Back

Rapid Social Transmission of Predator Location via Gaze Following in Pigeons

Delacoux, M.; Chimento, M.; Kano, F.

2025-10-30 animal behavior and cognition
10.1101/2025.10.29.685277 bioRxiv
Show abstract

Social information transmission is a key advantage of group living, particularly in uncertain or high-stakes contexts. During collective predator vigilance, individuals often initiate evasive action based on social cues--such as the escape behaviour of others--before detecting threats themselves. Yet these cues can be ambiguous or misleading, leading to costly false alarms. Gaze following, a key socio-cognitive skill widespread across species, may reduce this ambiguity by allowing individuals to localize threats with minimal effort and avoid unnecessary escapes. Although this vigilance function has long been hypothesized, direct evidence is lacking. Here we show that pigeons foraging in flocks detect a predator and socially transmit its location rapidly via gaze following, before any escape behaviour occurs. In experiments simulating predator attacks, we reconstructed predator-oriented gaze using high-resolution posture tracking and modelled social transmission with dynamic multi-network Bayesian models. Comparing models based on distinct social cues, we found that pigeons responded selectively to conspecific gaze toward the predator, rather than locomotion, head-up vigilance, or gaze elsewhere. Analyses of network structure further revealed that this transmission occurs through visually connected networks, particularly via peripheral rather than foveal vision. While individuals occasionally followed gaze to non-threatening locations, flocks reliably converged on the actual threat. These findings uncover a previously undocumented, cognition-based mechanism of collective detection. To our knowledge, this is the first evidence that gaze following serves an adaptive function in survival contexts, illustrating how individual-level cognition and group-level dynamics interact to the adaptive value of group living.

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.