What Is the Function of Confirmation Bias?

In the paper

“Since individuals within their social groups depend on their reputation so as to find collaborators, anticipating one's audience's responses should be and have been more adaptive than having a confirmation bias (ibid).”

What the sources say

• Partly supported

  Section 3.2 argues that anticipating an audience’s objections seems plausibly more adaptive than myside bias, and connects the issue to reputation. It does not establish that this was historically more adaptive or say that reputation matters specifically for finding collaborators; the claim states those points more firmly and specifically than the paper does. Page numbers are unavailable in our copy. (LLM, conf=0.86)

  cites Dutilh Novaes 2018 sourceExperimental — accuracy not yet measured

In the paper

“They can create new perspectives on these traits that may allow developing means to reduce the traits' potential negative effects (Roberts et al. 2012;Johnson et al. 2013).”

What the sources say

• Supported

  In “The Promise of Applying Evolutionary Psychology,” the authors argue that evolutionary insights can inform efforts to address behavioral problems and provide a framework for practical applications. “New Directions” gives examples of using those insights to reduce conflict and other harmful consequences. The claim appropriately describes this as a possibility, not an established effect. (LLM, conf=0.88)

  cites Roberts et al. 2012 source
• ?Not checked — no full text of the sourcecites Johnson et al. 2013 source

In the paper

“Moreover, these effects don't necessarily take much time to unfold but can happen swiftly in a single interaction (e.g., in interview settings; Word et al. 1974) right after the ascription (Turnwald et al. 2018: 49).”

What the sources say

• ?Not checked — no full text of the sourcecites Word et al. 1974 source
• Supported

  The paper says a mindset may be shaped almost immediately upon learning genetic risk. Its Methods describe participants receiving risk information and then completing the test in the same session, supporting the claim that effects can unfold swiftly after an ascription. The cited paper concerns genetic-risk disclosure; the interview-setting example is attributed to Word et al., not Turnwald et al. (LLM, conf=0.87)

  cites Turnwald et al. 2018 source

In the paper

“Moreover, some researchers have recently re-examined the relevant studies and found that confirmation bias is in fact less common and the evidence of it less robust than often assumed (Mercier 2016;Whittlestone 2017).”

What the sources say

• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites Mercier 2016
• Supported

  Section 2.2.5 says that, on closer examination, the case for confirmation bias is much weaker than generally appreciated and questions its characterization as persistent and ubiquitous. Section 2.5 concludes that the evidence is less convincing than it might seem, supporting the citing sentence’s summary. (LLM, conf=0.90)

  cites Whittlestone 2017

In the paper

“Dardenne and Leyens (1995) reasoned that when testing a hypothesis about the personality of another individual (e.g., their being introverted or extroverted), a preference for questions that match the hypothesis (e.g., that the subject is introverted) indicates social skill, conveying a feeling of being understood to the individual and contributing to a smooth conversation.”

What the sources say

• ?Could not verify

  The available passages are repository metadata and lists of related publications, not the paper’s argument or results. Searches for hypothesis-matching questions, feeling understood, and smooth conversation found no passage addressing those points, so the claim cannot be checked against our copy. (LLM, conf=0.74)

  cites Dardenne and Leyens 1995 sourceExperimental — accuracy not yet measured

In the paper

“While this overgeneralization tendency often leads to errors, it is thought to have evolved because failures to respond favorably to babies (i.e., false negatives) are evolutionarily more costly than overgeneralizing (i.e., false positives) (ibid).”

What the sources say

• Partly supported

  The “Impressions Created by Variations in Babyfaceness—Baby Face Overgeneralization” section (page unavailable) links favorable responses to babies with evolutionary importance and says prepared responses to infantile cues may be overgeneralized to non-babies. It does not say that failing to respond favorably to babies is evolutionarily more costly than overgeneralizing those responses. (LLM, conf=0.85)

  cites Zebrowitz and Montepare 2008 sourceExperimental — accuracy not yet measured

In the paper

“Dardenne and Leyens did find that matching questions pertaining to an introvert or an extrovert hypothesis were selected most by high self-monitors: socially skilled subjects displayed a stronger confirmatory tendency than less socially skilled subjects (ibid).”

What the sources say

• ?Could not verify

  The indexed copy provides a publication record (p. 1229–1239), not passages from the study’s methods or results. Searches for high self-monitors, introvert/extrovert matching questions, and social skill found no passage addressing the claimed comparison, so the claim cannot be verified from this copy. (LLM, conf=0.76)

  cites Dardenne and Leyens 1995 sourceExperimental — accuracy not yet measured

In the paper

“And this tendency is overgeneralized such that even adults are more readily viewed more favorably, treated as likeable (but also physically weak, and naïve) when they display babyface features.”

What the sources say

• Supported

  The “Baby Face Overgeneralization” section says adults with babyfaces are perceived as naïve, weak, and warm, and experience social treatment consistent with those impressions. It also describes favorable responses to infantile facial cues being overgeneralized to non-babies. Page numbers were not provided in our copy. (LLM, conf=0.91)

  cites Zebrowitz and Montepare 2008 source

In the paper

“Since the latter are shaping social reality via their impact on behavior and are more effective in doing so the more convinced people are about them (Kappes et al. 2020)”

What the sources say

• Partly supported

  The Abstract (page unavailable) says that confident judgments from others can influence behavior, illustrating this with a juror’s verdict and a patient’s treatment decision. It does not establish that ideological conceptions shape social reality or that people’s conviction in their own conceptions makes them more effective at doing so. The Discussion (page unavailable) addresses resistance to changing opinions, not that broader claim. (LLM, conf=0.82)

  cites Kappes et al. 2020 sourceExperimental — accuracy not yet measured

In the paper

“One of the issues is that since individuals reproduce faster than groups, a trait T that is an adaptation that is good for groups but bad for an individual's fitness won't spread, because the rate of proliferation of groups is undermined by the evolutionary disadvantage of T within groups (Pinker 2012).”

What the sources say

• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites Pinker 2012

In the paper

“Furthermore, more than half of the subjects who believed they were assessing someone else's arguments now rejected those that were in fact their own, and were more likely to do so for invalid than for valid ones.”

What the sources say

• ?Not checked — no full text of the sourcecites Trouche et al. (2016) source

In the paper

“These points equally apply when it comes to considering the origin of confirmation bias.They provide good reasons for searching for an evolutionary account of the bias.”

In the paper

“Evolutionary explanations might also stimulate novel, testable predictions that researchers who aren't evolutionarily minded would overlook (Ketelaar and Ellis 2000;De Bruine 2009).”

What the sources say

• ?Not checked — no full text of the sourcecites Ketelaar and Ellis 2000 source
• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites De Bruine 2009

In the paper

“After all, their members traveled less, interacted less with strangers, and were less economically dependent on other groups (Simpson and Beckes 2010: 37).This should have homogenized them with respect to race, culture, and background (Schuck 2001(Schuck : 1915)).”

What the sources say

• ?Not checked — no full text of the sourcecites Simpson and Beckes 2010 source
• ?Not checked — no full text of the sourcecites Schuck 2001 source

In the paper

“Presumably, subjects accepted that they favored a particular position, didn't know the reasons, and so were now looking for support that would justify their position.They displayed a confirmation bias.”

In the paper

“And even if it is granted that objections depend on private information, they also often derive from obvious interests and public knowledge, making an anticipation of them easy (ibid).”

What the sources say

• ?Not checked — no full text of the sourcecites Sterelny 2018 source

In the paper

“Prima facie, the bias thus seems maladaptive.”

In the paper

“This can result in significant developmental and epistemic benefits for us and other people, ensuring that over time we don't become epistemically disconnected from social reality but can navigate it more easily.”

In the paper

“While that might not be the only evolved function of confirmation bias, it is an important one that has so far been neglected in the theorizing on the bias.”

In the paper

“Even if confirmation bias is maladaptive for individuals, it might still be adaptive for groups.”

In the paper

“Social beliefs often lead to SFPs with beneficial outcomes.”

In the paper

“SFPs are, however, neither pervasive nor all-powerful (Jussim 2012), and there are various conditions for them to occur (Snyder and Klein 2007).”

What the sources say

• ?Not checked — no full text of the sourcecites Jussim 2012 source
• ?Not checked — no full text of the sourcecites Snyder and Klein 2007 source

In the paper

“And the operation of the former can be understood as motivated reasoning itself, because it too involves partial information processing to buttress a view that one wants to be true (ibid).”

What the sources say

• ?Not checked — no full text of the sourcecites Nickerson 1998 source

In the paper

“For there is evidence that, for instance, disrupting the fluency in information processing (Hernandez and Preston 2013) or priming subjects for distrust (Mayo et al. 2014) reduces the bias.”

What the sources say

• ?Not checked — no full text of the sourcecites Hernandez and Preston 2013 source
• ?Not checked — no full text of the sourcecites Mayo et al. 2014 source

In the paper

“Granted, even if the bias isn't an adaptation, we might still be able to explain why it is with us today.”

In the paper

“So confirmation bias is also importantly different from motivated reasoning, as it can take effect in the absence of a preferred view and might lead one to support even beliefs that one wants to be false (e.g., when one believes the catastrophic effects of climate change are unavoidable; Steel 2018).”

What the sources say

• ?Not checked — no full text of the sourcecites Steel 2018 source

In the paper

“Unmotivated confirmation bias, however, operates when people process data in one-sided, partial ways that support their predetermined views no matter whether they favor them.”

In the paper

“Despite overlapping with motivated reasoning, confirmation bias can thus plausibly be (and typically is) construed as a distinctive cognition.”

In the paper

“Moreover, since our ancesteral groups are (as noted above) likely to have been largely viewpoint homogenous, in supporting intention-alignment in these social environments, confirmation bias would have again facilitated group polarization, which is prima facie evolutionarily disadvantageous.”

In the paper

“Confirmation bias is typically taken to work against an individual's truth tracking (Mercier and Sperber 2017: 215;Peters 2018: 15), and indeed searching for information supporting one's beliefs and ignoring contradictory data is epistemically disadvantageous when what one takes to be reality is and stays different from what one believes it to be.”

What the sources say

• ?Not checked — no full text of the sourcecites Mercier and Sperber 2017 source
• ?Not checked — no full text of the sourcecites Peters 2018 source

In the paper

“By definition, dysfunctional traits should be more prone to extinction than functional ones (Nickerson 1998).”

What the sources say

• ?Not checked — no full text of the sourcecites Nickerson 1998 source

In the paper

“For instance, they tend to occur only when targets are able to change in accordance with the trait ascriptions, when the latter are believable rather than unrealistic (Alfano 2013: 91f), and when the ascriber holds more power than the ascribee (Copeland 1994: 264f).”

What the sources say

• ?Not checked — no full text of the sourcecites Alfano 2013 source
• ?Not checked — no full text of the sourcecites Copeland 1994 source

In the paper

“Goodwin et al. (2000) found just that: powerful people, in particular, tend to fail to take into account data that may contradict their social beliefs (capturing, e.g., stereotypes) about subordinates and attend more closely to information that supports their expectations.Relative to the powerless, powerful people displayed a stronger confirmation bias in their thinking about subordinates (ibid: 2”

What the sources say

• ?Not checked — no full text of the sourcecites Goodwin et al. 2000 source

In the paper

“First, the RM account is compatible with the view that confirmation bias was also selected for adaptive effects related to non-social beliefs.”

In the paper

“If, as the RM account proposes, confirmation bias was selected for facilitating the process of making reality match our beliefs, then the bias should be common and pronounced when (1) it comes to social beliefs, that is, beliefs (a) about oneself, (b) about other people, and (c) about social structures that the subject can determine, and when (2) social conditions are conducive to reality matching”

In the paper

“This insignificant confirmation bias with respect to beliefs about non-social matters contrasts with findings of a significant confirmation bias with respect to beliefs about people (Talaifar and Swann 2017;Goodwin et al. 2000;Marks and Fraley 2006;Darley and Gross 1983), and, as I shall argue now, social affairs whose reality the subject can determine.”

What the sources say

• ?Not checked — no full text of the sourcecites Talaifar and Swann 2017 source
• ?Not checked — no full text of the sourcecites Goodwin et al. 2000 source
• ?Not checked — no full text of the sourcecites Marks and Fraley 2006 source

In the paper

“Jill's biased processing is thus more adaptive than Jack's approach: a confirmation bias provides her with certainty and simpler information processing that simultaneously facilitates accurate predictions (via contributing to SFPs).”

In the paper

“If confirmation bias was selected for sustaining social beliefs for the sake of reality matching then the bias should also be particularly pronounced when it comes to beliefs about other people especially in situations conducive to reality matching.”

In the paper

“as mentioned above, motivated confirmation bias is an instance of motivated reasoning (Nickerson 1998)”

What the sources say

• ?Not checked — no full text of the sourcecites Nickerson 1998 source

In the paper

“Socially skilled people ('high self-monitors') should thus request 'matching questions', say, in an interview setting, for instance, when testing the introvert hypothesis, an interviewer could ask questions that are answered 'yes' by a typical introvert (e.g., 'Do you like to stay alone?'),confirming the presence of the hypothesized trait (ibid).”

In the paper

“the special role of social beliefs in our species (compared to others species) lends plausibility to the assumption that the costs of the bias' overgeneralizing might be lower than the costs of its failing to generalize.”

In the paper

“Similarly, if confirmation bias serves to facilitate social interaction by contributing to a match between beliefs and social reality then the bias should be increased with respect to trait attributions to other people in subjects who care about social interactions compared to other subjects.”

In the paper

“The RM account, on the other hand, proposes that the bias often facilitates social understanding: in making us less sensitive to our interlocutor's opposing perspective, it helps us remain confident about our social beliefs, which increases the probability of SFPs that in turn make people more predictable and mindreadable.”

In the paper

“This was problematic because the bias hinders us in anticipating people's objections, which weakens our argumentative skill and increases the risk of us appearing incompetent in argumentative exchanges.”

In the paper

“And it casts new light on the epistemic character of confirmation bias.”

In the paper

“When the bias targets positive social beliefs, this serves both the subject and the group, assisting them in their development (to become, e.g., more communicative or knowledgeable) while also making their social cognition more effective and tractable.”

In the paper

“If we attend to it, an account of the function of confirmation bias becomes available that coheres with data from across the psychological sciences, manages to avoid many of the shortcomings of competitor views, and has explanatory benefits that help advance the research on the function, nature, and epistemic character of the bias.”

In the paper

“The RM account is preferable to the group-cognition view because it doesn't rely on a notion of group selection but concerns primarily individual-level benefits, and it doesn't tie the adaptive effects of the bias to conditions of viewpoint diversity.”

In the paper

“When it targets negative social beliefs, in promoting reality matching, the bias might contribute to ethically problematic outcomes, but it can then still support social conformity and predictability, which were perhaps especially in ancestral tribal groups adaptive.”

In the paper

“For a general tendency to think that one's views rest on good reasons even when one doesn't know them is a tendency to favor and confirm these views while resisting balanced scrutiny of their basis.”

In the paper

“For instance, the RM account pertains to the function of confirmation bias, not folk psychology.”

In the paper

“It proposes instead that the adaptive SFP-related effects of the bias increase individuals' fitness (e.g., by facilitating their navigation of the social world, aligning them/others with their group's conceptions etc.) and can emerge whenever people hold beliefs about each other, interact, and fulfill social expectations.”

In the paper

“However, in ancestral, 'tribal' groups of highly interdependent subjects, even accumulative SFPs of, e.g., stereotypes would perhaps still have contributed to conformity and social stability.”

In the paper

“If it is a general default tendency to assume that any view that one holds rests on good reasons, then that would again suggest the presence of a confirmation bias.”

In the paper

“Second, it is consistent with, and provides new contributions to, different areas of evolutionary theorizing on human cognition.”

In the paper

“The RM account avoids this problem because it suggests that by disregarding objections or counterevidence to one's beliefs, one can remain convinced about them, which helps align social reality (not only, e.g., people's intentions) with them, producing the adaptive outcomes outlined above.”

In the paper

“The RM account avoids these problems: if confirmation bias was selected for reinforcing our preconceptions about people to promote SFPs then, since in one's own reasoning one only needs to justify one's beliefs to oneself, the first point one finds acceptable will suffice.”

In the paper

“Yet, it is doubtful that ancestral groups met this condition.”