Reference: Eliseev, E. D., & Marsh, E. J. (2023). Understanding why searching the internet inflates confidence in explanatory ability. Applied Cognitive Psychology.
Learning before and during the digital age
Imagine that you have invited some friends and family to your place for Thanksgiving dinner. At the table, someone brings up the 2024 presidential elections, which devolves into a heated discussion of the latest political developments in the country – and about the October federal government shutdown, in particular. Confusion arises as two of your friends disagree on whether courts still operate during shutdowns and, if so, for how long. To settle the debate, you set on finding the answer. But where can you find it?
In this day and age, you would probably look for it online. You would quickly tap the question into your phone’s browser, skim through a few links and report back to your friends. A couple of decades ago that would not have been an option. Back then, you might have turned to the quiet friend who conveniently left for the bathroom when the debate began or dusted off the old public administration encyclopedia you once bought for a college course.
Does the choice of where to find this information matter? Maybe not for settling the debate, but the way we seek out knowledge shapes how well we learn – and how well we think we’ve learned.
What happens when we seek knowledge online?
Researchers Eliseev and Marsh (2023) set out to explore the extent to which searching online for explanations influences how well we learn them, and how accurate our assessment of this learning is. In Experiment 1, participants in a control condition estimated how well they would answer several questions (e.g., why are there leap years?) using only their existing knowledge. Other participants provided the same ratings after searching online for specific reliable websites with the answers (“search” condition), and in a third condition participants merely read these websites’ answers (without having to search for them; “read” condition).
Results indicated that participants who searched for explanations rated themselves the highest in their ability to properly explain the phenomena they learned about. Those that read explanations (rather than searching for them) were also more confident in their ability to explain the phenomenon than participants in the control condition.
But did confidence in one’s knowledge translate into being better at providing explanations? To investigate whether participants’ perceived explanatory ability aligned with their actual understanding, Eliseev and Marsh (2023) had trained research assistants rate the accuracy of each participant’s written explanations which they wrote at the end of the experiment. As expected, participants in the control condition – who had not been exposed to any information on the topic – were less accurate in their explanations than both those that merely read and those that searched for explanations.
In an internal meta-analysis looking at the effects across the four experiments conducted by the authors, those that simply read explanations were also more accurate when generating their own explanations than those asked to search for the explanations online. Though searching online made people more confident in their explanatory ability, part of this confidence was unwarranted. They actually were not able to report the information as accurately from searching compared to the (less confident) participants that only read the explanations.
Why does searching online inflate perceptions of knowledge?
Participants who searched for explanations may have developed overinflated beliefs about their knowledge due to additional superficial information on the websites that they visited, such as images or visual cues. In contrast, participants who only read the explanations saw them in plain text, without these extra elements. To test this account, in Experiments 2A-2B, Eliseev and Marsh (2023) replaced the format of the explanation in the “read” condition, such that participants in this condition were presented with screenshots of the pages that those searching for the explanations would land at. Therefore, in this scenario, there were no visual differences in the explanations that participants read initially (i.e., in the “search” and “read” conditions), regardless of whether these were provided to them or whether they had to search them online.
If differences in the explanation format explained the results of the first experiment, in Experiments 2A-2B the results should be the same across “search” and “read” conditions. However, the original results held: While participants that searched for explanations believed themselves to be more capable of explaining the phenomenon than participants that read explanations without searching for them, the first group was less accurate in their actual explanations than the second. This means that differences in the format of the information and other superficial visual cues in websites do not explain the overinflated confidence of people who search online.
An alternative explanation could be that those who search online become more confident because they are exposed to additional information – not just information in a different format – such as partial snippets or multiple related webpages. These can prime the to-be-learned content and foster a sense of familiarity that translates into thinking that one has mastered the explanation. So, in Experiment 3, the authors replaced the control condition with one where participants were exposed to snippets of information prior to being handed the target explanation (as in the “read” condition). That is, they added a “preview” condition to the “search” and “read” conditions (and dropped the control condition).
If inflated perceptions of one’s knowledge arise because search snippets prime the relevant information, then participants in this new condition should feel more confident than those who only read the explanations, and their responses should resemble those who actively searched online, since they too were exposed to snippets. The actual results met these predictions: Participants in the “preview” and in the “search” conditions did not differ in their perceived explanatory ability and both rated their confidence higher than participants that read the explanations but were not exposed to snippets.
All in all, people consider themselves more capable of addressing questions presented to them after searching for the answers online. However, the act of looking up these answers online does not necessarily lead to being more capable of doing so – which means that people are often overconfident in their explanatory ability. This is because we become familiar with the information that we encounter later when searching online, as it is presented in search snippets. We incorrectly interpret this familiarity with being knowledgeable on the topic.
Take-home message
Next time that you have friends over and a debate ensues – beware! The means through which you learn the information you need to settle a dispute influence whether you accurately capture how much you have learned or not. Be on the lookout for overconfidence borne out of being exposed to bits of information, such as snippets or page previews, when navigating the internet. Be sure to engage with information deeply and attentively when searching online and monitor what you effectively learned. We may all be prone to overconfidence under the right circumstances!
Image credits:
Featured image: Photo by Nenad Stojkovic, under Creative Commons License CC BY 2.0. https://www.flickr.com/photos/nenadstojkovic/50202556601/
cogbites 