Distinctive Brain ‘Subnetwork’ Tied to Feeling Blue
Posted on by Dr. Francis Collins
Experiencing a range of emotions is a normal part of human life, but much remains to be discovered about the neuroscience of mood. In a step toward unraveling some of those biological mysteries, researchers recently identified a distinctive pattern of brain activity associated with worsening mood, particularly among people who tend to be anxious.
In the new study, researchers studied 21 people who were hospitalized as part of preparation for epilepsy surgery, and took continuous recordings of the brain’s electrical activity for seven to 10 days. During that same period, the volunteers also kept track of their moods. In 13 of the participants, low mood turned out to be associated with stronger activity in a “subnetwork” that involved crosstalk between the brain’s amygdala, which mediates fear and other emotions, and the hippocampus, which aids in memory.
The majority of volunteers with the amygdala-hippocampus subnetwork also scored high for baseline anxiety on a psychological test conducted before the study began. Interestingly, the subnetwork was absent in the remaining eight volunteers, all of whom had normal fluctuations in mood but a significantly lower tendency toward anxiety. Researchers speculate that mood changes may be driven by other factors in such individuals.
This work was supported by the Defense Advanced Research Projects Agency (DARPA) under the Brain Research through Advancing Innovative Neurotechnologies® (BRAIN) Initiative, which is a White House initiative launched in 2014 and co-led by NIH. The new findings represent yet another critical step toward achieving the goal of DARPA’s Systems-Based Neurotechnology for Emerging Therapies (SUBNETS) program: creation of implantable devices with the potential to help people with neuropsychiatric illnesses.
Researchers have long faced ethical and technological considerations that have slowed delving into the human neuroscience of mood in all its complexity. A big challenge has been that it didn’t seem possible to place electrodes in the brain to capture detailed recordings in real-time as people went about their daily activities.
But it turns out such studies are actually feasible in certain groups of patients. Among them are people with epilepsy who are awaiting surgery to remove seizure-inducing brain tissue. At the University of California, San Francisco’s (UCSF) Epilepsy Center, neurosurgeon Edward Chang prepares for such operations by placing 40 to 70 tiny electrodes directly onto the surfaces of a patient’s brain and within deep brain structures, such as the amygdala and hippocampus, up to two weeks prior to surgery. This procedure, known as intracranial electroencephalography (iEEG) recording, directly measures neural activity to pinpoint seizure-triggering areas of the brain.
In their study published in the journal Cell, Chang, Vikaas Sohal, Lowry Kirkby, and their colleagues decided to explore whether iEEG recording could also detect changes in communication among parts of the brain that control emotions. So, with their consent, the team used iEEG to look for such changes in the brains of 21 patients awaiting epilepsy surgery. Over the course of the week-long study, the patients also noted their mood swings on tablet devices.
To analyze the data, the UCSF researchers turned to machine learning. That is, they fed the data into sophisticated computer-based algorithms, allowing them to search through the information and “see” complex relationships between patterns of neural activity. This kind of artificial intelligence is comparable to how our own brain’s neural networks process information, learning to focus on certain salient features while ignoring others.
At first, the researchers had the computers explore patterns of interactions in the iEEG data involving different areas of the brain irrespective of mood. Then, in a subsequent round of machine learning, the algorithms searched for links between the most commonly recurring patterns of interactions in the brain and participants’ mood ratings.
Much to their surprise, the data crunching turned up a single common subnetwork that fluctuated at a specific frequency between the amygdala and hippocampus. The synchronized fluctuations indicated that the two parts of the brain were likely to be communicating.
The subnetwork was active in nearly two-thirds of participants and correlated with mood. In fact, the researchers could even predict from the data when a person began feeling down in the dumps!
Sohal says the existence of this subnetwork between the amygdala and the hippocampus suggests there may be a strong relationship between negative emotions felt in the present and a person’s recollection of negative experiences. The researchers speculate that, during periods of deep self-doubt, emotions such as loneliness, sadness, and fear in the amygdala might trigger bad memories, or vice versa.
The findings open the door for developing new ways to help anxiety-ridden people who are struggling with depression and other mood disorders. For instance, it may be possible to design new forms of responsive brain stimulation that treat depression, anxiety, and other conditions by interrupting the mood-altering patterns of activity.
With these new findings and many other ongoing efforts to revolutionize our understanding of the brain supported by NIH, DARPA, and our partners in the BRAIN Initiative, we’re on our way to learning some really amazing stuff.
 An amygdala-hippocampus subnetwork that encodes variation in human mood. Kirkby LA, Luongo FJ, Lee MB, Nahum M, Van Vleet TM, Rao VR, Dawes HE, Chang EF, Sohal VS. Cell. 2018 Oct 29. pii: S0092-8674(18)31313-31318,
Depression Basics (National Institute of Mental Health/NIH)
Vikaas Sohal (University of California, San Francisco)
Edward Chang (University of California, San Francisco)
The BRAIN Initiative® (NIH)
DARPA and the BRAIN Initiative® (Defense Advanced Research Projects Agency)