Swimming speed of schooling fish controls social interaction strength in open-loop immersive virtual reality

Collective motion in fish schools emerges from local social interactions, whose strength may vary with behavioural context. Swimming speed is known to influence manoeuvrability, spatial organization and information flow, yet its causal role in shaping social interaction strength remains poorly understood. Here, we experimentally test whether and how neighbour speed modulates social interactions in schooling fish. Using an immersive open-loop virtual-reality system, we exposed freely swimming rummy-nose tetras Hemigrammus rhodostomus to a single virtual conspecific moving along circular trajectories at different controlled speeds. This approach isolates the effect of partner speed while preserving ecologically relevant sensory cues. Across all speeds, real fish reliably followed the virtual partner and adjusted their own speed accordingly. However, spatial positioning, perception angles and leader-follower configurations changed systematically with partner speed. To interpret these patterns, we combined the experiments with a data-driven three-dimensional model of fish interactions. Model-based reconstructions show that increasing neighbour speed weakens horizontal and vertical attraction while strengthening alignment and wall repulsion. These results demonstrate that fish dynamically tune the strength of their social interactions as a function of swimming speed, identifying speed as a key control parameter governing coordination in minimal groups and transitions between collective regimes.