Figure 1.

State space portraits and energy landscapes for the memory and read-out networks. The coordinates of the state space are the firing rates of the neurons in the network. Only two of the dimensions of state space are depicted in the plane, with the height of the energy landscape as the third dimension. The dynamical trajectories in state space correspond to trajectories down the energy landscape. (A) All trajectories in the state space of the memory network flow toward a line attractor (thick line). The corresponding trajectories on the energy landscape flow down the walls of a trough, toward a line of minimum energy at the bottom. (B) All trajectories in the state space of the read-out network flow toward a point attractor. Input to the network controls the location of the point attractor, moving it along the dashed line. A point attractor can be visualized as the minimum of a bowl-shaped energy landscape.