Fig. 6

Effect of success bias, learning speed and learning schedule on the evolution of socially instructed learning. Evolved population mean number of learning episodes and the proportion of social learning (x- and y-axis, respectively) and the corresponding performance (colour) for 60 replicate simulations for each parameter setting (one data point corresponds to one replicate). Two left panels: All individuals used the ‘shuffled’ learning schedule, and the schedule could not evolve. Two right panels: The learning schedule could evolve. The most common schedule evolved in the population is indicated by the shape of the data point. Columns marked as ‘No success bias’ correspond to simulations in which learners chose demonstrators randomly. Columns ‘Success bias’ correspond to simulations in which learners chose demonstrators based on their lifetime energy gain as a proxy for success. Each row corresponds to different social learning speeds relative to individual learning (i.e. how much faster social learning is relative to individual learning). Environmental conditions are: frequency of change f = 0.1, magnitude of change m = 0.25 and width of the environmental profile σ = 0.25.