When humans learn new rules to make decisions, our brains appear to use the exact same information-gathering process that they use to process basic physical sensations. A recent study shows that a specific brain wave associated with collecting evidence readily adapts to track arbitrary, newly learned categories. The research was published as a preprint in The Journal of Neuroscience.
To understand how we make choices, neuroscientists often rely on a concept called evidence accumulation. This theory suggests that the brain acts like a bucket collecting drops of water. As sensory information comes in, the brain continuously gathers this evidence until it reaches a specific threshold, triggering a final decision.
Researchers can observe this accumulation process in real time using an electroencephalogram, or EEG. By placing sensors on a person’s scalp, scientists track electrical patterns known as brain waves. One specific pattern, called the centro-parietal positivity, reliably mirrors the evidence accumulation process.
The centro-parietal positivity presents as a gradual buildup of positive electrical voltage in the brain. This voltage climbs steadily while a person weighs their options. The electrical signal peaks just before the individual executes a response, like pressing a button.
Prior studies have shown that this brain wave tracks evidence from physical stimuli, such as a group of dots moving across a screen. As more dots move in the same direction, the electrical signal climbs faster. It also tracks information pulled from memory, like recalling trivia facts, or basic semantic knowledge.
Other research has found that the brain can accumulate evidence based on fixed, universally shared visual concepts. For instance, people naturally distinguish between vertical lines and diagonal lines. The brain uses these permanent visual frameworks to sort out incoming information.
Scientists did not know if the brain could apply this accumulation mechanism to entirely arbitrary rules. If a rule is newly invented and highly specific to one person, the evidence does not actually exist in the visual environment. Instead, the brain must compute the evidence internally by comparing what it sees against a newly learned, imaginary standard.
University of Nevada, Reno researchers Arianna Thoksakis and Edward F. Ester designed a study to test this question. They wanted to see if the brain’s decision-making machinery is truly flexible across different types of information. If so, the centro-parietal positivity should respond to abstract, newly learned rules just as it responds to direct sensory input.
The researchers recruited volunteers to complete a visual categorization task while hooked up to an EEG machine. Data from 38 participants was ultimately included in the analysis. The participants viewed circular images filled with hundreds of parallel lines.
During a training phase, participants had to categorize these images into two distinct groups by pressing specific keys on a keyboard. The researchers assigned a hidden, arbitrary dividing line for each participant. For example, a boundary might be set at exactly 73 degrees, completely invisible to the individual.
Any lines tilted counterclockwise to this specific angle belonged to the first category, while clockwise lines belonged to the second category. Through trial and error, guided by correct or incorrect feedback after every choice, the participants had to figure out their unique boundary. Most volunteers learned the invisible rule within a few short rounds.
Once the participants understood the rule, they moved on to the main task. The researchers presented lines tilted at specific angles relative to the participant’s hidden boundary. Some images featured lines tilted 45 degrees away from the boundary, making them easy to categorize. Other images were much harder, featuring lines tilted just two degrees away from the dividing line.
Behavioral results showed that participants were faster and more accurate when the lines were rotated further from their learned boundary. To connect this performance to the brain’s internal processes, the researchers used a mathematical framework called a drift-diffusion model. This approach separates the raw speed of a physical reaction from the cognitive process of weighing options.
The mathematical model estimates a specific metric known as the drift rate, which represents the speed of information gathering. The calculations confirmed that drift rates increased steadily as the lines moved further from the boundary. Essentially, the larger the angular distance from the hidden rule, the stronger the evidence became. This stronger evidence allowed the brain to accumulate information at a much faster pace, leading to quicker choices.
Next, the researchers examined the EEG data to see if the brain’s electrical signals matched this behavioral pattern. They measured the centro-parietal positivity buildup during the moments leading up to each participant’s button press. The electrical slope grew much steeper for images that were further from the category boundary.
The team then compared the behavioral math to the electrical brain recordings. They found a strong correlation across the participants. Individuals who showed a high behavioral sensitivity to the visual categories also displayed a highly sensitive electrical buildup in their brain waves.
This correlation suggests that the electrical signal is a direct reflection of the underlying decision variable. The brain’s machinery for accumulating physical sensations and its machinery for accumulating computed, abstract evidence are not separate systems. They appear to be a single, highly adaptable mechanism.
To verify that this electrical buildup was truly about decision-making, the researchers checked another region of the brain entirely. They analyzed beta waves over the motor cortex, which specifically control the physical movement of the hands and fingers. Because the right side of the brain controls the left hand and vice versa, researchers can track exactly when the brain prepares to push a button.
They needed to ensure the decision signals were not just the result of a participant flexing their muscles to press a key. While the motor cortex did show the expected activation as participants prepared to respond, this activity did not change based on the difficulty of the image. The motor preparation remained identical whether the lines were two degrees or 45 degrees from the boundary. This confirms that the centro-parietal positivity reflects the mental act of deciding, rather than the physical act of moving.
There are a few methodological details to consider regarding the study design. When an image suddenly appears on a screen, it causes a rapid burst of visual processing in the brain. Because participants responded relatively quickly, this initial visual response could overlap temporally with the decision-making brain waves being measured.
While this visual overlap is present, it is unlikely to fully explain the strong correlation seen between individual brain waves and computational drift rates. Future studies could separate the visual onset from the decision-making period to completely rule out any sensory interference.
Additionally, this experiment relied on a very specific type of visual feature. Participants evaluated a single dimension, which was the orientation of straight lines. The rule separating the categories was also absolute, relying on a hard dividing line.
In natural environments, categories are rarely this simple. Objects belong to categories based on a mixture of shapes, colors, and textures, and the boundaries are often probabilistic rather than absolute. Testing whether the brain’s evidence accumulation mechanism works the same way for these messier, real-world categories will require additional research.
The study, “Neural Measures of Human Decision Making Track Evidence Accumulation in Learned Space,” was authored by Arianna Thoksakis and Edward F. Ester.