Figure 7.

Dependence of (a) the average sensitivity to social influence 〈S〉 and (b) P_g m_g on the number of shared estimates τ, in the Sorted (black), Unsorted (blue) and Aggregated (red) treatments. Solid lines and empty circles in (a) are model simulations and dashed lines in (b) are fits produced by multiplying the analytical expression of P_g and m_g (equation (5.3)), using the parameter values estimated in figure 4a,b. Error bars in (a) were computed using a bootstrap procedure described in the Statistical methods, and roughly represent 1 s.e. Error bars in (b) were computed by multiplying $P_{\mathrm{g}}={P_{\mathrm{g}}}_{\exp }\pm {\mathrm{\Delta }}_{P_{\mathrm{g}}}$ and $m_{\mathrm{g}}={m_{\mathrm{g}}}_{\exp }\pm {\mathrm{\Delta }}_{m_{\mathrm{g}}}$, where P_gexp and m_gexp are the values estimated in figure 4a,b, and ${\mathrm{\Delta }}_{P_{\mathrm{g}}}$ and ${\mathrm{\Delta }}_{m_{\mathrm{g}}}$ the corresponding errors. The fitted saturation value 〈S〉_∞ is shown in each treatment in (b).