Changes of trait frequency over time in the multi-trait models of fashion.

<p>We simulate individuals interacting randomly in discrete time steps. Cultural traits are continuously introduced with a probability per individual per time step (a new trait is introduced, on average, every 10 time steps). Individuals may spontaneously change their preferences for existing traits, resetting them to random values between −1 and 1 at a rate of per individual per time step. Each individual has a probability of dying of per time step (average lifetime is 100 time steps). A dying individual is replaced by a naive individual who possesses no cultural traits (, ) and is maximally open to learning for others (, ). Simulations last 20,000 time steps (only 3,000 are shown in the figure, and a maximum of 40 traits is shown to maintain legibility).