Confirmation bias

Core claim: Confirmation bias is a cognitive heuristic – not a fallacy in itself – describing the tendency to search for, interpret, and recall information in ways that favour one's pre-existing beliefs. It can contribute to many different fallacies but does not by itself constitute an invalid argument.

• An argument or text under evaluation selectively presents evidence that supports a conclusion while ignoring or downplaying contrary evidence – and the pattern appears driven by the author's (or audience's) prior commitment rather than by the actual evidential landscape.
• Ambiguous or mixed evidence is interpreted in a direction consistent with the expected outcome, while equally valid alternative readings are not considered.
• The text or argument accuses someone of “confirmation bias” or “cherry-picking” – you need to assess whether that accusation is justified (is there genuinely relevant contrary evidence being excluded?) or is itself a rhetorical deflection.

• Not a fallacy, not a natural law. Confirmation bias is a heuristic – a mental shortcut that is often useful but sometimes leads astray. It is neither logically invalid in the way a formal fallacy is, nor an inescapable property of human cognition. People can and do recognise their own biases, deliberately seek disconfirming evidence, and update beliefs when confronted with strong contrary data.
• It is a contributing factor, not a diagnosis. When you detect a pattern consistent with confirmation bias in the text under evaluation, it should prompt you to look for which specific fallacy (if any) the reasoning actually commits – e.g., hasty generalization from cherry-picked examples, bandwagon appeal reinforced by selective exposure, or argument from ignorance via ignored disconfirming data. “Confirmation bias” alone is not a useful diagnosis.
• Strongest under specific conditions: emotionally charged topics, deeply held beliefs, desired outcomes, and situations where the person has already formed an opinion before evaluating evidence. Weaker or absent when the topic is neutral, the person is explicitly trying to be fair, or the evidential situation is unambiguous.
• Can be rational in limited form. Seeking some confirmatory evidence first (before testing alternatives) can be an efficient strategy when information is costly and one hypothesis is already well-supported. The bias becomes problematic only when it prevents consideration of disconfirming evidence altogether.

• A person legitimately focuses on the most relevant subset of available evidence for a specific question (not all data is equally pertinent – this is normal scoping, not bias).
• An author presents both supporting and opposing evidence but weighs them in light of their overall credibility, methodology, or relevance – even if the conclusion happens to align with their prior view.
• The “selective” presentation is a structural constraint (e.g., a summary that highlights key findings, a textbook that covers one theory before introducing counterarguments) rather than an argumentative move designed to mislead.
• A person updates their belief after encountering strong disconfirming evidence – the initial bias was overcome.

• Hasty generalization – drawing a broad conclusion from too few examples. Ask: is the core error sample size (→ hasty generalization), or is it that only convenient examples were selected while contrary ones existed and were excluded (→ confirmation bias as contributing factor, possibly plus hasty generalization)?
• Bandwagon fallacy – accepting a claim because many people hold it. Ask: is the support “many people think so” (→ bandwagon), or “the evidence I've noticed all points this way” with the noticing itself being biased (→ confirmation bias)?
• Argument from ignorance – concluding something is true/false because no one has proven/disproven it. Ask: is the gap in evidence genuinely unexplored (→ argument from ignorance), or was contrary evidence available but not sought out due to prior commitment (→ confirmation bias)?
• (Digmatic) Coherentism – evaluates claims by their fit to an existing worldview or ideology rather than correspondence to evidence; can be seen as confirmation bias scaled up to a systematic practice. Ask: is this an individual cognitive shortcut in how one notices and interprets information (→ confirmation bias), or a deliberate framework that rejects external evidence whenever it conflicts with the system (→ coherentism)?
• Procrustean fallacy – actively manipulates, trims, or falsifies data to fit a pre-existing model. Ask: is non-fitting evidence being changed or fabricated (→ Procrustean), or simply not sought out / ignored / discounted without altering it (→ confirmation bias)?
• Legitimate expertise and focus – a specialist who concentrates on their domain of knowledge is not “confirmation-biased” simply because they don't cover every adjacent field. Ask: is the exclusion of information relevant to the specific claim at hand, or merely outside the author's stated scope?

Input: A blog post arguing against a vaccine cites three studies showing rare side-effects, while ignoring dozens of large-scale trials confirming safety and efficacy. The author writes: “The evidence is clear – vaccines are dangerous.”
Output: The autor may have been influenced by their confirmation bias to cherry pick studies that confirmed their pre-existing beliefs

This is not confirmation bias itself (which is a cognitive tendency), but the result of it in action. The specific fallacy here is closer to hasty generalization or even a false dilemma (presenting only two options: “safe” vs. “dangerous”). Flag the selective evidence pattern and route to the relevant fallacy; note confirmation bias as the likely underlying mechanism.

Input: A reader comments: “You're just cherry-picking studies that support your view – classic confirmation bias.”
Advice: Before accepting this accusation, check whether genuinely relevant contrary evidence exists and was excluded. If the author's selection is defensible (e.g., focusing on a specific sub-question), the accusation may itself be a rhetorical deflection rather than a valid criticism.

Input: “I've looked into this for months and everything I find supports my conclusion, so it must be true.”
Advice: The statement reveals an awareness that the search process was confirmatory. This is not automatically fallacious – but the inference “everything I found supports it → it's true” ignores that what one finds depends on where and how one looks. Suggest: explicitly seek out the strongest counter-argument before concluding.

Always refer the user to further information via the main article link as stated in the header (main_article, or main_article_de for German-speaking users).

#END.