Editors' ChoicePsychiatry

Big behavior in the era of the brain

See allHide authors and affiliations

Science Translational Medicine  11 Nov 2020:
Vol. 12, Issue 569, eabf4683
DOI: 10.1126/scitranslmed.abf4683


A smart box learns to distinguish the behavioral profiles of stressed and nonstressed mice.

Translational neuroscience has focused on neural circuit mapping to identify abnormalities that contribute to psychiatric and neurological disease. Although increasingly advanced methods for circuit mapping have facilitated the study of psychiatric disorders, studies of behavior have brought many advances in neuroscience and psychiatry, and methods that reliably capture complex behaviors may facilitate translation.

Lorsch et al. demonstrate such potential by characterizing a host of behaviors among stress-exposed mice. Although this is not a mouse model of depression per se, the behaviors themselves have likely parallels in human depressive symptoms. Mice were exposed to an aggressive older male mouse who “defeated” them through physical and psychological intimidation over several days. This is known to produce a long-lasting change in social behavior in a subset of vulnerable mice, whereas other resilient mice show no change. The neural circuitry involved in resilience can be established by comparing the vulnerable versus resilient mice. Such research has been important in establishing brain mechanisms that contribute to stress-related depression and alcohol use. Lorsch et al. extend prior work by demonstrating a rich data capture and analysis method to investigate a range of behaviors that may be influenced after stress exposure. Various challenges were presented to the mouse in a contained environment, while video cameras and machine learning captured several thousand features of the mouse’s behavior. Stressed mice showed changes in clustered features reflecting activity, grooming, free rearing, and freezing, relative to controls. Finally, a multivariate profile of behaviors that distinguished stressed mice from controls was recovered to normal after antidepressant treatment. The system is interesting because it provides repeatable behavioral challenges that could facilitate replication from lab to lab and information about subtle behaviors.

In the future, such an approach could be used to validate novel drugs for depression. Lorsch et al. provided a proof of concept, only among males and with an established drug, and further research will be needed to assess generalizability. However, similar behavioral assays are being developed for use in humans, with machine learning of behavioral features in mobile data (mobility, activity, sleep, word use). Validation of comparable behaviors in both rodents and humans may help overcome barriers in translating central neurobiology to psychiatric treatments.

Highlighted Article

View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article