Back

Tracing information propagation in fish schools with a conditioned escape response

Jadhav, V.; Aimon, C.; Lamshana, F.; Trendafilov, D.; Escobedo, R.; Sire, C.; Theraulaz, G.; Guttal, V.

2026-08-05 animal behavior and cognition
10.64898/2026.08.04.742699 bioRxiv
Show abstract

Collective escape responses are widespread in animal groups, yet how threat-related information spreads from a few informed individuals to the collective remains poorly understood because the identities of the individuals who initially detect danger are rarely experimentally controlled. Here, we combined aversive conditioning, behavioural tracking, and computational modelling to investigate how escape information propagates through schools of Puntigrus tetrazona. We conditioned selected fish to associate a green light with an aversive stimulus. Subsequently, we tested the escape response of schools containing one informed and four naive fish to the green light. One informed fish was sufficient to trigger a collective escape response. Upon stimulus onset, conditioned fish accelerated and crossed the hurdle, after which naive fish sequentially increased their speed and followed. Temporal correlations in speed revealed a hierarchical leader-follower structure mostly aligned with hurdle-crossing order. Neither initial distance, viewing angle, nor relative orientation to informed fish predicted the order of the escape sequence. An agent-based model incorporating local alignment and distance regulation reproduced experimental observations and suggested that the informed individual likely remains partly attentive to its neighbours during escape. Overall, our work provides a general framework for experimentally dissecting information transfer in animal groups and identifying the behavioural mechanisms underlying collective escape dynamics.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.