Fig 7.

Agents that learn without forgetting (γ = 0): (a) Response probabilities grow towards unity. (b) The collective motion tends to be strongly aligned, with z(t) fixed at ±1 for long times. This is noteworthy since the scenario presented here has relatively low density, with the same parameter values for N and B that resulted in completely disordered motion in agents that do forget (cf. Fig 4a). As one moves to higher densities, the tendency towards long-term collective alignment becomes even more pronounced. Parameters: number of agents N = 10, world size W = 80, sensory range r = 5 (number of bins B = 8).