Identifying a Brain Network Associated with Variation in Human Mood
Post by Elisa Guma
What's the science?
Human emotions arise in part from interactions between brain areas within the limbic system, which includes the amygdala, hippocampus, insula, and cingulate cortex. Most research on brain networks encoding emotion involves non-invasive imaging such as functional magnetic resonance imaging or positron emission tomography. These techniques, however, are not sensitive to rapid changes in brain activity, and measure it indirectly. Little is known about the way in which human brain networks contribute to real-time changes in mood. This week in Cell, Kirkby and colleagues aimed to identify brain networks associated with rapid variations in human mood, using intracranial brain recordings from limbic regions and self-reported mood.
How did they do it?
The authors made use of uniquely rich multi-site data, whose original purpose was to aid in seizure localization and treatment for 21 patients with epilepsy. This data included intracranial electroencephalography recordings over multiple days from regions of the human limbic system, as well as self-reported mood. Since the amygdala is highly implicated in mood and emotion, the authors only included subjects who had an electrode in the amygdala, and at least three other limbic regions known to connect to the amygdala (ventral hippocampus, cingulate cortex, insular cortex, orbitofrontal cortex, or subtemporal cortex). A custom-made questionnaire was used to measure subjective mood several times per day; a higher score indicated a more positive mood. To identify limbic subnetworks, authors first looked for correlations between all pairs of recording sites in four different frequency bands associated with brain function: theta, alpha, beta, and gamma. To do so, they used an independent component analysis and confirmed that the network patterns they observed occur more often than expected by chance. These patterns corresponded to limbic subnetworks. To identify the most mood-predictive network, they used a regression analysis to measure the strength of association between mood scores and the networks they identified in a subset of subjects with a sufficient number of mood scores (>10). The authors used a cross-validation analysis, in which they tested whether this mood-predictive network identified using the first subset of subjects could predict mood in in the remaining subset of subjects (who had <10 mood scores). Lastly, they investigated how psychological traits may have influenced the presence or absence of the identified mood-predictive network, focusing on anxiety-like, and depressive-like traits.
What did they find?
First, the authors identified 9 distinct networks based on pairwise correlations between brain regions. The most common network they identified was between electrodes in the amygdala and ventral hippocampus, oscillating at the beta frequency. This was present in 62% (13 out of 21) of their subjects. Further, the authors showed that their identified networks exceeded chance levels of covariation, confirming the robustness of their findings. Interestingly, they found that this network was also most predictive of changes in mood over time, with higher variation in the activity correlating with worsening mood. Moreover, the authors show that network activity was always present in individuals with high levels of trait anxiety, and often absent (only present in half) of individuals with low trait anxiety. Finally, the authors also confirmed that epileptiform activity in both the hippocampus and amygdala did not correlate with mood and did not confound the significance of the relationship between the mood-predictive network and mood.
What's the impact?
This study was the first to identify a specific subnetwork (beta frequency amygdala-hippocampus subnetwork) associated with rapid changes in mood in humans. Further, the authors identified that stronger activity in this network may be linked with higher trait anxiety. Previous studies have used imaging methods that do not have the temporal resolution to detect real-time changes in both brain activity and mood. This work provides a deeper understanding of how mood and anxiety are encoded by the brain and may aid in revealing potential biomarkers for diagnosis and treatment of mood and anxiety disorders.
Kirkby et al., An Amygdala-Hippocampus Subnetwork that Encodes Variation in Human Mood. Cell (2018). Access the original scientific publication here