2026-09-23 カリフォルニア大学サンディエゴ校（UCSD）
The glassfish Danionella schools in groups using its sense of vision. Here, a single real fish (foreground) schools with a group of virtual fish programmed to move on a video screen (background). Photo credit: Jo-Hsien Yu and Geoff T. Meyerhof
＜関連情報＞
- https://today.ucsd.edu/story/social-brains-allow-animal-groups-to-escape-danger
- https://www.nature.com/articles/s41586-026-11041-1
社会的行動の神経学的検出が集団逃避行動を指示する Neuronal detection of social actions directs collective escape behaviour
Jo-Hsien Yu (游若嫺), Geoff T. Meyerhof, Jimjohn Milan, Julia L. Napoli & Matthew Lovett-Barron
Nature Published:23 September 2026
DOI:https://doi.org/10.1038/s41586-026-11041-1
Abstract
Animals in groups obtain information from social partners to engage in adaptive behaviour1,2,3. Social information transmission is observed in fish schools4,5,6, bird flocks7,8 and human groups9,10, but the neural representation of such socially acquired information is poorly understood3,11,12. Here we show that, in the schooling glassfish Danionella cerebrum13,14,15, collective escape from danger can be mediated by an individual’s visual perception of other escaping animals. To understand the neural basis of socially transmitted escape behaviour, we imaged neural activity from adult glassfish viewing the actions of virtual conspecifics. Visual neurons in the midbrain optic tectum16,17 and thalamus18 increased their activity when virtual conspecifics escaped. Escape-responsive neurons also responded to the sudden disappearance of virtual fish, yet were unaffected by the disappearance of stimuli moving with non-biological linear motion. Behaviourally, fish retreated from virtual schools that escaped or disappeared, but only those swimming with biological burst-and-glide motion. Neural encoding of this rapid social offset allows fish to infer danger from social information alone, a potentially effective strategy for animals that are capable of rapid movement but have a limited visual range14,19. These results show how the neural computations of individuals enable rapid information sharing in collectives.
