Figure 6.

Overlaps O_p and A_p (color encoded, see Eqs. (13) and (14)), for a N = 100 network, of the neural activities with the N_p = 7 attractor ruins, as a function of time t. The sparseness of the binary patterns defining the attractor ruins is α = 0.3. Here the average activity α of the attractor relicts and the target mean activity level μ = 0.3 have been selected to be equal, resulting in a limiting cycle with clean latching dynamics.