Figure 5
(a) The mean asymmetry over h and \(\eta \), \({\langle {\rm{\Lambda }}({\eta }_{R},0,\eta )-{\rm{\Lambda }}({\eta }_{R},\eta ,0)\rangle }_{h,\eta }\), as a function of network connectivity k, defined as the number of neighbors. Here a population of size \(N=100\) is set on a lattice with periodic boundary condition and 1st up to 5th nearest neighbor interactions. (b) The mean asymmetry as a function of representation noise \({\eta }_{R}\). The mean asymmetry increases with \({\eta }_{R}\) up to high noise levels and drops for very large representation noises. (c) The mean asymmetry as a function of the number of states n. The mean asymmetry increases with n until it saturates to a constant value for large n. (d) Non-stationary behavior. The mean asymmetry as a function of time T. Asymmetry is zero for \(T=1\), and rapidly increases and tends to a stationary limit for large times.
