The emergence of cooperation in wolf-pack hunting is studied using a simple, homogeneous, particle-based computational model. Wolves and prey are modelled as particles that interact through attractive and repulsive forces. Realistic patterns of wolf aggregation readily emerge in numerical simulations, even though the model includes no explicit wolf-wolf attractive forces, showing that the form of cooperation needed for wolf-pack hunting can take place even among strangers. Simulations are used to obtain the stationary states and equilibria of the wolves and prey system and to characterize their stability. Different geometric configurations for different pack sizes arise. In small packs, the stable configuration is a regular polygon centred on the prey, while in large packs, individual behavioural differentiation occurs and induces the emergence of complex behavioural patterns between privileged positions. Stable configurations of large wolf-packs include travelling and rotating formations, periodic oscillatory behaviours and chaotic group behaviours. These findings suggest a possible mechanism by which larger pack sizes can trigger collective behaviours that lead to the reduction and loss of group hunting effectiveness, thus explaining the observed tendency of hunting success to peak at small pack sizes. They also explain how seemingly complex collective behaviours can emerge from simple rules, among agents that need not have significant cognitive skills or social organization.