Summary: New research shows that people often know which option is statistically best but still choose alternatives based on recent success, habit, or intuition rather than acting on what they have learned.
Source: Ohio State University
People sometimes ignore what they know to be the best choice, opting instead for the option that felt right or worked most recently, a new study finds.
According to Ian Krajbich, co-author of the study and an associate professor of psychology and economics at Ohio State University, people frequently rely on gut feelings, habits, or the outcome of a single recent experience rather than on consistent knowledge of what works most often.
This finding challenges the common assumption that suboptimal choices stem mainly from a lack of knowledge. “In our study, people knew what worked most often. They just didn’t use that knowledge,” Krajbich said.
The research, published in Nature Communications, was led by Arkady Konovalov, a former graduate student at Ohio State who is now at the University of Zurich.
Krajbich gave a simple everyday example: suppose Main Street is typically the fastest way home. Yesterday a special event slowed Main Street, so you tried Spruce Street and arrived faster than usual. Today, do you return to Main Street, which is usually quicker, or do you take Spruce Street again because it worked yesterday?
The study suggests many people will repeat the route that worked most recently, even when they know statistically that an alternative is usually better. “There’s this tension between doing what you should do from a statistical perspective versus doing what worked out well recently,” Krajbich explained.
To examine this behavior experimentally, the researchers designed a simple computer task in which recognizing and exploiting patterns increased rewards. Participants chose between two symbols at the top of the screen, moved the cursor toward the bottom, and then clicked on the symbol that appeared on either the left or right to reveal their reward. The task was repeated dozens of times.
The team tracked mouse movements to infer what participants expected to happen next. The cursor trajectories revealed where participants anticipated the next symbol would appear, allowing the researchers to determine whether participants had learned the task structure even when their final choices did not reflect that knowledge.
“Nearly everyone — 56 of 57 participants — learned the pattern. That was no problem for our participants,” Krajbich noted.
However, the experiment included conditions where the pattern that usually produced the highest reward failed between 10 and 40 percent of the time. After a trial in which the usual pattern failed, researchers observed how participants reacted: did they stick with the statistically better option or switch to the option that had just succeeded?
When the highest-paying pattern worked only part of the time (for example, 60 percent of trials), participants followed the statistically superior strategy on only about 20 percent of those critical trials. In versions of the task where the best pattern worked consistently, participants followed it roughly twice as often — about 40 percent of the time.
These results indicate that people often possess accurate structural knowledge about their environment but do not always use it to guide choices. The study demonstrates a separation between learning the structure of a task and applying that knowledge when making decisions.
Why do people fail to apply what they know? The researchers did not provide a definitive answer, but Krajbich suggested that consistent, model-based decision making demands mental effort and planning. When the statistical advantage of one choice over another is small, the cognitive cost of always choosing the statistically better option may outweigh perceived benefits. In addition, outcomes alone can be misleading: a good decision can lead to a bad outcome and a poor decision can produce a lucky win, making it harder to judge which strategy is actually superior.
This tension between statistical reasoning and short-term experience shows up frequently in everyday contexts, including sports strategy. Coaches deciding whether to go for it on fourth down or managers choosing whether to intentionally walk a batter often face options where the statistically better choice produces only marginally higher success. In such cases, recent success or intuitive judgment can easily override long-term evidence.
Krajbich summarized the practical implication: people often learn what works best, but they do not always put that knowledge into practice. Recognizing this gap between knowledge and action could help in designing training, feedback systems, or decision aids that encourage more consistent use of learned structure in real-world choices.
About this neuroscience research article
Source:
Ohio State University
Media contact:
Jeff Grabmeier – Ohio State University
Image source:
The image is in the public domain.
Original research (open access):
“Mouse tracking reveals structure knowledge in the absence of model-based choice” by Arkady Konovalov & Ian Krajbich. Nature Communications. DOI: 10.1038/s41467-020-15696-w
Abstract
Mouse tracking reveals structure knowledge in the absence of model-based choice
Evidence indicates that people use two broad strategies when learning in complex environments: model-free learning (simple reinforcement of rewarded actions) and model-based learning (using knowledge of the environment’s structure). Recent work suggested limited model-based behavior unless it yielded higher rewards. Using mouse tracking, this study examined learning in stochastic and deterministic environments. Participants’ mouse movements revealed they had learned structural features of the tasks even when traditional behavior-based measures did not indicate such learning. Mouse tracking therefore can expose structure knowledge that is necessary, but not sufficient, for model-based choice.
Feel free to share this psychology research summary.