Consequences of overconfidence
There are many negative consequences of knowledge miscalibration for both individual and societal wellbeing. As we have stated, people tend to be overconfident in their knowledge, know less than they think, underestimate or neglect true underlying complexity, and confuse exterior knowledge and information for their own.
Research from decision affect theory has shown that this overconfidence can lead to real affective consequences for decision makers. In two studies where recreational basketball players made predictions about their shooting accuracy, McGraw et al. (2004) showed that overconfidence leads to less pleasure in response to predicted successes (because they are less surprising) and more pain in response to unexpected failures (because they are more surprising).These affective consequences were the result of overconfidence in a trivial task, but overconfidence and knowledge miscalibration can also lead to underestimation of risks, with significant material consequences. Research on consumer financial decision making has shown that higher self-assessments of financial knowledge lead to higher likelihoods of people making risky investments, and are correlated with lower subjective ratings of risk (but not objective risk) (Hadar et al., 2013; Long et al., 2018;Ward, Grillo, and Fernbach, 2019). For most, losing money as a result of underestimating financial risk would be affectively painful on its own, and could potentially cause additional stress if paying for bills becomes difficult.Overconfidence among people with objectively high levels of knowledge is often manifested as excessive attitude certainty. Instead of overestimating their knowledge, they rely too heavily on it or reject external information that may be relevant or helpful. Wood and Lynch (2002) found that people with high prior product knowledge committed memory encoding errors and had lower motivation to search for new information, caused by superficial information processing.
These findings are consistent with work by Alba and Hutchinson (1987) who argued that overconfidence can lead experts to shorten the search for answers, find false causal connections, and misinterpret objective facts. In the domain of medical decision making the consequences of overconfidence can be a matter of life and death. Dawes, Faust, and Meehl (1989) argued that medical experts make poorer choices than simple computer programs based on those experts' decision processes, which, they posit, is partially a consequence of the human propensity to overattend to information consistent with hypotheses and underattend to contradictory information (Greenwald, Pratkanis, Leippe, and Baumgardner, 1986). Among political and military leaders, overconfidence can lead to initiation of too many wars, military overcommitment, and loss of life (D. D. P Johnson, 2004).These phenomena are important due to the connection between higher perceptions of knowledge and the tendency to rely on the knowledge (Krosnick and Petty, 1995).There are also several streams of research pointing to extremity of beliefs as a consequence of knowledge miscalibration. When people overestimate their knowledge they are more likely to hold extreme views and fail to realize that they do not have enough knowledge required to justify their extreme positions (Fernbach, Rogers, et al., 2013). In three experiments, Fernbach et al. (2013) showed that asking people to explain policies in detail reduced subjective perceptions of understanding and led to more moderate political attitudes, which suggests that political polarization is at least partially caused by people's overestimation of their knowledge of the issues (also D. R. Johnson, Murphy, and Messer, 2016; but see Crawford, n.d.).
People also hold more complex cognitive representations of their own groups than of others, which leads them to evaluate others more extremely (Linville 1982). In this literature on the relationship between complexity and extremity of evaluations (see also Linville 1985, 1987), complexity is operationalized as the perceived number of non-redundant features or dimensions of people, objects, or issues—the more perceived dimensions, the higher perceived complexity. Linville (1982) manipulated the complexity of knowledge structures: participants in one study were instructed to think of either two (simple condition) or six (complex condition) attributes of a cookie before rating it.
The participants in the simple condition gave higher ratings to the cookies they liked and lower ratings to the cookies they disliked, while participants in the complex condition gave more moderate ratings for all cookies. The fact that the complexity— extremity effect is present when people evaluate policies, people, and simple objects suggests that it is broadly generalizable and is a product of how humans mentally represent and store knowledge.The mechanism for these extremity effects is that it is more difficult for someone to think of any person, object, or issue in black and white when cognitively representing them with more dimensionality.With increasingly more labels to attach to something, the weight each label can possibly contribute to an overall evaluation mathematically decreases.Because both knowledge overconfidence and extreme beliefs are at least partially caused by oversimplified mental models, lack of intellectual humility and extreme beliefs often coincide. Additional evidence for the link comes from work on opposition to applications of genetic engineering. Fernbach, Light et al. (2019) examined the relationship between self-assessed knowledge, objective knowledge, and extremity of participants' opposition to two applications of genetic engineering (on which there are scientific consensuses of safety): genetically modified foods and gene therapy. For both issues, they found that the most extreme opponents thought they knew the most about genetically modified foods (or gene therapy), but actually knew the least, scoring the lowest on a battery of true-false science and genetics questions. Similar patterns of results have been found in the domains of anti-establishment voting in the Netherlands (van Prooijen and Krouwel, 2018), and opposition to vaccination (Motta, Callaghan, and Sylvester, 2018). More generally, work on metacognition showed that participants with extreme political beliefs exhibited less insight into the correctness of their assessments of simple visual displays (Rollwage, Dolan, and Fleming, 2018).These examples suggest that extreme views often reflect low objective knowledge paired with high self-assessments of knowledge.This is doubly problematic, since gaining knowledge frequently has the effect of revealing nuance and complexity to those who gain it (Rozenblit and Keil, 2002), which in turn reduces extremity of belief.
Taken together, these findings suggest that overconfidence can lead to inaccurate predictions, lost money, wasted time, dissatisfaction, and entrenched false beliefs.We believe they also underscore an underappreciated obstacle: those who are the most overconfident or miscalibrated in their knowledge are the most likely to hold extreme views and the most in need of humilityinducing knowledge, but also the least likely to be receptive to learning. As previously discussed, knowledge overconfidence is correlated with decreased openness to new information (Wood and Lynch, 2002), and people are more likely to reconsider their positions on policies or issues when they have less confidence in their knowledge (Krosnick and Petty, 1995).To make matters worse, increasing the difficulty of a task or decision process has also been shown to increase overconfidence (Larrick, Burson, and Soll, 2007), so people may be most likely to overestimate their understanding and make judgement errors in the most complex and critical situations.This suggests that a necessary condition for increasing intellectual humility is getting people to realize gaps in their knowledge.
35.4