Set (R,H) = (2,-2). Imagine that you initialize the system at the only fixed point, the global minimum of V(x) around x = -1.8. Now increase H while leaving R unchanged. If H is increased sufficiently slowly, the system would stay close to the fixed point and "track" it as the fixed point shifts to the right because the fixed point is stable. If you continue to increase H, a new stable fixed point appears around x = 1 once the (R,H) point passes through the lower bifurcation curve (roughly H = -1.09). This new fixed point is irrelevant to the system's behavior at the moment. The system was initialized at the x < 0 fixed point, and it follows that point as H increases because the fixed point is stable. If you increase H even more, you'll notice the value of the potential at the x < 0 fixed point increasing. The fixed point the system has been tracking appears to be becoming "less stable". Finally, if you increase H so much that the (R,H) point passes through the upper bifurcation curve (roughly H = 1.09), the x < 0 fixed point suddenly disappears completely. The system is no longer at a fixed point, so it immediately starts evolving toward the stable x > 0 fixed point. This is the catastrophe. A small change in a parameter (increasing H from 1.08 to 1.09) causes a large, sudden change in the state of the system. This is analogous to the act of continually stacking small weights on a glass table, where the total weight on the table is like H and the state of the table is like the state of the system. The table will be able to hold some number of weights without anything happening to it. Eventually though, one small weight will be added to the pile on the table and it will suddenly shatter.