What Is Motivated Reasoning?
Motivated reasoning is the influence of goals, desires, commitments, and valued identities on how people search for information, retrieve memories, interpret evidence, evaluate arguments, and arrive at beliefs. A person may want to reach an accurate conclusion, defend a cherished belief, preserve a favorable self-image, maintain membership in a group, or avoid an emotionally costly possibility. These goals can influence which questions seem worth asking and which explanations seem credible. The decisive scientific question is how changes in goals alter the process and quality of judgment when relevant information and evaluation standards are held constant.
In the influential review The Case for Motivated Reasoning, Ziva Kunda (1990) distinguished accuracy goals, which orient people toward conclusions they believe are justified by evidence, from directional goals, which orient people toward particular preferred conclusions. A directional goal typically works through available cognitive resources: accessible memories, plausible explanations, standards of proof, or counterarguments. People can construct a justification that feels coherent even when its selection was shaped by the outcome they hoped to reach.
Motivated reasoning is a research framework for studying goal-sensitive cognition, not a diagnosis, a personality disorder, a synonym for lying, or proof that an individual's belief is false. A true conclusion can be reached through a biased process, and an incorrect conclusion can result from honest, impartial reasoning under weak information. The object of study is the process connecting motivation, evidence, inference, and judgment.
The Historical Development of the Concept
From cognitive consistency to motivated cognition
Earlier research on cognitive consistency, dissonance, and self-protective judgment asked how people handle information that conflicts with commitments. Studies of selective exposure and attitude change suggested that information can have emotional and social value in addition to its informational value. This intellectual background prepared the distinction between goals directed at accuracy and goals directed at a preferred outcome.
Lord, Ross, and Lepper (1979) exposed 48 college students who supported or opposed capital punishment to two purported studies about deterrence, one supporting and the other undermining the participants' initial position. Participants generally rated the congenial study as more convincing. This famous experiment established a vivid demonstration of biased evidence evaluation in a small and highly specific sample. It did not establish that every encounter with mixed evidence increases polarization, nor did it by itself identify a unique motivational cause rather than differences in interpretation or prior beliefs.
Kunda's 1990 synthesis offered a general mechanism: motivation shapes the accessibility and deployment of otherwise familiar reasoning operations. This formulation remains important because it treats reasoning as an active, constrained construction. A desired conclusion cannot always be reached; it must still be supported by arguments the thinker finds plausible. Kunda (1990).
Political judgment and the contemporary evidence debate
A later program investigated motivated skepticism, partisan identity, and argument evaluation. Taber and Lodge (2006) found that participants evaluating arguments about affirmative action and gun control were more critical of arguments that challenged existing views and were more likely to seek supporting information when given a choice. The observed tendencies included selective scrutiny, congenial information search, and changes in expressed attitudes. Yet subsequent methodological debates emphasized the need to control the actual diagnostic value of messages, confidence in prior beliefs, and incentives for responding accurately.
Modern research asks whether observed polarization results from motivated processing, differing priors, different assessments of source reliability, expressive responding, or some combination. Those explanations make distinguishable predictions and should be tested against explicit benchmarks rather than inferred from political disagreement alone. Druckman and McGrath (2019); Tappin, Pennycook, and Rand (2020).
Accuracy Goals and Directional Goals
Accuracy motivation
Accuracy motivation refers to wanting a judgment that tracks the best available evidence. A person with this goal may search more widely, compare sources, pay attention to uncertainty, and revise a prior belief after credible disconfirmation. Accuracy motivation does not ensure correct conclusions: attention, time, expertise, probabilistic understanding, and evidence quality can still limit performance. Its effects must be observed by comparing judgments and behavior under experimentally varied incentives or instructions.
Directional motivation
Directional motivation means wanting the conclusion to fall in a favored direction. Protecting a worldview, justifying an earlier choice, maintaining group loyalty, reducing anxiety, and defending personal competence are possible motives. Researchers distinguish a person's preferred conclusion from their actual prior belief: someone can confidently expect an unwelcome result, or strongly desire a result they regard as unlikely. Studies that measure only party identity or existing opinion may confound those two variables.
A common sequence begins with a personally significant claim. The person readily retrieves congruent examples, notices weaknesses in incongruent studies, asks more questions of unwelcome evidence, and settles on a conclusion that seems adequately justified. Each operation can also occur for legitimate reasons. The diagnostic evidence for motivated reasoning comes from asymmetric standards tied to experimentally manipulated desirability, with competing explanations held as constant as possible.
Multiple motives in one decision
Accuracy and direction often operate together. A clinician may genuinely want the correct interpretation of a test while preferring reassurance; an investor may seek an accurate forecast while hoping an existing position succeeds; a voter may seek trustworthy information while protecting group loyalty. Their relative influence changes with stakes, accountability, social context, expertise, and the ease of verifying facts. A single stable trait score cannot capture every motivated episode.
Psychological Mechanisms
Selective retrieval and accessible reasons
Motivation can influence what information comes to mind. Remembered successes may be easier to retrieve when defending a decision; counterexamples may become more salient when challenging an opponent. Researchers should differentiate retrieval bias from the underlying distribution of memories: people exposed to different information may genuinely have different knowledge. The key causal test manipulates a goal while keeping the information base and retrieval opportunity comparable.
Asymmetric evidence evaluation
Motivated skepticism occurs when congruent claims are accepted with relatively little scrutiny while conflicting claims provoke methodological objections, alternative interpretations, and demands for higher certainty. In Taber and Lodge's studies, the asymmetry appeared in evaluations and counterarguing. The methodological challenge is ensuring that pro and con arguments are matched in actual quality and that experts, not only partisans, assess their strength. Taber and Lodge (2006).
Selective exposure and information avoidance
People may seek congenial sources before evidence is even evaluated. The meta-analysis by Hart and colleagues (2009) found a moderate overall preference for supportive over challenging information (d = 0.36), with meaningful variation by conviction, confidence, information quality, and goals. That effect estimates selective information preference in the studies synthesized; it is not an effect size for every form of motivated reasoning. In some circumstances, people prefer contradictory information because obtaining it supports a practical accuracy goal.
Affect, threat, and emotional regulation
An argument can convey fear, shame, pride, relief, or belonging alongside factual content. These feelings may influence attention and evaluation, especially when accepting an implication threatens a valued self-concept. Emotional intensity by itself does not establish motivated reasoning: researchers need evidence that the specific motivational state caused systematic shifts in evidence processing. Affective and motivational mechanisms can interact but should remain distinguishable constructs.
Identity and group membership
Political, professional, and cultural identifications create incentives to interpret information in identity-consistent ways. Van Bavel and Pereira (2018) developed an identity-based model integrating research on partisan evaluation, perception, and memory. Their account helps generate predictions about when a belief carries group-signaling value. It remains a theoretical model; a group difference in conclusions is insufficient on its own to prove identity-protective reasoning.
Confidence, uncertainty, and standards of proof
Confidence can determine how extensively evidence is inspected. A highly confident person may have less reason to search, whether their confidence is warranted or not; a threatened belief may motivate more counterarguing. An especially informative outcome is a change in the standard of proof: what sample size, control group, or reliability is considered sufficient for a claim supporting versus challenging the desired conclusion? If the same methodological flaw is disqualifying only in the unwelcome case, the asymmetry has direct process relevance.
Motivated Reasoning and Neighboring Concepts
Motivated reasoning versus confirmation bias
Confirmation bias is a tendency to search for, interpret, remember, or weigh information in ways that favor an existing belief or hypothesis. Motivated reasoning identifies how goals can influence those operations. Confirmation bias can arise from hypothesis testing habits, asymmetric evidence availability, or prior probability judgments without a strong desire for the hypothesis to be true. Motivated reasoning can also favor a desired conclusion that the thinker initially doubts. They overlap in many experiments but are not interchangeable explanations.
Motivated reasoning versus belief perseverance
Belief perseverance concerns continued endorsement after initial support has been weakened or discredited. Motivated reasoning can contribute by supplying counterarguments or replacement explanations, but persistence can also reflect memory, confusion about the correction, independent evidence, or source distrust. The relevant outcome is belief after disconfirmation; a mechanism claim requires further measurement or manipulation.
Motivated reasoning versus selective exposure
Selective exposure is behavior: choosing what information to encounter. Motivated reasoning covers the broader causal processes of judgment before, during, and after exposure. A person may intentionally read challenging evidence yet still evaluate it asymmetrically. Conversely, someone may consume congenial material because of availability or entertainment rather than a desire to protect a belief. Hart et al. (2009) illustrates the value of studying selection and motives separately.
Motivated reasoning versus cognitive dissonance
Cognitive dissonance theories investigate discomfort and regulation when cognitions or actions conflict. Motivated reasoning describes how goals can recruit particular inferential strategies. A dissonance-producing inconsistency can create a directional goal, but not every motivated belief concerns dissonance, and not all dissonance reduction occurs through evaluating evidence.
Motivated reasoning versus logical fallacy, deception, and symptom
A logical fallacy is an invalid inferential form or argument pattern. Motivated reasoning is a psychological account of why evidence or arguments may be selected and evaluated asymmetrically. Deception requires misrepresentation with an intention to mislead; motivated reasoning may occur without conscious awareness. Motivated interpretation is also not, by itself, a clinical symptom, psychotic delusion, or mental disorder. Clinical questions require their own diagnostic evidence and criteria.
Normative Benchmarks: When Is Reasoning Biased?
Correctness of a conclusion and quality of an update
A scientific assessment needs a target. For a factual claim with verifiable truth, accuracy can be measured by correct answers or a proper scoring rule. For a probabilistic judgment, researchers can examine calibration, discrimination, and revision in response to diagnostic information. For an argument evaluation task, blinded expert judgments can estimate evidential quality, although expert disagreement and uncertainty should be measured.
Bayesian updating as an explicit comparison
When a hypothesis H has prior probability P(H) and evidence E has a known likelihood ratio, posterior odds equal prior odds multiplied by the likelihood ratio: P(H|E)/P(not H|E) = [P(H)/P(not H)] × [P(E|H)/P(E|not H)]. This equation provides a normative benchmark only when priors, hypotheses, evidence, and likelihoods are defined. Different priors can legitimately yield different posteriors after exposure to the same evidence. Opposite-direction belief change is not automatically irrational when participants attach different likelihoods to an ambiguous message.
If someone starts at 20% probability for a hypothesis, their prior odds are 0.25. Evidence with a likelihood ratio of 3 raises posterior odds to 0.75, corresponding to a posterior probability of about 42.9%. The same information will produce a different final probability for someone with a different prior. The important measurement is whether people use the evidential diagnosticity appropriately relative to their own priors, not whether everyone converges on one belief.
The benchmark may be contested
Many social claims do not have objectively provided likelihoods or fully agreed source reliability. Participants may dispute what the data sample represents, whether an expert is trustworthy, and which alternative hypothesis is relevant. Researchers can vary these features experimentally or estimate them independently. Tappin, Pennycook, and Rand (2020) directly addressed problems in earlier studies by comparing political belief updates to Bayesian benchmarks and taking specific priors into account. A headline-level partisan difference and a motivational distortion of Bayesian updating are distinct findings.
How Researchers Measure Motivated Reasoning
Identical-evidence and randomized-goal designs
One method presents otherwise identical evidence to groups that differ in whether a conclusion benefits them. Another randomly assigns accuracy incentives, identity cues, outcome desirability, or accountability while holding the evidential material constant. Key dependent measures include credibility, argumentative strength, confidence, estimated probability, willingness to seek disconfirming evidence, and posterior belief. Strong designs preregister hypotheses, show that the motivational manipulation worked, and test alternative mechanisms.
Tracking information search
Researchers measure article choice, search order, browsing time, information avoidance, and requests for additional evidence. Such behavior identifies selection preferences. To attribute selection to motivated reasoning, the study must control relevance, usefulness, prior knowledge, source familiarity, and the available information menu. Search behavior may improve accuracy when a person deliberately inspects a trusted source rather than reading every item indiscriminately.
Counterarguing and process traces
Some studies ask participants to record arguments for and against a claim, or analyze comments that participants offer spontaneously. They can reveal disproportionate scrutiny of disliked evidence. Think-aloud protocols can also change reasoning, and self-reports are incomplete measures of unconscious processes. Response time, eye movements, and physiological indices may complement behavioral tests but rarely identify motivation without experimental context.
Accuracy incentives and expressive responding
A survey answer may express identity rather than a privately held factual probability. Bullock, Gerber, Hill, and Huber (2015) found that small incentives for correct answers, together with opportunities to acknowledge uncertainty, reduced apparent partisan gaps on factual questions. This finding does not demonstrate that political disagreement is merely performative; it shows why incentives and response formats must be considered when interpreting survey differences.
Design quality and reproducibility
Reliable inference depends on adequate sample size, random assignment, balanced argument strength, transparent exclusion rules, pre-registration when appropriate, and replication across materials and populations. Measuring initial beliefs after presenting treatments can contaminate the very baseline used to claim updating. Researchers should report the change in judgment and the distribution of effects, rather than just whether one group's average reached a significance threshold.
The Evidence: Classical Findings, Meta-Analyses, and Challenges
Biased assimilation and the polarization question
The classic capital-punishment experiment showed asymmetric ratings of studies aligned with prior attitudes. It became a foundation for thinking about attitude polarization. Its sample of 48 students, simulated studies, and narrow setting make it a demonstration of possibility rather than a universal estimate. Evidence that participants regard congenial arguments more favorably can coexist with overall information-driven learning; changes in attitudes should be measured directly, including initial position, final position, and uncertainty. Lord et al. (1979).
Partisan evaluation in a synthesis of experiments
Ditto and colleagues (2019) meta-analyzed 51 experiments involving more than 18,000 participants. The overall partisan-bias association was r = .245, and average estimates were similar for liberal and conservative groups in the analyzed tasks (r = .235 and r = .255). This is evidence of a recurring tendency to evaluate congenial material more favorably under those experimental designs. It does not establish that all ideological groups are equivalent on every factual question, topic, outcome, historical period, or exposure environment. Critiques also emphasize interpretation of the normative baseline and task selection.
Selective exposure and a measurable average
The meta-analysis by Hart et al. (2009) estimated an overall preference for attitudinally congenial information of d = 0.36. Moderator analyses indicated sensitivity to strength of attitudes and the relative appeal of defense and accuracy goals. An average across studies is not a fixed human parameter: study design, available sources, urgency, personal relevance, and the quality of the information change the strength and sometimes the direction of preference.
An alternative account: accuracy-seeking with different priors
In their review of climate change attitudes, Druckman and McGrath (2019) argued that much apparent directional motivated reasoning could also be explained by efforts to reach accurate beliefs combined with divergent judgments of source credibility. Their proposal does not rule out identity or directional motives. It requires investigators to test credible alternatives before assigning motivational causation to disagreement.
Analytical thinking: motivated sophistication or better discernment?
A strong version of the motivated-reasoning hypothesis predicts that additional reasoning ability sometimes enables more elaborate rationalization. But Pennycook and Rand (2019), studying evaluations of true and false headlines, found that higher analytic thinking was associated with better truth discernment even for politically congenial items. In those studies, reflection did not systematically magnify partisan distortion. Results about some political argument tasks should therefore not be applied indiscriminately to all information-processing environments.
Bayesian comparisons to partisan belief updating
Tappin, Pennycook, and Rand (2020) highlighted that inferences based on party-aligned judgments may miss specific prior beliefs and the diagnostic value of new evidence. Their experiments supplied a more explicit Bayesian comparison. The result strengthens the general methodological rule: sophisticated reasoning and politically different posteriors do not by themselves demonstrate a motivated departure from rational updating.
Correction and the backfire debate
An influential early study reported circumstances in which corrective messages were followed by increased political misperceptions among certain participants Nyhan and Reifler (2010). The claim that corrections generally backfire has since been disputed and is not a sound universal rule. Researchers should distinguish a failed correction, incomplete belief updating, temporary resistance, and the stronger claim of an increase in the specific false belief. The existence of motivational resistance is compatible with many successful corrections; future effects depend on correction quality, source, timing, prior attitudes, and measurement.
Where Motivated Reasoning Appears
Personal decisions and self-evaluation
A person choosing among careers may highlight advantages of a preferred option while raising the evidentiary bar for unwelcome alternatives. After investing heavily in a project, an individual may interpret ambiguous results optimistically. These examples are hypotheses about a possible mechanism, not proof of motivated reasoning in any particular person. Research designs should distinguish it from information asymmetry, attachment to sunk costs, and realistic domain expertise.
Science, organizations, and professional judgment
Scientists, managers, clinicians, and legal professionals work with evidence while facing reputational, financial, and social pressures. When the same study design or testimony is judged differently depending on whether it supports a preferred conclusion, motivated evaluation becomes a testable concern. Professional standards, blinded review, preregistration, independent replication, and structured decision rules can reduce some sources of discretionary interpretation. Specific clinical or forensic recommendations must be based on the evidence and institutional context of the relevant discipline.
Public debate and political beliefs
Group identity can motivate a person to defend claims associated with their preferred group. Yet misinformation belief, party loyalty, truth discernment, social expression, and actual voting behavior are different dependent variables. A survey gap should not automatically be presented as an individual psychological defect. The literature supports studying both symmetrical processing tendencies under comparable tasks and asymmetric real-world exposure, communication, or factual conditions. Ditto et al. (2019); Van Bavel and Pereira (2018).
Health information and risk
Desired reassurance can affect how someone weighs an unfavorable test result; fear can create strong expectations of danger. Evidence-driven health decisions therefore benefit from understandable numerical risk estimates, transparent uncertainty, and appropriately qualified professional advice. Motives are one part of the picture alongside health literacy, clinical context, access to reliable sources, and the quality of communication. No diagnostic label follows from ordinary motivated judgment.
Human–AI systems and information environments
Search engines, recommendation systems, and AI assistants make selected information easy to access and reframe. Someone seeking reassurance may repeatedly request supportive explanations while ignoring probabilistic caveats; someone seeking a reliable answer may use the same tools to compare competing interpretations. AI outputs can sound coherent regardless of evidential quality. Useful safeguards include showing sources and uncertainty, separating a user's desired outcome from the evidence, and evaluating both aligned and contrary explanations against the same criteria. The motivational hypothesis concerns human interpretation and interaction; it does not establish any claim about a machine's subjective experience.
Individual Differences, Social Context, and Cultural Generalizability
No single percentage states how often motivated reasoning occurs. Some tasks concern political partisans in the United States, some use college students, and others use online convenience samples. Personal relevance, belief certainty, available evidence, educational background, social accountability, and source trust can change behavior. Without comparable multilingual stimuli and measurement invariance, findings from US political argument tasks cannot be automatically generalized across cultures or institutions.
Knowledge and analytic skill may support correction when participants have clear accuracy goals, while domain knowledge may also supply more arguments for a valued position. Which tendency predominates is a testable interaction among task, motive, and evidence. A high reflection score is neither immunity to bias nor proof of skillful rationalization. Social identity itself differs in salience and meaning across groups and situations. Generalizations require representative sampling and independent replications with culturally appropriate measures.
Can Motivated Reasoning Be Reduced?
Clarify the decision question before searching
Specify the claim, what would count as supporting or disconfirming evidence, which sources are trustworthy, and what threshold would change the conclusion. Applying the same rules to welcome and unwelcome evidence limits opportunistic shifts in standards. This is a reasoned decision procedure; the strength of its real-world effect depends on how consistently it is implemented.
Consider the opposite
Ask what an informed critic would say and which observation would undermine the favored interpretation. In controlled social-judgment studies, Lord, Lepper, and Preston (1984) found that prompts to consider opposing possibilities reduced particular judgment biases. This provides experimental support for a targeted strategy, not a guarantee of durable transfer to every high-stakes choice.
Use symmetric evaluation rubrics
Create an evidence table in which every argument receives the same assessment of sample size, design, measurement validity, replication, competing causes, and relevance to the question. Blind the direction of results when possible. The practical point is to make a standard stable before learning whether a finding is congenial; once the preferred answer is known, a person can unconsciously change which methodological defects feel decisive.
Seek accuracy incentives and allow uncertainty
Correctness rewards can sometimes reduce expressive response gaps. In Bullock et al. (2015), small payments and explicit uncertainty options narrowed partisan differences on factual survey questions. This is a task-specific effect on reported beliefs, and should not be mistaken for proof that incentive payments change enduring worldviews.
Assess evidence quality rather than just exposure balance
Reading an opposing argument is beneficial only insofar as the content is reliable and relevant. Mechanical balance between well-supported findings and demonstrably weak claims can distort judgment. Ask which evidence has stronger methods, which uncertainty remains, and whether a disagreement reflects prior knowledge or testable motivational asymmetry. A credible source can still be wrong on a particular claim; disagreement with it is not automatically a bias.
External decision aids and accountability
Structured forecasts, pre-committed decision rules, calibration feedback, independent review, and clearly defined falsification conditions can create opportunities to detect inconsistency. Such approaches are plausible safeguards, but evidence from one laboratory bias does not establish that generic training transfers to job performance, political beliefs, or clinical outcomes. Direct field evaluations and transparent outcome measures remain necessary.
Evaluating Your Own Beliefs: A Research-Informed Procedure
Begin with a concrete probability or confidence statement, not only a categorical position. Write the current best case for your conclusion and the strongest alternative explanation. Identify the observation that would change your assessment, seek a source capable of producing that observation, and apply the same evidentiary rule regardless of whether the finding would be welcome. Record what actually changed after the information arrived, including confidence and unresolved uncertainty.
For example, before reading a report on a favored intervention, specify which outcome measure, comparison group, and magnitude would count as persuasive. Then evaluate supportive and unfavorable trials with the same rubric. If an unfavorable study suddenly seems too small while a supportive study of the same size seems definitive, that asymmetry is a useful signal to revisit the reasoning process. The test concerns the consistency of evaluation, not the moral worth of holding a position.
Frequently Asked Questions
Is motivated reasoning always unconscious?
No. Some goals are explicit, such as wanting one's proposal to succeed; other aspects of selective retrieval and argument evaluation may occur without awareness. Researchers should measure awareness rather than assume either complete intention or complete automaticity.
Is motivated reasoning the same as being irrational?
The term describes goal-sensitive processing. A claim of irrationality requires a specified benchmark: for example, inconsistent application of evidence standards, deviation from a defined probability update, or lower accuracy in a verifiable task. Preference for a source can be rational when that source is more reliable.
Can intelligent people engage in motivated reasoning?
Yes, but the direction of the association between cognitive sophistication and bias depends on the task. Reasoning ability may help detect misinformation and may also support detailed defenses of a preferred position. Evidence from political argument evaluation should not be merged indiscriminately with headline accuracy experiments.
How is motivated reasoning different from confirmation bias?
Confirmation bias describes belief-consistent information selection or evaluation. Motivated reasoning specifies a possible goal-based explanation for such patterns and includes attempts to reach conclusions people desire even when those conclusions are not initially believed.
Does showing contradictory evidence make people more extreme?
Sometimes people discount unwelcome evidence, and some experimental designs have found polarization. The stronger claim that contradictory evidence routinely increases extremity is not warranted. Outcomes depend on evidence quality, prior beliefs, credibility, and whether a study measures argument ratings, factual beliefs, or behavioral choices.
Are liberals or conservatives more vulnerable?
In a meta-analysis of comparable partisan-evaluation experiments, both groups showed similar average congeniality effects Ditto et al. (2019). This finding is limited to the sampled tasks and outcomes; it does not settle every political information or behavior question.
Can motivated reasoning affect scientific or professional decisions?
It can influence interpretation wherever preferred outcomes and ambiguous evidence intersect. Demonstrating that influence in a specific organization requires evidence about task structure, standards, incentives, and competing explanations. Blinded evaluation and precommitted procedures can help constrain discretionary interpretation.
Can AI make motivated reasoning worse?
Personalized systems may make congenial information and persuasive justifications easier to obtain; systems can also support counterargument search and calibration. The outcome depends on interface design, evidential reliability, user goals, and how the tool is used. Broad empirical claims require direct human–AI interaction studies.
What is the most important way to check a cherished belief?
Specify in advance what evidence could change your mind and give that evidence the same quality threshold you would apply if it supported your position. Record an explicit update in confidence. This procedure makes motivated shifts in the standard of proof easier to detect.
Related Articles
Cluster Cognitive Biases, Judgment and Decision-Making — The scientific research chapter connecting judgment, belief formation, and decision models.
Cognitive Bias in Criminal Investigations: Tunnel Vision and Confirmation Bias — An applied investigation of evidence interpretation and competing hypotheses in criminal inquiries.
Forensic Evidence and Cognitive Bias: How Context Can Contaminate Judgment — Context effects and safeguards when evaluating forensic evidence.
Judge Decision-Making: Psychology, Expertise, and Bias — Decision standards, expertise, and cognitive mechanisms in judicial contexts.
AI as Authority: Trust, Expertise, Automation Bias, and Human Decision-Making — Trust calibration when people assess information offered by AI systems.
References
Bullock, J. G., Gerber, A. S., Hill, S. J., and Huber, G. A. (2015). Partisan bias in factual beliefs about politics. Quarterly Journal of Political Science, 10, 519–578. https://doi.org/10.1561/100.00014074.
Ditto, P. H., Liu, B. S., Clark, C. J., Wojcik, S. P., Chen, E. E., Grady, R. H., Celniker, J. B., and Zinger, J. F. (2019). At least bias is bipartisan: A meta-analytic comparison of partisan bias in liberals and conservatives. Perspectives on Psychological Science, 14, 273–291. https://doi.org/10.1177/1745691617746796.
Druckman, J. N., and McGrath, M. C. (2019). The evidence for motivated reasoning in climate change preference formation. Nature Climate Change, 9, 111–119. https://doi.org/10.1038/s41558-018-0360-1.
Hart, W., Albarracín, D., Eagly, A. H., Brechan, I., Lindberg, M. J., and Merrill, L. (2009). Feeling validated versus being correct: A meta-analysis of selective exposure to information. Psychological Bulletin, 135, 555–588. https://doi.org/10.1037/a0015701.
Kunda, Z. (1990). The case for motivated reasoning. Psychological Bulletin, 108, 480–498. https://doi.org/10.1037/0033-2909.108.3.480.
Lord, C. G., Lepper, M. R., and Preston, E. (1984). Considering the opposite: A corrective strategy for social judgment. Journal of Personality and Social Psychology, 47, 1231–1243. https://doi.org/10.1037/0022-3514.47.6.1231.
Lord, C. G., Ross, L., and Lepper, M. R. (1979). Biased assimilation and attitude polarization: The effects of prior theories on subsequently considered evidence. Journal of Personality and Social Psychology, 37, 2098–2109. https://doi.org/10.1037/0022-3514.37.11.2098.
Nyhan, B., and Reifler, J. (2010). When corrections fail: The persistence of political misperceptions. Political Behavior, 32, 303–330. https://doi.org/10.1007/s11109-010-9112-2.
Pennycook, G., and Rand, D. G. (2019). Lazy, not biased: Susceptibility to partisan fake news is better explained by lack of reasoning than by motivated reasoning. Cognition, 188, 39–50. https://doi.org/10.1016/j.cognition.2018.06.011.
Taber, C. S., and Lodge, M. (2006). Motivated skepticism in the evaluation of political beliefs. American Journal of Political Science, 50, 755–769. https://doi.org/10.1111/j.1540-5907.2006.00214.x.
Tappin, B. M., Pennycook, G., and Rand, D. G. (2020). Bayesian or biased? Analytic thinking and political belief updating. Cognition, 204, 104375. https://doi.org/10.1016/j.cognition.2020.104375.
Van Bavel, J. J., and Pereira, A. (2018). The partisan brain: An identity-based model of political belief. Trends in Cognitive Sciences, 22, 213–224. https://doi.org/10.1016/j.tics.2018.01.004.