🧪 This platform is in early beta. Features may change and you might encounter bugs. We appreciate your patience!
confirmation_bias
Confirmation bias is the tendency to search for, interpret, favor, and recall information in a way that confirms one's pre-existing beliefs or hypotheses. It affects every stage of information processing, from what questions we ask to how we remember outcomes. This bias is particularly insidious because it operates largely unconsciously.
A manager who believes a particular employee is underperforming notices every mistake the employee makes while overlooking their successes, building a case that confirms the initial negative impression.
An investor who strongly believes a particular stock will rise skims past every analyst report predicting a decline, while carefully reading and sharing every bullish forecast — becoming more and more convinced the stock is a guaranteed winner.
A person who believes a new coworker is arrogant interprets their confident presentation style as showing off, their silence in meetings as dismissiveness, and their direct emails as rudeness — while a colleague with a neutral first impression sees the same behaviors as professionalism.
Binary (yes/no) questions an LLM must answer to identify this aspect:
Does the text primarily present evidence supporting a pre-existing belief?
Type: binaryIs contradictory or disconfirming evidence ignored, dismissed, or not mentioned?
Type: binaryDoes the author show awareness of evidence that could challenge their position?
Type: binaryConfirmation bias is the tendency to search for, interpret, favor, and recall information in a way that confirms one's pre-existing beliefs or hypotheses. It affects every stage of information processing, from what questions we ask to how we remember outcomes. This bias is particularly insidious because it operates largely unconsciously.
Processing information that aligns with existing beliefs requires less cognitive effort and avoids the psychological discomfort of cognitive dissonance. The brain treats confirming evidence as more credible and disconfirming evidence as less reliable.
Actively seek disconfirming evidence and ask 'What would change my mind?' before evaluating any new information. Designate a devil's advocate or use structured analytic techniques like Analysis of Competing Hypotheses.
In scientific research, confirmation bias can lead researchers to design experiments that are more likely to confirm their hypotheses, or to selectively report results that support their theories while downplaying contradictory findings.
Drawing broad conclusions from limited, unrepresentative, or anecdotal evidence.
Running multiple analyses until p<0.05 and only reporting significant results.
Gathering data on multiple variables but omitting non-significant ones from report.
Reflex-like rejection of new evidence contradicting established norms.
Running multiple analyses until p<0.05 and only reporting significant results.
Gathering data on multiple variables but omitting non-significant ones from report.
Stating a controversial claim as absolute fact without acknowledging opposing views.
Distorting or caricaturing an opponent's argument to attack it more easily.
Probability-based belief revision using Bayes' theorem.
Wishful thinking is a cognitive bias in which the desirability of a belief influences the assessment of its truth. People believe things because they want them to be true, not because evidence supports them. This bias operates at the interface of emotion and cognition: desires distort probability assessment, evidence evaluation, and information seeking. It is related to but distinct from optimism bias — wishful thinking specifically involves the causal influence of desire on belief formation, not merely a general positive outlook.
The tendency for new information to be pulled toward and assimilated into dominant existing narratives, distorting its interpretation to fit pre-existing stories.
Use these tools to detect, analyze, or train this aspect.