To better understand decision-making, researchers can create computational models—groups of equations that aim to predict what decisions people would make when faced with a set of choices. For example, a model might estimate how people would respond when given the choice between receiving a guaranteed amount of money or a chance to win a greater amount of money.
These models can shed light on the calculations the human brain employs to make decisions, how those calculations may change under certain scenarios, and how that might impact how we make important decisions, such as those around medical treatments or finances.
To build the models, researchers must input numerical data, such as the amount of money in the previous modeling example. However, many decisions we make in real life don’t involve precise numbers.
Now, for the first time, Yale researchers have modeled more nuanced decision-making, choices that are based on the description of available outcomes rather than hard numbers.
In a new study published in PLOS Computational Biology, the researchers gave participants a hypothetical medical scenario and asked them to choose between different treatments.
They then used this data to build a model that not only performed well on this type of qualitative decision-making, but also outperformed traditional models based on quantitative—number-based—datasets.
This new model will allow neuroscientists to investigate how people make many different kinds of decisions and how our brains value various outcomes. For example, scientists may study whether individuals are more risk-averse under certain scenarios, such as when making financial or medical decisions.
The team included first author Nachshon Korem, Ph.D., associate research scientist in psychiatry, and senior author Ifat Levy, Ph.D., Elizabeth Mears and House Jameson Professor of Comparative Medicine.
“We want to get to the underlying mechanisms for decision-making—the algorithm that the brain is using to make decisions,” says Levy. “This kind of modeling approach allows us to look at very different types of decisions, and then, through neuroscientific investigations, try to see how our values of different factors are represented in the brain.”
Quality vs. quantity
In the study, participants made a series of choices between a guaranteed outcome and a chance for a better outcome. These choices came with varying degrees of risk and ambiguity. Researchers refer to risk as the likelihood of a potential outcome. For example, a person might have a 50% chance of winning the better outcome. Ambiguity, on the other hand, is when the probability of the better outcome is unknown.
The researchers first asked participants to make quantitative decisions—the more traditional approach for modeling based on objective data, such as amounts of money. For instance, they might choose between having a guaranteed $5 or a 75% chance of winning $8. In other instances, the lottery was ambiguous—the chance of winning the higher amount was unknown.
Participants also made qualitative decisions about medical treatments. The researchers gave them a hypothetical scenario in which they suffered a spinal injury from a car accident. The participants could choose a known treatment with guaranteed slight improvement. Or they could choose experimental treatments that offered a chance of greater improvement.
The team verbally described each outcome to the participants. For example, they might describe a treatment offering “moderate improvement” as one that enables the patient to stand and walk with assistance, such as crutches or a walker. But they can only go very short distances and need to rest frequently.
Once again, the outcomes of the experimental treatments varied—either offering moderate improvement, major improvement, or complete recovery—and had different degrees of risk and ambiguity.
New model outperforms traditional decision-making models
Based on the qualitative data, the team created a model that estimated how highly the participants valued each medical treatment option. The researchers then tested the model on the decisions participants made when presented with different monetary options and compared how their new model performed against existing decision-making models.
“We’ve shown that our model works better than the common models that we and many other scientists have been using,” says Levy. “For quantitative data, even though our model doesn’t use numbers as values, it actually works better to explain behavior.”
The model also works on data in which one category is not inherently greater than another.
“For example, we can ask participants whether they prefer a banana or an apple, and using this same model, we can estimate how much they value each option,” says Korem.
The team plans to continue exploring how to best model these sorts of subjective preferences. They are also interested in investigating how emotions like fear can affect decision-making.
“For instance, in a scenario where someone needs to make a decision about a surgery, we want to know how stressful emotions affect their valuations compared to when they are relaxed,” says Korem.
More information:
Nachshon Korem et al, Modeling decision-making under uncertainty with qualitative outcomes, PLOS Computational Biology (2025). DOI: 10.1371/journal.pcbi.1012440
Citation:
Innovative model captures how we make decisions without numbers (2025, April 15)
retrieved 15 April 2025
from https://medicalxpress.com/news/2025-04-brain-decisions.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.
To better understand decision-making, researchers can create computational models—groups of equations that aim to predict what decisions people would make when faced with a set of choices. For example, a model might estimate how people would respond when given the choice between receiving a guaranteed amount of money or a chance to win a greater amount of money. These models can shed light on the calculations the human brain employs to make decisions, how those calculations may change under certain scenarios, and how that might impact how we make important decisions, such as those around medical treatments or finances. To build the models, researchers must input numerical data, such as the amount of money in the previous modeling example. However, many decisions we make in real life don’t involve precise numbers. Now, for the first time, Yale researchers have modeled more nuanced decision-making, choices that are based on the description of available outcomes rather than hard numbers. In a new study published in PLOS Computational Biology, the researchers gave participants a hypothetical medical scenario and asked them to choose between different treatments. They then used this data to build a model that not only performed well on this type of qualitative decision-making, but also outperformed traditional models based on quantitative—number-based—datasets. This new model will allow neuroscientists to investigate how people make many different kinds of decisions and how our brains value various outcomes. For example, scientists may study whether individuals are more risk-averse under certain scenarios, such as when making financial or medical decisions. The team included first author Nachshon Korem, Ph.D., associate research scientist in psychiatry, and senior author Ifat Levy, Ph.D., Elizabeth Mears and House Jameson Professor of Comparative Medicine. “We want to get to the underlying mechanisms for decision-making—the algorithm that the brain is using to make decisions,” says Levy. “This kind of modeling approach allows us to look at very different types of decisions, and then, through neuroscientific investigations, try to see how our values of different factors are represented in the brain.” In the study, participants made a series of choices between a guaranteed outcome and a chance for a better outcome. These choices came with varying degrees of risk and ambiguity. Researchers refer to risk as the likelihood of a potential outcome. For example, a person might have a 50% chance of winning the better outcome. Ambiguity, on the other hand, is when the probability of the better outcome is unknown. The researchers first asked participants to make quantitative decisions—the more traditional approach for modeling based on objective data, such as amounts of money. For instance, they might choose between having a guaranteed $5 or a 75% chance of winning $8. In other instances, the lottery was ambiguous—the chance of winning the higher amount was unknown. Participants also made qualitative decisions about medical treatments. The researchers gave them a hypothetical scenario in which they suffered a spinal injury from a car accident. The participants could choose a known treatment with guaranteed slight improvement. Or they could choose experimental treatments that offered a chance of greater improvement. The team verbally described each outcome to the participants. For example, they might describe a treatment offering “moderate improvement” as one that enables the patient to stand and walk with assistance, such as crutches or a walker. But they can only go very short distances and need to rest frequently. Once again, the outcomes of the experimental treatments varied—either offering moderate improvement, major improvement, or complete recovery—and had different degrees of risk and ambiguity. Based on the qualitative data, the team created a model that estimated how highly the participants valued each medical treatment option. The researchers then tested the model on the decisions participants made when presented with different monetary options and compared how their new model performed against existing decision-making models. “We’ve shown that our model works better than the common models that we and many other scientists have been using,” says Levy. “For quantitative data, even though our model doesn’t use numbers as values, it actually works better to explain behavior.” The model also works on data in which one category is not inherently greater than another. “For example, we can ask participants whether they prefer a banana or an apple, and using this same model, we can estimate how much they value each option,” says Korem. The team plans to continue exploring how to best model these sorts of subjective preferences. They are also interested in investigating how emotions like fear can affect decision-making. “For instance, in a scenario where someone needs to make a decision about a surgery, we want to know how stressful emotions affect their valuations compared to when they are relaxed,” says Korem. More information: Citation: This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
Quality vs. quantity
New model outperforms traditional decision-making models
Nachshon Korem et al, Modeling decision-making under uncertainty with qualitative outcomes, PLOS Computational Biology (2025). DOI: 10.1371/journal.pcbi.1012440
Innovative model captures how we make decisions without numbers (2025, April 15)
retrieved 15 April 2025
from https://medicalxpress.com/news/2025-04-brain-decisions.html
part may be reproduced without the written permission. The content is provided for information purposes only.
