Cognitive fixation from Google searches hurts a team’s ability to innovate

When it comes to brainstorming, internet access might be more of a double-edged sword than a creative superpower. A new study published in Memory & Cognition suggests that while using Google can help individuals come up with more ideas for alternative uses of everyday objects, it can actually hinder creativity at the group level. The findings reveal a surprising cost of internet use: groups of people working independently without online assistance tended to produce a more diverse and novel set of ideas than groups with internet access.

The research was prompted by growing interest in how digital tools influence cognition, particularly creative thinking. Previous studies have explored the effects of the internet on memory, attention, and metacognition. Some researchers, such as Mercedes T. Oliva, Benjamin C. Storm, have even found that online search can help individuals come up with ideas they might not have thought of on their own.

But exposure to examples—especially ones retrieved via search engines—can also make people conform to a narrower range of responses. This phenomenon, sometimes referred to as “fixation,” limits the range of ideas that emerge, especially when people are working in groups and drawing from the same pool of online content.

“I had a student who was interested in the topic and was asking me a lot of questions I couldn’t answer,” said study author Danny Oppenheimer, a professor of decision sciences and psychology at Carnegie Mellon University and author of Psychology: The Comic Book Introduction.

“When we started the project, right in the middle of COVID, nobody really knew what using the internet was doing to our creative thinking. Oliva and Storm’s data hadn’t come out yet. When they published their data, we had already been thinking about creativity for a while, and we realized that while it spoke to the question of individual creativity, it didn’t explore collective creativity, so we wanted to keep working on the project to follow up on what happened to groups.”

Free daily newsletter

To investigate this issue, the research team conducted a large experiment with 256 undergraduate students. Participants were randomly assigned to one of two conditions: some were allowed to use Google during a classic brainstorming task known as the alternative uses task, while others were instructed to stay offline.

Each participant had three minutes to list as many nonstandard uses as they could for either a shield or an umbrella. These two objects were chosen because they varied in how helpful Google would likely be for generating suggestions. For instance, Google searches for umbrellas tended to produce many results, while searches for shields offered only a couple of standard ideas.

Importantly, the researchers ran the study in person and monitored participants’ internet use to ensure compliance. The students typed their answers into a text box during the allotted time. After the task, researchers categorized the responses based on whether they were conceptually distinct. This allowed the team to analyze how many unique ideas were generated across different conditions.

At the individual level, results showed that internet access only boosted idea generation when Google offered many suggestions. For instance, participants brainstorming uses for umbrellas came up with significantly more ideas if they had internet access. But for shields—where online search offered little help—there was no difference in idea fluency between the groups. These results replicated findings from an earlier study by Oliva and Storm, though the current study improved upon the design by removing the five-idea cap used in that earlier research.

However, the picture changed when researchers aggregated individuals into “nominal groups”—a technique often used in brainstorming studies to assess collective output by combining the responses of individuals who worked separately. When looking at group-level creativity, the researchers found that larger nominal groups without internet access consistently outperformed those with access. This was true even when Google provided helpful suggestions.

The reason for this pattern appears to be that Google users tended to converge on the same ideas. For example, multiple participants in the Google condition not only listed the same uses for umbrellas but often listed them in the same order. In contrast, participants without internet access generated a wider range of distinct and often unusual ideas. As group size increased, the benefits of this idea diversity outweighed the individual boost provided by Google. The group without Google ended up producing more unique uses overall.

To further probe these findings, the team also analyzed the frequency of unique versus common responses. In the umbrella condition, non-Google users were responsible for a greater share of one-of-a-kind ideas. Even when internet access led to more ideas per person, those ideas were often clustered around the same common examples, reducing the group’s overall creative diversity.

The study also explored other ways of measuring creativity, beyond just counting ideas. Independent coders rated each idea on how novel, effective, and creative it seemed. Across most comparisons, groups without internet access generated more highly rated ideas, particularly when creativity was defined as a combination of novelty and usefulness. While the ratings varied slightly depending on who did the coding and which threshold was used, the broader pattern held: larger groups without Google access typically produced higher-quality ideas.

To test the robustness of these findings, the researchers reanalyzed data from the original Oliva and Storm study. Even though that earlier study limited each participant to five ideas, the current team applied the same nominal group technique and found a consistent pattern: as group size increased, the advantage of offline brainstorming grew stronger. In fact, across all 20 analyses they conducted with these archival data, groups without Google access generated more distinct ideas than those with access.

Perhaps the most striking result came from an analysis of the “best” ideas—those rated a perfect 5 out of 5 on creativity. Of the 20 top-rated ideas in the Oliva and Storm dataset, 95% came from participants who had not used Google. This suggests that the internet may help people generate good ideas quickly, but it may also prevent the truly exceptional ones from emerging.

The authors argue that these results align with past research on cognitive fixation. When people are shown examples, even unintentionally, their thinking becomes constrained, making it harder to break out of conventional patterns. The internet functions much like a giant set of examples, guiding users toward solutions that others have already thought of. This can be helpful for routine problems—but it may be counterproductive for creative tasks that require original thinking.

The study highlights a potential “tragedy of the commons” for creativity. While individuals may benefit from using Google during idea generation, collective innovation may suffer if everyone relies on the same digital tools. This has implications for workplaces, classrooms, and any setting where brainstorming and creative collaboration are encouraged.

“Using the internet is a great way to help come up with ideas – especially if you’re thinking about topics that lots of other people have thought about and so there’s a large record of ideas on the internet to sample from,” Oppenheimer told PsyPost. “But the internet will help you think of ideas that other people have already thought of, and will guide you to think in ways that are similar to other internet users.”

“This means that if everybody is using the internet, everybody comes up with the same ideas. The collective comes up with fewer ideas even as individuals within that collective come up with more ideas. When brainstorming on teams, it may be better to avoid using the internet, at least for early rounds of brainstorming.”

Still, the researchers caution that their findings have limitations. The participants were all university students, which may limit how broadly the results apply. Only a small set of objects were tested, and the coding of ideas into categories introduced some subjectivity. In addition, the study used a short three-minute time limit, which may not reflect how people use the internet in real-world creative work.

“There are many ways that a person could use the internet,” Oppenheimer said. “The ways that people seem to use naturally right now yield the effects we find, but that doesn’t mean that there might not be other ways that people could be trained to use the internet that might facilitate collective decision making. We don’t know what those might be, but digital technology (especially AI) is advancing so rapidly, and the way people use such technology is evolving so quickly that these results could look very different in a decade or two.”

Even so, the consistent pattern across different samples, coders, and measures suggests that the effect is real and worth paying attention to. The new findings also align closely with the results of a recent experiment led by Min Tang and colleagues, which compared how different forms of collaboration affect creative thinking.

In both studies, the use of internet tools—whether through search engines like Google or generative artificial intelligence like ChatGPT—tended to produce more homogenous and less novel ideas than collaboration with other people. Tang’s research showed that pairs of humans consistently generated more original and clever ideas than individuals working with AI or using Google.

Looking ahead, Oppenheimer said that he and his colleagues are now “exploring how we can help people be more effective users of the internet and LLMs. Hopefully, this is just the first step in a series of studies that ultimately helps us be more effective rather than just pointing out where we are currently ineffective.”

The study, “Thinking outside the box means thinking outside the search engine,” was authored by Daniel M. Oppenheimer and Mark T. Patterson.