Confirmation Bias and How It Shapes What We Believe

Human thinking can be modified and influenced by a range of deviations and distortions. Systematic, repeated errors in our everyday mental processes are known as cognitive shortcuts, distortions, or biases. They lead to flawed conclusions about the information we process, and cause a whole range of errors in decision-making and argumentation. [1] There are several causes behind them, tied to how data and information are created in the first place, and then to how we perceive, process, and turn them into knowledge, along with other processes of information behaviour. [2] Emotional, ethical, normative, and other influences play a role too. People are often unaware of these errors in their own judgement. [1]

One of these errors is so-called confirmation bias. Someone under its influence favours, at every stage of their information behaviour, the information that confirms their existing mental model, opinion, or belief. Information that conflicts with their view of the world, on the other hand, gets avoided, is remembered less accurately, gets interpreted differently from its original meaning, or the person simply isn’t willing to share it further. [3]

Illustration by Mike Pick (Creative Commons)

The brain already behaves selectively at several stages of thinking. Because of the phenomenon of selective attention, a person only pays attention to communication they consider acceptable given their existing mental model. Selective perception then means they only decode what’s consistent with their beliefs. In other words: we see what we want to see. [4] Selective perception was demonstrated back in the 1950s by a case study in which spectators at an American college football match “saw” the game differently depending on which side they supported. [5] Remembering the information we’ve taken in works similarly. The phenomenon of selective retention means a person recalls what fit their pre-existing concept, and forgets the rest. [6]

In research published in the journal Nature, participants were shown evidence that contradicted their political beliefs. The part of the brain associated with perceiving physical pain became more active when they encountered these opposing views. [7] That points to the fact that it’s harder for the brain to accept opinions it disagrees with. Confirmation bias is therefore closely tied to so-called cognitive dissonance.

Cognitive psychologist Peter Cathcart Wason was the first to introduce the concept of confirmation bias, in 1960. His study showed that most people test hypotheses incorrectly — they don’t try to disprove them. He gave participants a numerical sequence to continue (for example, 2–4–6). For every subsequent number they suggested, he’d answer yes or no depending on whether the sequence could continue that way. Most participants continued with 8–10–12, then 14–16–18. They always got positive feedback. After a few rounds, they became convinced they’d figured out the correct rule — whether that was a sequence of even/odd numbers, or some fixed gap between each number. In reality, though, any number higher than the previous one would have fit the rule. Most participants never discovered this, because they never tried to falsify their own hypothesis. Wason repeated a similar experiment a few more times over the course of the 1960s, and showed that people look for information that supports what they already believe. [8]

While Wason focused purely on cognition, another branch of researchers looked at how attitudes change or persist. According to Catherine A. Sanderson, confirmation bias helps reinforce stereotypes. People remember information that confirms a stereotype better. Information that contradicts it either gets ignored entirely, or is quickly forgotten. [9]

In the 1970s, a study at Stanford University tried to find out how confirmation bias affects how we evaluate information. Researchers selected an equal number of supporters and opponents of the death penalty and had them read different (fictional) comparative studies — one pointing to the positive effects of the death penalty, the other to negative ones. Participants were then asked to rate the studies. Death-penalty supporters rated the study that supported their view as more credible and better designed methodologically; opponents, in turn, preferred the study showing the negative effects. The study also found that after reading both papers, participants ended up even more strongly convinced of their original opinion. [10]

This phenomenon — where participants confronted with evidence supporting the opposite view hold on to their original opinion and become even more convinced of it — is known as the backfire effect. [11] The British fact-checking organisation Full Fact, however, considers the effect a myth, based on a review of seven experimental studies. Cases where the backfire effect was found tended to involve especially contentious topics, or cases where the fact being tested was objectively ambiguous to begin with. [12]

A study by Deanna Kuhn and Joseph Lao also challenges the Stanford study’s conclusion that exposure to mixed evidence on a topic leads to polarised attitudes. Based on their own research among university students, whom they asked about the death penalty in the same way as Lord et al. [10], they instead propose that the polarisation phenomenon involves two distinct patterns of change: a shift from a mildly held position to a more extreme one, and a shift from a neutral position to a mild one. Their data also supports the idea that cognitive engagement, rather than mere exposure to new evidence, is the effective part of such interventions. [13]

The core concept of confirmation bias itself — an inclination toward information consistent with our existing beliefs — is widely accepted and backed by plenty of research. Its weak point, besides the mixed results on polarisation, may be that it’s overly simplified, particularly from a psychological standpoint. The original concept comes from a cognitive paradigm, and doesn’t account for social factors, such as our relationship with the person delivering the information. Other factors also shape how we accept information, process it, or make decisions based on it — the environment the user is in, or their prior experience. [14] Klayman therefore recommends talking about a whole family of confirmation biases, rather than a single phenomenon. [15]

Although the concept of confirmation bias was originally the domain of psychology, it has since made its way into other fields too, such as political science or economics. Its ties to information behaviour are visible at various stages of interacting with information. Confirmation bias can be observed, for example, in information seeking, as mentioned in the section on Wason’s experiment, and it can similarly show up in information searching. Klayman describes this as a positive-testing strategy — the tendency to ask questions in a way that leads to finding data confirming the hypothesis. [14] Confirmation bias also affects other ways we work with information, such as evaluation and memory, as the studies mentioned above show. It’s also worth focusing today on sharing information on social media, avoiding information, and other elements of information behaviour.

Sources

[1] AMBROZOVÁ, Eva and Vratislav POKORNÝ. Kognitivní management: pro prostředí 4.0 a další (Cognitive management: for the 4.0 environment and beyond). 1st ed., Brno: Newton Academy, 2018. ISBN 978-80-87325-13-1. (in Czech)

[2] CEJPEK, Jiří. Informace, komunikace a myšlení: úvod do informační vědy (Information, communication, and thinking: an introduction to information science). 1st ed., Prague: Charles University, 2005. ISBN 978-80-246-1037-5. (in Czech)

[3] CASAD, Bettina. Confirmation bias [online]. Britannica: 2019. [accessed 15.03.2021]. Available from: https://www.britannica.com/science/confirmation-bias

[4] DVOŘÁK, Tomáš (ed.). Kapitoly z dějin a teorie médií (Chapters from the history and theory of media). Prague: Academy of Fine Arts in Prague, Research Department, 2010. ISBN 978-80-87108-16-8. (in Czech)

[5] HASTORF, Albert and Hadley CANTRIL. They saw a game: A case study. The Journal of Abnormal and Social Psychology, 1954, vol. 49, no. 1, pp. 129–134. DOI: https://doi.org/10.1037/h0057880 Also available from: https://www2.psych.ubc.ca/~schaller/Psyc590Readings/Hastorf1954.pdf

[6] APA Dictionary of Psychology. Selective retention [online]. American Psychological Association: 2020. [accessed 15.03.2021]. Available from: https://dictionary.apa.org/confirmation-bias

[7] KAPLAN, Jonas T., Sarah I. GIMBEL and Sam HARRIS. Neural correlates of maintaining one’s political beliefs in the face of counterevidence. Scientific Reports. 2016, vol. 6, no. 1, p. 39589. ISSN 2045-2322. DOI: https://doi.org/10.1038/srep39589

[8] OSVALD, Margarit and Stefan GROSJEAN. Confirmation bias. In: RÜDIGER, Pohl (ed.). Cognitive illusions: a handbook on fallacies and biases in thinking, judgement and memory. 1st ed. Hove: Psychology Press, 2004, pp. 79–85. ISBN 9781841693514.

[9] SANDERSON, Catherine A. Social Psychology. 1st ed., United Kingdom: John Wiley and Sons, 2010. ISBN 9780471250265.

[10] LORD, Charles (et al.). Biased Assimilation and Attitude Polarization: The Effects of Prior Theories on Subsequently Considered Evidence. Journal of Personality and Social Psychology. 1979, vol. 37, no. 11, pp. 2098–2109. DOI: 10.1037/0022-3514.37.11.2098 Also available from: https://www.researchgate.net/publication/232555483_Biased_Assimilation_and_Attitude_Polarization_The_Effects_of_Prior_Theories_on_Subsequently_Considered_Evidence

[11] NYHAN, Brendan and Jason REIFLER. When Corrections Fail: The persistence of political misperceptions. Political Behaviour. 2010, vol. 32, no. 2, pp. 303–330. DOI: 10.1007/s11109-010-9112-2. ISSN 1573-6687

[12] SIPPIT, Amy. The backfire effect: Does it exist? And does it matter for factcheckers? 1st ed., London: Full Fact, 2019. Available from: https://fullfact.org/media/uploads/backfire_report_fullfact.pdf

[13] KUHN, Deanna and Joseph LAO. Effects of evidence on attitudes: Is polarization the norm? Psychological Science. 1996, vol. 7, no. 2, pp. 115–120. DOI: 10.1111/j.1467-9280.1996.tb00340.x

[14] KLAYMAN, Joshua and Young-Won HA. Confirmation, disconfirmation, and information in hypothesis testing. Psychological Review. 1987, vol. 94, no. 2, pp. 211–228. DOI: 10.1037/0033-295X.94.2.211

[15] KLAYMAN, Joshua. Varieties of Confirmation Bias. In BUSEMEYER, Jerome, HASTIE, Reid and Douglas MEDIN (eds.), Decision Making from a Cognitive Perspective. 1st ed., San Diego, CA: Academic Press, 1995. ISBN 0080863833