Climate

The way back is a different one

An essay on hysteresis, attractors and the memory of complex systems.

5 min read

Translated from the German original.

IThe word

The Greek hysterein means to lag behind, to come too late. The Scottish physicist James Alfred Ewing gave the phenomenon its name in the 1880s, when he magnetised iron and noticed something strange.

Switch off the external magnetic field, and the iron does not return to its original state. It stays magnetic. To erase the magnetisation, you have to turn the field not just down to zero but beyond it, into the opposite direction.

The iron remembers. Its state depends not only on what is acting on it now, but also on what acted on it before.

Every hard disk, every magnetic stripe, every tape recording relies on this principle. What is memory in technology becomes a trap in nature.

IITwo paths

Imagine a dial: temperature, nutrients, CO₂, salinity. You turn it up slowly. The system follows, good-naturedly for a long time, barely noticeably. Then, at a certain point, it jumps. It falls into a different state.

Now you turn the dial back, exactly to the point where the jump happened. And nothing happens. The system stays where it is. You have to turn much further back, far beyond the old value, before it jumps back – if it does at all.

Plot this course and you get a loop: the way there and the way back do not lie on top of each other. In between lies a range in which, under the same conditions, two different states are possible. Physicists call this bistability.

The band at the top of this page shows exactly this movement: a point follows the upper branch, drops off the edge, travels back along the lower branch and only jumps up again much later.

IIIAttractors

A picture helps: a landscape with two valleys, and a ball lying in one of them. The valleys are the stable states of the system, its attractors. Small disturbances make the ball roll back and forth a little, but it always finds its way back to the bottom of its valley.

The dial deforms the landscape. Turn it up, and the valley in which the ball lies becomes shallower. The restoring force weakens. The ball takes longer to recover from disturbances. This is critical slowing down, the early warning signal before a system tips.

At some point the valley disappears altogether. The ball rolls over the edge into the other valley. Mathematics calls this moment a saddle-node or fold bifurcation.

Turn the dial back, and the old valley reappears. But the ball now lies in the new one, and between the two stands a hill. The new valley has its own stability, its own inner logic. It holds on to whatever has fallen into it.

IVMemory

Hysteresis means that the present alone does not explain the state. Two systems can be under exactly the same conditions and still behave in fundamentally different ways, because they have a different history.

This contradicts a deep-seated intuition. We like to think in functions: same cause, same effect. Take back the cause and you take back the effect. In linear systems that holds. In systems with feedback it often does not.

For it is the feedback loops that stabilise the state – and they stabilise whatever state the system happens to be in, the old one as well as the new one. The system is not sluggish. It defends itself.

A system with hysteresis does not ask where you stand now. It asks where you came from.

VLake, ice, ocean

The lake. A shallow lake is clear; water plants grow all the way to the bottom. They bind nutrients and hold the sediment in place. If more fertiliser gets in, the lake stays clear for a long time. Then it tips: algae take the light, the plants die, the water turns murky. Lower the nutrients back to the level before the shift, and the lake stays murky. The stirred-up sediment and the missing light lock in the new state. Marten Scheffer and his colleagues have shown on such lakes how far back you have to go – often far below the old value.

The ice. The Greenland ice sheet is up to three kilometres thick. Its surface reaches into cold layers of air and so keeps itself cold. As it melts, the surface sinks into warmer air and melts even faster. Once it has gone, it will not grow back at today's temperatures. For that, it would have to become considerably colder than it was when the ice sheet formed.

The ocean. As early as 1961, Henry Stommel showed with a simple model of two boxes of water that the overturning circulation of the Atlantic can have two stable states: on or off. The current transports the salt that drives it. If it shuts down, the salt that could restart it is missing. What this would mean for Europe is described in the essay on the AMOC.

VIThe hysteresis of society

Social systems remember, too. In 1986 the economists Olivier Blanchard and Lawrence Summers described why unemployment in Europe did not fall back to its old level after crises, even though the causes had long disappeared: people who had been out of work for a long time lost skills, contacts and confidence. The crisis was over; its trace remained.

Trust works in a similar way. It grows slowly, through many small confirmations, and can break in an instant. It does not return just because the reason for the break has gone. Institutions, languages, craft traditions, a whole body of knowledge about how something is done: once torn, they cannot simply be restored by returning to the old conditions.

Societies are path-dependent systems. Their state is the sediment of their history.

VIICoda: The price of return

The width of the loop is a measure. It tells you how much more it costs to come back than it would have cost not to leave.

In some systems the loop is narrow. A mistake can be corrected with some extra effort. In others it is so wide that the way back lies beyond what is possible. Then the tipping point is a point of no return.

From this follows a simple, uncomfortable rule: before the threshold, caution is cheap. After the threshold, it is priceless.

Hysteresis is not a special case in nature. It is the norm wherever feedback is at work. The linear world, in which every path is reversible, is the exception. Once you have understood this, you look at the present and the future differently: not as the setting of a dial, but as a trace you leave behind.

Sources and further reading

  • Ewing, J. A. (1885): Experimental Researches in Magnetism. Philosophical Transactions of the Royal Society of London. — The investigations in which the term hysteresis was coined.
  • Stommel, H. (1961): Thermohaline convection with two stable regimes of flow. Tellus. — The two-box model of ocean circulation.
  • Blanchard, O. & Summers, L. (1986): Hysteresis and the European Unemployment Problem. NBER Macroeconomics Annual.
  • Scheffer, M. et al. (2001): Catastrophic shifts in ecosystems. Nature. DOI: 10.1038/35098000
  • Scheffer, M. (2009): Critical Transitions in Nature and Society. Princeton University Press. — Tipping points, hysteresis and early warning signals, very readable.
  • Robinson, A., Calov, R. & Ganopolski, A. (2012): Multistability and critical thresholds of the Greenland ice sheet. Nature Climate Change. DOI: 10.1038/nclimate1449
  • Strogatz, S. H. (2015): Nonlinear Dynamics and Chaos. 2nd ed., Westview Press. — Bifurcations and hysteresis, mathematically accessible.