Fish pay attention to one influential neighbour in virtual reality

Life

Observing interactions between real and virtual rummy-nose tetra fish shows that, within a group, each individual takes cues from only one influential neighbour

1 November 2026

A virtual reality tailored to fish reveals that each animal may be paying attention to only a single, most influential neighbour. This finding may shed light on how animals move as groups without becoming overwhelmed by all the activity and interactions around them.

Fish form schools, milling about seamlessly and acting as a collective. But how does a single fish know what to do? Guy Theraulaz at the University of Toulouse in France and his colleagues attempted to answer the question by using virtual reality (VR) to better study interactions between five rummy-nose tetra fish (Hemigrammus rhodostomus).

“A fish in a group can perceive, simultaneously, many, many individuals, but perception is not equivalent to influence,” he says. “Does a real fish use all this information at once, or does it just focus on a few neighbours?”

The researchers put a single tetra fish in a hemispherical bowl and tracked its swimming with a camera while projecting four virtual tetra fish onto the bowl’s interior in such a way that they appeared real to the tetra. The projections and the camera were connected into a feedback loop, allowing the virtual tetras to react and adjust to the real fish. Theraulaz says that he and his colleagues previously studied tetras in observational experiments and developed a mathematical model for how they might interact. The virtual reality setup allowed them to test that model.

Specifically, they tested three hypotheses: that a single fish would have the most influence on the real tetra’s behaviour; that two tetras would jointly be the most influential fish; and that the real tetra would pay attention to all four virtual fish. The team found clear evidence favouring the first case: as far as the real tetra was concerned, there was only one most influential fish at a time. This may be the case simply because it is easier on the fish, says Theraulaz. “By focusing on just one neighbour, the fish reduce their sensory and cognitive demand,” he says.

Iain Couzin at the Max Planck Institute of Animal Behavior in Germany says that the new study confirms past ideas about tetra behaviour, which were largely pioneered by Theraulaz and his colleagues. It also underscores how powerful VR could be for eventually getting into the heads of animals as they make decisions, he says.

Studies of collective animal behaviour are difficult to perform because an individual animal in a group can’t be isolated without disturbing group behaviours that the researchers want to study, which makes it tricky to test specific hypotheses. But making some animals virtual and fully under researchers’ control sidesteps this problem. Overall, VR leads to endless opportunities for experimenting, says Guy Amichay at Northwestern University in Illinois.

“You could technically design any virtual world you want,” he says. Couzin says that it would be exciting to apply the new method to other species of fish for whom no mathematical models exist.

Therauluz and his colleagues now want to advance their experiment by making some of the virtual fish act stressed, as if they were facing a predator, and then see whether this would affect the real tetra through their interactions.

Topics: animal behaviour / Life