A recent study published in the journal Motivation Science demonstrates how our initial impressions can have a lasting impact on our decisions, often leading us to persistently choose inferior options even when better ones are available. Researchers found that when people form an early preference for a particular choice, they tend to stick with it, ignoring more rewarding alternatives.
Previous research has shown that people tend to seek out rewarding outcomes more frequently than non-rewarding ones, which inherently biases their experiences. This can lead to a skewed perception of the effectiveness of their choices. The new study aimed to demonstrate that this bias towards rewarding outcomes can create an idiosyncratic subset of experiences, causing individuals to persist in choosing suboptimal options even when better alternatives are available.
“My driving question for this line of research is why people have persisting erroneous beliefs, such as stereotypes or superstitions, even when they seemingly should have plenty of opportunities to put their beliefs to the test,” said study author Chris Harris, an assistant professor at Utrecht University.
“I believe that one crucial factor is that the evidence that we encounter as a result of our actions is already highly biased. Next to looking at cognitive and motivational processes, we have to look at what evidence people have available to work with. In previous experiments (Harris et al., 2020, JEP:G), we had demonstrated that initial biases persist over many trials (repeated choices) as long as people encounter sufficiently many positive outcomes.”
“For this paper, we were interested how robust this finding was and asked ourselves whether a bias towards one option would persist even when the alternative was actually better? As it turns out, that can be the case. In other words, even when we get to make many repeated choices and even when the alternative is better, we might still continue wrongly thinking that we have the best option. I believe this research has implications for why stereotypes are so pervasive.”
Harris and his colleagues conducted two experiments to investigate how initial biases influence decision-making and lead to persistent suboptimal choices, even when better alternatives are available.
In the first experiment, the researchers aimed to understand how initial biases influence decision-making even when one option is objectively better. They recruited 100 participants through Prolific Academic, an online crowdsourcing platform.
The experiment was structured into four distinct phases. The first phase, the induction phase, was designed to establish an initial bias. Participants were presented with two bags, A and B, from which they could draw yellow or blue balls. One color would earn points, while the other would lose points.
During this phase, one bag (the frequent option) was shown more often than the other, though it had a slightly lower chance of yielding a reward (75% for the frequent option versus 80% for the infrequent option). This setup aimed to induce a bias towards the frequent option due to its higher visibility, despite its lower reward probability.
Following the induction phase, participants entered the first estimate phase. Here, they were asked to indicate which bag they believed was more likely to yield a reward based on their experiences in the induction phase. This phase served as a manipulation check to ensure the bias induction was successful.
In the subsequent free sampling phase, participants could freely choose between the two bags over 83 trials. Their goal was to maximize their points, which would later be converted into a monetary reward. The researchers examined the participants’ choices to see if they would persist in choosing the biased option or shift to the objectively better alternative.
Finally, in the final estimate phase, participants were again asked to estimate the reward probabilities for each bag based on their experiences during the free sampling phase.
Initially, participants displayed a strong preference for the frequently presented but inferior option. This bias persisted throughout the free sampling phase, even though the alternative had a higher probability of rewards. The data showed that frequent positive outcomes from the biased choice reinforced the initial preference, leading to a persistent suboptimal choice pattern.
Interestingly, participants were divided into two groups based on their initial estimates. Those with a bias towards the frequent option continued to choose it more often, whereas participants without this initial bias were more likely to adjust their choices towards the better option.
In the second experiment, the researchers sought to test whether increasing the difference in reward probabilities between the two options would help participants overcome their initial biases more readily. They recruited another 100 participants, similar to the first experiment, and followed a comparable procedure with some modifications.
The induction phase in Experiment 2 used a more extreme distribution of evidence. The frequent option resulted in a positive outcome on 67% of the trials, while the infrequent option did so on 80% of the trials. This increased discrepancy aimed to make the superior option more apparent and test if participants could overcome their initial bias more effectively.
After the induction phase, participants again went through the first estimate phase, where they indicated their beliefs about the reward probabilities of the two bags. The free sampling phase followed, with participants choosing between the two bags over 83 trials to maximize their points. Finally, the final estimate phase asked participants to reassess the reward probabilities based on their experiences.
The findings from Experiment 2 indicated that while the more pronounced difference in reward probabilities did help some participants adjust their choices, initial biases still played a significant role. Participants who did not show an initial preference bias managed to shift towards the better option more effectively and even exceeded the probability matching baseline, indicating a preference for the objectively better choice.
However, those with an initial bias towards the frequent but inferior option took longer to adjust and often failed to reach optimal decision-making. Despite the increased difference in expected values, the initial bias induced by the frequent presentation still led to persistent suboptimal choices for many participants.
“I was most surprised about the second experiment: Participants did learn eventually which option was better but it took them a very long time,” Harris told PsyPost. “I would have expected a more clear-cut pattern; either, that they don’t learn or that they learn faster that the alternative is better.”
Both experiments highlighted the powerful influence of initial biases on decision-making. Experiment 1 showed that even a slight initial bias could lead to persistent suboptimal choices, reinforced by frequent positive outcomes. Experiment 2 demonstrated that increasing the discrepancy in reward probabilities between options helped some participants overcome their biases but was not sufficient for all.
“Going for the best choice alternative (the best lottery, the best sandwich place in town, the fun group to hang out with, …) often means that by interacting you learn more about this option, but that you don’t learn (enough) about alternatives,” Harris explained. “Say you preferably interact with one group of people, this often means that you then interact less with other groups. But as a result, it will be much more difficult to realize whether this other group is equally fun to hang out with or perhaps even more fun.”
“Our choices affect what evidence we might possibly encounter and therefore what we can learn. If things are going well for you and you can afford a suboptimal decision, perhaps occasionally check out some promising alternatives.”
The controlled experimental setting may not fully capture the complexity of real-world decision-making, where multiple external factors such as social influences, emotions, and environmental contexts play significant roles. Additionally, the short-term nature of the study did not account for long-term consequences of persistent biases.
“The most obvious caveat is that both experiments were limited to 100 trials,” Harris noted. “We think this is quite long (how often in real life do you make the same decision 100 times?), but it might be that people always overcome initial biases but are simply slower than we thought (and we would need more trials to test this).”
Examining interventions that encourage exploration over exploitation could help identify strategies to mitigate the persistence of suboptimal choices, potentially leading to better decision-making processes in everyday life.
“I believe that many biases have their origin already in the evidence that people accrued, so even before mental processes come into play,” Harris explained. “It’s in how the environment is structured (e.g., we are almost by definition more likely to encounter people of a majority group as opposed to a minority group), but it’s also in how our behavior creates or upholds skewed distributions of evidence. I am interested in understanding better how this happens and then how we can intervene.”
“Of course, there are also good reasons to quickly go for a (supposedly) best option. Exploring alternatives can be costly, either because we have to invest (e.g., time, energy, money) or because by choosing a suboptimal option we forfeit potential wins. If you go on a first date, for example, and it went very poorly, chances are higher that you won’t go on a repeat date. You probably don’t want to invest the time (or energy) to go on multiple extra dates just to be really sure that you are no match.”
“But at the same time, this can result in us making premature conclusions, especially when your choice seems promising. If you care about picking any decent option, you are probably fine,” Harris continued. “But if you want the best option or you want to have an accurate representation about the choice alternatives, then perhaps it could at times be wise to check out alternatives a bit more or give second chances.”
“I would like to recommend the book Sampling in Judgment and Decision Making (2023) edited by Fiedler, Juslin, and Denrell. It gives a great overview of current research and theories on decisions by experience, the larger framework in which my work also falls.”
The study, “Missing Out by Pursuing Rewarding Outcomes: Why Initial Biases Can Lead to Persistent Suboptimal Choices,” was authored by Chris Harris, Henk Aarts, Klaus Fiedler, and Ruud Custers.
