Why Explaining One Thing Can Lower Confidence About Another

Using a zipper every day is not the same as being able to explain, step by step, how its parts bring two sides together. Attempting the explanation can reveal a gap between familiarity and a detailed causal account. A stranger possibility is that the attempt could also change how well someone thinks they understand snow formation. No new meteorological fact need have been learned. What changes may be a judgment about knowledge in a different domain.

Ethan A. Meyers and colleagues tested this broader adjustment in a 2023 paper. Across three experiments, people rated their understanding of several phenomena, attempted an explanation of one, and rated all of them again. Reassessing the unexplained items made the design distinctive. It could investigate whether an explanatory experience changed only the judgment attached to its immediate target, or also judgments about other things the participant believed they understood. [1]

Separating the target of explanation from the target of judgment

The illusion of explanatory depth concerns believing that one understands a mechanism in greater detail than one can actually explain. Saying what an object does and describing the causal sequence by which its components work together are different demands. An explanation request can make the second demand explicit after a person has based an initial judgment on a more general impression.

In a standard task, people rate understanding, explain the target phenomenon and rate that same phenomenon again. A lower second rating can be interpreted as recognizing gaps in the explanation. That interpretation alone does not determine whether the adjustment is topic-specific. Perhaps the experience leads people to reassess their knowledge more broadly. The new experiments separated these possibilities by collecting second ratings for both explained and unexplained phenomena. [1]

In the first experiment, 240 online adult participants rated six mechanical devices on a seven-point understanding scale. Each attempted to explain one assigned device, then rated all six again. The explained item and the five unexplained items were consequently compared within the same participants. They were not wholly separate groups of people performing unrelated tasks. [1]

Why the first result left an important question open

On average, understanding ratings declined for both explained and unexplained devices. But the first experiment had no group that made two judgments without generating an explanation. A second opportunity to consider the question might itself lower a rating. Alternatively, the experience of explaining might change the meaning of understanding: an initially broad sense of familiarity could become a stricter requirement to produce a causal chain.

Those alternatives do not make the initial finding uninteresting. They identify what the next comparison needs to address. A difference between a first and second rating cannot immediately be assigned entirely to the intervening activity. Repetition, additional reflection or a changed interpretation of the scale might contribute. The second experiment altered the procedure to examine those possibilities. [1]

Copying a paragraph instead of generating an explanation

One group again explained a device. A control group manually transcribed a descriptive paragraph about a room from an image into a text box, preventing a simple copy-and-paste response. The control therefore involved reading and typing, but not generating an account of how a mechanism worked. It was a different activity, not an empty period in which nothing happened.

The rating instructions were also clarified at the outset. Participants were asked to consider their ability to explain the causal chain when assessing understanding. This reduced the opportunity for that requirement to appear for the first time only after the explanation task. It did not guarantee that every participant interpreted every word identically, but made a specific scale-reinterpretation explanation less compelling. [1]

Ratings fell in the control group too. Nevertheless, the decline was greater following explanation, including for devices that had not been explained. Repeated judgment could not account for the entire observed pattern. Equally, it would be wrong to describe the control group as unchanged. Understanding the result requires retaining both observations rather than choosing the more convenient one. [1]

Beyond related mechanical devices

The third experiment replaced four of the six items, adding natural phenomena such as snow formation and earthquakes alongside economic and political systems. This tested whether the broader adjustment was restricted to a collection of mechanical devices. On average, attempting an explanation also reduced subsequent ratings for previously rated phenomena in other domains. [1]

This was not a uniform rule for every person and item. For some phenomena, participants' ratings increased after explaining that phenomenon while their ratings of other phenomena decreased. Nor was it clear that explaining the same topic always produced a larger decline than explaining a different one. Treating the aggregate pattern as an identical response in every case would conceal variation explicitly noted by the authors.

Most importantly, a lower rating for an unrelated topic is not evidence that actual knowledge of that topic deteriorated. The outcome concerns self-assessment. A lower rating is also not automatically a more accurate rating. Establishing improved calibration would require comparing the judgment with an appropriate criterion for actual knowledge. Feeling less knowledgeable and judging one's knowledge more accurately are different claims.

Possible accounts without a single settled mechanism

One interpretation links otherwise separate illusions. Difficulty explaining one topic might encourage a person to question apparent understanding of others. Another account treats the pattern as an expression of general overconfidence. The experiments did not conclusively distinguish those possibilities. Nor does including a control condition definitively eliminate every influence of participants' expectations about how they should respond. [1]

The studies were preregistered, but the paper explicitly reports presenting simpler analyses following an editorial suggestion instead of adhering unchanged to the original analysis plan. The reported tests were directional, or one-tailed. Imbalances between conditions also raised the possibility of selective attrition. Preregistration should therefore not be used as shorthand for every reported analysis being exactly the one originally planned. These distinctions belong in evaluating the evidence.

There is a further structural issue in comparing explained with unexplained items. A participant explained one item but rerated five others. Averaging across those five produces a different amount of information and variability from the single explained item. Larger aggregate changes or test statistics should not casually become a claim that explaining something unrelated is universally the stronger intervention. The authors themselves cautioned against that interpretation. [1]

Reassessment is not a guarantee of lasting humility

The rerated phenomena had all been presented before the explanation attempt. The experiments did not establish that a topic introduced for the first time afterward would receive the same adjustment. They also did not measure how long the effect lasted. Whether a teacher, parent or peer asking similar questions would produce the same response outside the experimental arrangement remained open. [1]

The findings are therefore not a demonstrated program for permanent intellectual humility or a method guaranteed to resolve disagreements. They show something more specific: attempting to explain one mechanism can alter reported understanding beyond that immediate target, over and above a smaller change associated with repeated judgment.

That distinction makes explanation useful as an object of investigation without turning lower confidence into an achievement by definition. The central questions remain what someone knows and what that person thinks they know. An explanation attempt can make the distance between those questions more visible. The experiments do not guarantee that the activity always closes it. [1]

Sources

[1] Meyers, E. A., Gretton, J. D., Budge, J. R. C., Fugelsang, J. A., and Koehler, D. J. (2023). Broad effects of shallow understanding: Explaining an unrelated phenomenon exposes the illusion of explanatory depth. Judgment and Decision Making, 18, e24. https://doi.org/10.1017/jdm.2023.24

The source study is licensed under Creative Commons Attribution 4.0: https://creativecommons.org/licenses/by/4.0/ This is an original explanatory assessment, not an official translation.