Scientists from Columbia University and Yonsei University tested whether the volume of gray and white matter in the brain can predict how strongly a person prefers immediate gain over future gain. The work was published in the journal JAMA Psychiatry. The study involved 146 students aged about 22.6 years, of whom 75 were women and 71 were men.
The basis is the concept of "delay discounting." This is the tendency to subjectively devalue a reward as the waiting time increases. For example, a person chooses 20 dollars today instead of 50 in a month. The stronger such devaluation, the higher the impulsivity and the tendency toward short-sighted decisions. Psychologists use this indicator as a behavioral measure of impulsivity.
Participants underwent a behavioral discounting test and a standard MRI (magnetic resonance imaging) in T1 mode, which provides a detailed three-dimensional image of brain anatomy. The researchers then used an atlas to mark regions of interest and measured the volume of gray and white matter in each. To account for differences in head size, the volumes were normalized by total intracranial volume.
To train and test the algorithms, the data were randomly divided into five sets: 80% of participants went into training, 20% into testing. First, using generalized linear models (a statistical method for assessing relationships between variables), they selected regions whose volumes were statistically associated with discounting after adjustment for age and sex. Then they iterated through all possible combinations of these regions and chose the best ones. These combinations were fed into a multilayer perceptron, a type of artificial neural network that captures complex patterns in data.
On average, the model based on white matter explained 22% of the variation in discounting indicators (R-squared 0.22, standard deviation 0.11), and the model based on gray matter explained 19% (R-squared 0.19, standard deviation 0.10). R-squared shows the share of the indicator's variance that the model was able to explain. The accuracy is moderate, but the authors call it significant for an attempt to predict a complex behavioral trait solely from brain structure.
Among the most informative zones in both models were the right accumbens area, a structure deep in the ventral striatum involved in processing reward and motivation, and the right frontal pole, located at the anterior edge of the frontal lobe and associated with planning for the future and evaluating long-term consequences. The white matter model also relied on four regions, predominantly sensory and visual, while the gray matter model relied on seven more, including self-control areas and parts of the visual cortex. The article does not state whether a larger or smaller volume of these zones corresponded to stronger discounting.
The authors emphasize that the study is a proof-of-concept. The sample included only young people, scanning was performed on a single machine, and the models were tested on internal rather than independent external data. In addition, the design is cross-sectional: measurements were made at a single point in time, so it cannot be asserted that brain volume causes a tendency toward quick rewards or vice versa. This concerns statistical associations, not causal relationships.
"Our results do not mean that brain scanning can predict the behavior of a specific person," PsyPost quotes one of the authors, Wi Hoon Jung, as saying. "The models identify statistical patterns associated with discounting, not fixed biological traits." The scientists plan to test the approach on clinical groups, including people with substance use disorders, eating disorders and ADHD (attention deficit hyperactivity disorder).