Probabilistic Sensory Constraints Shape Swarm Cohesion in Echolocating Bats

Cohesive animal groups rely on continuous behavioural updating to regulate spacing, alignment, and collision avoidance, yet all sensory-guided interaction is constrained by finite signal propagation delays, processing time, and motor latency. In actively sensing species such as echolocating bats, these constraints raise a fundamental ecological question: what limits the stability and density of cohesive groups under increasing interaction load?

I develop a constraint-based framework in which neighbour-based interaction is treated as a closed-loop process that is only feasible when sensory updates can be acquired, processed, and acted upon within a finite temporal budget. Building on an asynchronous swarm simulation grounded in echo-timed biosonar control, I formalise two receiver-side feasibility constraints that become critical in dense groups: (i) temporal overlap between conspecific calls and the echo-processing window, and (ii) level dominance of the tracked neighbour’s echo over competing conspecific signals.

Both constraints emerge as probabilistic feasibility boundaries rather than binary conditions for perceptual success or failure. Across a broad parameter space of responsivity, group density, and flight speed, the fraction of call events supporting reliable neighbour-based updates declines smoothly with compounded interaction load. Temporal overlap is organised by a single compound term integrating local density, neighbourhood call rate, call duration, and echo delay, while level dominance operates in a marginal regime where masking is frequent but intermittent.

Simulation results show that stable, collision-averse swarm cohesion can persist across wide ranges of density and motion without explicit coordination or interference avoidance, provided that sufficiently frequent informative updates remain available. Breakdown occurs not because echoes become undetectable, but because the probability of obtaining timely, behaviourally relevant updates falls below what is required to sustain closed-loop control.

Together, these findings identify temporal feasibility and interaction dominance as general constraints shaping cohesion, spacing, and fragmentation in actively sensing animal groups. Rather than invoking acoustic jamming as a failure mode, the framework reframes interference as a background condition that regulates the statistics of actionable information. This closed-loop perspective provides a unifying ecological principle for collective behaviour in active sensing systems and generates testable predictions for when cohesion should persist or fail under increasing sensory and interactional demand.