Cognitive Bias - Confirmation Bias

jashburn8020 · Wed, 17 Jul 2024

psychology   bias

Confirmation bias (also confirmatory bias, myside bias, or congeniality bias) is the tendency to search for, interpret, favour, and recall information in a way that confirms or supports one’s prior beliefs or values.

@startmindmap * Confirmation bias * Types * Biased search for information * Principle of compatibility * Personality traits * Biased interpretation of information * Disconfirmation bias * Biased memory recall of information * Evidence remembered selectively to reinforce expectations * Information processing explanations * Cognitive versus motivational * Pollyanna principle * Cost-benefit * Manifestation of social skill * Exploratory versus confirmatory * Make-believe * Habit of rationalisation of false beliefs * Effects * Attitude polarisation * Belief perseverance * Irrational primacy effect * Illusory correlation * Debiasing * Awareness * Avoid using precedent as reference point * Start with neutral fact base * Prove that you are wrong * Continue assessing with new information * Obtain information independently in a group * Debate to challenge our views * "Preference-inconsistent" recommendation system @endmindmap

Note: The information on this page is mainly summarised from the references listed below.

Confirmation Bias

  • People display this bias when they
    • select information that supports their views, ignoring contrary information, or
    • interpret ambiguous evidence as supporting their existing attitudes.
  • Effect is strongest for:
    • desired outcomes,
    • emotionally charged issues, and
    • deeply entrenched beliefs.
  • Insuperable (impossible to overcome) for most people, but they can manage it, e.g., by education and training in critical thinking skills.
  • 3 types of confirmation bias:
    • Biased search for information: People tend to search for evidence consistent with their current hypothesis.
    • Biased interpretation of information: People tend to accept “confirming” evidence at face value while subjecting “disconfirming” evidence to critical evaluation.
    • Biased memory recall of information: People remember evidence selectively to reinforce their expectations.
  • 4 effects of confirmation bias:
    • Attitude polarisation: When a disagreement becomes more extreme even though the different parties are exposed to the same evidence.
    • Belief perseverance: When beliefs persist after the evidence for them is shown to be false.
    • Irrational primacy effect: A greater reliance on information encountered early in a series.
    • Illusory correlation: When people falsely perceive an association between 2 events or situations.
  • A result of automatic, unintentional strategies rather than deliberate deception.

Types of Confirmation Bias

Biased Search for Information

  • People tend to test hypotheses in a one-sided way, by searching for evidence consistent with their current hypothesis:
    • phrase questions to receive an affirmative answer that supports their theory; and
    • look for the consequences that they would expect if their hypothesis was true, rather than what would happen if it was false.
  • A small change in a question’s wording can affect how people search through available information, and hence the conclusions they reach.
    • Principle of compatibility:
      • The weighting of inputs is enhanced by their compatibility with output.
      • The positive dimensions of options (pros) loom larger when one is choosing, and the negative dimensions (cons) loom larger when one is rejecting.
    • E.g., with a fictional child custody case:
      • Problem:
        • Parent A was moderately suitable to be the guardian in multiple ways.
        • Parent B had a mix of salient positive (e.g., a close relationship with the child) and negative (e.g., a job that would take them away for long periods of time) qualities.
      • Questions:
        • “Which parent should have custody of the child?” the majority of participants chose Parent B.
        • “Which parent should be denied custody of the child?” the majority answered that Parent B should be denied custody. (Shafir, 1993)
  • Personality traits influence and interact with biased search processes.
    • High confidence levels: More readily seek out contradictory information to their personal position to form an argument.
    • Low confidence levels: Do not seek out contradictory information and prefer information that supports their personal position.

Biased Interpretation of Information

  • Even if two individuals have the same information, the way they interpret it can be biased.
  • People set higher standards of evidence for hypotheses that go against their current expectations: disconfirmation bias.
    • They are apt to accept “confirming” evidence (preference-consistent information) at face value.
      • The evidence/information appears convincing by itself and so they do not feel the need for justification.
    • They subject “disconfirming” evidence (preference-inconsistent information) to critical evaluation.
      • The evidence/information appears less valid and this leads to a deliberate search for further information to refute dissenting arguments.
    • As a result, they draw undue support for their initial positions from mixed or random empirical findings.
    • 48 undergraduates supporting and opposing capital punishment were exposed to 2 purported studies:
      • One seemingly confirming and one seemingly disconfirming their existing beliefs about the deterrent efficacy of the death penalty.
      • Both proponents and opponents of capital punishment rated those results and procedures that confirmed their own beliefs to be the more convincing and probative ones.
    • The result of exposing contending factions in a social dispute to an identical body of relevant empirical evidence may be not a narrowing of disagreement but rather an increase in polarisation. (Lord et al., 1979)
  • Biases in belief interpretation are persistent, regardless of intelligence level.

Biased Memory Recall of Information

  • People may remember evidence selectively to reinforce their expectations, even if they gather and interpret evidence in a neutral manner.
    • This effect is called “selective recall”, “confirmatory memory”, or “access-biased memory”.
  • A selective memory effect has been shown in experiments that manipulate the desirability of personality types.
    • A group of participants were shown evidence that extroverted people are more successful than introverts, while another group were told the opposite.
    • Participants were asked to recall events from their lives in which they had been either introverted or extroverted.
    • Each group of participants provided more memories connecting themselves with the more desirable personality type, and recalled those memories more quickly.
  • Memories for emotional responses are partially reconstructed based on current appraisals of events.
    • Levine et al. (2001) investigated memory for emotions and how it changes over time.
      • People were asked about their initial emotional reactions to the O.J. Simpson verdict, and then how they felt and what they thought about the verdict months and over a year later.
    • Memories of emotions can be unstable.
      • People’s memories of how strongly they felt - happy, angry, or surprised - about the verdict changed over time.
    • Current beliefs can influence memories of emotions.
      • The more the person’s current beliefs about O.J. Simpson’s guilt or innocence changed, the less stable their memories of the initial emotional intensity were.
    • Memories of emotions can be reconstructed.
      • The way people remembered feeling about the verdict after a long time seemed to be influenced by their current beliefs about the case, rather than their actual emotions at the time.

Information processing explanations

Cognitive versus Motivational

  • Most biased evidence processing occurs through a combination of “cold” (cognitive) and “hot” (motivated) mechanisms.
  • Cognitive explanations for confirmation bias are based on limitations in people’s ability to handle complex tasks and the shortcuts (heuristics) that they use.
  • Motivational explanations involve an effect of desire on belief.
    • People prefer positive thoughts over negative ones: Pollyanna principle.
      • This could explain why desired conclusions are more likely to be believed true.
  • It is possible that motivation creates the bias and cognitive factors determine the size of the effect.

Cost-Benefit

  • People do not primarily aim at truth in testing hypotheses, but try to avoid the most costly errors.
  • People compare the 2 different kinds of error: accepting a false hypothesis or rejecting a true hypothesis.
    • E.g., someone who underestimates a friend’s honesty might treat them suspiciously and so undermine the friendship.
      • Overestimating the friend’s honesty may also be costly, but less so.
      • It would be rational to seek, evaluate or remember evidence of their honesty in a biased way.
  • Dardenne and Leyens (1995) showed that people might exhibit a form of confirmation bias as a manifestation of a social skill when forming impressions of others.
    • In an experiment where participants ask questions to determine whether another person possesses a particular personality trait,
      • participants with high self-monitoring (ability to regulate behaviour to accommodate social situations) asked questions that confirm their initial hypothesis (matching questions) about someone’s personality (e.g., introverted or extroverted),
      • which come across as more empathic,
      • when the context (e.g., status of the other person) stresses their relevance.

Exploratory versus confirmatory

  • 2 different kinds of thinking process:
    • Exploratory thought: Neutrally considers multiple points of view and tries to anticipate all possible objections to a particular position.
      • Used when the other parties are well-informed, genuinely interested in the truth, and whose views they do not already know.
    • Confirmatory thought: Seeks to justify a specific point of view.
      • Used when people expect to justify their position to others whose views they already know.
      • They will tend to adopt a similar position to those people, and then use confirmatory thought to bolster their own credibility.
  • If the external parties are overly aggressive or critical, people will
    • disengage from thought altogether, and
    • simply assert their personal opinions without justification.
  • The conditions for exploratory thought rarely exist and so most people use confirmatory thought most of the time.

Make-Believe

  • Many of people’s beliefs and biases are formed early in life when children begin to distinguish between fantasy and reality.
  • Parents commonly encourage young children to engage in pretend play.
    • Little life scenarios, like playing house, that help to reinforce cultural norms and beliefs and aid in assimilation as they age.
    • Children also learn that sometimes it is OK to make believe things are true, even though they know they are not.
  • In adolescence, people develop critical thinking skills and some begin to question what they were taught as children.
    • Despite evidence to the contrary, some are willing to rationalise their parents’ false beliefs in order to avoid upsetting their parents.
  • The habit of rationalisation of false beliefs can become unconscious over the years.
    • Serves as a framework for processing information in adulthood.
    • Easier to cling to a simple fiction than a complicated reality. (“Why We’re Susceptible to Fake News, How to Defend Against It,” 2018)

Effects of Confirmation Bias

Attitude Polarisation

  • When people with opposing views interpret new information in a biased way, their views can move even further apart.
    • Does not necessarily occur when people simply hold opposing positions, but rather when they openly commit to them.
  • In a study using the emotionally charged topics of gun control and affirmative action:
    • Participants’ attitudes towards these issues were measured before and after reading arguments on each side of the debate.
    • 2 groups of participants showed attitude polarisation:
      • those with strong prior opinions, and
      • those who were politically knowledgeable.
    • Participants chose which information sources to read from a list prepared by the experimenters.
      • Even when instructed to be even-handed, participants were more likely to read arguments that supported their existing attitudes than arguments that did not (biased search for information).

Belief perseverance

  • Confirmation biases provide one plausible explanation for the persistence of beliefs when the initial evidence for them is removed or when they have been sharply contradicted.
  • In a study, participants read job performance ratings of two firefighters, along with their responses to a risk aversion test.
    • Some participants were told that a risk-taking firefighter did better, while others were told they did less well than a risk-averse colleague.
    • Participants found the case studies to be subjectively persuasive about firefighters in general.
    • The participants were then thoroughly debriefed and informed that there was no link between risk taking and performance.
    • Participants’ belief in a link diminished, but around half of the original effect remained.
    • Participants seemed to trust the debriefing (that the case studies were fictional), but regarded the discredited information as irrelevant to their personal belief.

Irrational Primacy Effect

  • Information is weighted more strongly when it appears early in a series, even when the order is unimportant.
  • Seeing the initial evidence, people form a working hypothesis that affects how they interpret the rest of the information (biased interpretation).
  • An experiment was conducted involving a slide show of a single object seen as just a blur at first and in slightly better focus with each succeeding slide.
    • After each slide, participants state their best guess of what the object was.
    • Participants whose early guesses were wrong persisted with those guesses, even when the picture was sufficiently in focus that the object was readily recognisable to other people.

Illusory Correlation

  • Illusory correlation is the tendency to see non-existent correlations in a set of data.
  • A study recorded the symptoms experienced by arthritic patients, along with weather conditions over a 15-month period.
    • Nearly all the patients reported that their pains were correlated with weather conditions, although the real correlation was zero.
  • A kind of biased interpretation.
    • Objectively neutral or unfavourable evidence is interpreted to support existing beliefs.
  • Related to biases in hypothesis-testing behaviour.
    • In judging whether two events are correlated, people rely heavily on the number of positive-positive cases.
    • They pay relatively little attention to the other kinds of observation (of no pain and/or good weather).
    • Parallels the reliance on positive tests in hypothesis testing.
  • May also reflect selective recall.
    • People may have a sense that two events are correlated because it is easier to recall times when they happened together.

Examples

  • In social media, confirmation bias is amplified by the use of filter bubbles, or “algorithmic editing”.
    • Displays to individuals only information they are likely to agree with, while excluding opposing views.
  • Some scientists have resisted new discoveries by selectively interpreting or ignoring unfavourable data.
    • Several studies have shown that scientists rate studies that report findings consistent with their prior beliefs more favourably than studies reporting findings inconsistent with their previous beliefs.
    • Data that conflict with the experimenter’s expectations may be more readily discarded as unreliable, producing the file-drawer effect (also known as publication bias).
    • The peer review process may be susceptible to biases.
      • Research presenting controversial results frequently receives harsh peer review.
  • Psychic readings
    • By making a large number of ambiguous statements in each sitting, the psychic gives the client more opportunities to find a match.
    • Listeners apply a confirmation bias which fits the psychic’s statements to their own lives.
  • Seattle windshield pitting epidemic
    • Confirmation bias can play a key role in the propagation of mass delusions.
    • A “pitting epidemic” in which windshields were damaged due to an unknown cause.
    • As news of the apparent wave of damage spread, more and more people checked their windshields, discovered that their windshields too had been damaged, thus confirming belief in the supposed epidemic.
    • In fact, the windshields were previously damaged, but the damage went unnoticed until people checked their windshields as the delusion spread.
  • The murder trial of Fred van der Vyver in South Africa
    • Accused of murdering his girlfriend, Inge Lotz, in June 2005.
    • The police determined the hypothesis early on that Fred was their man - apparently due to strange behaviour by Fred after the murder.
    • The entire police investigation and the prosecution was based on this hypothesis.
      • Every piece of evidence that could possibly exonerate Fred was ignored.
        • Including Fred van der Vyver’s alibi - eyewitness testimony that he was working in his office at Old Mutual at the time of the murder.
      • Any evidence that supported the hypothesis or that was ambiguous was interpreted to mean that he was guilty.
    • Ultimately acquitted, but not before his parents spent millions of rands on defending him and his career at Old Mutual was ruined.
    • Many followers of this story still believe Fred is guilty notwithstanding the objective exculpatory evidence.
      • The public perception of his guilt was so great that he has left the country. (Confirmation Bias and the Law, 2016)
  • Life imprisonment of Joy Rahman
    • In September 1994, home-help worker Joy Rahman was sentenced to life imprisonment for the murder of an elderly woman in Stockholm.
    • Reversed by a unanimous Court of Appeal almost 8 years later.
    • The causes of this outcome can be traced back to the preliminary investigation of the murder.
      • Rahman was established as the prime suspect at an early stage of the investigation.
      • He had bought coffee at the exact same time and place where two bottles of lighting fluid were sold.
        • The lighting fluid was of the same brand that was used in an attempt to burn the victim’s body
      • The weapon used to strangle the victim came from the apartment of another elderly woman who Rahman had visited on the day of the murder.
      • The son of the victim claimed that his mother had a large sum of money in the apartment, and Rahman was said to be in need of money (the motive).
    • The following facts that were brought up in the Court of Appeal illustrate that the investigation of the murder was seriously flawed.
      • It was not established whether it was actually Rahman who bought the bottles of lighting fluid or whether they were identical to those found at the crime scene.
      • It was not established whether the murder weapon had been taken from the other woman’s apartment on the day of the murder or whether Rahman was the only person who could possibly have taken it.
      • The crime scene showed no signs that anyone had searched the apartment for money.
        • None of Rahman’s colleagues at the home-help service knew about any money.
        • No evidence except for the claim of the victim’s son that the money even existed.
        • No support from an examination of Rahman’s assets for the claims that he was in need of money.
        • Rahman and his wife had a large amount of money saved in bank accounts.
          • The prosecutor argued that the savings were intended to finance the education of Rahman’s children and that his cultural background (Bangladeshi) made it morally impossible to spend the money; a claim based on ethnic stereotypes. (Ask & Granhag, 2005)

Debiasing

  • Be aware that it is a problem.
    • By understanding its effect and how it works, we are more likely to identify it in our decision-making.
  • Even if we have complied with similar requests in the past, we should not use that precedent as a reference point.
    • Recall past actions and ask yourself: “Knowing what I know now, if I could go back in time, would I make the same commitment?”
  • Since the bias is most likely to occur early in the decision-making process, focus on starting with a neutral fact base.
    • Diversify where we get our information from, and having multiple sources.
    • Reach for reputable, neutral outlets.
  • Set your hypothesis and look for instances to prove that you are wrong.
    • True definition of self-confidence: the ability to look at the world without the need to look for instances that please your ego.
  • Having reached a conclusion, continue reassessing whether our conclusion is correct as new information becomes available.
  • For group decision-making, obtain information from each member in a way that they are independent.
    • Avoid discussions between group members prior to providing the information to prevent unbiased members from influencing each other.
  • When hypotheses are being drawn from assembled data, consider having interpersonal discussions that explicitly aim at identifying individual cognitive bias in the hypothesis selection and evaluation.
    • Engage in debate as a way to challenge our views and expose ourselves to information we may have otherwise avoided.
  • Use a “preference-inconsistent” recommendation system/process.
    • Preference-inconsistent information is information that is in disagreement with a person’s prior preference on a topic.
    • Having preference-inconsistent recommendations on a controversial topic can lead to:
      • more moderate views on the topic,
      • better recall of the dissenting information, and
      • stimulation of divergent thinking in the form of increased argument generation and the number of novel arguments. (Schwind et al., 2012)

References

Twitter, Facebook