Efforts to increase intellectual humility

If lacking intellectual humility is caused in part by lacking knowledge, an intuitive recom­mendation for mitigating knowledge overconfidence might be to simply educate people about important issues on which their knowledge is lacking.

There is another seemingly intuitive approach to mitigating intellectual overconfidence in the domain of public acceptance of science.The thinking is that if teaching people the facts that they lack does not cause meaningful attitude change, maybe simply communicating that there are scientific consensuses on these important issues (and what those consensuses are) will make an impact. Recent research has tested this approach in a randomized online survey experiment (Landrum, Hallman, and Jamieson, 2019). Specifically, this work tested the effect of different messages on subjects’ evaluations and acceptance of the scientific consensus on the safety of genetically modified organisms.

These attempts and the previously mentioned findings regarding the prevalence of high self-assessments of knowledge paired with low objective knowledge among extremists sug­gest that the challenge of improving intellectual humility may not just be a matter of inform­ing people about facts, but of encouraging them to appreciate that issues are complex, and they know less about them than they think.There have been some initial inroads along these lines. One team of researchers explored the connection between knowledge and perceived absoluteness of information surrounding contentious topics.Their findings show that people shift their understanding of the nature of truth to match certain goals in interactive sce­narios (Fisher, Knobe, Strickland, and Keil, 2017). Specifically, Fisher et al. (2017) found that people view the truth of contentious issues as less absolute when adopting a goal to learn when arguing, but more absolute when adopting a goal to win the argument (or with no experimentally manipulated goal at all). Because the authors found no significant differences in their dependent measure between the argue-to-win and control conditions, they con­cluded that a learning goal in an argument caused participants to develop a more subjective view of incoming information. The authors cited previous work showing that goal-directed cognitive processes are more likely to acted upon when a corresponding goal is active (Xu and Wyer, 2008), and proposed that the mechanism for their overall findings is that a learn­ing goal causes people to place higher information value on incoming messages.

We believe that the way in which humans store and access knowledge provides additional evidence for why argument goal interventions may be promising for increasing intellectual humility surrounding contentious issues. Evaluating the validity of incoming information about processes, domains, or truth in general entails evaluating and drawing on one's own knowledge to use as a comparison (Hofer, 2017). Fox and Tversky (1995) found that people avoid options about which they know less in a joint evaluation, but do not when options are evaluated sepa- rately.This finding suggests that people may be led to realize their ignorance of a topic if assess­ing their knowledge in comparison with another topic about which they know comparatively more (“comparative ignorance” Fox and Tversky, 1995).Assessments of knowledge are also fun­damentally social (Chinn, Buckland, and Samarapungavan, 2011; S. Sloman and Fernbach, 2017). Evidence for this comes from the previously mentioned idea that knowledge miscalibration is partially driven by people confusing their access to others' knowledge for their own internally held knowledge, which allows them to deal with a world of infinitely deep complexity by out­sourcing it to others (Keil et al. 2004). Knowledge is also inherently goal-directed (Chinn et al., 2011; Sandoval, 2017). People seek knowledge in order to do things, and rarely obtain knowl­edge for its own sake. If humans are hardwired to use others to store knowledge and expertise, and seek knowledge in order to achieve goals, it makes sense that adopting a learning goal in an argument—with an explicit aim of receiving previously outsourced knowledge from others to be put to use—would positively affect people's openness to new information and perceptions of their own knowledge.

In terms of interventions designed to increase intellectual humility and decrease knowl­edge miscalibration, we have discussed education attempts, appeals to expert consensus, and goal interventions.

In addition to these interventions that have already been tested, we believe that several others may be promising. The first potential intervention is to attempt to draw people's attention to the distributed nature of knowledge (e.g., by asking them to identify who has specific pieces of knowledge). Experimental subjects could be asked to indicate who has—and where they would find—the information necessary to make the most informed decision regarding a specific policy, technology, topic, etc. A more abstract version of this type of intervention would be to simply remind people that they may not know as much as they think about particular issues, but that, in a real situation, they would have access to broad expertise. In both cases, experiments would measure perceived understanding and attitude strength after the interventions in order to test their effectiveness at making subjects more humble and less extreme in their positions.

Given that the IOED manipulations reduce perceptions of understanding/knowledge but would be difficult to implement on a larger scale outside of a lab setting, our second proposed direction would be to design similar mechanism and analogy interventions that target specific false beliefs. In these studies, we would collect individual difference measures that indicate the categories of concerns that drive attitudes, and customize interventions to address those con­cerns.

We also believe that there is a broader opportunity to design interventions that manipulate subjects' perceptions of complexity. Specifically, there may be promise in using manipulated complexity to make domains, technologies, or issues seem more or less similar to each other by experimentally equating them on number of dimensions, attributes, or themes. Evers, Inbar, and Zeelenberg (2014) found that people prefer sets of items that are less complex (with products within sets perceived to be more similar to each other) than more complex sets (with products perceived to be less similar to each other), even if the more complex set contains a more pre­ferred item. These studies suggest that the constructs of complexity and similarity are highly negatively correlated. As a result, complexity manipulations could help encourage the realiza­tion of the complexity of contentious scientific policies or technologies by equating them on complexity with more accepted complex issues. On the other hand, complexity manipulations could be deployed to encourage the belief that an issue or technology is not similar to others that are unpopular with specific groups.

35.7