Scientists from the Yale School of Medicine studied fMRI data (functional magnetic resonance imaging, a method that shows brain activity through changes in blood flow) from 890 healthy young people aged 22 to 35. Among them were 459 women and 431 men. The work was published in the International Journal of Neuropsychopharmacology.
The focus was a small area in the front of the brain, the subgenual anterior cingulate cortex (sgACC for short). It is involved in regulating emotions and bodily reactions. Disruptions in its activity have been linked to depression, and this area serves as a target for stimulation-based treatments.
To understand the direction of influence, the researchers used Granger causality analysis. This is a statistical method that uses time shifts in signals to assess whether activity in one area precedes activity in another. It does not prove that one area directly controls another, but only indicates the likely direction of influence.
The scientists compared the sgACC's connections with 268 other brain areas, separately assessing incoming and outgoing signals. Then, using clustering (a method of grouping by similarity of features), they divided the participants into groups with different connectivity patterns and compared these with their symptoms of depression, anxiety and other disorders.
It turned out that signals to the sgACC come from many areas: frontal and parietal regions (responsible for planning and attention), the anterior insula (which processes bodily and emotional information), temporal zones and the cerebellum. The sgACC itself, judging by the estimates, sends signals mainly to motor areas, the brainstem and the cerebellum.
Clustering by incoming connections divided the participants into five groups. The level of depressive symptoms rose from the first group to the fifth, although only the group with the lowest scores differed significantly from the three groups with the highest scores. People with weak incoming signals from several frontal, parietal, temporal and occipital areas were more often in groups with higher depression scores. The strongest connections involved the superior frontal gyrus and the temporal poles.
Because the participants were not patients with diagnosed depression, the differences in absolute scores were moderate: the gap between the extreme groups was small or medium. The groups differed not only in depression but also in anxiety, somatic complaints, attention problems, aggression and rule-breaking. The group with the highest depressive symptoms also had the greatest severity of internal problems (such as anxiety and depression) and external ones (aggression, behavioral problems).
A machine-learning model based on incoming connections assigned participants to the correct group with about 87% accuracy, and after stricter tuning, about 92%. But because the groups were identified from the same data, these figures only show how clearly the groups separate in this dataset, not the accuracy of diagnosing depression in new people.
Clustering by outgoing connections produced two groups with no differences in symptoms, and traditional measures of joint activity of areas did not allow meaningful groups to be identified. The authors caution that this difference may be explained by different reliability of measuring different types of connections, rather than by incoming signals being more important for mood than outgoing ones.
The work has limitations. Participants were not selected by psychiatric diagnoses, and most had mild symptoms, so the analysis cannot show that these connectivity patterns detect major depressive disorder or predict response to treatment. The data come from a single dataset, without independent replication, and were collected at one point in time. In addition, Granger causality analysis on fMRI assesses directional patterns in blood-flow signals and does not prove that one brain area actually controls another.