role of analytical reasoning on fake news detection. In line with our prediction, real news accuracy, in contrast, was not influenced by analytical reasoning ability. As real news is more common in everyday life than fake news, detection of real news may not be as resource-demanding than detection of fake news, possibly underlying the moderating effect of analytical reasoning on fake but not real news detection. The Naturalistic Decision Making framework (Klein, 2008 , 2015 ) highlights the role of relatively automatic (intuitive) and experience-based successful decision making in naturalistic real-world settings. This framework may be particularly fruitful in future research on determining the mechanisms underlying news evaluation. As touched on earlier, we believe that our full-length article approach is more representative of how news articles are typically encountered in real life (e.g., with rich contextual information), thus allowing to better capture complex cognitive processes involved in naturalistic news evaluation. To further improve ecological validity, future research could leverage real or simulated social media platforms (e.g., Twitter, Facebook), where people directly interact with the news (see Lin et al., 2019 , for a similar approach in email phishing detection). This approach would also be in line with research demonstrating the importance of using ecological valid task formats to improve performance (Evans, 2011 ; Mercier & Sperber, 2011 ). The present study constitutes a first important step in this direction.

Further, consistent across both studies, higher analytical reasoning was associated with better detection of fake than real news. One could argue that better detection of fake compared to real news with higher analytical reasoning may simply reflect a response bias (i.e., tendency to overclaim news as fake, which could be an artifact of task instructions). However, results from an additional analysis we conducted that controlled for sensitivity and response bias did not support this interpretation. Instead, this finding may reflect an enhanced ability to detect deceptive cues inherent in fake news stories among individuals who engage in higher levels of analytical reasoning. That is, diagnostic cues and details in the full-length fake news articles used in this study such as pertaining to formatting, grammar issues, general writing style (and that may not be present in real news articles) may have facilitated fake news detection among individuals who engage in deeper processing (i.e., higher analytical reasoning). These explanations are rather speculative and warrant research that uses natural language processing machine learning approaches (Gilda, 2017 ; Oshikawa et al., 2018 ), for example, to determine deception-related diagnostic cues in fake (relative to real) news and to further clarify the interplay between these cues and analytical reasoning ability in news detection.

Lower analytical reasoning enhances detection of real news paired with credible sources

We found that lower analytical reasoning was associated with better detection of real news paired with credible sources, while news source credibility did not influence accuracy for fake news across levels of analytical reasoning. To date only a small number of studies have examined the impact of source credibility on news detection accuracy. Luo et al. ( 2020 ) showed that reliability of the source (indexed by a high number of Facebook likes) increased the detection of real news but decreased the detection of fake news. In contrast, Schaewitz et al. ( 2020 ) found no effect of source credibility (i.e., fictitious news sources that were rated on credibility) on fake news accuracy. Furthermore, Pennycook and Rand ( 2020 ) reported a negative association between analytical reasoning and susceptibility to fake news, regardless of whether a news source was present or absent, suggesting no moderating effect of source credibility on the relationship between analytical reasoning and fake news detection (also see Dias et al., 2020 for similar results).

Our study contributes to this literature and is the first to suggest that news source credibility may influence news detection as a function of analytical reasoning in full-length real (but not fake) news articles. However, this finding only emerged in Study 2 but not in Study 1 and thus needs to be interpreted with caution. It is possible that lower analytical reasoning reflects greater reliance on source heuristics. In fact, our results are consistent with predictions from the Elaboration Likelihood Model (Petty & Cacioppo, 1986 ), which proposes that peripheral cues such as the credibility of the source of a message, more likely influence individuals low in cognitive resources as they engage in less elaborative or systematic processing; a possible explanation that can be systematically explored in future work. Also, as the three-way Veracity × CRT × Source interaction was only significant in Study 2, which comprised a larger sample size, but not in Study 1, a future replication of this effect in a sample of at least the size as in Study 2 is warranted to corroborate the finding. Additionally, because of study duration related constraints and our preference for keeping our news article material uniform across participants (i.e., each participant viewed the same real and fake news articles), the credibility of the source the news articles were paired with was manipulated between participants in this project. This design feature may have reduced statistical power to detect significant effects related to source credibility (i.e., one would expect greater sensitivity of a factor that is manipulated within-subjects (in this case, veracity) than one that is manipulated between-subjects (in this case, news source credibility)). Future studies could employ a within-subjects design to investigate this possibility.

Beyond accuracy, perceived credibility as an additional route to study cognitive mechanisms underlying news evaluation

Overall, perceived credibility for real news was higher than perceived credibility for fake news in both studies. Furthermore, and again consistent across both studies, higher analytical reasoning was associated with lower perceived credibility for fake news, while perceived credibility for real news did not vary by level of analytical reasoning.

Somewhat in contrast to our findings pertaining to accuracy, news source did not moderate the effect of analytical reasoning on perceived credibility of real vs. fake news. Specifically, participants who relied more on analytical reasoning were better at detecting fake news and rated fake news as less credible. Importantly, the credibility of the news source did not affect accuracy or perceived credibility of fake news in individuals high on analytical reasoning. This finding may suggest that individuals high on analytical reasoning utilize diagnostic cues and contextual features provided within the fake news article itself (e.g., sentiment, formatting style, grammar issues, general writing style).

If this interpretation is true, then this highlights two important implications for future research. First, future research may benefit from using full-length news articles because headlines only contain a finite amount of diagnostic cues and may strip away important information to discern between real and fake news. Given that our current results (using full-length articles) align with past research that used only headlines, future research needs to directly compare full-length articles with headlines only and by systematically manipulating news source among individuals with varying levels of analytical reasoning to better assess these claims. Second, the aforementioned pattern emerged clearer by collecting novel outcome measures (i.e., perceived credibility of the news), thus, supporting the need for future research to explore other (sensitive) outcome measures (e.g., news content related questions) that may help gain a more complete understanding of the phenomenological process individuals engage in when detecting fake news.

Additionally, the possibility that participants may have directed their attention primarily towards the news stories and its central content (e.g., sentiment, language style) rather than peripheral cues (e.g., the news source) can be further investigated using eye-tracking. This technique will allow determination of eye gaze patterns as well as physiological reactions associated with arousal levels (e.g., pupil dilation) when interacting with news stories. These innovative methodological approaches would not only help identifying candidate cognitive mechanisms but could also inform targeted interventions (e.g., eye-tracking guided reading intervention to train people to process information relevant to detection of deceptive cues). This rich data will also lend itself particularly well to computational modeling approaches to describe decision-making processes underlying deception detection (see Hakim et al., 2020 for a computational modeling approach to phishing email detection).

Future research directions

Our study, like the majority of previous work, focused on a rather homogeneous (e.g., in terms of race/ethnicity and age) sample. Based on growing evidence that sensitivity for detection of deceptive cues decreases with chronological age (Ebner et al., 2020 ; Grilli et al., in press; Zebrowitz et al., 2018 ) as well as varies by gender and marital status (Alves & Wilson, 2008 ), education (Wood et al., 2018 ), and income (James et al., 2014 ), we propose examining fake news detection using more diverse samples to move this research forward (Pehlivanoglu et al., 2020 ). For example, older compared to younger individuals were more likely to share fake news (Grinberg et al., 2019 ; Guess et al., 2019 ). A recent narrative review by Brashier and Schacter ( 2020 ) argues that susceptibility to fake news with age may not only depend on cognitive decline, but may also be related to age-related changes in socioemotional functioning (e.g., increase in positive emotion and interpersonal trust) as well as in expertise with online news media platforms. Thus, examining the role of expertise with online news media outlets (e.g., indexed by digital literacy; Sengpiel & Dittberner, 2008 , and news media literacy; Maksl et al., 2015 ) on the relationship between analytical reasoning and real vs. fake news evaluation in a sample of adults varying in age (college students vs. middle-aged adults vs. older adults) is a fruitful future research direction. These future age-comparative studies would also be helpful to identify mechanisms that may render certain groups particularly vulnerable to fake news and would open tremendous potential for interventional approaches, including particular at-risk populations (Ebner, et al., in press).

Future studies should also set out to determine the specific dynamics of the impact of analytical reasoning on real and fake news evaluation. For example, it is possible that news related variables such as news topics/content (e.g., politics vs. pop culture) differentially call on analytical reasoning ability when evaluating real and fake news articles. In addition, it is possible that individuals can flexibly allocate their resources and switch between processing modes (e.g., effortful vs. non-effortful thinking; shallow vs. deep processing) for improved news evaluation. Utilizing neuroimaging techniques (e.g., fMRI) could help outline the neurocognitive mechanisms underlying news evaluation. Event-related potentials could help determine temporal dynamics of engagement in different levels of reasoning during news evaluation (e.g., whether engagement in analytic reasoning changes during early vs. late stages of processing; whether one reasoning mode is replaced by the other over time; whether news-related variables such as source credibility moderates these processes).

This study is the first to demonstrate a positive association between analytical reasoning and fake news detection accuracy using full-length news articles, as a relatively more ecologically valid approach in research on news evaluation. The study is also first in supporting a moderating role of news source credibility in the endeavor to delineate cognitive mechanisms underlying news evaluation; and it advances knowledge pertaining to perceived credibility of news as an alternative outcome variable to accuracy. Across two independent studies, findings from this research underline the importance of both individual differences and news-related characteristics when evaluating news. Our research has potential for theoretical advancement regarding relative contributions of rational vs. more naturalistic decision making in the applied context of fake news detection. Employing full-length news articles, novel findings reported here spur future research hypotheses regarding the (neuro)cognitive mechanisms involved in detection of deceptive cues in news evaluation as well as possible intervention designs to tackle the major and daily growing threat of misinformation from fake news, at both individual and societal levels.

Availability of data and materials

The stimulus set, complete datasets used in the analyses, and analysis code are available in the OSF repository, https://osf.io/yrabp/ . None of the experiments were preregistered.

Reading time data (in seconds; averaged across real and fake news articles) showed that participants took more than 60 s on average [Mean = 106.15 (Study 1), 101.53 (Study 2); Median = 89.25 (Study 1), 87.26 (Study 2); SD  = 48.69 (Study 2), 48.13 (Study 2); Range = 62.15–363.15 (Study 1), 61.93–404.48 (Study 2)], suggesting that the news articles were processed adequately.

Both Study 1 and Study 2 also included the Gullibility Scale (Teunisse et al., 2020 ) and the short form of the Need for Cognition Scale (Cacioppo et al., 1984 ). Study 2 additionally included measures for religiosity (Batson & Schoenrade, 1991 ), spirituality (Büssing et al., 2007 ), conservatism (Everett, 2013 ), and media consumption habits (adopted from Maksl et al., 2015 ). These additional constructs were outside the scope of this report and were therefore not included in the statistical analysis for parsimony.

To support findings for accuracy, we conducted parallel analyses on confidence ratings and report the results in Additional file 2 : Appendix B. News sharing was not analysed as an outcome measure due to floor effects in “yes” responses (see Table B1 in Additional file 2 : Appendix B).

To ensure that our results regarding accuracy were not confounded by response bias, based on signal detection theory (Macmillan & Creelman, 2004 ), we computed sensitivity (d’ = z(Hit rate)—z(False alarm rate)) and response bias (c = − 0.5[z(Hit rate) + z(False alarm rate)]) for each participant, in both Study 1 and Study 2. Then, we added the scores for sensitivity and response bias as covariates and re-ran the analyses pertaining to accuracy. This re-analysis resulted in the same findings as our original analysis. In particular, the Veracity × CRT interaction was significant in both studies [Study 1: χ 2 (1)  = 21.92, p  < 0.001; Study 2: χ 2 (1) ) = 10.01 p  = 0.002]. The three-way interaction between Veracity × CRT × Source was not significant in Study 1 ( χ 2 (1)  = 0.03, p  = 0.857), but was significant in Study 2 ( χ 2 (1)  = 5.8, p  = 0.016).

Abbreviations

Allcott, H., & Gentzkow, M. (2017). Social media and fake news in the 2016 election. Journal of Economic Perspectives, 31 (2), 211–236.

Article   Google Scholar  

Allen, J., Howland, B., Mobius, M., Rothschild, D., & Watts, D. J. (2020). Evaluating the fake news problem at the scale of the information ecosystem. Science Advances , 6 (14).

Alves, L., & Wilson, S. (2008). The effects of loneliness on telemarketing fraud vulnerability among older adults. Journal of Elder Abuse & Neglect, 20 (1), 63–85.

Bago, B., Rand, D. G., & Pennycook, G. (2020). Fake news, fast and slow: deliberation reduces belief in false (but not true) news headlines. Journal of Experimental Psychology: General , 1–18.

Batson, C. D., & Schoenrade, P. A. (1991). Measuring religion as quest: 2) reliability concerns. Journal for the Scientific Study of Religion, 30 (4), 430–447.

Brashier, N. M., & Schacter, D. L. (2020). Aging in an era of fake news. Current Directions in Psychological Science, 29 (3), 316–323.

Article   PubMed   Google Scholar  

Bronstein, M. V., Pennycook, G., Bear, A., Rand, D. G., & Cannon, T. D. (2019). Belief in fake news is associated with delusionality, dogmatism, religious fundamentalism, and reduced analytic thinking. Journal of Applied Research in Memory and Cognition, 8 (1), 108–117.

Büssing, A., Ostermann, T., & Matthiessen, P. F. (2007). Distinct Expressions of Vital Spirituality "The ASP Questionnaire as an Explorative Research Tool". Journal of Religion and Health , 267–286.

Cacioppo, J. T., Petty, R. E., Chuan, F. K., & Rodriguez, R. (1986). Central and peripheral routes to persuasion. An individual difference perspective. Journal of Personality and Social Psychology, 51 (5), 1032–1043.

Cacioppo, J. T., Petty, R. E., & Kao, C. F. (1984). The efficient assessment of need for cognition. Journal of Personality Assessment, 48 (3), 306–307.

Carpenter, C. J. (2015). A meta-analysis of the Elm’s argument quality × processing type predictions. Human Communication Research, 41 (4), 501–534.

De Neys, W. (2012). Bias and conflict: A case for logical intuitions. Perspectives on Psychological Science, 7 (1), 28–38.

Dias, N., Pennycook, G., & Rand, D. G. (2020). Emphasizing publishers does not effectively reduce susceptibility to misinformation on social media. Harvard Kennedy School Misinformation Review , 1 (1).

Ebner, N. C., Ellis, D. M., Lin, T., Rocha, H. A., Yang, H., Dommaraju, S., Soliman, A., Woodard, D. L., Turner, G. R., Spreng, R. N., & Oliveira, D. S. (2020). Uncovering susceptibility risk to online deception in aging. The Journals of Gerontology: Series B Psychological Sciences and Social Sciences, 75 (3), 522–533.

Ebner, N. C., Pehlivanoglu, D., Polk, R., Turner, G. R., & Spreng, R. N. (in press). Aging online: Rethinking the aging decision maker in a digital era . In Y. Hanoch & S. Wood (Eds). A fresh look at fraud: theoretical and applied approaches (behavioral economics and healthy behaviors). (Routledge Taylor Francis).

Evans, J. S. B. T. (2007). Dual-processing accounts of reasoning, judgment, and social cognition. Annual Review of Psychology, 59 (1), 255–278.

Evans, J. S. B. T. (2011). Reasoning is for thinking, not just for arguing. Behavioral and Brain Sciences, 34 (2), 77–78.

Everett, J. A. C. (2013). The 12 item social and economic conservatism scale (SECS). PLoS ONE, 8 (12), e82131.

Article   PubMed   PubMed Central   Google Scholar  

Ferreira, M. B., Garcia-Marques, L., Sherman, S. J., & Sherman, J. W. (2006). Automatic and controlled components of judgment and decision making. Journal of Personality and Social Psychology .

Frederick, S. (2005). Cognitive reflection and decision making. Journal of Economic Perspectives, 19 (4), 25–42.

Gelman, A., & Hill, J. (2007). Data Analysis Using Regression and Multilevel/Hierarchical Models . . Cambridge University Press.

Google Scholar  

Gigerenzer, G. (2007). Gut feelings: The intelligence of the unconscious . . Penguin.

Gilda, S. (2017). Notice of violation of IEEE publication principles: Evaluating machine learning algorithms for fake news detection. In 2017 IEEE 15th student conference on research and development (SCOReD) , (pp. 110–115).

Grilli, M. D., McVeigh, K. S., Hakim, Z. M., Wank, A. A., Getz, S. J., Levin, B. E., Ebner, N. C., & Wilson, R. C. (in press). Is this phishing? Older age is associated with greater difficulty discriminating between safe and malicious emails. Journal of Gerontology: Psychological Sciences.

Grinberg, N., Joseph, K., Friedland, L., Swire-Thompson, B., & Lazer, D. (2019). Fake news on Twitter during the 2016 US presidential election. Science, 363 (6425), 374–378.

Guess, A., Nagler, J., & Tucker, J. (2019). Less than you think: Prevalence and predictors of fake news dissemination on Facebook. Asian-Australasian Journal of Animal Sciences, 32 (2), 1–9.

Hakim, Z. M., Ebner, N. C., Oliveira, D. S., Getz, S. J., Levin, B. E., Lin, T., Grilli, M. D., & Wilson, R. C. (2020). The Phishing Email Suspicion Test (PEST) a lab-based task for evaluating the cognitive mechanisms of phishing detection. Behavior Research Methods, 1–11.

Hox, J. J. (2010). Multilevel analysis: Techniques and applications . . Routledge.

Book   Google Scholar  

James, B. D., Boyle, P. A., & Bennett, D. A. (2014). Correlates of susceptibility to scams in older adults without dementia. Journal of Elder Abuse & Neglect, 26 (2), 107–122.

Kahneman, D. (2011). Thinking, fast and slow . . Macmillan. https://doi.org/10.2307/1914185 .

Kahneman, D., Slovic, S. P., Slovic, P., & Tversky, A. (1982). Judgment under uncertainty: Heuristics and biases . . Cambridge University Press.

Klein, G. (2008). Naturalistic decision making. Human Factors: The Journal of the Human Factors and Ergonomics Society, 50 (3), 456–460.

Klein, G. (2015). A naturalistic decision making perspective on studying intuitive decision making. Journal of Applied Research in Memory and Cognition, 4 (3), 164–168.

Lazer, D. M., Baum, M. A., Benkler, Y., Berinsky, A. J., Greenhill, K. M., Menczer, F., Metzger, M. J., Nyhan, B., Pennycook, G., Rothschild, D., & Schudson, M. (2018). The science of fake news. Science, 359 (6380), 1094–1096.

Lin, T., Capecci, D. E., Ellis, D. M., Rocha, H. A., Dommaraju, S., Oliveira, D. S., & Ebner, N. C. (2019). Susceptibility to spear-phishing emails: Effects of internet user demographics and email content. ACM Transactions on Computer-Human Interaction, 26 (5), 1–28.

Luo, M., Hancock, J. T., & Markowitz, D. M. (2020). Credibility perceptions and detection accuracy of fake news headlines on social media: Effects of truth-bias and endorsement cues. Communication Research , 0093650220921321.

Macmillan, N. A., & Creelman, C. D. (2004). Detection theory: A user’s guide . . Psychology Press.

Maksl, A., Ashley, S., & Craft, S. (2015). Measuring news media literacy. Journal of Media Literacy Education, 6 (3), 29–45.

Mercier, H., & Sperber, D. (2011). Why do humans reason? Arguments for an argumentative theory. Behavioral and Brain Sciences, 34 (2), 57–74.

Murray, L. (2020). National enquirer. Encyclopedia Britannica, 17 Dec. 2013, https://www.britannica.com/topic/National-Enquirer . Retrived July 20, 2020.

Open Science Collaboration. (2015). Estimating the reproducibility of psychological science. Science, 349 (6251), aac4716.

Oshikawa, R., Qian, J., & Wang, W. Y. (2018). A survey on natural language processing for fake news detection. https://arxiv.org/abs/1811.00770 .

Pehlivanoglu, D., Lin, T., Chi, K., Perez, E., Polk, R., Cahill, B., Lighthall, N., Ebner, N. C. (2020). News veracity detection among older adults during the COVID-19 pandemic: The role of analytical reasoning, mood, news consumption, and news content. https://doi.org/10.31234/osf.io/3kgq9

Pennycook, G., Fugelsang, J. A., & Koehler, D. J. (2015). Everyday consequences of analytic thinking. Current Directions in Psychological Science, 24 (6), 425–432.

Pennycook, G., & Rand, D. G. (2019a). Fighting misinformation on social media using crowdsourced judgments of news source quality. Proceedings of the National Academy of Sciences of the United States of America, 116 (7), 2521–2526.

Pennycook, G., & Rand, D. G. (2019b). Lazy, not biased: Susceptibility to partisan fake news is better explained by lack of reasoning than by motivated reasoning. Cognition, 188 , 39–50.

Pennycook, G., & Rand, D. G. (2020). Who falls for fake news? The roles of bullshit receptivity, overclaiming, familiarity, and analytic thinking. Journal of Personality, 88 (2), 185–200.

Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. In Communication and persuasion: Central and peripheral routes to attitude change (pp. 1–24).

Associated Press. (2019). AP-NORC/USAFacts poll: Americans struggle to ID true facts. AP News. Retrieved from https://apnews.com/article/c762f01370ee4bbe8bbd20f5ddf2adbe

Ratneshwar, S., & Chaiken, S. (1991). Comprehension’s role in persuasion: The case of its moderating effect on the persuasive impact of source cues. Journal of Consumer Research, 18 (1), 52–62.

Schaewitz, L., Kluck, J. P., Klösters, L., & Krämer, N. C. (2020). When is disinformation (in)credible? Experimental findings on message characteristics and individual differences. Mass Communication and Society, 23 (4), 484–509.

Sengpiel, M., & Dittberner, D. (2008). The computer literacy scale (CLS) for older adults-development and validation. Mensch und Computer 2008-Tagungsband.

Silverman, C., Strapagiel, L., Shaban, H., Hall, E., & Singer-Vine, J. (2016). Hyperpartisan Facebook pages are publishing false and misleading information at an alarming rate. Buzzfeed News, 20 , 68.

Stanovich, K. E. (2009). Distinguishing the reflective, algorithmic, and autonomous minds: Is it time for a tri-process theory? In In two minds: Dual processes and beyond. (pp. 55–88).

StataCorp. . (2019). Stata statistical software: Release 16 . . StataCorp LP.

Teunisse, A. K., Case, T. I., Fitness, J., & Sweller, N. (2020). I Should have known better: Development of a self-report measure of gullibility. Personality and Social Psychology Bulletin, 46 (3), 408–423.

Vosoughi, S., Roy, D., & Aral, S. (2018). News on-line. Science, 1151 , 1146–1151.

Wood, S., Liu, P.-J.J., Hanoch, Y., Xi, P. M., & Klapatch, L. (2018). Call to claim your prize: Perceived benefits and risk drive intention to comply in a mass marketing scam. Journal of Experimental Psychology: Applied, 24 (2), 196–206.

PubMed   Google Scholar  

Zebrowitz, L. A., Ward, N., Boshyan, J., Gutchess, A., & Hadjikhani, N. (2018). Older adults’ neural activation in the reward circuit is sensitive to face trustworthiness. Cognitive, Affective and Behavioral Neuroscience, 18 (1), 21–34.

Download references

Acknowledgements

We appreciate the willingness of our participants to contribute to this research.

This research was funded by the Department of Psychology, College of Liberal Arts and Science, University of Florida and NIH/NIA Grant 1R01AG057764.

Author information

Senior authorship: Natalie C. Ebner and Brian S. Cahill

Authors and Affiliations

Department of Psychology, University of Florida, 945 Center Dr, Gainesville, FL, 32603, USA

Didem Pehlivanoglu, Tian Lin, Farha Deceus, Amber Heemskerk, Natalie C. Ebner & Brian S. Cahill

Department of Aging and Geriatric Research, Institute on Aging, University of Florida, Gainesville, USA

Natalie C. Ebner

Florida Institute for Cybersecurity, University of Florida, Gainesville, USA

Evelyn F. McKnight Brain Institute, University of Florida, Gainesville, USA

You can also search for this author in PubMed   Google Scholar

Contributions

BC, NCE, and FD designed the study, developed the stimuli, and collected the data. DP, TL, NCE, and BC developed the formal analytical strategy. DP processed the data, conducted analyses, and reported the findings. DP, BC, FD, TL, and NCE wrote the Methods. DP, BC, and NCE wrote the introduction and the discussion. All authors contributed to manuscript conceptualization and editing and approved the final manuscript.

Corresponding author

Correspondence to Didem Pehlivanoglu .

Ethics declarations

Ethics approval and consent to participate.

The Project was approved by University of Florida Institutional Review Board (IRB) (approval #201902194, #20200960). Electronic informed consent was obtained at enrollment from all participants using an IRB-approved consent form.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file1. appendix a:.

A full set of the news articles used in the current project.

Additional file 2. Appendix B:

Confidence ratings and sharing responses for the news articles.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Pehlivanoglu, D., Lin, T., Deceus, F. et al. The role of analytical reasoning and source credibility on the evaluation of real and fake full-length news articles. Cogn. Research 6 , 24 (2021). https://doi.org/10.1186/s41235-021-00292-3

Download citation

Received : 01 August 2020

Accepted : 17 March 2021

Published : 31 March 2021

DOI : https://doi.org/10.1186/s41235-021-00292-3

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Analytical reasoning
• Source credibility
• Dual-process theory
• Naturalistic decision-making

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here .

Loading metrics

Open Access

Peer-reviewed

Research Article

A systematic review on fake news research through the lens of news creation and consumption: Research efforts, challenges, and future directions

Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing

Affiliation School of Intelligence Computing, Hanyang University, Seoul, Republic of Korea

Roles Conceptualization, Formal analysis, Investigation, Methodology, Supervision, Writing – original draft, Writing – review & editing

Affiliation College of Information Sciences and Technology, Pennsylvania State University, State College, PA, United States of America

Roles Funding acquisition, Methodology, Project administration, Supervision, Writing – original draft, Writing – review & editing

Roles Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing

* E-mail: [email protected]

• Bogoan Kim, 
• Aiping Xiong, 
• Dongwon Lee, 
• Kyungsik Han

• Published: December 9, 2021
• https://doi.org/10.1371/journal.pone.0260080
• Reader Comments

28 Dec 2023: The PLOS One Staff (2023) Correction: A systematic review on fake news research through the lens of news creation and consumption: Research efforts, challenges, and future directions. PLOS ONE 18(12): e0296554. https://doi.org/10.1371/journal.pone.0296554 View correction

Although fake news creation and consumption are mutually related and can be changed to one another, our review indicates that a significant amount of research has primarily focused on news creation. To mitigate this research gap, we present a comprehensive survey of fake news research, conducted in the fields of computer and social sciences, through the lens of news creation and consumption with internal and external factors.

We collect 2,277 fake news-related literature searching six primary publishers (ACM, IEEE, arXiv, APA, ELSEVIER, and Wiley) from July to September 2020. These articles are screened according to specific inclusion criteria (see Fig 1). Eligible literature are categorized, and temporal trends of fake news research are examined.

As a way to acquire more comprehensive understandings of fake news and identify effective countermeasures, our review suggests (1) developing a computational model that considers the characteristics of news consumption environments leveraging insights from social science, (2) understanding the diversity of news consumers through mental models, and (3) increasing consumers’ awareness of the characteristics and impacts of fake news through the support of transparent information access and education.

We discuss the importance and direction of supporting one’s “digital media literacy” in various news generation and consumption environments through the convergence of computational and social science research.

Citation: Kim B, Xiong A, Lee D, Han K (2021) A systematic review on fake news research through the lens of news creation and consumption: Research efforts, challenges, and future directions. PLoS ONE 16(12): e0260080. https://doi.org/10.1371/journal.pone.0260080

Editor: Luigi Lavorgna, Universita degli Studi della Campania Luigi Vanvitelli, ITALY

Received: March 24, 2021; Accepted: November 2, 2021; Published: December 9, 2021

Copyright: © 2021 Kim et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data are within the manuscript.

Funding: This research was supported by the Institute of Information & communications Technology Planning & Evaluation (IITP) grant funded by the Korea government (2019-0-01584, 2020-0-01373).

Competing interests: The authors have declared that no competing interests exist.

1 Introduction

The spread of fake news not only deceives the public, but also affects society, politics, the economy and culture. For instance, Buzzfeed ( https://www.buzzfeed.com/ ) compared and analyzed participation in 20 real news and 20 fake news articles (e.g., likes, comments, share activities) that spread the most on Facebook during the last three months of the 2016 US Presidential Election. According to the results, the participation rate of fake news (8.7 million) was higher than that of mainstream news (7.3 million), and 17 of the 20 fake news played an advantageous role in winning the election [ 1 ]. Pakistan’s ministry of Defense posted a tweet fiercely condemning Israel after coming to believe that Israel had threatened Pakistan with nuclear weapons, which was later found to be false [ 2 ]. Recently, the spread of the absurd rumor that COVID-19 propagates through 5G base stations in the UK caused many people to become upset and resulted in a base station being set on fire [ 3 ].

Such fake news phenomenon has been rapidly evolving with the emergence of social media [ 4 , 5 ]. Fake news can be quickly shared by friends, followers, or even strangers within only a few seconds. Repeating a series of these processes could lead the public to form the wrong collective intelligence [ 6 ]. This could further develop into diverse social problems (i.e., setting a base station on fire because of rumors). In addition, some people believe and propagate fake news due to their personal norms, regardless of the factuality of the content [ 7 ]. Research in social science has suggested that cognitive bias (e.g., confirmation bias, bandwagon effect, and choice-supportive bias) [ 8 ] is one of the most pivotal factors in making irrational decisions in terms of the both creation and consumption of fake news [ 9 , 10 ]. Cognitive bias greatly contributes to the formation and enhancement of the echo chamber [ 11 ], meaning that news consumers share and consume information only in the direction of strengthening their beliefs [ 12 ].

Research using computational techniques (e.g., machine or deep learning) has been actively conducted for the past decade to investigate the current state of fake news and detect it effectively [ 13 ]. In particular, research into text-based feature selection and the development of detection models has been very actively and extensively conducted [ 14 – 17 ]. Research has been also active in the collection of fake news datasets [ 18 , 19 ] and fact-checking methodologies for model development [ 20 – 22 ]. Recently, Deepfake, which can manipulate images or videos through deep learning technology, has been used to create fake news images or videos, significantly increasing social concerns [ 23 ], and a growing body of research is being conducted to find ways of mitigating such concerns [ 24 – 26 ]. In addition, some research on system development (i.e., a game to increase awareness of the negative aspects of fake news) has been conducted to educate the public to avoid and prevent them from the situation where they could fall into the echo chamber, misunderstandings, wrong decision-making, blind belief, and propagating fake news [ 27 – 29 ].

While the creation and consumption of fake news are clearly different behaviors, due to the characteristics of the online environment (e.g., information can be easily created, shared, and consumed by anyone at anytime from anywhere), the boundaries between fake news creators and consumers have started to become blurred. Depending on the situation, people can quickly change their roles from fake news consumers to creators, or vice versa (with or without their intention). Furthermore, news creation and consumption are the most fundamental aspects that form the relationship between news and people. However, a significant amount of fake news research has positioned in news creation while considerably less research focus has been placed in news consumption (see Figs 1 & 2 ). This suggests that we must consider fake news as a comprehensive aspect of news consumption and creation .

• PPT PowerPoint slide
• PNG larger image
• TIFF original image

https://doi.org/10.1371/journal.pone.0260080.g001

The papers were published in IEEE, ACM, ELSEVIER, arXiv, Wiley, APA from 2010 to 2020 classified by publisher, main category, sub category, and evaluation method (left to right).

https://doi.org/10.1371/journal.pone.0260080.g002

In this paper, we looked into fake news research through the lens of news creation and consumption ( Fig 3 ). Our survey results offer different yet salient insights on fake news research compared with other survey papers (e.g., [ 13 , 30 , 31 ]), which primarily focus on fake news creation. The main contributions of our survey are as follows:

• We investigate trends in fake news research from 2010 to 2020 and confirm a need for applying a comprehensive perspective to fake news phenomenon.
• We present fake news research through the lens of news creation and consumption with external and internal factors.
• We examine key findings with a mental model approach, which highlights individuals’ differences in information understandings, expectations, or consumption.
• We summarize our review and discuss complementary roles of computer and social sciences and potential future directions for fake news research.

We investigate fake news research trend (Section 2), and examine fake news creation and consumption through the lenses of external and internal factors. We also investigate research efforts to mitigate external factors of fake news creation and consumption: (a) indicates fake news creation (Section 3), and (b) indicates fake news consumption (Section 4). “Possible moves” indicates that news consumers “possibly” create/propagate fake news without being aware of any negative impact.

https://doi.org/10.1371/journal.pone.0260080.g003

2 Fake news definition and trends

There is still no definition of fake news that can encompass false news and various types of disinformation (e.g., satire, fabricated content) and can reach a social consensus [ 30 ]. The definition continues to change over time and may vary depending on the research focus. Some research has defined fake news as false news based on the intention and factuality of the information [ 4 , 15 , 32 – 36 ]. For example, Allcott and Gentzkow [ 4 ] defined fake news as “news articles that are intentionally and verifiably false and could mislead readers.” On the other hand, other studies have defined it as “a news article or message published and propagated through media, carrying false information regardless of the means and motives behind it” [ 13 , 37 – 43 ]. Given this definition, fake news refers to false information that causes an individual to be deceived or doubt the truth, and fake news can only be useful if it actually deceives or confuses consumers. Zhou and Zafarani [ 31 ] proposed a broad definition (“Fake news is false news.”) that encompasses false online content and a narrow definition (“Fake news is intentionally and verifiably false news published by a news outlet.”). The narrow definition is valid from the fake news creation perspective. However, given that fake news creators and consumers are now interchangeable (e.g., news consumers also play a role of gatekeeper for fake news propagation), it has become important to understand and investigate the fake news through consumption perspectives. Thus, in this paper, we use the broad definition of fake news.

Our research motivation for considering news creation and consumption in fake news research was based on the trend analysis. We collected 2,277 fake news-related literature using four keywords (i.e., fake news, false information, misinformation, rumor) to identify longitudinal trends of fake news research from 2010 to 2020. The data collection was conducted from July to September 2020. The criteria of data collection was whether any of these keywords exists in the title or abstract. To reflect diverse research backgrounds/domains, we considered six primary publishers (ACM, IEEE, arXiv, APA, ELSEVIER, and Wiley). The number of papers collected for each publisher is as follows: 852 IEEE (37%), 639 ACM (28%), 463 ELSEVIER (20%), 142 arXiv (7%), 141 Wiley (6%), 40 APA (2%). We excluded 59 papers that did not have the abstract and used 2,218 papers for the analysis. We then randomly chose 200 papers, and two coders conducted manual inspection and categorization. The inter-coder reliability was verified by the Cohen’s Kappa measurement. The scores for each main/sub-category were higher than 0.72 (min: 0.72, max: 0.95, avg: 0.85), indicating that the inter-coder reliability lies between “substantial” to “perfect” [ 44 ]. Through the coding procedure, we excluded non-English studies (n = 12) and reports on study protocol only (n = 6), and 182 papers were included in synthesis. The PRISMA flow chart depicts the number of articles identified, included, and excluded (see Fig 1 ).

The papers were categorized into two main categories: (1) creation (studies with efforts to detect fake news or mitigate spread of fake news) and (2) consumption (studies that reported the social impacts of fake news on individuals or societies and how to appropriately handle fake news). Each main category was then classified into sub-categories. Fig 4 shows the frequency of the entire literature by year and the overall trend of fake news research. It appears that the consumption perspective of fake news still has not received sufficient attention compared with the creation perspective ( Fig 4(a) ). Fake news studies have exploded since the 2016 US Presidential Election, and the trend of increase in fake news research continues. In the creation category, the majority of papers (135 out of 158; 85%) were related to the false information (e.g., fake news, rumor, clickbait, spam) detection model ( Fig 4(b) ). On the other hand, in the consumption category, much research pertains to data-driven fake news trend analysis (18 out of 42; 43%) or fake content consumption behavior (16 out of 42; 38%), including studies for media literacy education or echo chamber awareness ( Fig 4(c) ).

We collected 2,277 fake news related-papers and randomly chose and categorized 200 papers. Each marker indicates the number of fake news studies per type published in a given year. Fig 4(a) shows a research trend of news creation and consumption (main category). Fig 4(b) and 4(c) show a trend of the sub-categories of news creation and consumption. In Fig 4(b), “Miscellaneous” includes studies on stance/propaganda detection and a survey paper. In Fig 4(c), “Data-driven fake news trend analysis” mainly covers the studies reporting the influence of fake news that spread around specific political/social events (e.g., fake news in Presidential Election 2016, Rumor in Weibo after 2015 Tianjin explosions). “Conspiracy theory” refers to an unverified rumor that was passed on to the public.

https://doi.org/10.1371/journal.pone.0260080.g004

3 Fake news creation

Fake news is no longer merely propaganda spread by inflammatory politicians; it is also made for financial benefit or personal enjoyment [ 45 ]. With the development of social media platforms people often create completely false information for reasons beyond satire. Further, there is a vicious cycle of this false information being abused by politicians and agitators.

Fake news creators are indiscriminately producing fake news while considering the behavioral and psychological characteristics of today’s news consumers [ 46 ]. For instance, the sleeper effect [ 47 ] refers to a phenomenon in which the persuasion effect increases over time, even though the pedigree of information shows low reliability. In other words, after a long period of time, memories of the pedigree become poor and only the content tends to be remembered regardless of the reliability of the pedigree. Through this process, less reliable information becomes more persuasive over time. Fake news creators have effectively created and propagated fake news by targeting the public’s preference for news consumption through peripheral processing routes [ 35 , 48 ].

Peripheral routes are based on the elaboration likelihood model (ELM) [ 49 ], one of the representative psychological theories that handles persuasive messages. According to the ELM, the path of persuasive message processing can be divided into the central and the peripheral routes depending on the level of involvement. On one hand, if the message recipient puts a great deal of cognitive effort into processing, the central path is chosen. On the other hand, if the process of the message is limited due to personal characteristics or distractions, the peripheral route is chosen. Through a peripheral route, a decision is made based on other secondary cues (e.g., speakers, comments) rather than the logic or strength of the argument.

Wang et al. [ 50 ] demonstrated that most of the links shared or mentioned in social media have never even been clicked. This implies that many people perceive and process information in only fragmentary way, such as via news headlines and the people sharing news, rather than considering the logical flow of news content.

In this section, we closely examined each of the external and internal factors affecting fake news creation, as well as the research efforts carried out to mitigate the negative results based on the fake news creation perspective.

3.1 External factors: Fake news creation facilitators

We identified two external factors that facilitate fake news creation and propagation: (1) the unification of news creation, consumption, and distribution, (2) the misuse of AI technology, and (3) the use of social media as a news platform (see Fig 5 ).

We identify two external factors—The unification of news and the misuse of AI technology—That facilitate fake news creation.

https://doi.org/10.1371/journal.pone.0260080.g005

3.1.1 The unification of news creation, consumption, and distribution.

The public’s perception of news and the major media of news consumption has gradually changed. The public no longer passively consumes news exclusively through traditional news organizations with specific formats (e.g., the inverted pyramid style, verified sources) nor view those news simply as a medium for information acquisition. The public’s active news consumption behaviors began in earnest with the advent of citizen journalism by implementing journalistic behavior based on citizen participation [ 51 ] and became commonplace with the emergence of social media. As a result, the public began to prefer interactive media, in which new information could be acquired, their opinions can be offered, and they can discuss the news with other news consumers. This environment has motivated the public to make content about their beliefs and deliver the content to many people as “news.” For example, a recent police crackdown video posted in social media quickly spread around the world that influenced protesters and civic movements. Then, it was reported later by the mainstream media [ 52 ].

The boundaries between professional journalists and amateurs, as well as between news consumers and creators, are disappearing. This has led to a potential increase in deceptive communications, making news consumers suspicious and misinterpreted the reality. Online platforms (e.g., YouTube, Facebook) that allow users to freely produce and distribute content have been growing significantly. As a result, fake news content can be used to attract secondary income (e.g., multinational enterprises’ advertising fees), which contributes to accelerating fake news creation and propagation. An environment in which the public can only consume news that suits their preferences and personal cognitive biases has made it much easier for fake news creators to achieve their specific purposes (e.g., supporting a certain political party or a candidate they favor).

3.1.2 The misuse of AI technology.

The development of AI technology has made it easier to develop and utilize tools for creating fake news, and many studies have confirmed the impact of these technologies— (1) social bots, (2) trolls, and (3) fake media —on social networks and democracy over the past decade.

3.1.2.1 Social bots . Shao et al. [ 53 ] analyzed the pattern of fake news spread and confirmed that social bots play a significant role in fake news propagation and social bot-based automated accounts were largely affected by the initial stage of spreading fake news. In general, it is uneasy for the public to determine whether such accounts are people or bots. In addition, social bots are not illegal tools and many companies legally purchase them as a part of marketing, thus it is not easy to curb the use of social bots systematically.

3.1.2.2 Trolls . The term “trolls” refers to people who deliberately cause conflict or division by uploading inflammatory, provocative content or unrelated posts to online communities. They work with the aim of stimulating people’s feelings or beliefs and hindering mature discussions. For example, the Russian troll army has been active in social media to advance its political agenda and cause social turmoil in the US [ 54 ]. Zannettou et al. [ 55 ] confirmed how effectively the Russian troll army has been spreading fake news URLs on Twitter and its significant impact on making other Twitter users believe misleading information.

3.1.2.3 Fake media . It is now possible to manipulate or reproduce content in 2D or even 3D through AI technology. In particular, the advent of fake news using Deepfake technology (combining various images on an original video and generating a different video) has raised another major social concern that had not been imagined before. Due to the popularity of image or video sharing on social media, such media types have become the dominant form of news consumption, and the Deepfake technology itself is becoming more advanced and applied to images and videos in a variety of domains. We witnessed a video clip of former US President Barack Obama criticizing Donald Trump, which was manipulated by the US online media company BuzzFeed to highlight the influence and danger of Deepfake, causing substantial social confusion [ 56 ].

3.2 Internal factors: Fake news creation purposes

We identified three main purposes for fake news creation— (1) ideological purposes, (2) monetary purposes, and (3) fear/panic reduction .

3.2.1 Ideological purpose.

Fake news has been created and propagated for political purposes by individuals or groups that positively affect the parties or candidates they support or undermine those who are not on the same side. Fake news with this political purpose has shown to negatively influence people and society. For instance, Russia created a fake Facebook account that caused many political disputes and enhanced polarization, affecting the 2016 US Presidential Election [ 57 ]. As polarization has intensified, there has also been a trend in the US that “unfriending” people who have different political tendencies [ 58 ]. This has led the public to decide whether to trust the news or not regardless of its factuality and has resulted in worsening in-group biases. During the Brexit campaign in the UK, many selective news articles were exposed on Facebook, and social bots and trolls were also confirmed as being involved in creating public opinions [ 59 , 60 ].

3.2.2 Monetary purpose.

Financial benefit is another strong motivation for many fake news creators [ 34 , 61 ]. Fake news websites usually reach the public through social media and make profits through posted advertisements. The majority of fake websites are focused on earning advertising revenue by spreading fake news that would attract readers’ attention, rather than political goals. For example, during the 2016 US Presidential Election in Macedonia, young people in their 10s and 20s used content from some extremely right-leaning blogs in the US to mass-produce fake news, earning huge advertising revenues [ 62 ]. This is also why fake news creators use provocative titles, such as clickbait headlines, to induce clicks and attempt to produce as many fake news articles as possible.

3.2.3 Fear and panic reduction.

In general, when epidemics become more common around the world, rumors of absurd and false medical tips spread rapidly in social media. When there is a lack of verified information, people feel great anxious and afraid and easily believe such tips, regardless of whether they are true [ 63 , 64 ]. The term infodemic , which first appeared during the 2003 SARS pandemics, describes this phenomenon [ 65 ]. Regarding COVID-19, health authorities have recently announced that preventing the creation and propagation of fake news about the virus is as important as alleviating the contagious power of COVID-19 [ 66 , 67 ]. The spread of fake news due to the absence of verified information has become more common regarding health-related social issues (e.g., infectious diseases), natural disasters, etc. For example, people with disorders affecting cognition (e.g., neurodegenerative disorder) are tend to easily believe unverified medical news [ 68 – 70 ]. Robledo and Jankovic [ 68 ] confirmed that many fake or exaggerated medical journals are misleading people with Parkinson’s disease by giving false hopes and unfounded fake articles. Another example is a rumor that climate activists set fire to raise awareness of climate change quickly spread as fake news [ 71 ], when a wildfire broke out in Australia in 2019. As a result, people became suspicious and tended to believe that the causes of climate change (e.g., global warming) may not be related to humans, despite scientific evidence and research data.

3.3 Fake news detection and prevention

The main purpose of fake news creation is to make people confused or deceived regardless of topic, social atmosphere, or timing. Due to this purpose, it appears that fake news tends to have similar frames and structural patterns. Many studies have attempted to mitigate the spread of fake news based on these identifiable patterns. In particular, research on developing computational models that detect fake information (text/images/videos), based on machine or deep learning techniques has been actively conducted, as summarized in Table 1 . Other modeling studies include the credibility of weblogs [ 84 , 85 ], communication quality [ 88 ], susceptibility level [ 90 ], and political stance [ 86 , 87 ]. The table was intended to characterize a research scope and direction of the development of fake information creation (e.g., the features employed in each model development), not to present an exhaustive list.

https://doi.org/10.1371/journal.pone.0260080.t001

3.3.1 Fake text information detection.

Research has considered many text-based features, such as structural (e.g., website URLs and headlines with all capital letters or exclamations) and linguistic information (e.g., grammar, spelling, and punctuation errors) about the news. Research has also considered the sentiments of news articles, the frequency of the words used, user information, and who left comments on the news articles, and social network information among users (who were connected based on activities of commenting, replying, liking or following) were used as key features for model development. These text-based models have been developed for not only fake news articles but also other types of fake information, such as clickbaits, fake reviews, spams, and spammers. Many of the models developed in this context performed a binary classification that distinguished between fake and non-fake articles, with the accuracy of such models ranging from 86% to 93%. Mainstream news articles were used to build most models, and some studies used articles on social media, such as Twitter [ 15 , 17 ]. Some studies developed fake news detection models by extracting features from images, as well as text, in news articles [ 16 , 17 , 75 ].

3.3.2 Fake visual media detection.

The generative adversary network (GAN) is an unsupervised learning method that estimates the probability distribution of original data and allows an artificial neural network to produce similar distributions [ 109 ]. With the advancement of GAN, it has become possible to transform faces in images into those of others. However, photos of famous celebrities have been misused (e.g., being distorted into pornographic videos), increasing concerns about the possible misuse of such technology [ 110 ] (e.g., creating rumors about a certain political candidate). To mitigate this, research has been conducted to develop detection models for fake images. Most studies developed binary classification models (fake image or not), and the accuracy of fake image detection models was high, ranging from 81% to 97%. However, challenges still exist. Unlike fake news detection models that employ fact-checking websites or mainstream news as data verification or ground-truth, fake image detection models were developed using the same or slightly modified image datasets (e.g., CelebA [ 97 ], FFHQ [ 99 ]), asking for the collection and preparation of a large amount of highly diverse data.

4 Fake news consumption

4.1 external factors: fake news consumption circumstances.

The implicit social contract between civil society and the media has gradually disintegrated in modern society, and accordingly, citizens’ trust in the media began to decline [ 111 ]. In addition, the growing number of digital media platforms has changed people’s news consumption environment. This change has increased the diversity of news content and the autonomy of information creation and sharing. At the same time, however, it blurred the line between traditional mainstream media news and fake news in the Internet environment, contributing to polarization.

Here, we identified three external factors that have forced the public to encounter fake news: (1) the decline of trust in the mainstream media, (2) a high-choice media environment, and (3) the use of social media as a news platform .

4.1.1 Fall of mainstream media trust.

Misinformation and unverified or biased reports have gradually undermined the credibility of the mainstream media. According to the 2019 American mass media trust survey conducted by Gallup, only 13% of Americans said they trusted traditional mainstream media: newspapers or TV news [ 112 ]. The decline in traditional media trust is not only a problem for the US, but also a common concern in Europe and Asia [ 113 – 115 ].

4.1.2 High-choice media environment.

Over the past decade, news consumption channels have been radically diversified, and the mainstream has shifted from broadcasting and print media to mobile and social media environments. Despite the diversity of news consumption channels, personalized preferences and repetitive patterns have led people to be exposed to limited information and continue to consume such information increasingly [ 116 ]. This selective news consumption attitude has enhanced the polarization of the public in many multi-media environments [ 117 ]. In addition, the commercialization of digital platforms have created an environment in which cognitive bias can be easily strengthened. In other words, a digital platform based on recommended algorithms has the convenience of providing similar content continuously after a given type of content is consumed. As a result, it may be easy for users to fall into the echo chamber because they only access recommended content. A survey of 1,000 YouTube videos found that more than two-thirds of the videos contained content in favor of a particular candidate [ 118 ].

News consumption in social media does not simply mean the delivery of messages from creators to consumers. The multi-directionality of social media has blurred the boundaries between information creators and consumers. In other words, users are already interacting with one another in various fashions, and when a new interaction type emerges and is supported by the platform, users will display other types of new interactions, which will also influence ways of consuming news information.

4.1.3 Use of social media as news platform.

Here we focus on the most widely used social media platforms—YouTube, Facebook, and Twitter—where each has characteristics of encouraging limited news consumption.

First, YouTube is the most unidirectional of social media. Many YouTube creators tend to convey arguments in a strong, definitive tone through their videos, and these content characteristics make viewers judge the objectivity of the information via non-verbal elements (e.g., speaker, thumbnail, title, comments) rather than facts. Furthermore, many comments often support the content of the video, which may increase the chances of viewers accepting somewhat biased information. In addition, a YouTube video recommendation algorithm causes users who watch certain news to continuously be exposed to other news containing the same or similar information. This behavior and direction on the part of isolated content consumption could undermine the viewer’s media literacy, and is likely to create a screening effect that blocks the user’s eyes and ears.

Second, Facebook is somewhat invisible regarding the details of news articles because this platform ostensibly shows only the title, the number of likes, and the comments of the posts. Often, users have to click on the article and go to the URL to read the article. This structure and consumptive content orientation on the part of Facebook presents obstacles that prevent users from checking the details of their posts. As a result, users have become likely to make limited and biased judgments and perceive content through provocative headlines and comments.

Third, the largest feature of Twitter is anonymity because Twitter asks users to make their own pseudonyms [ 119 ]. Twitter has a limited number of letters to upload, and compared to other platforms, users can produce and spread indiscriminate information anonymously and do not know who is behind the anonymity [ 120 , 121 ]. On the other hand, many accounts on Facebook operate under real names and generally share information with others who are friends or followers. Information creators are not held accountable for anonymous information.

4.2 Internal factors: Cognitive mechanism

Due to the characteristics of the Internet and social media, people are accustomed to consuming information quickly, such as reading only news headlines and checking photos in news articles. This type of news consumption practice could lead people to consider news information mostly based on their beliefs or values. This practice can make it easier for people to fall into an echo chamber and further social confusion. We identified two internal factors affecting fake news consumption: (1) cognitive biases and (2) personal traits (see Fig 6 ).

https://doi.org/10.1371/journal.pone.0260080.g006

4.2.1 Cognitive biases.

Cognitive bias is an observer effect that is broadly recognized in cognitive science and includes basic statistical and memory errors [ 8 ]. However, this bias may vary depending on what factors are most important to affect individual judgments and choices. We identified five cognitive biases that affect fake news consumption: confirmation bias, in-group bias, choice-supportive bias, cognitive dissonance, and primacy effect.

Confirmation bias relates to a human tendency to seek out information in line with personal thoughts or beliefs, as well as to ignore information that goes against such beliefs. This stems from the human desire to be reaffirmed, rather than accept denials of one’s opinion or hypothesis. If the process of confirmation bias is repeated, a more solid belief is gradually formed, and the belief remains unchanged even after encountering logical and objective counterexamples. Evaluating information with an objective attitude is essential to properly investigating any social phenomenon. However, confirmation bias significantly hinders this. Kunda [ 122 ] discussed experiments that investigated the cognitive processes as a function of accuracy goals and directional goals. Her analysis demonstrated that people use different cognitive processes to achieve the two different goals. For those who pursue accuracy goals (reaching a “right conclusion”), information is used as a tool to determine whether they are right or not [ 123 ], and for those with directional goals (reaching a desirable conclusion), information is used as a tool to justify their claims. Thus, biased information processing is more frequently observed by people with directional goals [ 124 ].

People with directional goals have a desire to reach the conclusion they want. The more we emphasize the seriousness and omnipresence of fake news, the less people with directional goals can identify fake news. Moreover, their confirmation bias through social media could result in an echo chamber, triggering a differentiation of public opinion in the media. The algorithm of the media platform further strengthens the tendency of biased information consumption (e.g., filter bubble).

In-group bias is a phenomenon in which an individual favors a group that he or she belongs to. The causes of in-group bias are two [ 125 ]. One is a categorization process, which exaggerates the similarities between members within one category (the internal group) and differences with others (the external groups). Consequently, positive reactions towards the internal group and negative reactions (e.g., hostility) towards the external group are both increased. The other reason is self-respect based on social identity theory. To positively evaluate the internal group, a member tends to perceive that other group members are similar to himself or herself.

In-group bias has a significant impact on fake news consumption because of radical changes in the media environment [ 126 ]. The public recognizes and forms groups based on issues through social media. The emotions and intentions of such groups of people online can be easily transferred or developed into offline activities, such as demonstrations and rallies. Information exchanges within such internal groups proceeds similarly to the situation with confirmation bias. If confirmation bias is keeping to one’s beliefs, in-group bias equates the beliefs of my group with my beliefs.

Choice-supportive bias refers to an individual’s tendency to justify his or her decision by highlighting the evidence that he or she did not consider in making the decision [ 127 ]. For instance, people sometimes have no particular purpose when they purchase a certain brand of products or service, or support a particular politician or political party. They emphasize that their choices at the time were right and inevitable. They also tend to focus more on positive aspects than negative effects or consequences to justify their choice. However, these positive aspects can be distorted because they are mainly based on memory. Thus, choice-supportive bias, can be regarded as the cognitive errors caused by memory distortion.

The behavioral condition of choice-supportive bias is used to justify oneself, which usually occurs in the context of external factors (e.g., maintaining social status or relationships) [ 7 ]. For example, if people express a certain political opinion within a social group, people may seek information with which to justify the opinion and minimize its flaws. In this procedure, people may accept fake news as a supporting source for their opinions.

Cognitive dissonance was based on the notion that some psychological tension would occur when an individual had two perceptions that were inconsistent [ 128 ]. Humans have a desire to identify and resolve the psychological tension that occurs when a cognitive dissonance is established. Regarding fake news consumption, people easily accept fake news if it is aligned with their beliefs or faith. However, if such news is seen as working against their beliefs or faith, people define even real news as fake and consume biased information in order to avoid cognitive dissonance. This is quite similar to cognitive bias. Selective exposure to biased information intensifies its extent and impact in social media. In these circumstances, an individual’s cognitive state is likely to be formed by information from unclear sources, which can be seen as a negative state of perception. In that case, information consumers selectively consume only information that can be in harmony with negative perceptions.

Primacy effect means that information presented previously will have a stronger effect on the memory and decision-making than information presented later [ 129 ]. The “interference theory [ 130 ]” is often referred to as a theoretical basis for supporting the primacy effect, which highlights the fact that the impression formed by the information presented earlier influences subsequent judgments and the process of forming the next impression.

The significance of the primary effect for fake news consumption is that it can be a starting point for biased cognitive processes. If an individual first encounters an issue in fake news and does not go through a critical thinking process about that information, he or she may form false attitudes regarding the issue [ 131 , 132 ]. Fake news is a complex combination of facts and fiction, making it difficult for information consumers to correctly judge whether the news is right or wrong. These cognitive biases induce the selective collection of information that feels more valid for news consumers, rather than information that is really valid.

4.2.2 Personal traits.

We two aspects of personal characteristics or traits can influence one’s behaviors in terms of news consumption: susceptibility and personality.

4.2.2.1 Susceptibility . The most prominent feature of social media is that consumers can be also creators, and the boundaries between the creators and consumers of information become unclear. New media literacy (i.e., the ability to critically and suitably consume messages in a variety of digital media channels, such as social media) can have a significant impact on the degree of consumption and dissemination of fake news [ 133 , 134 ]. In other words, the higher new media literacy is, the higher the probability that an individual is likely to take a critical standpoint toward fake news. Also, the susceptibility level of fake news is related to one’s selective news consumption behaviors. Bessi et al. [ 35 ] studied misinformation on Facebook and found that users who frequently interact with alternative media tend to interact with intentionally false claims more often.

Personality is an individual’s traits or behavior style. Many scholars have agreed that the personality can be largely divided into five categories (Big Five)—extraversion, agreeableness, neuroticism, openness, and conscientiousness [ 135 , 136 ]—and used them to understand the relationship between personality and news consumption.

Extroversion is related to active information use. Previous studies have confirmed that extroverts tend to use social media and that their main purpose of use is to acquire information [ 137 ] and better determine the factuality of news on social media [ 138 ]. Furthermore, people with high agreeableness, which refers to how friendly, warm, and tactful, tend to trust real news than fake news [ 138 ]. Neuroticism refers to a broad personality trait dimension representing the degree to which a person experiences the world as distressing, threatening, and unsafe. People with high neuroticism usually show negative emotions or information sharing behavior [ 139 ]. Neuroticism is positively related to fake news consumption [ 138 ]. Openness refers to the degree of enjoying new experiences. High openness is associated with high curiosity and engagement in learning [ 140 ], which enhances critical thinking ability and decreases negative effects of fake news consumption [ 138 , 141 ]. Conscientiousness refers to a person’s work ethic, being orderly, and thoroughness [ 142 ]. People with high conscientiousness tend to regard social media use as distraction from their tasks [ 143 – 145 ].

4.3 Fake news awareness and prevention

4.3.1 decision-making support tools..

News on social media does not go through the verification process, because of its high degree of freedom to create, share, and access information. The study reported that most citizens in advanced countries will have more fake information than real information in 2022 [ 146 ]. This indicates that potential personal and social damage from fake news may increase. Paradoxically, many countries that suffer from fake news problems strongly guarantee the freedom of expression under their constitutions; thus, it would be very difficult to block all possible production and distribution of fake news sources through laws and regulations. In this respect, it would be necessary to put in place not only technical efforts to detect and prevent the production and dissemination of fake news but also social efforts to make news consumers aware of the characteristics of online fake information.

Inoculation theory highlights that human attitudes and beliefs can form psychological resistance by being properly exposed to arguments against belief in advance. To have the ability to strongly protest an argument, it is necessary to expose and refute the same sort of content with weak arguments first. Doris-Down et al. [ 147 ] asked people who were from different political backgrounds to communicate directly through mobile apps and investigated whether these methods alleviated their echo-chamberness. As a result, the participants made changes, such as realizing that they had a lot in common with people who had conflicting political backgrounds and that what they thought was different was actually trivial. Karduni et al. [ 148 ] provided comprehensive information (e.g., connections among news accounts and a summary of the location entities) to study participants through the developed visual analytic system and examined how they accepted fake news. Another study was conducted to confirm how people determine the veracity of news by establishing a system similar to social media and analyzing the eye tracking of the study participants while reading fake news articles [ 28 ].

Some research has applied the inoculation theory to gamification. A “Bad News” game was designed to proactively warn people and expose them to a certain amount of false information through interactions with the gamified system [ 29 , 149 ]. The results confirmed the high effectiveness of inoculation through the game and highlighted the need to educate people about how to respond appropriately to misinformation through computer systems and games [ 29 ].

4.3.2 Fake information propagation analysis.

Fake information tends to show a certain pattern in terms of consumption and propagation, and many studies have attempted to identify the propagation patterns of fake information (e.g., the count of unique users, the depth of a network) [ 150 – 153 ].

4.3.2.1 Psychological characteristics . The theoretical foundation of research intended to examine the diffusion patterns of fake news lies in psychology [ 154 , 155 ] because psychological theories explain why and how people react to fake news. For instance, a news consumer who comes across fake news will first have doubts, judge the news against his background knowledge, and want to clarify the sources in the news. This series of processes ends when sufficient evidence is collected. Then the news consumer ends in accepting, ignoring, or suspecting the news. The psychological elements that can be defined in this process are doubts, negatives, conjectures, and skepticism [ 156 ].

4.3.2.2 Temporal characteristics . Fake news exhibits different propagation patterns from real news. The propagation of real news tends to slowly decrease over time after a single peak in the public’s interest, whereas fake news does not have a fixed timing for peak consumption, and a number of peaks appear in many cases [ 157 ]. Tambuscio et al. [ 151 ] proved that the pattern of the spread of rumors is similar to the existing epidemic model [ 158 ]. Their empirical observations confirmed that the same fake news reappears periodically and infects news consumers. For example, rumors that include the malicious political message that “Obama is a Muslim” are still being spread a decade later [ 159 ]. This pattern of proliferation and consumption shows that fake news may be consumed for a certain purpose.

5 A mental-model approach

We have examined news consumers’ susceptibility to fake news due to internal and external factors, including personal traits, cognitive biases, and the contexts. Beyond an investigation on the factor level, we seek to understand people’s susceptibility to misinformation by considering people’s internal representations and external environments holistically [ 5 ]. Specifically, we propose to comprehend people’s mental models of fake news. In this section, we first briefly introduce mental models and discuss their connection to misinformation. Then, we discuss the potential contribution of using a mental-model approach to the field of misinformation.

5.1 Mental models

A mental model is an internal representation or simulation that people carry in their minds of how the world works [ 160 , 161 ]. Typically, mental models are constructed in people’s working memory, in which information from long-term memory and the environments are combined [ 162 ]. They also indicate that individuals represent complex phenomena with somewhat abstraction based on their own experiences and understanding of the contexts. People rely on mental models to understand and predict their interactions with environments, artifacts and computing systems, as well as other individuals [ 163 , 164 ]. Generally, individuals’ ability to represent the continually changing environments is limited and unique. Thus, mental models tend to be functional and dynamic but not necessarily accurate or complete [ 163 , 165 ]. Mental models also differ between various groups and in particular between experts and novices [ 164 , 166 ].

5.2 Mental models and misinformation

Mental models have been proposed to understand human behaviors in spatial navigation [ 167 ], learning [ 168 , 169 ], deductive reasoning [ 170 ], mental presentations of real or imagined situations [ 171 ], risk communication [ 172 ], and usable cybersecurity and privacy [ 166 , 173 , 174 ]. People use mental models to facilitate their comprehension, judgment, and actions, and can be the basis of individual behaviors. In particular, the connection between a mental-model approach and misinformation has been revealed in risk communication regarding vaccines [ 175 , 176 ]. For example, Downs et al. [ 176 ] interviewed 30 parents from three US cities to understand their mental models about vaccination for their children aged 18 to 23 months. The results revealed two mental models about vaccination: (1) heath oriented : parents who focused on health-oriented topics trusted anecdotal communication more than statistical arguments; and (2) risk oriented : parents with some knowledge about vaccine mechanisms trusted communication with statistical arguments more than anecdotal information. Also, the authors found that many parents, even those favorable to vaccination, can be confused by ongoing debate, suggesting somewhat incompleteness of their mental models.

5.3 Potential contributions of a mental-model approach

Recognizing and dealing with the plurality of news consumers’ perception, cognition and actions is currently considered as key aspects of misinformation research. Thus, a mental model approach could significantly improve our understanding of people’s susceptibility to misinformation, as well as inform the development of mechanisms to mitigate misinformation.

One possible direction is to investigate the demographic differences in the context of mental models. As more Americans have adopted social media, the social media users have become more representative for the population. Usage by older adults has increased in recent years, with the use rate of about 12% in 2012 to about 35% in 2016 ( https://www.pewresearch.org/internet/fact-sheet/social-media/ ). Guess et al. (2019) analyzed participants’ profiles and their sharing activity on Facebook during the 2016 US Presidential campaign. A strong age effect was revealed. While controlled the effects of ideology and education, their results showed that Facebook users who are over 65 years old were associated with sharing nearly seven times as many articles from fake news domains on Facebook as those who are between 18–29 years old, or about 2.3 times as many as those in the age between 45 to 65.

Besides older adults, college students were shown more susceptibility to misinformation [ 177 ]. We can identify which mental models a particular age group ascribes to, and compare the incompleteness or incorrectness of the mental models by age. On the other hand, such comparison might be informative to design general mechanisms to mitigate misinformation independent of the different concrete mental models possessed by different types of users.

Users’ actions and decisions are directed by their mental models. We can also explore news consumers’ mental models and discover unanticipated and potentially risky human system interactions, which will inform the development and design of user interactions and education endeavors to mitigate misinformation.

A mental-model approach supplies an important, and as yet unconsidered, dimension to fake news research. To date, research on people’s susceptibility to fake news in social media has lagged behind research on computational aspect research on fake news. Scholars have not considered issues of news consumers’ susceptibility across the spectrum of their internal representations and external environments. An investigation from the mental model’s perspective is a step toward addressing such need.

6 Discussion and future work

In this section, we highlight the importance of balancing research efforts on fake news creation and consumption and discuss potential future directions of fake news research.

6.1 Leveraging insights of social science to model development

Developing fake news detection models has achieved great performance. Feature groups used in the model are diverse including linguistics, vision, sentiment, topic, user, and network, and many models used multiple groups to increase the performance. By using datasets with different size and characteristics, research has demonstrated the effectiveness of the models through a comparison analysis. However, much research has considered and used the features that are easily quantifiable, and many of them tend to have unclear justification or rationale of being used in modeling. For example, what is the relationship between the use of question (?), exclamation (!), or quotation marks (“…”) and fake news?, what does it mean by a longer description relates to news trustworthiness?. There are also many important aspects that can be used as additional features for modeling and have not yet found a way to be quantified. For example, journalistic styles are important characteristics that determine a level of information credibility [ 156 ], but it is challenging to accurately and reliably quantified them. There are many intentions (e.g., ideological standpoint, financial gain, panic creation) that authors may implicitly or explicitly display in the post but measuring them is uneasy and not straightforward. Social science research can play a role in here coming up with a valid research methodology to measure such subjective perceptions or notions considering various types and characteristics of them depending on a context or environment. Some research efforts in this research direction include quantifying salient factors of people’s decision-making identified in social science research and demonstrating the effectiveness of using the factors in improving model performance and interpreting model results [ 70 ]. Yet more research that applies socio-technical aspects in model development and application would be needed to better study complex characteristics of fake news.

6.1.1 Future direction.

Insights from social science may help develop transparent and applicable fake news detection models. Such socio-technical models may allow news consumers to have a better understanding of fake news detection results and its application as well as to take more appropriate actions to control fake news phenomenon.

6.2 Lack of research on fake news consumption

Regarding fake news consumption, we confirmed that only few studies involve the development of web- or mobile-based technology systems to help consumers aware possible dangers of fake news. Those studies [ 28 , 29 , 147 , 148 ] tried to demonstrate the feasibility of developed self-awareness systems through user studies. However, due to the limited number of study participants (min: 11, max: 60) and their lack of demographic diversity (i.e., recruited only college students of one school, the psychology research pool at the authors’ institution), the generalization and applicability of these systems are still questionable. On the other hand, research that involves the development of fake news detection models or network analysis to identify the pattern of fake news propagation has been relatively active. These results can be used to identify people (or entities) who intentionally create malicious fake content; however, it is still challenging to restrict people who originally had not shown any behaviors or indications of sharing or creating fake information but later manipulated real news to fake or disseminated fake news with their malicious intention or cognitive biases.

In other words, although fake news detection models have shown great, promising performance, the influence of the models may be exerted in limited cases. This is because fake news detection models heavily rely on the data that were labeled as fake by other fact-checking institutions or sites. If someone manipulates the news that were not covered by fact-checking, the format or characteristics of the manipulated news may be different from those (i.e., conventional features) that are identified and managed in the detection model. Such differences may not be captured by the model. Therefore, to prevent fake news phenomenon more effectively, research needs to consider changes of news consumption.

6.2.1 Future direction.

It may be desirable to support people recognizing that their news consumption behaviors (e.g., like, comment, share) can have a significant ripple effect. Developing a system that tracks activities of people’s news consumption and creation, measures similarity and differences between those activities, and presents behaviors or patterns of news consumption and creation to people would be helpful.

6.3 Limited coverage of fact-checking websites and regulatory approach

Some of the well-known fact-checking websites (e.g., snopes.com, politifact.com) cover news shared mostly on the Internet and label the authenticity or deficiencies of the content (e.g., miscaptioned, legend, misattributed). However, these fact-checking websites may show limited coverage in that they are only used for those who are willing to check the veracity of certain news articles. Social media platforms have been making continuous efforts to mitigate the spread of fake news. For example, Facebook shows that content that has been falsely assessed by fact-checkers is relatively less exposed to news feeds or shows warning indicators [ 178 ]. Instagram has also changed the way that warning labels are displayed when users attempt to view the content that has been falsely assessed [ 179 ]. However, this type of an interface could lead news consumers to relying on algorithmic decision-making rather than self-judgment because these ostensible regulations (e.g., warning labels) tend to lack transparency of the decision. As we explained previously, this is related to filter bubbles. Therefore, it is important to provide a more clear and transparent communicative interface for news consumers to access and understand underlying information of the algorithm results.

6.3.1 Future direction.

It is necessary to create a news consumption circumstance that gives a wider coverage of fake news and more transparent information of algorithmic decisions on news credibility. This will help news consumers preemptively avoid fake news consumption and contribute more to preventing fake news propagation. Consumers also make more proper and accurate decisions based on their understanding of the news.

6.4 New media literacy

With the diversification of news channels, we can easily consume news. However, we are also in a media environment that asks us to self-critically verify news content (e.g., whether the news title reads like a clickbait, whether the news title and content are related), which in reality is hard to be done. Moreover, in social media, news consumers can be news creators or reproducers. During this process, news information could be changed based on a consumer’s beliefs or interests. A problem here is that people may not know how to verify news content or not be aware of whether the information could be distorted or biased. As the news consumer environment changes rapidly and faces modern media deluge, the importance of media literacy education is high. Media literacy refers to the ability to decipher media content, but in a broad sense, to understand the principles of media operation and media content sensibly and critically, and in turn to the ability to utilize and creatively reproduce content. Being a “lazy thinker” is more susceptible to fake news than having a “partisan bias” [ 32 ]. As “screen time” (i.e., time spent looking at smartphone, computer, or television screens) has become more common, people are consuming only stimulating (e.g., sensual pleasure and excitement) information [ 180 ]. This could gradually lower one’s ability of critical, reasonable thinking, leading to making wrong judgments and actions. In France, when fake news problem became more serious, and a great amount of efforts were made to create “European Media Literacy Week” in schools [ 181 ]. The US is also making legislative efforts to add media literacy to the general education curriculum [ 182 ]. However, the acquisition of new media literacy through education may be limited to people in school (e.g., young students) and would be challenging to be expanded to wider populations. Thus, there is also a need for supplementary tools and research efforts to support more people to critically interpret and appropriately consume news.

In addition, more critical social attention is needed because visual content (e.g., images, videos), which had been naturally accepted as facts, can be easily manipulated in a malicious fashion and looked very natural. We have seen that people prefer to watch YouTube videos for news consumption rather than reading news articles. This visual content makes it relatively easy for news consumers to trust the content compared to text-based information and makes it easier to obtain information simply by playing the video. Since visual content will become a more dominant medium in future news consumption, educating and inoculating news consumers about potential threats of fake information in such news media would be important. More attention and research are needed on the technology supporting fake visual content awareness.

6.4.1 Future direction.

Research in both computer science and social science should find ways (e.g., developing a game-based education system or curriculum) to help news consumers aware of their practice of news consumption and maintain right news consumption behaviors.

7 Conclusion

We presented a comprehensive summary of fake news research through the lenses of news creation and consumption. The trends analysis indicated a growing increase in fake news research and a great amount of research focus on news creation compared to news consumption. By looking into internal and external factors, we unpacked the characteristics of fake news creation and consumption and presented the use of people’s mental models to better understand people’s susceptibility to misinformation. Based on the reviews, we suggested four future directions on fake news research—(1) a socio-technical model development using insights from social science, (2) in-depth understanding of news consumption behaviors, (3) preemptive decision-making and action support, and (4) educational, new media literacy support—as ways to reduce the gap between news creation and consumption and between computer science and social science research and to support healthy news environments.

Supporting information

S1 checklist..

https://doi.org/10.1371/journal.pone.0260080.s001

• View Article
• Google Scholar
• 2. Goldman R. Reading fake news, Pakistani minister directs nuclear threat at Israel. The New York Times . 2016;24.
• PubMed/NCBI
• 6. Lévy P, Bononno R. Collective intelligence: Mankind’s emerging world in cyberspace. Perseus Books; 1997.
• 11. Jamieson KH, Cappella JN. Echo chamber: Rush Limbaugh and the conservative media establishment. Oxford University Press; 2008.
• 14. Shu K, Cui L, Wang S, Lee D, Liu H. defend: Explainable fake news detection. In: In Proc. of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (KDD); 2019. p. 395–405.
• 15. Ruchansky N, Seo S, Liu Y. Csi: A hybrid deep model for fake news detection. In: In Proc. of the 2017 ACM on Conference on Information and Knowledge Management (CIKM); 2017. p. 797–806.
• 16. Cui L, Wang S, Lee D. Same: sentiment-aware multi-modal embedding for detecting fake news. In: In Proc. of the 2019 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM); 2019. p. 41–48.
• 17. Wang Y, Ma F, Jin Z, Yuan Y, Xun G, Jha K, et al. Eann: Event adversarial neural networks for multi-modal fake news detection. In: In Proc. of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data mining (KDD); 2018. p. 849–857.
• 18. Nørregaard J, Horne BD, Adalı S. Nela-gt-2018: A large multi-labelled news for the study of misinformation in news articles. In: In Proc. of the International AAAI Conference on Web and Social Media (ICWSM). vol. 13; 2019. p. 630–638.
• 20. Nguyen AT, Kharosekar A, Krishnan S, Krishnan S, Tate E, Wallace BC, et al. Believe it or not: Designing a human-ai partnership for mixed-initiative fact-checking. In: In Proc. of the 31st Annual ACM Symposium on User Interface Software and Technology (UIST); 2018. p. 189–199.
• 23. Brandon J. Terrifying high-tech porn: creepy’deepfake’videos are on the rise. Fox News . 2018;20.
• 24. Nguyen TT, Nguyen CM, Nguyen DT, Nguyen DT, Nahavandi S. Deep Learning for Deepfakes Creation and Detection. arXiv . 2019;1.
• 25. Rossler A, Cozzolino D, Verdoliva L, Riess C, Thies J, Nießner M. Faceforensics++: Learning to detect manipulated facial images. In: IEEE International Conference on Computer Vision (ICCV); 2019. p. 1–11.
• 26. Nirkin Y, Keller Y, Hassner T. Fsgan: Subject agnostic face swapping and reenactment. In: In Proc. of the IEEE International Conference on Computer Vision (ICCV); 2019. p. 7184–7193.
• 28. Simko J, Hanakova M, Racsko P, Tomlein M, Moro R, Bielikova M. Fake news reading on social media: an eye-tracking study. In: In Proc. of the 30th ACM Conference on Hypertext and Social Media (HT); 2019. p. 221–230.
• 35. Horne B, Adali S. This just in: Fake news packs a lot in title, uses simpler, repetitive content in text body, more similar to satire than real news. In: In Proc. of the 11th International AAAI Conference on Web and Social Media (ICWSM); 2017. p. 759–766.
• 36. Golbeck J, Mauriello M, Auxier B, Bhanushali KH, Bonk C, Bouzaghrane MA, et al. Fake news vs satire: A dataset and analysis. In: In Proc. of the 10th ACM Conference on Web Science (WebSci); 2018. p. 17–21.
• 37. Mustafaraj E, Metaxas PT. The fake news spreading plague: was it preventable? In: In Proc. of the 9th ACM Conference on Web Science (WebSci); 2017. p. 235–239.
• 40. Jin Z, Cao J, Zhang Y, Luo J. News verification by exploiting conflicting social viewpoints in microblogs. In: In Proc. of the 13th AAAI Conference on Artificial Intelligence (AAAI); 2016. p. 2972–2978.
• 41. Rubin VL, Conroy N, Chen Y, Cornwell S. Fake news or truth? using satirical cues to detect potentially misleading news. In: In Proc. of the Second Workshop on Computational Approaches to Deception Detection ; 2016. p. 7–17.
• 45. Kahneman D, Tversky A. Prospect theory: An analysis of decision under risk. In: Handbook of the fundamentals of financial decision making: Part I. World Scientific; 2013. p. 99–127.
• 46. Hanitzsch T, Wahl-Jorgensen K. Journalism studies: Developments, challenges, and future directions. The Handbook of Journalism Studies . 2020; p. 3–20.
• 48. Osatuyi B, Hughes J. A tale of two internet news platforms-real vs. fake: An elaboration likelihood model perspective. In: In Proc. of the 51st Hawaii International Conference on System Sciences (HICSS); 2018. p. 3986–3994.
• 49. Cacioppo JT, Petty RE. The elaboration likelihood model of persuasion. ACR North American Advances. 1984; p. 673–675.
• 50. Wang LX, Ramachandran A, Chaintreau A. Measuring click and share dynamics on social media: a reproducible and validated approach. In Proc of the 10th International AAAI Conference on Web and Social Media (ICWSM). 2016; p. 108–113.
• 51. Bowman S, Willis C. How audiences are shaping the future of news and information. We Media . 2003; p. 1–66.
• 52. Hill E, Tiefenthäler A, Triebert C, Jordan D, Willis H, Stein R. 8 Minutes and 46 Seconds: How George Floyd Was Killed in Police Custody; 2020. Available from: https://www.nytimes.com/2020/06/18/us/george-floyd-timing.html .
• 54. Carroll O. St Petersburg ‘troll farm’ had 90 dedicated staff working to influence US election campaign; 2017.
• 55. Zannettou S, Caulfield T, Setzer W, Sirivianos M, Stringhini G, Blackburn J. Who let the trolls out? towards understanding state-sponsored trolls. In: Proc. of the 10th ACM Conference on Web Science (WebSci); 2019. p. 353–362.
• 56. Vincent J. Watch Jordan Peele use AI to make Barack Obama deliver a PSA about fake news. The Verge . 2018;17.
• 58. Linder M. Block. Mute. Unfriend. Tensions rise on Facebook after election results. Chicago Tribune . 2016;9.
• 60. Howard PN, Kollanyi B. Bots, #StrongerIn, and #Brexit: computational propaganda during the UK-EU referendum. arXiv . 2016; p. arXiv–1606.
• 61. Kasra M, Shen C, O’Brien JF. Seeing is believing: how people fail to identify fake images on the Web. In Proc of the 2018 CHI Conference Extended Abstracts on Human Factors in Computing Systems (CHI). 2018; p. 1–6.
• 62. Kirby EJ. The city getting rich from fake news. BBC News . 2016;5.
• 63. Hu Z, Yang Z, Li Q, Zhang A, Huang Y. Infodemiological study on COVID-19 epidemic and COVID-19 infodemic. Preprints . 2020; p. 2020020380.
• 71. Knaus C. Disinformation and lies are spreading faster than Australia’s bushfires. The Guardian . 2020;11.
• 72. Karimi H, Roy P, Saba-Sadiya S, Tang J. Multi-source multi-class fake news detection. In: In Proc. of the 27th International Conference on Computational Linguistics ; 2018. p. 1546–1557.
• 73. Wang WY. “Liar, liar pants on fire”: A new benchmark dataset for fake news detection. arXiv . 2017; p. arXiv–1705.
• 74. Pérez-Rosas V, Kleinberg B, Lefevre A, Mihalcea R. Automatic Detection of Fake News. arXiv . 2017; p. arXiv–1708.
• 75. Yang Y, Zheng L, Zhang J, Cui Q, Li Z, Yu PS. TI-CNN: Convolutional Neural Networks for Fake News Detection. arXiv . 2018; p. arXiv–1806.
• 76. Kumar V, Khattar D, Gairola S, Kumar Lal Y, Varma V. Identifying clickbait: A multi-strategy approach using neural networks. In: In Proc. of the 41st International ACM SIGIR Conference on Research & Development in Information Retrieval (SIGIR); 2018. p. 1225–1228.
• 77. Yoon S, Park K, Shin J, Lim H, Won S, Cha M, et al. Detecting incongruity between news headline and body text via a deep hierarchical encoder. In: Proc. of the AAAI Conference on Artificial Intelligence. vol. 33; 2019. p. 791–800.
• 78. Lu Y, Zhang L, Xiao Y, Li Y. Simultaneously detecting fake reviews and review spammers using factor graph model. In: In Proc. of the 5th Annual ACM Web Science Conference (WebSci); 2013. p. 225–233.
• 79. Mukherjee A, Venkataraman V, Liu B, Glance N. What yelp fake review filter might be doing? In: In Proc. of The International AAAI Conference on Weblogs and Social Media (ICWSM); 2013. p. 409–418.
• 80. Benevenuto F, Magno G, Rodrigues T, Almeida V. Detecting spammers on twitter. In: In Proc. of the 8th Annual Collaboration , Electronic messaging , Anti-Abuse and Spam Conference (CEAS). vol. 6; 2010. p. 12.
• 81. Lee K, Caverlee J, Webb S. Uncovering social spammers: social honeypots+ machine learning. In: In Proc. of the 33rd International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR); 2010. p. 435–442.
• 82. Li FH, Huang M, Yang Y, Zhu X. Learning to identify review spam. In: In Proc. of the 22nd International Joint Conference on Artificial Intelligence (IJCAI); 2011. p. 2488–2493.
• 83. Wang J, Wen R, Wu C, Huang Y, Xion J. Fdgars: Fraudster detection via graph convolutional networks in online app review system. In: In Proc. of The 2019 World Wide Web Conference (WWW); 2019. p. 310–316.
• 84. Castillo C, Mendoza M, Poblete B. Information credibility on twitter. In: In Proc. of the 20th International Conference on World Wide Web (WWW); 2011. p. 675–684.
• 85. Jo Y, Kim M, Han K. How Do Humans Assess the Credibility on Web Blogs: Qualifying and Verifying Human Factors with Machine Learning. In: In Proc. of the 2019 CHI Conference on Human Factors in Computing Systems (CHI); 2019. p. 1–12.
• 86. Che X, Metaxa-Kakavouli D, Hancock JT. Fake News in the News: An Analysis of Partisan Coverage of the Fake News Phenomenon. In: In Proc. of the 21st ACM Conference on Computer-Supported Cooperative Work and Social Computing (CSCW); 2018. p. 289–292.
• 87. Potthast M, Kiesel J, Reinartz K, Bevendorff J, Stein B. A Stylometric Inquiry into Hyperpartisan and Fake News. arXiv . 2017; p. arXiv–1702.
• 89. Popat K, Mukherjee S, Strötgen J, Weikum G. Credibility assessment of textual claims on the web. In: In Proc. of the 25th ACM International on Conference on Information and Knowledge Management (CIKM); 2016. p. 2173–2178.
• 90. Shen TJ, Cowell R, Gupta A, Le T, Yadav A, Lee D. How gullible are you? Predicting susceptibility to fake news. In: In Proc. of the 10th ACM Conference on Web Science (WebSci); 2019. p. 287–288.
• 91. Gupta A, Lamba H, Kumaraguru P, Joshi A. Faking sandy: characterizing and identifying fake images on twitter during hurricane sandy. In: In Proc. of the 22nd International Conference on World Wide Web ; 2013. p. 729–736.
• 92. He P, Li H, Wang H. Detection of fake images via the ensemble of deep representations from multi color spaces. In: In Proc. of the 26th IEEE International Conference on Image Processing (ICIP). IEEE; 2019. p. 2299–2303.
• 93. Sun Y, Chen Y, Wang X, Tang X. Deep learning face representation by joint identification-verification. Advances in Neural Information Processing Systems . 2014; p. 1–9.
• 94. Huh M, Liu A, Owens A, Efros AA. Fighting fake news: Image splice detection via learned self-consistency. In: In Proc. of the European Conference on Computer Vision (ECCV); 2018. p. 101–117.
• 95. Dang H, Liu F, Stehouwer J, Liu X, Jain AK. On the detection of digital face manipulation. In: In Proc. of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR); 2020. p. 5781–5790.
• 96. Tariq S, Lee S, Kim H, Shin Y, Woo SS. Detecting both machine and human created fake face images in the wild. In Proc of the 2nd International Workshop on Multimedia Privacy and Security (MPS). 2018; p. 81–87.
• 97. Liu Z, Luo P, Wang X, Tang X. Deep learning face attributes in the wild. In: In Proc. of the IEEE International Conference on Computer Vision (ICCV); 2015. p. 3730–3738.
• 98. Wang R, Ma L, Juefei-Xu F, Xie X, Wang J, Liu Y. Fakespotter: A simple baseline for spotting ai-synthesized fake faces. arXiv . 2019; p. arXiv–1909.
• 99. Karras T, Laine S, Aila T. A style-based generator architecture for generative adversarial networks. In: In Proc. of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR); 2019. p. 4401–4410.
• 100. Yang X, Li Y, Qi H, Lyu S. Exposing GAN-synthesized faces using landmark locations. In Proc of the ACM Workshop on Information Hiding and Multimedia Security (IH&MMSec). 2019; p. 113–118.
• 101. Zhang X, Karaman S, Chang SF. Detecting and simulating artifacts in gan fake images. In Proc of the 2019 IEEE International Workshop on Information Forensics and Security (WIFS). 2019; p. 1–6.
• 102. Amerini I, Galteri L, Caldelli R, Del Bimbo A. Deepfake video detection through optical flow based cnn. In Proc of the IEEE International Conference on Computer Vision Workshops (ICCV). 2019; p. 1205–1207.
• 103. Li Y, Lyu S. Exposing deepfake videos by detecting face warping artifacts. arXiv . 2018; p. 46–52.
• 104. Korshunov P, Marcel S. Deepfakes: a new threat to face recognition? assessment and detection. arXiv . 2018; p. arXiv–1812.
• 105. Jeon H, Bang Y, Woo SS. Faketalkerdetect: Effective and practical realistic neural talking head detection with a highly unbalanced dataset. In Proc of the IEEE International Conference on Computer Vision Workshops (ICCV). 2019; p. 1285–1287.
• 106. Chung JS, Nagrani A, Zisserman A. Voxceleb2: Deep speaker recognition. arXiv . 2018; p. arXiv–1806.
• 107. Songsri-in K, Zafeiriou S. Complement face forensic detection and localization with faciallandmarks. arXiv . 2019; p. arXiv–1910.
• 108. Ma S, Cui L, Dai D, Wei F, Sun X. Livebot: Generating live video comments based on visual and textual contexts. In Proc of the AAAI Conference on Artificial Intelligence (AAAI). 2019; p. 6810–6817.
• 109. Goodfellow I, Pouget-Abadie J, Mirza M, Xu B, Warde-Farley D, Ozair S, et al. Generative adversarial nets. Advances in Neural Information Processing Systems . 2014; p. arXiv–1406.
• 110. Metz R. The number of deepfake videos online is spiking. Most are porn; 2019. Available from: https://cnn.it/3xPJRT2 .
• 111. Strömbäck J. In search of a standard: Four models of democracy and their normative implications for journalism. Journalism Studies . 2005; p. 331–345.
• 112. Brenan M. Americans’ Trust in Mass Media Edges Down to 41%; 2019. Available from: https://bit.ly/3ejl6ql .
• 114. Ladd JM. Why Americans hate the news media and how it matters. Princeton University Press; 2012.
• 116. Weisberg J. Bubble trouble: Is web personalization turning us into solipsistic twits; 2011. Available from: https://bit.ly/3xOGFqD .
• 117. Pariser E. The filter bubble: How the new personalized web is changing what we read and how we think. Penguin; 2011.
• 118. Lewis P, McCormick E. How an ex-YouTube insider investigated its secret algorithm. The Guardian . 2018;2.
• 120. Kavanaugh AL, Yang S, Li LT, Sheetz SD, Fox EA, et al. Microblogging in crisis situations: Mass protests in Iran, Tunisia, Egypt; 2011.
• 121. Mustafaraj E, Metaxas PT, Finn S, Monroy-Hernández A. Hiding in Plain Sight: A Tale of Trust and Mistrust inside a Community of Citizen Reporters. In Proc of the 6th International AAAI Conference on Weblogs and Social Media (ICWSM) . 2012; p. 250–257.
• 125. Tajfel H. Human groups and social categories: Studies in social psychology. Cup Archive ; 1981.
• 127. Correia V, Festinger L. Biased argumentation and critical thinking. Rhetoric and Cognition: Theoretical Perspectives and Persuasive Strategies . 2014; p. 89–110.
• 128. Festinger L. A theory of cognitive dissonance. Stanford University Press; 1957.
• 136. John OP, Srivastava S, et al. The Big Five trait taxonomy: History, measurement, and theoretical perspectives. Handbook of Personality: theory and research . 1999; p. 102–138.
• 138. Shu K, Wang S, Liu H. Understanding user profiles on social media for fake news detection. In: 2018 IEEE Conference on Multimedia Information Processing and Retrieval (MIPR). IEEE; 2018. p. 430–435.
• 142. Costa PT, McCrae RR. The NEO personality inventory. Psychological Assessment Resources; 1985.
• 146. Panetta K. Gartner top strategic predictions for 2018 and beyond; 2017. Available from: https://gtnr.it/33kuljQ .
• 147. Doris-Down A, Versee H, Gilbert E. Political blend: an application designed to bring people together based on political differences. In Proc of the 6th International Conference on Communities and Technologies (C&T). 2013; p. 120–130.
• 148. Karduni A, Wesslen R, Santhanam S, Cho I, Volkova S, Arendt D, et al. Can You Verifi This? Studying Uncertainty and Decision-Making About Misinformation Using Visual Analytics. In Proc of the 12th International AAAI Conference on Web and Social Media (ICWSM). 2018;12(1).
• 149. Basol M, Roozenbeek J, van der Linden S. Good news about bad news: gamified inoculation boosts confidence and cognitive immunity against fake news. Journal of Cognition . 2020;3(1).
• 151. Tambuscio M, Ruffo G, Flammini A, Menczer F. Fact-checking effect on viral hoaxes: A model of misinformation spread in social networks. In Proc of the 24th International Conference on World Wide Web (WWW). 2015; p. 977–982.
• 152. Friggeri A, Adamic L, Eckles D, Cheng J. Rumor cascades. In Proc of the 8th International AAAI Conference on Weblogs and Social Media (ICWSM) . 2014;8.
• 153. Lerman K, Ghosh R. Information contagion: An empirical study of the spread of news on digg and twitter social networks. arXiv . 2010; p. arXiv–1003.
• 155. Cantril H. The invasion from Mars: A study in the psychology of panic. Transaction Publishers; 1952.
• 158. Bailey NT, et al. The mathematical theory of infectious diseases and its applications. Charles Griffin & Company Ltd; 1975.
• 159. on Religion PF, Life P. Growing Number of Americans Say Obama Is a Muslim; 2010.
• 160. Craik KJW. The nature of explanation. Cambridge University Press; 1943.
• 161. Johnson-Laird PN. Mental models: Towards a cognitive science of language, inference, and consciousness. 6. Harvard University Press; 1983.
• 162. Johnson-Laird PN, Girotto V, Legrenzi P. Mental models: a gentle guide for outsiders. Sistemi Intelligenti . 1998;9(68).
• 164. Rouse WB, Morris NM. On looking into the black box: Prospects and limits in the search for mental models. Psychological Bulletin . 1986;100(3).
• 166. Wash R, Rader E. Influencing mental models of security: a research agenda. In Proc of the 2011 New Security Paradigms Workshop (NSPW). 2011; p. 57–66.
• 167. Tversky B. Cognitive maps, cognitive collages, and spatial mental models. In Proc of European conference on spatial information theory (COSIT). 1993; p. 14–24.
• 169. Mayer RE, Mathias A, Wetzell K. Fostering understanding of multimedia messages through pre-training: Evidence for a two-stage theory of mental model construction. Journal of Experimental Psychology: Applied . 2002;8(3).
• 172. Morgan MG, Fischhoff B, Bostrom A, Atman CJ, et al. Risk communication: A mental models approach. Cambridge University Press; 2002.
• 174. Kang R, Dabbish L, Fruchter N, Kiesler S. “My Data Just Goes Everywhere:” User mental models of the internet and implications for privacy and security. In Proc of 11th Symposium On Usable Privacy and Security . 2015; p. 39–52.
• 178. Facebook Journalism Project. Facebook’s Approach to Fact-Checking: How It Works; 2020. https://bit.ly/34QgOlj .
• 179. Sardarizadeh S. Instagram fact-check: Can a new flagging tool stop fake news?; 2019. Available from: https://bbc.in/33fg5ZR .
• 180. Greenfield S. Mind change: How digital technologies are leaving their mark on our brains. Random House Incorporated ; 2015.
• 181. European Commission. European Media Literacy Week; 2020. https://bit.ly/36H9MR3 .
• 182. Media Literacy Now. U.S. media literacy policy report 2020; 2020. https://bit.ly/33LkLqQ .

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts

Latest science news, discoveries and analysis

Weird signal that baffled seismologists traced to mega-landslide in Greenland

The brain aged more slowly in monkeys given a cheap diabetes drug

AI chatbot gets conspiracy theorists to question their convictions

First private spacewalk a success! What the SpaceX mission means for science

Why do we crumble under pressure science has the answer, famed pacific island’s population 'crash' debunked by ancient dna, us election debate: what harris and trump said about science, why some people enter menopause early — and how that could affect their cancer risk, europe sidelines alzheimer’s drug: lessons must be learnt henrik zetterberg.

How we slashed our lab’s carbon footprint

Red light, green light: flickering fluorophores reveal biochemistry in cells

The human costs of the research-assessment culture

The baseless stat that could be harming Indigenous conservation efforts

Academics say flying to meetings harms the climate — but they carry on, ancient dna debunks rapa nui ‘ecological suicide’ theory, human embryo models are getting more realistic — raising ethical questions, why does heart disease affect so many young south asians.

How to support Indigenous Peoples on biodiversity: be rigorous with data

Mpox: apply COVID lessons to control outbreak in Africa

Data on SDGs are riddled with gaps. Citizens can help

Wildfires are spreading fast in canada — we must strengthen forests for the future, why the next pandemic could come from the arctic — and what to do about it christian sonne, current issue.

Bioprospecting marine microbial genomes to improve biotechnology

Spectroscopic confirmation of two luminous galaxies at a redshift of 14, one month convection timescale on the surface of a giant evolved star, research analysis.

Rapa Nui’s population history rewritten using ancient DNA

Menopause age shaped by genes that influence mutation risk

Swirling star bubbles offer a glimpse of the Sun’s future

Future optoelectronics unlocked by ‘doping’ strategy

Long-lasting heart-failure treatment could be a game-changer, lipid recycling by macrophage cells drives the growth of brain cancer, thread, read, rewind, repeat: towards using nanopores for protein sequencing, cell-to-cell tunnels rescue neurons from degeneration.

The grassroots organizations continuing the fight for Ukrainian science

How a struggling biotech company became a university ‘spin-in’

I make fake eyes for those who need them, massive attack’s science-led drive to lower music’s carbon footprint, books & culture.

How influencers and algorithms mobilize propaganda — and distort reality

Consider the finches: Books in brief

Why repairing forests is not just about planting trees

Candidate 1143172 cover letter: junior pot scrubber, nature podcast.

Latest videos

Nature briefing.

An essential round-up of science news, opinion and analysis, delivered to your inbox every weekday.

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Early Dark Energy Could Resolve Cosmology’s Two Biggest Puzzles

Michael Boylan-Kolchin and others show "early dark energy" might help solve the Hubble Tension and explain why there are more early galaxies than expected.

• Share on Facebook
• Share on LinkedIn
• Show and hide the URL copier tool.
• Share via email
• Cosmology, Theoretical Physics and Gravity
• McDonald Observatory

Take a Tour of UT Through the Lens of Science

September 4, 2024 • by Sowmya Sridhar

Early Galaxies Weren’t Too Big for Their Britches After All

August 26, 2024 • by Marc Airhart

Announcements

Observatory Remembers Harlan J. Smith on His 100th Birthday

August 25, 2024 • by Staff Writer

• Diversity & Inclusion
• Community Values
• Visiting MIT Physics
• People Directory
• Faculty Directory
• Faculty Awards
• History of MIT Physics
• Policies and Procedures
• Departmental Committees
• Academic Programs Team
• Finance Team
• Meet the Academic Programs Team
• Prospective Students
• Requirements
• Employment Opportunities
• Research Opportunities
• Graduate Admissions
• Doctoral Guidelines
• Financial Support
• Graduate Student Resources
• PhD in Physics, Statistics, and Data Science
• MIT LEAPS Program
• Physics Student Groups
• for Undergraduate Students
• for Graduate Students
• Mentoring Programs Info for Faculty
• Non-degree Programs
• Student Awards & Honors
• Astrophysics Observation, Instrumentation, and Experiment
• Astrophysics Theory
• Atomic Physics
• Condensed Matter Experiment
• Condensed Matter Theory
• High Energy and Particle Theory
• Nuclear Physics Experiment
• Particle Physics Experiment
• Plasma Physics
• Quantum Gravity and Field Theory
• Quantum Information Science
• Strong Interactions and Nuclear Theory
• Center for Theoretical Physics
• Affiliated Labs & Centers
• Program Founder
• Competition
• Donor Profiles
• Patrons of Physics Fellows Society
• Giving Opportunties
• Latest Physics News
• Physics Journal: Fall 2023 Edition
• Events Calendar
• Physics Colloquia
• Search for: Search

Study: Early dark energy could resolve cosmology’s two biggest puzzles

In the universe’s first billion years, this brief and mysterious force could have produced more bright galaxies than theory predicts..

A new study by MIT physicists proposes that a mysterious force known as early dark energy could solve two of the biggest puzzles in cosmology and fill in some major gaps in our understanding of how the early universe evolved.

One puzzle in question is the “Hubble tension,” which refers to a mismatch in measurements of how fast the universe is expanding. The other involves observations of numerous early, bright galaxies that existed at a time when the early universe should have been much less populated.

Now, the MIT team has found that both puzzles could be resolved if the early universe had one extra, fleeting ingredient: early dark energy. Dark energy is an unknown form of energy that physicists suspect is driving the expansion of the universe today. Early dark energy is a similar, hypothetical phenomenon that may have made only a brief appearance, influencing the expansion of the universe in its first moments before disappearing entirely.

Some physicists have suspected that early dark energy could be the key to solving the Hubble tension, as the mysterious force could accelerate the early expansion of the universe by an amount that would resolve the measurement mismatch.

The MIT researchers have now found that early dark energy could also explain the baffling number of bright galaxies that astronomers have observed in the early universe. In their new study, reported today in the Monthly Notices of the Royal Astronomical Society , the team modeled the formation of galaxies in the universe’s first few hundred million years. When they incorporated a dark energy component only in that earliest sliver of time, they found the number of galaxies that arose from the primordial environment bloomed to fit astronomers’ observations.

“ You have these two looming open-ended puzzles,” says study co-author Rohan Naidu , a postdoc in MIT’s Kavli Institute for Astrophysics and Space Research. “We find that in fact, early dark energy is a very elegant and sparse solution to two of the most pressing problems in cosmology.”

The study’s co-authors include lead author and Kavli postdoc Xuejian (Jacob) Shen, and MIT professor of physics Mark Vogelsberger , along with Michael Boylan-Kolchin at the University of Texas at Austin, and Sandro Tacchella at the University of Cambridge.

Big city lights

Based on standard cosmological and galaxy formation models, the universe should have taken its time spinning up the first galaxies. It would have taken billions of years for primordial gas to coalesce into galaxies as large and bright as the Milky Way.

But in 2023, NASA’s James Webb Space Telescope (JWST) made a startling observation. With an ability to peer farther back in time than any observatory to date, the telescope uncovered a surprising number of bright galaxies as large as the modern Milky Way within the first 500 million years, when the universe was just 3 percent of its current age.

“The bright galaxies that JWST saw would be like seeing a clustering of lights around big cities, whereas theory predicts something like the light around more rural settings like Yellowstone National Park,” Shen says. “And we don’t expect that clustering of light so early on.”

For physicists, the observations imply that there is either something fundamentally wrong with the physics underlying the models or a missing ingredient in the early universe that scientists have not accounted for. The MIT team explored the possibility of the latter, and whether the missing ingredient might be early dark energy.

Physicists have proposed that early dark energy is a sort of antigravitational force that is turned on only at very early times. This force would counteract gravity’s inward pull and accelerate the early expansion of the universe, in a way that would resolve the mismatch in measurements. Early dark energy, therefore, is considered the most likely solution to the Hubble tension.

Galaxy skeleton

The MIT team explored whether early dark energy could also be the key to explaining the unexpected population of large, bright galaxies detected by JWST. In their new study, the physicists considered how early dark energy might affect the early structure of the universe that gave rise to the first galaxies. They focused on the formation of dark matter halos — regions of space where gravity happens to be stronger, and where matter begins to accumulate.

“We believe that dark matter halos are the invisible skeleton of the universe,” Shen explains. “Dark matter structures form first, and then galaxies form within these structures. So, we expect the number of bright galaxies should be proportional to the number of big dark matter halos.”

The team developed an empirical framework for early galaxy formation, which predicts the number, luminosity, and size of galaxies that should form in the early universe, given some measures of “cosmological parameters.” Cosmological parameters are the basic ingredients, or mathematical terms, that describe the evolution of the universe.

Physicists have determined that there are at least six main cosmological parameters, one of which is the Hubble constant — a term that describes the universe’s rate of expansion. Other parameters describe density fluctuations in the primordial soup, immediately after the Big Bang, from which dark matter halos eventually form.

The MIT team reasoned that if early dark energy affects the universe’s early expansion rate, in a way that resolves the Hubble tension, then it could affect the balance of the other cosmological parameters, in a way that might increase the number of bright galaxies that appear at early times. To test their theory, they incorporated a model of early dark energy (the same one that happens to resolve the Hubble tension) into an empirical galaxy formation framework to see how the earliest dark matter structures evolve and give rise to the first galaxies.

“What we show is, the skeletal structure of the early universe is altered in a subtle way where the amplitude of fluctuations goes up, and you get bigger halos, and brighter galaxies that are in place at earlier times, more so than in our more vanilla models,” Naidu says. “It means things were more abundant, and more clustered in the early universe.”

“A priori, I would not have expected the abundance of JWST’s early bright galaxies to have anything to do with early dark energy, but their observation that EDE pushes cosmological parameters in a direction that boosts the early-galaxy abundance is interesting,” says Marc Kamionkowski, professor of theoretical physics at Johns Hopkins University, who was not involved with the study. “I think more work will need to be done to establish a link between early galaxies and EDE, but regardless of how things turn out, it’s a clever — and hopefully ultimately fruitful — thing to try.”

“ We demonstrated the potential of early dark energy as a unified solution to the two major issues faced by cosmology. This might be an evidence for its existence if the observational findings of JWST get further consolidated,” Vogelsberger concludes. “In the future, we can incorporate this into large cosmological simulations to see what detailed predictions we get.”

This research was supported, in part, by NASA and the National Science Foundation.

• Paper: “Early Galaxies and Early Dark Energy: A Unified Solution to the Hubble Tension and Puzzles of Massive Bright Galaxies revealed by JWST”

Related News

Novel Architecture Makes Neural Networks More Understandable

Atoms on the edge

Study predicts a new quantum anomalous crystal in fractionally filled moiré superlattices

Prof Vijay Subramanian awarded $7.5M MURI to rethink game theory in dynamic environments

The interactions of today’s world are increasingly complex, as humans regularly interface with semi- and fully-autonomous artificial intelligence (AI) systems. Michigan is taking the lead on improving our understanding, and predicting the outcomes, of these interactions through a $7.5M, five year Multidisciplinary University Research Initiatives (MURI) called New Game Theory for New Agents: Foundations and Learning Algorithms for Decision-Making Mixed-Agents.

“There are lots of different agents that are interacting, including the usual players––humans––which could be big entities, like corporations, governments, or other institutions,” explained Vijay Subramanian , professor of Electrical and Computer Engineering and project director. “But in today’s world, we have these new AI agents as well. What we want to understand is: how do these computational agents interact?”

Game theory models how individuals strategize and make decisions, either collaboratively or competitively. Each player should attempt to maximize their progress toward an individual or shared goal, using the information available to them. This information could include the rules of the game––such as in poker, Go, or trading in the stock market––as well as any knowledge about the other players’ goals or intentions. When none of the players can improve their outcomes by changing their decisions alone, the game has reached a state called equilibrium.

Over several decades, researchers in economics, mathematics, computer science,  engineering, and even biology have developed game theory to predict the outcomes and equilibria of various scenarios. Now, AI systems are overtaking humans in their ability to quickly handle and process huge amounts of data, adding an element of the unknown into these assessments.

Our goal is to transcend existing theory and develop new theory that can address this mixture of autonomous, semi-autonomous, algorithmic, and human agents. Vijay Subramanian

“The existing theory makes very stringent assumptions on the computing or reasoning capabilities of agents––and the AI agents that I mentioned need not have all of those,” said Subramanian. “Our goal is to transcend that and develop new theory that can address this mixture of autonomous, semi-autonomous, algorithmic, and human agents.”

If the research team can predict the outcomes of interactions that involve AI agents, they can design environments and projects to be carried out more efficiently and accurately.

One real-world example of a scenario that would benefit from this type of analysis is the rescue and cleanup operations in a disaster zone––say, after an earthquake or airstrike. In a modern disaster zone, humans may work together with robots to clear debris from the area and provide medical care to injured survivors. 

“In this case, those robots are AI agents, but they get signals from and have to follow the humans. And the humans have to react to these agents as well,” Subramanian said. “It’s important to understand how such systems would perform and come up with an algorithm to get the system to achieve your goals.”

“You will have some agents that are more capable and some that are less capable,” he added, “Can the more capable agents direct the systems toward achieving their objectives more often?”

In addition to the complexities introduced by the presence of multiple types of agents, the players must anticipate or react to any environmental changes produced by their actions. For example, in the context of rescue and cleanup operations in a disaster zone, as the area is cleared, it may become easier for humans and robots to move around; conversely, further obstacles could be created by falling debris that restricts movement or alters the number of workers.

These types of complex scenarios have presented challenges to existing game theory. Subramanian’s team aims to bring together the many years of game theory development that incorporate dynamic settings with the mixed capabilities of today’s AI agents.

Other examples of modern multi-agent systems include combatting poachers ; assessing the likelihood of and thereafter preventing systemic failures in the financial system, like the Great Depression (1930s) and the Great Recession (2000s); and deploying fleets of automated cars.

“We are thinking of the methodology being composed of three core components,” Subramanian said, “Agents have to form the models of each other, the environment, and themselves. Based on that, we have to create algorithms that estimate those models and make decisions. And thereafter, we must understand what the outcomes result in. These three things together predict equilibria––their interplay will determine what happens in the game.”

These steps happen in a loop, helping the researchers predict the outcomes of their modeled scenarios. If the outcome doesn’t satisfy their goals, they can change the algorithms, the communication between agents, or the incentives to direct the result toward preferred configurations.

The research will be conducted with MURI collaborators Dirk Bergemann (Yale University), Avrim Blum (Toyota Technological Institute Chicago), Rahul Jain (University of Southern California), Elchanan Mossel (Massachusetts Institute of Technology), Milind Tambe (Harvard University), Omer Tamuz (California Institute of Technology), and Eva Tardos (Cornell University).

New Research Debunks Violent “Steppe” Invasion Theory in Iberia 4,200 Years Ago

A new study disputes the theory that Steppe warriors violently replaced Iberian males 4,200 years ago, suggesting instead a peaceful integration with an already weakened local population.

A new study from the Universitat Autònoma de Barcelona and University of Murcia challenges the idea that warrior groups with ‘Steppe’ ancestry violently replaced Iberian Peninsula males 4,200 years ago. Instead, it suggests these groups mixed with already demographically weakened local populations, presenting a more nuanced view of the genetic shifts

In the paper, published in the Journal of Archaeological Science: Reports , the research team explored how society and populations changed in southeast Spain 4,200 years ago, during the transition from the Copper to the Bronze Age. To this end, they focused on one of the best-known aspects of this transition: the shift from communal burials in the Copper Age to the single and double tombs of the Bronze Age El Argar society. The team looked at a large sample of radiocarbon (C14) dates from human bones discovered in these different types of graves.

Key Findings: A Demographic Shift

The first result of the analysis is chronological and suggests that the change from communal to individual tombs happened quickly. But it is the second result that has arguably greater implications. By examining a large sample of radiocarbon dates from human remains in southeast Iberia, they observed a peak in the number of buried dead during 2550-2400 BCE, followed by a sudden drop in 2300-2250 BCE.

The authors interpret this data from a demographic perspective. “It is likely that the inhabitants of southeastern Iberia were already very few, around 4,300 or 4,200 years ago, just before the arrival of populations with new genetic components, labeled ‘Steppe’. When individuals with Steppe ancestry were found in southeastern Iberia, around 2200-2000 BC, they simply mixed with small local groups or occupied uninhabited areas,” says Rafael Micó, professor at the UAB and co-director of the Mediterranean Social Archaeoecology Research Group (ASOME-UAB), which carried out the study.

Along with these results, the team also cites previous archaeogenetic studies that point to the absence of a ‘male bias’ among peninsular groups with Steppe ancestry. “This allows us to propose a different historical scenario, which does not contemplate invading hordes of ‘Steppe’ warriors who would have annihilated the local men and formed a male elite with exclusive access to local women,” says Cristina Rihuete Herrada, also a professor at the UAB and co-author of the study.

A period of abrupt change, but with a progressive ‘steppe’ genetic influence

4,200 years ago, between the Late Copper Age and the Early Bronze Age, major social disruptions occurred in Central and Western Europe. Archaeologists are still debating over their exact sources, and explanations range from drought to large-scale violent migrations or the spread of contagious diseases.

“In recent years it has been argued that populations with what is known as ‘Steppe ancestry’ migrated westwards from the region around the Black Sea, aided by the horse and wheel as new technologies, and brutally raided Western Europe,” explains Camila Oliart, UAB researcher and co-author of the study. “In the case of Iberia, it has been suggested that men arriving from the East had preferential access to women and discriminated or eliminated local males, in what is a very impactful ‘invasionist’ interpretation in the media, but perhaps also a too hasty one.”

A Decline in Local Societies Before Steppe Arrival

In the study now published, the research team outlines a context that may have important implications for understanding the transition between the Chalcolithic and the Bronze Age in south Iberia 4,200 years ago, and in the southeast in particular. Over the two centuries prior to this date, the social landscape may have been quite far from that of a thriving Copper Age. It was probably characterised by smaller settlements and low population density. From this perspective, the ‘collapse’ of the Copper Age 4,200 years ago was not a rapid, massive, and disruptive event affecting a densely populated and powerful society, but the culmination of two centuries of declining local dynamics.

This new scenario does not involve the mass elimination of men or the subjugation of local women after an alleged conquest, as the study points out. “The inhabitants of southern Iberia were already few in number at the end of the Copper Age and mixed with groups of Steppe genetic ancestry without the need for a large-scale invasion. We should start to consider alternative explanations,” suggests Miguel Valério, UAB researcher and co-author of the study. “We cannot ignore the fact that violence was an ingredient of social life in the Copper Age, but so far nothing proves that its end was the consequence of a generalized conflict between genetically distinct populations.”

Still, the team emphasizes that more high-precision radiocarbon dating and genetic analysis on human samples from the latest Copper Age and the earliest Bronze Age (El Argar) burials are needed. “Such data is absolutely crucial to gain a better understanding of the nature, scale, and pace of the changes taking place in the formation of Bronze Age societies,” they concluded.

To carry out the study, some 450 radiocarbon dates corresponding to individuals buried in tombs from the Copper Age and Early Bronze Age in Almería (La Atalaya, Las Churuletas, Los Millares, El Argar, El Barranquete, Fuente Álamo, Gatas, Llano del Jautón, Loma del Campo and Loma de Belmonte), Murcia (Camino del Molino), Granada (Cerro de la Virgen, Panoría), Jaén (Marroquíes Bajos), Seville (Valencina de la Concepción), and Évora (Perdigões) were analyzed.

Reference: “Tracing social disruptions over time using radiocarbon datasets: Copper and Early Bronze Ages in Southeast Iberia” by Rafael Micó, Eva Celdrán Beltrán, Joaquín Lomba Maurandi, Camila Oliart Caravatti, Cristina Rihuete Herrada and Miguel Valério, 8 September 2024, Journal of Archaeological Science: Reports . DOI: 10.1016/j.jasrep.2024.104692

Along with Rafael Micó, Camila Oliart, Cristina Rihuete Herrada and Miguel Valério, researchers at ASOME-UAB, the authorship of the study also includes Eva Celdrán Beltrán and Joaquín Lomba Maurandi, from the University of Murcia.

Related Articles

7,000-year-old deer bones reveal secrets of ancient neolithic tool manufacturing, 7,500-year-old pottery reveals secrets of ancient european diets, 7,200 years ago: how neolithic shepherds navigated the complexities of early agriculture, rewriting history: new discoveries challenge assumptions about pre-agricultural societies in europe, how prehistoric humans adapted to intense climate change – revealed by marine mollusk shells, million-year-old ash in south african cave yields evidence of cooking, exposing 8,000 years of civilization with satellites, humans implicated in africa’s deforestation 3,000 years ago, archaeologists inch closer to understanding stonehenge.

Save my name, email, and website in this browser for the next time I comment.

Type above and press Enter to search. Press Esc to cancel.

Study: Early dark energy could resolve cosmology’s two biggest puzzles

A new study by MIT physicists proposes that a mysterious force known as early dark energy could solve two of the biggest puzzles in cosmology and fill in some major gaps in our understanding of how the early universe evolved.

One puzzle in question is the “Hubble tension,” which refers to a mismatch in measurements of how fast the universe is expanding. The other involves observations of numerous early, bright galaxies that existed at a time when the early universe should have been much less populated.

Now, the MIT team has found that both puzzles could be resolved if the early universe had one extra, fleeting ingredient: early dark energy. Dark energy is an unknown form of energy that physicists suspect is driving the expansion of the universe today. Early dark energy is a similar, hypothetical phenomenon that may have made only a brief appearance, influencing the expansion of the universe in its first moments before disappearing entirely.

Some physicists have suspected that early dark energy could be the key to solving the Hubble tension, as the mysterious force could accelerate the early expansion of the universe by an amount that would resolve the measurement mismatch.

The MIT researchers have now found that early dark energy could also explain the baffling number of bright galaxies that astronomers have observed in the early universe. In their new study, reported today in the Monthly Notices of the Royal Astronomical Society , the team modeled the formation of galaxies in the universe’s first few hundred million years. When they incorporated a dark energy component only in that earliest sliver of time, they found the number of galaxies that arose from the primordial environment bloomed to fit astronomers’ observations.

“ You have these two looming open-ended puzzles,” says study co-author Rohan Naidu, a postdoc in MIT’s Kavli Institute for Astrophysics and Space Research. “We find that in fact, early dark energy is a very elegant and sparse solution to two of the most pressing problems in cosmology.”

The study’s co-authors include lead author and Kavli postdoc Xuejian (Jacob) Shen, and MIT professor of physics Mark Vogelsberger, along with Michael Boylan-Kolchin at the University of Texas at Austin, and Sandro Tacchella at the University of Cambridge.

Big city lights

Based on standard cosmological and galaxy formation models, the universe should have taken its time spinning up the first galaxies. It would have taken billions of years for primordial gas to coalesce into galaxies as large and bright as the Milky Way.

But in 2023, NASA’s James Webb Space Telescope (JWST) made a startling observation. With an ability to peer farther back in time than any observatory to date, the telescope uncovered a surprising number of bright galaxies as large as the modern Milky Way within the first 500 million years, when the universe was just 3 percent of its current age.

“The bright galaxies that JWST saw would be like seeing a clustering of lights around big cities, whereas theory predicts something like the light around more rural settings like Yellowstone National Park,” Shen says. “And we don’t expect that clustering of light so early on.”

For physicists, the observations imply that there is either something fundamentally wrong with the physics underlying the models or a missing ingredient in the early universe that scientists have not accounted for. The MIT team explored the possibility of the latter, and whether the missing ingredient might be early dark energy.

Physicists have proposed that early dark energy is a sort of antigravitational force that is turned on only at very early times. This force would counteract gravity’s inward pull and accelerate the early expansion of the universe, in a way that would resolve the mismatch in measurements. Early dark energy, therefore, is considered the most likely solution to the Hubble tension.

Galaxy skeleton

The MIT team explored whether early dark energy could also be the key to explaining the unexpected population of large, bright galaxies detected by JWST. In their new study, the physicists considered how early dark energy might affect the early structure of the universe that gave rise to the first galaxies. They focused on the formation of dark matter halos — regions of space where gravity happens to be stronger, and where matter begins to accumulate.

“We believe that dark matter halos are the invisible skeleton of the universe,” Shen explains. “Dark matter structures form first, and then galaxies form within these structures. So, we expect the number of bright galaxies should be proportional to the number of big dark matter halos.”

The team developed an empirical framework for early galaxy formation, which predicts the number, luminosity, and size of galaxies that should form in the early universe, given some measures of “cosmological parameters.” Cosmological parameters are the basic ingredients, or mathematical terms, that describe the evolution of the universe.

Physicists have determined that there are at least six main cosmological parameters, one of which is the Hubble constant — a term that describes the universe’s rate of expansion. Other parameters describe density fluctuations in the primordial soup, immediately after the Big Bang, from which dark matter halos eventually form.

The MIT team reasoned that if early dark energy affects the universe’s early expansion rate, in a way that resolves the Hubble tension, then it could affect the balance of the other cosmological parameters, in a way that might increase the number of bright galaxies that appear at early times. To test their theory, they incorporated a model of early dark energy (the same one that happens to resolve the Hubble tension) into an empirical galaxy formation framework to see how the earliest dark matter structures evolve and give rise to the first galaxies.

“What we show is, the skeletal structure of the early universe is altered in a subtle way where the amplitude of fluctuations goes up, and you get bigger halos, and brighter galaxies that are in place at earlier times, more so than in our more vanilla models,” Naidu says. “It means things were more abundant, and more clustered in the early universe.”

“A priori, I would not have expected the abundance of JWST’s early bright galaxies to have anything to do with early dark energy, but their observation that EDE pushes cosmological parameters in a direction that boosts the early-galaxy abundance is interesting,” says Marc Kamionkowski, professor of theoretical physics at Johns Hopkins University, who was not involved with the study. “I think more work will need to be done to establish a link between early galaxies and EDE, but regardless of how things turn out, it’s a clever — and hopefully ultimately fruitful — thing to try.”

“ We demonstrated the potential of early dark energy as a unified solution to the two major issues faced by cosmology. This might be an evidence for its existence if the observational findings of JWST get further consolidated,” Vogelsberger concludes. “In the future, we can incorporate this into large cosmological simulations to see what detailed predictions we get.”

This research was supported, in part, by NASA and the National Science Foundation.

• Kansas State University

Status of ITS resources

• K-State home
• » K-State News
• » Sept. 2024
• » Shamir publishes study supporting century-old theory that challenges Big Bang

K-State News

• K-State Today
• Seek research magazine
• Graduation/honors lists

K-State News Kansas State University 128 Dole Hall 1525 Mid-Campus Dr North Manhattan, KS 66506

785-532-2535 [email protected]

Shamir publishes study supporting century-old theory that challenges Big Bang

Wednesday, Sept. 11, 2024

Lior Shamir, associate professor of computer science, has published findings from an observational study that support the century-old "Tired Light" theory. 

MANHATTAN — A Kansas State University engineer recently published results from an observational study in support of a century-old theory that directly challenges the validity of the Big Bang theory. Lior Shamir , associate professor of computer science, used imaging from a trio of telescopes and more than 30,000 galaxies to measure the redshift of galaxies based on their distance from Earth. Redshift is the change in the frequency of light waves that a galaxy emits, which astronomers use to gauge a galaxy's speed. Shamir's findings lend support to the century-old "Tired Light" theory instead of the Big Bang. "In the 1920s, Edwin Hubble and George Lemaitre discovered that the more distant the galaxy is, the faster it moves away from Earth," Shamir said. "That discovery led to the Big Bang theory, suggesting that the universe started to expand around 13.8 billion years ago. At around the same time, preeminent astronomer Fritz Zwicky proposed that galaxies that were more distant from Earth did not really move faster." Zwicky's contention was that the redshift observed from Earth is not because the galaxies move but because the light photons lose their energy as they travel through space. The longer the light travels, the more energy it loses, leading to an illusion that galaxies that are more distant from Earth also move faster. "The Tired Light theory was largely neglected, as astronomers adopted the Big Bang theory as the consensus model of the universe," Shamir said. "But the confidence of some astronomers in the Big Bang theory started to weaken when the powerful James Webb Space Telescope saw first light. The JWST provided deep images of the very early universe, but instead of showing an infant early universe as astronomers expected, it showed large and mature galaxies. If the Big Bang happened as scientists initially believed, these galaxies are older than the universe itself." While new imaging casts doubt on the Big Bang, Shamir's study used the constant rotational velocity of the Earth around the center of the Milky Way to examine the redshift of galaxies that move in different velocities relative to Earth and to test how the change in the redshift responds to the change in velocity. "The results showed that galaxies that rotate in the opposite direction relative to the Milky Way have lower redshift compared to galaxies that rotate in the same direction relative to the Milky Way," Shamir said. "That difference reflects the motion of the Earth as it rotates with the Milky Way. But the results also showed that the difference in the redshift increased when the galaxies were more distant from Earth. "Because the rotational velocity of the Earth relative to the galaxies is constant, the reason for the difference can be the distance of the galaxies from Earth. That shows that the redshift of galaxies changes with the distance, which is what Zwicky predicted in his Tired Light theory." Shamir's research was recently published in Particles , a quarterly international, open-access, peer-reviewed journal covering all aspects of nuclear physics, particle physics and astrophysics science.

Media contact

Division of Communications and Marketing 785-532-2535 [email protected]

" An Empirical Consistent Redshift Bias: A Possible Direct Observation of Zwicky’s TL Theory "

Grant Guggisberg 785-532-6715 [email protected]

• Statements and disclosures
• Accessibility

• Manhattan, KS 66506
• 785-532-6011
• © Kansas State University
• Updated: 9/11/24

Cornell Chronicle

• Architecture & Design
• Arts & Humanities
• Business, Economics & Entrepreneurship
• Computing & Information Sciences
• Energy, Environment & Sustainability
• Food & Agriculture
• Global Reach
• Health, Nutrition & Medicine
• Law, Government & Public Policy
• Life Sciences & Veterinary Medicine
• Physical Sciences & Engineering
• Social & Behavioral Sciences
• Coronavirus
• News & Events
• Public Engagement
• New York City
• Photos of the Week
• Big Red Sports
• Freedom of Expression
• Student Life
• University Statements
• Around Cornell
• All Stories
• In the News
• Expert Quotes
• Cornellians

AI succeeds in combatting conspiracy theories

Arguing with a conspiracy theorist that the moon landing wasn’t staged is usually a futile effort, but ChatGPT might have better luck, according to new research by Cornell, American University and Massachusetts Institute of Technology psychologists.

In a new paper, “Durably Reducing Conspiracy Beliefs Through Dialogues With AI,” which publishes Sept. 13 in Science ,  the researchers show that conversations with large language models can effectively reduce individuals’ belief in conspiracy theories – and that these reductions last for at least two months – a finding that offers new insights into the psychological mechanisms behind the phenomenon as well as potential tools to fight conspiracies’ spread.

The persistence of conspiracy theories in the face of counter-evidence has led many researchers to conclude that they fulfill deep-seated psychological needs, rendering them impervious to facts and logic. But for researchers Thomas Costello, lead author and assistant professor at American University; David Rand, MIT Sloan School of Management professor; and  Gordon Pennycook , associate professor of psychology and Himan Brown Faculty Fellow in Cornell’s College of Arts and Sciences, who have conducted extensive research on the spread and uptake of misinformation, that conclusion didn’t ring true. Instead, they suspected a simpler explanation was at play – that people just hadn’t been exposed to convincing-enough evidence.

Effectively debunking conspiracy theories, in other words, would require two things: personalized arguments and access to vast quantities of information – both now readily available through generative AI. 

To test their theory, Pennycook, Costello and Rand harnessed the power of GPT-4 Turbo, OpenAI’s most advanced large language model, to engage more than 2,000 conspiracy believers in personalized, evidence-based dialogues. Participants were first asked to identify and describe a conspiracy theory they believed in using their own words, along with the evidence supporting their belief.

GPT-4 Turbo then used this information to generate a personalized summary of the participant's belief and initiate a dialogue. The AI was instructed to persuade users that their beliefs were untrue, adapting its strategy based on each participant’s unique arguments and evidence.

These conversations, lasting an average of 8.4 minutes, allowed the AI to directly address and refute the specific evidence supporting each individual’s conspiratorial beliefs, an approach that was impossible to test at scale prior to the technology’s development. 

The results of the intervention were striking. On average, the AI conversations reduced the average participant’s belief in a chosen conspiracy theory by about 20%, and about one in four participants – all of whom believed the conspiracy beforehand – disavowed the conspiracy after the conversation.

The AI conversation’s effectiveness was not limited to specific types of conspiracy theories. It successfully challenged beliefs across a wide spectrum, including conspiracies that potentially hold strong political and social salience, like those involving COVID-19 and fraud during the 2020 election. 

“This research indicates that evidence matters much more than we thought it did – so long as it is actually related to people’s beliefs,” Pennycook said. “This has implications far beyond just conspiracy theories: Any number of beliefs based on poor evidence could, in theory, be undermined using this approach.”

While the intervention was less successful among participants who reported that the conspiracy was central to their worldview, it did still have an impact, with little variance across demographic groups. 

“I was quite surprised at first, but reading through the conversations made much me less skeptical. The AI provided page-long, highly detailed accounts of why the given conspiracy was false in each round of conversation -- and was also adept at being amiable and building rapport with the participants,” Costello said.

Notably, the impact of the AI dialogues extended beyond mere changes in belief. Participants also demonstrated shifts in their behavioral intentions related to conspiracy theories. They reported being more likely to unfollow people espousing conspiracy theories online, and more willing to engage in conversations challenging those conspiratorial beliefs.

The researchers note the need for continued responsible AI deployment, since the technology could potentially be used to convince users to believe in conspiracies as well as to abandon them.

Nevertheless, the potential for positive applications of AI to reduce belief in conspiracies is significant. For example, AI tools could be integrated into search engines to offer accurate information to users searching for conspiracy-related terms. 

“Although much ink has been spilled over the potential for generative AI to supercharge disinformation, our study shows that it can also be part of the solution,” Rand said. “Large language models like GPT-4 have the potential to counter conspiracies at a massive scale.”

The team has developed a website where visitors can try out the software.

Media Contact

Becka bowyer.

Get Cornell news delivered right to your inbox.

You might also like

Gallery Heading

Trump attends 9/11 event with a far-right activist who pushed a conspiracy theory about the tragedy

Donald Trump attended a Sept. 11 remembrance Wednesday alongside a far-right activist who has pushed a false narrative that the terrorist attack was "an inside job."

Laura Loomer, a Trump ally , joined him at a fire station in lower Manhattan, where he and his running mate, Sen. JD Vance of Ohio, met and took photos with firefighters. Some members of the station died responding to the attacks on Sept. 11, 2001.

Throughout the day Wednesday, Loomer posted videos to social media documenting Trump's visits to commemorate 9/11. She also attended the presidential debate Tuesday night and traveled on Trump’s plane.

Just last year, Loomer posted a lengthy message on X that included a video that falsely said that "9/11 was an Inside Job!"

Loomer did not immediately respond to a request for comment Wednesday night about her stop with Trump and whether her views on 9/11 have changed.

In a lengthy post on X after NBC News asked for comment, Loomer said, "To the many reporters who are calling me and obsessively asking me to talk to them today, the answer is no."

"I’m not sure why this is so hard for people to understand, but I believe in unconditional loyalty to those who are deserving," she said in her post. "And there is nobody more deserving of our loyalty and unwavering support than Donald Trump."

NBC News asked the Trump campaign for details about the decision to have Loomer travel with Trump, the extent of her conversations with him and whether the campaign knew she has promoted 9/11 conspiracy theories before it invited her to accompany Trump.

The campaign did not respond, instead providing a statement about Trump's 9/11 commemoration.

"Today, President Trump put politics aside and stood beside Kamala Harris and Joe Biden to honor those who lost their lives during the worst terrorist attack in our nation’s history," a campaign official said. "The day wasn’t about anyone other than the souls who are no longer with us, their families, and the heroes who courageously stepped up to save their fellow Americans on that fateful day."

The Harris campaign declined to comment.

Loomer has a history of pushing a variety of conspiracy theories. She said last week that school shootings have been allowed to happen to help Democrats. She has pushed conspiracy theories about undocumented immigrants' registering to vote in November, and she touted the baseless story about Haitian immigrants' eating cats. She has falsely said Taylor Swift and Travis Kelce are in an "arranged relationship" to "influence the 2024 election," among other mistruths.

Loomer's closeness to Trump doesn't appear to be sitting well with all his allies.

On Wednesday, Rep. Marjorie Taylor Greene, R-Ga., called for Loomer to delete an “extremely racist” post targeting Vice President Kamala Harris.

“This is appalling and extremely racist. It does not represent who we are as Republicans or MAGA,” Greene, a fellow Trump ally, said Wednesday night on X . “This does not represent President Trump. This type of behavior should not be tolerated ever. @LauraLoomer should take this down.”

In a post on Sunday, Loomer launched into a racist tirade against Harris, saying that if she won, “the White House will smell like curry & White House speeches will be facilitated via a call center and the American people will only be able to convey their feedback through a customer satisfaction survey at the end of the call that nobody will understand.”

Harris is the first person of Indian descent to be a major party’s nominee for president. She is also the first Black woman to be a major party’s nominee.

Following Greene’s tweet, Loomer attacked her in a series of posts, adding that she would not delete the initial post.

Greene has promoted antisemitic conspiracy theories and made controversial remarks about the Holocaust, comparing mask mandates to the Holocaust before apologizing. She also previously expressed support for the far-right conspiracy theory QAnon and been criticized for a series of racist comments. She has used derogatory language to refer to George Floyd, a Black man killed by Minneapolis police, calling him "Convicted Felon George Floyd" and saying Democrats were "worshiping" him.

Megan Lebowitz is a politics reporter for NBC News.

Vaughn Hillyard is a correspondent for NBC News. 

Alec Hernández is a 2024 NBC News campaign embed.

AI Conversations Help Conspiracy Theorists Change Their Views

Summary: AI-powered conversations can reduce belief in conspiracy theories by 20%. Researchers found that AI provided tailored, fact-based rebuttals to participants’ conspiracy claims, leading to a lasting change in their beliefs.

In one out of four cases, participants disavowed the conspiracy entirely. The study suggests that AI has the potential to combat misinformation by engaging people directly and personally.

Key Facts :

• AI reduced conspiracy beliefs by 20% after short conversations.
• One in four participants fully disavowed the conspiracy theory they believed in.
• The effect persisted for two months after the AI conversation.

Source: American University

‘They’re so far down the rabbit hole of conspiracy theories that they’re lost for good’ is common thinking when it comes to conspiracy theorists. This generally accepted notion is now crumbling.

In a pathbreaking research study, a team of researchers from American University, Massachusetts Institute of Technology and Cornell University show that conspiracy theorists changed their views after short conversations with artificial intelligence.

Study participants believing some of the most deeply entrenched conspiracies, including those about the COVID-19 pandemic and fraud in the 2020 U.S. presidential election, showed large and lasting reductions in conspiracy belief following the conversations. Stoked by polarization in politics and fed by misinformation and social media, conspiracy theories are a major issue of public concern. They often serve as a wedge between theorists and their friends and family members. 

YouGov survey results from last December show that large shares of Americans believe various conspiratorial falsehoods.

In the field of psychology, the widespread view the findings challenge is that conspiracy theorists adhere to their beliefs because of the significance to their identities, and because the beliefs resonate with underlying drives and motivations, says Thomas Costello, assistant professor of psychology at American University and lead author of the new study published in the journal  Science .

In fact, most approaches have focused on preventing people from believing conspiracies in the first place.

“Many conspiracy believers were indeed willing to update their views when presented with compelling counterevidence,” Costello said.

“I was quite surprised at first, but reading through the conversations made much me less skeptical. The AI provided page-long, highly detailed accounts of why the given conspiracy was false in each round of conversation — and was also adept at being amiable and building rapport with the participants.”

More than 2,000 self-identified conspiracy believers participated in the study. The AI conversations reduced the average participant’s belief in their chosen conspiracy theory by about 20 percent, and about 1 in 4 participants — all of whom believed the conspiracy beforehand — disavowed the conspiracy after the conversation. 

Until now, delivering persuasive, factual messages to a large sample of conspiracy theorists in a lab experiment has proved challenging. For one, conspiracy theorists are often highly knowledgeable about the conspiracy—often more so than skeptics. Conspiracies also vary widely, such that evidence backing a particular theory can differ from one believer to another. 

AI as an intervention

The new study comes as society debates the promise and peril of AI. Large language models driving generative AI are powerful reservoirs of knowledge. Researchers emphasize that the study demonstrates one way that these reservoirs of knowledge can be used for good: by helping people have more accurate beliefs.

The ability of artificial intelligence to connect across diverse topics of information within seconds makes it possible to tailor counterarguments to specific conspiracies of a believer in ways that aren’t possible for a human to do. 

“ Previous efforts to debunk dubious beliefs have a major limitation: One needs to guess what people’s actual beliefs are in order to debunk them – not a simple task,” said Gordon Pennycook, associate professor of psychology at Cornell University and a paper co-author.

“In contrast, the AI can respond directly to people’s specific arguments using strong counterevidence. This provides a unique opportunity to test just how responsive people are to counterevidence.”

Researchers designed the chatbot to be highly persuasive and engage participants in such tailored dialogues. GPT-4, the AI model powering ChatGPT, provided factual rebuttals to participants’ conspiratorial claims. In two separate experiments, participants were asked to describe a conspiracy theory they believe in and provide evidence to support.

Participants then engaged in a conversation with an AI. The AI’s goal was to challenge beliefs by addressing specific evidence. In a control group, participants discussed an unrelated topic with the AI.

To tailor the conversations, researchers provided the AI with participants’ initial statement of belief and the rationale. This setup allowed for a more natural dialogue, with the AI directly addressing a participant’s claims. The conversation averaged 8.4 of the participant’s minutes and involved three rounds of interaction, excluding the initial setup.

Ultimately, both experiments showed a reduction in participants’ beliefs in conspiracy theories. When the researchers assessed participants two months later, they found that the effect persisted.

While the results are promising and suggest a future in which AI can play a role in diminishing conspiracy belief when used responsibly, further studies on long-term effects, using different AI models, and practical applications outside of a laboratory setting will be needed.   

“Although much ink has been spilled over the potential for generative AI to supercharge disinformation, our study shows that it can also be part of the solution,” said David Rand, a paper co-author and MIT Sloan School of Management professor. “Large language models like GPT4 have the potential to counter conspiracies at a massive scale.”

Additionally, members of the public interested in this ongoing work can visit a  website  and try out the intervention for themselves.

About this AI and psychology research news

Author: Rebecca Basu Source: American University Contact: Rebecca Basu – American University Image: The image is credited to Neuroscience News

Original Research: The findings will appear in Science

Manipulating Brain Waves During Sleep With Sound

High Doses of ADHD Meds Linked to Increased Psychosis Risk

Key Neurons Found to Predict Memory of People and Places